Use of textual elaboration with literary texts in intermediate Spanish

University of Iowa
Iowa Research Online
Theses and Dissertations
2005
Use of textual elaboration with literary texts in
intermediate Spanish
Mary E. O'Donnell
University of Iowa
Copyright 2005 Mary E O'Donnell
This dissertation is available at Iowa Research Online: http://ir.uiowa.edu/etd/102
Recommended Citation
O'Donnell, Mary E.. "Use of textual elaboration with literary texts in intermediate Spanish." PhD (Doctor of Philosophy) thesis,
University of Iowa, 2005.
http://ir.uiowa.edu/etd/102.
Follow this and additional works at: http://ir.uiowa.edu/etd
Part of the First and Second Language Acquisition Commons
USE OF TEXTUAL ELABORATION WITH LITERARY TEXTS IN
INTERMEDIATE SPANISH
by
Mary E. O'Donnell
An Abstract
Of a thesis submitted in partial fulfillment
of the requirements for the Doctor of
Philosophy degree in Second Language Acquisition
in the Graduate College of
The University of Iowa
May 2005
Thesis Supervisors:
Associate Professor Judith E. Liskin-Gasparro
Associate Professor Michael E. Everson
1
ABSTRACT
The purpose of this study was to analyze textual elaboration as a modification
option for literary texts in an intermediate Spanish language course. This study suggests
that literary texts might obtain a more predominant place in second language instruction,
especially for readers with lower levels of language proficiency, if learners were supplied
with additional tools to facilitate comprehension. Elaboration is one of these tools that
could ease comprehension difficulty that occur due to as text failure; that is, the
difficulties readers encounter when the text’s structure is so obtuse or unfamiliar to the
reader. Here, readers’ response to textual elaboration was tested and measured. Readers’
comprehension was assessed via three instruments: (a) a recall protocol, (b) multiplechoice questions and (c) short constructed response questions.
The study findings indicate that elaborative modifications to literary texts
influence the amount and kind of information that intermediate-level Spanish learners
comprehend about the texts. Limited-proficiency L2 readers of short literary texts that
have been modified elaboratively can recall more information about the texts and are able
to identify more vocabulary that appeared within the text. They cannot, however, infer
meaning from these texts to a greater degree than the readers of unmodified versions of
the same texts. Moreover, level of text difficult plays a pivotal role in relations to the
effectiveness of elaborative modifications. In term of comprehension, students who read
unmodified texts that are relatively easy to comprehend benefit the least from elaborative
modifications.
2
Abstract Approved: ____________________________________
Thesis Supervisor
____________________________________
Title and Department
____________________________________
Date
____________________________________
Thesis Supervisor
____________________________________
Title and Department
____________________________________
Date
USE OF TEXTUAL ELABORATION WITH LITERARY TEXTS IN
INTERMEDIATE SPANISH
by
Mary E. O'Donnell
A thesis submitted in partial fulfillment
of the requirements for the Doctor of
Philosophy degree in Second Language Acquisition
in the Graduate College of
The University of Iowa
May 2005
Thesis Supervisors:
Associate Professor Judith E. Liskin-Gasparro
Associate Professor Michael E. Everson
Graduate College
The University of Iowa
Iowa City, Iowa
CERTIFICATE OF APPROVAL
_______________________
PH.D. THESIS
_______________
This is to certify that the Ph.D. thesis of
Mary E. O'Donnell
has been approved by the Examining Committee
for the thesis requirement for the Doctor of Philosophy
degree in Second Language Acquisition at the May 2005 graduation.
Thesis Committee: ___________________________________
Judith E. Liskin-Gasparro, Thesis Supervisor
___________________________________
Michael E. Everson, Thesis Supervisor
___________________________________
L. Kathy Heilenman
___________________________________
Chuanren Ke
___________________________________
Anne DiPardo
ACKNOWLEDGMENTS
A number of individuals helped me to successfully complete this project. First
and foremost, I would like to acknowledge the assistance offered by my dissertation
directors, Professors Judy Liskin-Gasparro and Michael Everson. Both gave generously
of their time and provided me with practical suggestions as well as words of
encouragement during difficult periods.
Additionally, I would like to thank those who provided technical assistance and
support. In order to successfully collect data, I relied on the personnel in the University of
Iowa’s Language Media Center, particularly the help provided by Director Sue Otto,
Video and Instructional Computing Specialist, Becky Bohde, Audiovisual and
Collections Coordinator, Carolyn Goddard, and Computer Consultant, Greg Johnson. All
four went out of their way to support and assist me. In addition, I would like to thank
Aprille Clarke and Kyle Gassiott, in Academic Technologies who patiently helped me
figure out how to use Web Course Tools (WebCT) to collect and download my data. I
also received invaluable support with data analysis from Lili Zhao and Rui Qin, Research
Assistances in the Statistical Counseling Center and from Ye Tong, who helped to
interpret the more complex statistical data. In addition, Mindy Tuttle from Informational
Technology Office Support Services was wonderfully patient as I struggled with
formatting issues.
A number of my colleagues also generously gave of their time. They include,
Jennifer Cabrelli, Susan Hildenbrandt, Elaine Shenk, Joshua Thoms, and Amy VonCanon
who served as raters for my data, Anne Cummings, who answered a variety of small but
important questions, Curtis Bauer, who was my official English-language expert and
Idoia Elola who was always there to support me no matter what I needed.
Finally but most importantly, I’d like to thank my son Daniel. Thank you, Dan,
for putting up with a mom who has been a full-time student for nearly half of your 21
ii
years of life. Due to your support more than that of any other person, I have been able to
complete this work.
iii
TABLE OF CONTENTS
LIST OF TABLES .......................................................................................................viii
LIST OF FIGURES......................................................................................................... x
CHAPTER I MOTIVATION FOR THE STUDY............................................................ 1
Introduction .................................................................................................. 1
Literary Texts: Appropriateness for L2 Instruction........................................ 2
Problems with Literature for L2 Instruction: Global Issues............................ 4
Global Issues: Historical ........................................................................ 4
Global Issues: Pedagogical..................................................................... 4
Global Issues: Programmatic.................................................................. 5
Problems with Literature for L2 Instruction: Local Issues ............................. 6
Local Issues: Instructor Acceptance ....................................................... 6
Local Issues: Beneficial to Readers ........................................................ 7
Local Issues: Reader Preparedness ......................................................... 7
Problems with Literature for L2 Instructions: Summary ................................ 8
Increasing the Use of Literature at Lower Proficiency Levels........................ 9
Present Study .............................................................................................. 11
Research Questions..................................................................................... 12
Scope and Limitations of this Study ............................................................ 13
CHAPTER II REVIEW OF THE LITERATURE.......................................................... 14
Introduction ................................................................................................ 14
Organization of the Chapter ........................................................................ 14
Reading: Cognitive Process......................................................................... 15
Reading: The Relation Between L1 and L2 Reading ................................... 17
Reading: Similarities Between L1 and L2 Reading ..................................... 18
Similarities: Rate and Fluency of Reading............................................ 19
Similarities: Reader Knowledge ........................................................... 19
Similarities: Text Features.................................................................... 19
Similarities: Text Processing................................................................ 20
Similarities: Summary.......................................................................... 20
Reading: Dissimilarities Between L1 and L2 Reading................................. 21
Dissimilarities: Variability in Access to Phonetic Code ........................ 21
Dissimilarities: L1 Syntactic and Semantic Interference ....................... 22
Dissimilarities: Language Processing Speed......................................... 22
Dissimilarities: Lexical Storage Issues ................................................. 23
Dissimilarities: Affective and Motivational Factors.............................. 23
Dissimilarities: Learner Maturation Factors.......................................... 24
Dissimilarities: Summary..................................................................... 24
Reading Factors That Contribute to L2 Reading Difficulty.......................... 25
Text-Based Factors: Syntax.................................................................. 26
Text-Based Factors: Text Organization ................................................ 29
Text-Based Factor: Level of Vocabulary Density or Difficulty............. 33
Text-Based Factor: Text Length........................................................... 36
Text-Based Factors: Summary ............................................................. 38
Reader-Based Factors: Background Knowledge/Schemata ................... 39
Reader-Based Factors: Cultural Expectation/Cultural Bias .................. 40
iv
Reader-Based Factors: L2 Proficiency ................................................ 41
Reader-Based Factors: Summary ......................................................... 42
Why Literary Texts? ................................................................................... 42
Value of Literary Texts: Cultural ......................................................... 43
Value of Literary Texts: Linguistic ...................................................... 43
Value of Literary Texts: Educational.................................................... 43
Reasons Literary Texts Present a Problem in L2 Instruction........................ 44
Making L2 Literary Texts Easier to Read.................................................... 45
Text Modification: Available Choices.................................................. 45
Text Modification: Linguistic versus Conversational/Interactional ....... 46
Linguistic Modification: Genesis/Roots ............................................... 46
Linguistic Modification: Simplification................................................ 47
Simplification: Positive Research Findings .......................................... 47
Simplification: Negative Research Findings......................................... 49
Conversational Modification: Genesis/Roots........................................ 50
Conversational Modification: Elaboration ............................................ 52
Elaboration: Procedure................................................................................ 53
Elaboration Feature: Message Maintenance.......................................... 53
Elaboration Feature: Message Clarification .......................................... 54
Elaboration Feature: Text Organization with Signaling ........................ 54
Elaboration Feature: Repetition and Redundancy ................................. 55
Elaboration: Research Findings................................................................... 56
Elaboration Research Findings: Parker and Chaudron (1987) .............. 56
Elaboration Research Findings: Brown (1987) .................................... 57
Elaboration Research Findings: Ross et al. (1991)............................... 58
Elaboration Research Findings: Oh (2001) .......................................... 60
Present Study: Rationale for Exclusion of Simplified Text .......................... 61
Incidental Vocabulary Acquisition Through Reading .................................. 61
Incidental Vocabulary Acquisition: Definition ..................................... 62
Incidental Vocabulary Acquisition: Text-Based Factors That
Influence Lexical Learning .................................................................. 63
Incidental Vocabulary Acquisition: Text Modification ......................... 66
Incidental Vocabulary Acquisition: Marginal Glosses .......................... 66
Measuring Incidental Vocabulary Learning.......................................... 67
Assessing Reading Comprehension............................................................. 69
Assessing Reading Comprehension: General Assumptions................... 70
Definition of Terms.............................................................................. 71
Text-Explicit versus Text-Implicit Assessment Measures.............. 71
Memory For a Text versus Learning From a Text.......................... 72
Assessing Reading Comprehension: Most Common Methods Used
in L2 Reading ...................................................................................... 73
Multiple-choice Questions/Selected Response............................... 73
Recall/ Summary Protocols ........................................................... 75
Short constructed Response Questions .......................................... 76
Assessing Reading Comprehension: Implications ....................................... 77
Chapter II Summary: What is Known and Unknown................................... 78
CHAPTER III METHOD .............................................................................................. 81
Introduction ................................................................................................ 81
Organization of the Chapter ........................................................................ 82
Study Design: General Description ............................................................. 82
Study Design: Operationalization................................................................ 82
Study Design: Participants ................................................................... 82
v
Study Design: Reading Texts ............................................................... 83
Reading Texts: Elaborative Modification Preparation.................... 85
Reading Texts: Elaborative Modification Procedure...................... 87
Study Design: Data Collection Procedure............................................. 91
Assessing Reading: Testing Methods Used in this Study............................. 94
Testing Methods: Assessment Instrument #1............................................... 94
Recall Protocols for Text-Explicit Information: Rationale for Use....... 94
Recall Protocols: Preparation of Texts for Scoring ............................... 96
Recall Protocols: Scoring ..................................................................... 98
Testing Methods: Assessment Instrument #2............................................. 100
Multiple-choice Questions for Text-Implicit Information:
Rationale for Selection/Use................................................................ 100
Multiple-Choice Questions: Preparation and Piloting ......................... 102
Multiple-Choice Questions: Data Collection Procedure...................... 104
Multiple-Choice Questions: Scoring................................................... 105
Testing Methods: Assessment Instrument #3............................................. 105
Short Constructed Responses for Vocabulary Knowledge:
Rationale for Selection/ Use............................................................... 105
Short Constructed Responses: Preparation and Piloting...................... 106
Short Constructed Responses: Data Collection ................................... 109
Short Constructed Responses: Scoring ............................................... 110
Chapter III Summary ................................................................................ 110
CHAPTER IV RESULTS............................................................................................ 111
Introduction .............................................................................................. 111
Results of Recall of Text-Based Information............................................. 112
Number of Words per Text Version ................................................... 112
General Estimated Equations for Effect of Elaboration....................... 113
Number of Words per Text Version per Reading Text........................ 114
Idea Unit Totals per Text Version (Representative Sample) ............... 118
Idea Unit per Reading Text (Representative Sample) ......................... 119
Idea Unit Means per Text Version per Reading Text
(Represenative Sample) ..................................................................... 120
Essential Idea Unit per Text Version (Representative Sample) ........... 123
Results of Multiple-choice Inferential Information.................................... 124
Correct Responses per Text Version................................................... 124
General Estimated Equations for Effect of Elaboration....................... 125
Correct Responses per Text Version per Reading Text....................... 125
Correct Responses per Question Type ................................................ 128
Correct Responses per Question Type per Text Version ..................... 129
Correct Responses per Question Type per Text Version per
Reading Text ..................................................................................... 130
Results of Vocabulary Knowledge ............................................................ 132
Correct Responses per Text Version................................................... 132
General Estimated Equations for Effect of Elaboration....................... 133
Correct Response per Text Version per Reading Text ........................ 134
Correct Response for Glossed Items per Text Version........................ 137
Summary Chapter IV ................................................................................ 138
vi
CHAPTER V DISCUSSION....................................................................................... 139
Introduction .............................................................................................. 139
Summary of Results:Recall Protocol Quantitative ..................................... 139
Summary of Results:Recall Protocol Qualitative....................................... 143
Summary of Results: Multiple-Choice Questions ...................................... 145
Summary of Results: Incidental Vocabulary Learning............................... 146
Summary of Results: Influence of Spanish Glossing ................................. 150
Overall Findings ....................................................................................... 153
Study Limitations...................................................................................... 153
Pedagogical Implications .......................................................................... 154
Recommendations for Future Research ..................................................... 154
Conclusion................................................................................................ 155
APPENDIX A ............................................................................................................. 156
APPENDIX B ............................................................................................................. 162
APPENDIX C ............................................................................................................. 167
APPENDIX D ............................................................................................................. 174
APPENDIX E ............................................................................................................. 182
APPENDIX F.............................................................................................................. 186
REFERENCES............................................................................................................ 190
vii
LIST OF TABLES
Table
3.1.
Missing Participant Numbers per Section per Reading and per
Assessment Instrument: Fiebre de lotto ............................................................... 93
3.2.
Missing Participant Numbers per Section per Reading and per
Assessment Instrument: Tiempo libre .................................................................. 93
3.3.
Missing Participant Numbers per Section per Reading and per
Assessment Instrument: Apocalipsis .................................................................... 94
3.4.
Categorical Breakdown of Targeted Lexical Items............................................. 108
4.1.
Total Word Count per Text Version................................................................... 113
4.2.
General Estimated Equations for Difference in Word Count Between
Text Versions .................................................................................................... 114
4.3.
Word Count Totals per Text Version per Reading Text...................................... 115
4.4.
Independent Samples t-Test for Word Count Totals: Fiebre de lotto .................. 115
4.5.
Independent Samples t-Test for Word Count Totals: Tiempo libre ..................... 116
4.6.
Independent Samples t-Test for Word Count Totals: Apocalipsis ....................... 117
4.7.
Recall Length as Compared to Sentence Length................................................. 117
4.8.
Idea Unit Totals per Text Version (Representative Sample) ............................... 118
4.9.
Independent Samples-t Test, Total Idea Units per Text Version
(Representative Sample) .................................................................................... 119
4.10. Descriptive Statistics for Idea Units per Reading Text
(Representative Sample) .................................................................................... 119
4.11. ANOVA Results for Idea Units per Reading Text
(Representative Sample) .................................................................................... 120
4.12. More-Essential versus Less-Essential Recalled Idea Units per
Text Version (Representative Sample) ............................................................... 123
4.13. Independent Samples t-test for More-Essential versus Less-Essential
Recalled Idea Units per Text Version (Representative Sample).......................... 124
4.14. Correct Multiple-Choice Responses per Text Version........................................ 124
4.15. General Estimated Equations for Difference in MC Results Between
Text Versions .................................................................................................... 125
viii
4.16. Multiple-Choice Totals per Text Version per Reading Text ............................... 126
4.17. Independent Samples t-Test for MC Responses: Fiebre de lotto......................... 128
4.18. Average Correct Responses to Multiple-Choice Questions
per Question Type ............................................................................................. 128
4.19. Average Correct Responses to Multiple-choice Questions
per Question Type per Text Version .................................................................. 129
4.20. Correct Vocabulary Responses per Text Version ............................................... 133
4.21. General Estimated Equations for Difference in Vocabulary
Knowledge Between Text Versions ................................................................... 134
4.22. Vocabulary Scores per Text Version per Reading Text (In Percent) ................... 134
4.23. Vocabulary Scores per Text Version per Reading Text ...................................... 135
4.24. Independent Samples t-test for Vocabulary Results per Text Version:
Fiebre de lotto ................................................................................................... 135
4.25. Independent Samples t-test for Vocabulary Results per Text Version:
Tiempo libre ...................................................................................................... 136
4.26. Independent Samples t-test for Vocabulary Results per Text Version:
Apocalipsis ........................................................................................................ 136
4.27. Correct Response for Glossed Items per Text Version and per
Reading Text ..................................................................................................... 138
ix
LIST OF FIGURES
Figure
3.1.
Raters’ Pre- and Post- Idea Unit Scoring Totals ................................................. 100
4.1.
Idea Unit Means per Text Version per Reading Text (Representative
Sample) ............................................................................................................. 121
4.2.
Mean Number of Idea Units per Text Version per Short, Average and
Long Recall (Representative Sample) ................................................................ 122
4.3.
Correct Multiple-choice Responses per Text Version per Reading Text ............. 127
4.4.
Mean Number of Correct Responses for MC Questions per Question
Type: Fiebre de lotto ......................................................................................... 131
4.5.
Mean Number of Correct Responses for MC Questions per Question
Type: Tiempo libre ............................................................................................ 131
4.6.
Mean Number of Correct Responses for MC Questions per Question Type:
Apocalipsis ........................................................................................................ 132
x
1
CHAPTER I
MOTIVATION FOR THE STUDY
Introduction
Second language (L2) educators have long recognized the potential benefits
associated with the use of literary texts for language instruction. Many educators
appreciate the power inherent in works of literature. Not only do such works serve
learners as models of linguistic excellence and high-quality literary devices, but literature
also has the potential to transmit cultural knowledge and sensitivity to readers who
otherwise may never experience the culture first hand. In addition, it can be argued that
authentic literary works; that is, texts produced within the L2 culture, are not only bona
fide representations of the language and culture, but also the types of texts with which L2
readers eventually have to deal, either within the culture itself or as future students of
literature.
However, the use of literary texts for language learning, especially at the lower
proficiency levels, is an issue that has been subjected to considerable discussion (Barnett,
1991; Bernhardt, 1991b, 1995; Kramsch, 1985; Schulz, 1981; Shanahan, 1997). Over the
past few decades, and with the advent of methods of language teaching ranging from
audiolingual to communicative language instruction, literary texts have come to hold a
position that Patrikis (2002) refers to as “less than privileged” (p. 51). This less-thanprivileged status has resulted not only from a change in methodological focus in language
instruction but also from frustrations encountered by both educators and readers dealing
with literary material, especially those at lower levels of language proficiency (Barnett,
1991; Davis, 1992; Kramsch, 1985; Schulz, 1981). These educators contend that the
introduction of literary material to students not equipped to handle its linguistic and
sociocultural complexities frequently results in learners who are turned off to L2 reading.
2
Frequently, literary texts first appear in intermediate-level courses where, as
Kramsch (1985) noted, “learners perceive an unfair gap between the literary selections of
the second year and the readings they were offered at the elementary level, where
meaning of the text seemed coextensive with the dictionary translation” (p. 356). Some
educators report not only a high level of reader frustration associated with the
introduction and use of literary texts, (Gottwald & Liskin-Gasparro, 1995; Kramsch,
1985) but even anger towards instructors for assigning such reading; an emotional
combination that often results in students abandoning their language study in favor of a
less frustrating discipline (McGuigan, 1979).
Literary Texts: Appropriateness for L2 Instruction
Despite the above-mentioned problems, throughout the past few decades,
language educators have called for a greater integration of literary texts within the
language classroom setting (Barnett, 1991; Kramsch, 1987, 1993; Rice, 1991; Schulz,
1981; Shanahan, 1997). Bernhardt (1995) contends that the literature curriculum holds
the key to cultural competence for L2 students and, therefore, a growing need exists to
determine the most beneficial use of literature in the language classroom. Shanahan
(1997) asserts that literature, in and of itself, has something deeply significant to offer to
the process of language learning no matter what the ultimate goals of the learner may be.
He contends that through the blending of language and culture, literature creates a
positive affective experience that may serve as an inducement to the learner’s success.
Indeed, many language educators argue that the use of literature in the L2 classroom
provides a unique opportunity to combine the best models of language usage with the
cultural values, intentions, and beliefs that are inherently embedded in literary texts
(Kramsch, 1985; Schulz, 1981).
In addition to literature’s cultural appeal, literary texts can be presented early in
the language curriculum to serve a utilitarian end. Besides fostering awareness and
3
appreciation for the literature of the target language culture, it is hoped that familiarity
with such texts will increases student interest and, consequently, enrollments in upperdivision literature courses (Harper, 1988, as cited in Gottwald & Liskin-Gasparro, 1995,
p. 41). According to Schulz (1981), citing college language programs’ own
documentation as evidence, a major goal of most undergraduate L2 programs is to
develop familiarity and admiration for the literature of the target language culture. She
adds, “anyone contesting this assertion should examine course offerings and sequences in
current college catalogs” (Schulz, 1985, p. 43).
Above and beyond the role of literature in promoting cultural competence and
easing the transition into upper-level courses, many believe that the literature of a people
is so closely tied to language and culture that the three are inseparable (Kramsch, 1985;
Schultz, 1981; Shanahan, 1997). Barnett expresses this notion as follows:
I see language and literature as allegorical figures caught up in
some arcane ritualistic dance, in which it is hard to determine
who’s partnering whom; and I return to a question close to the
hearts of many of us who have literary training and inclinations as
well as day-to-day language classes and enthusiasms: Why do
language and literature often seem at odds with each other, though
closely bound together? (1991, p. 7)
Barnett’s statement touches upon another, albeit less student-centered, reason that
many instructors are proponents of the use of literature in language classrooms: Many
educators are themselves students of and advocates for literature. Educators who have
experienced and appreciated the “ritualistic dance” (Barnett, 1991, p. 7) and know
literature’s value as a means of entry into the target language culture are frequently the
most avid supporters of early literary exposure. As Ladousse (2001) notes, “teachers who
‘indulge’ in [literature] do so because they love it themselves” (p. 27).
Aside from the many potential benefits that the use of literature in early language
instruction may offer, its inclusion can create a number of potential problems. For the
purpose of this discussion problems related to the use of literary texts for L2 language
instruction are broken down into two categories: (a) problems that are global; that is, tied
4
to historical, pedagogic or programmatic issues; and (b) problems that are local, or
related to individual student or instructor concerns.
Problems with Literature for L2 Instruction: Global Issues
With so many impassioned advocates for the early introduction of literary texts in
L2 instruction, the question arises as to why the practice is not more widespread.
Partially, this question can be answered by focusing attention on programs and
institutions within the United States where second languages are taught.
Global Issues: Historical
Historically speaking, in U.S. institutions of higher education, the study of
literature, especially the classical languages such as Greek and Latin, has always been
held in high esteem (Rudolph, 1962). In truth, in addition to rote memorization of
grammar rules, translation of the classical languages was for many decades the
predominant form of second language instruction, if not acquisition (Richards &
Rodgers, 2001). In recent decades, however, the translation method has fallen out of
favor with most language educators, and instead focus has turned to efforts that increase
students’ ability to communicative competently in a L2. Kramsch (1985) described this
trend by stating “communicative approaches to language teaching and the current focus
on oral proficiency are calling for a reassessment of the use of literary texts in language
classrooms” (p. 356). Despite the fact that Kramsch made this comment 20 years ago,
the polemic between literary and communicative based language teaching still exists
today.
Global Issues: Pedagogical
Although not all educators have abandoned the use of literature, many feel that its
inclusion is at cross-purposes with communicative expectations. Shanahan (1997)
describes the disparity between curricular outcomes as two pedagogical camps, “one
5
basing its emphasis on communicative competence, the other on the importance of
exposure to culture and, especially, literature” (p. 164). At first glance, little seems to
unite these two camps, either theoretically or practically.
Tied to the goal to produce communicatively competent language learners is the
utilitarian approach to language acquisition and instruction. At times, the study of
literature may seem at odds with student outcomes that emphasize professional
preparedness, a prevailing curricular stance articulated on some college campuses
(Shanahan, 1997). Shanahan (1997) describes the career-orientated approach to language
instruction as the prevalent attitude in the United States. He argues that the utilitarian
goals of many foreign language departments are the predominating factor in the
development of the language curriculum, especially in regard to methods and materials.
In such environments, educators who advocate for instruction that includes the use of
literature are hard pressed to describe the useful function such instruction might
ultimately serve. Shanahan (1997) asks, “what is it that convinces us that literature has, in
and of itself, something deeply significant to contribute to the process of language
learning and how do we articulate that ‘something’ in a way that establishes us on firm
ground in the contemporary professional environment?” (p. 165). The lack of ability to
articulate the “something” may be one of the primary reasons that the study of literature
is not a mainstay of many L2 early language programs.
Global Issues: Programmatic
Building on the lack of a compelling argument in regard to what literature has to
offer students and returning to the idea that literary studies do not have a well-defined
practical purpose, another concern relates to how language programs define student
outcomes. Rice (1991) asks if it would not be appropriate for programs to be built, first
and foremost, on what educators expect students to be able to do with the language after
formal instruction. If outcomes are kept in mind, what role does literature play? If,
6
according to Rice, educators expect students to read a play or a poem as an interactive
experience, without the need for the instructor to tell them what it is about, then programs
and materials should be constructed to that end.
Problems with Literature for L2 Instruction: Local Issues
Institutional obstacles are not the only reason that literary texts are frequently
avoided or under-used during early language instruction. Local issues also determine
when literary texts are introduced to L2 learners. Three local issues involve: (a) instructor
acceptance; (b) concerns regarding benefit to readers; and (c) issues of reader
preparedness.
Local Issues: Instructor Acceptance
Many educators feel that the needs of low-proficiency learners are better met
through readings and activities dealing with the here and now and that the abstract and
complex nature of literary readings are best avoided (Goodman, 2003). A number of
language instructors have legitimate concerns regarding the usefulness or appropriateness
of having lower-proficiency language students reading literary texts. Goodman (2003)
makes this point when he states that “reading material in early [L2] language instruction
should probably avoid special language uses such as literature and focus on mundane,
situationally related language” (p. 252).
Earlier work of Goodman (1973) is cited by some educators to add theoretical
support to the position that early introduction of literary texts should be avoided. It was
Goodman’s contention that if first language (L1) readers are presented with an inadequate
number of familiar words in an unfamiliar context, a context that does not permit
reasonable anticipation of linguistic and semantic elements, students revert to a word-byword decoding process. Schulz (1981) contended that educators should work to prevent
readers from resorting to word-by-word decoding, because the process “contributes
7
neither to the development of global reading comprehension, enjoyment of the text, nor
to the encouragement of continued reading in the foreign language” (p. 44).
Local Issues: Beneficial to Readers
An additional comprehension concern that results from readers reading beyond
their readiness is that they miss the aesthetic value or intent or tone of the text. Davis
(1992) argues that the aesthetic effect that a L1 reader experiences is especially difficult
to achieve for most L2 readers of literature. In a study Davis conducted with 25 students
of sixth-semester French reading the first chapter of Candide, he notes that, “while many
students accurately recalled a large number of ideas…none of them recognized
[Candide’s] intent” (1992, p. 363). The result, he contends, is disillusioned and frustrated
students who frequently expend considerable time and effort deciphering texts that,
although perhaps comprehensible on a certain level, basically remain a mystery.
Responding to students’ inability to engage literary texts in the same manner as
native readers do, instructors try to make texts more accessible. Unfortunately, literary
texts continue to be taught, as described by Kramsch (1985), as “finished products, to be
unilaterally decoded, analyzed and explained” (p. 356). But although Kramsch calls for a
more discursive and interpretive approach to the use of literature in the language
classroom, the fact remains that literary readings are more likely to be explained to
students than to be discussed by them.
Local Issues: Reader Preparedness
Often lower-proficiency readers are simply not properly equipped or prepared to
handle literary readings. Storme and Siskin contend that some instructors, intent on
getting beyond the surface level of text interpretation, place the blame for lack of
comprehension squarely on the shoulders of students (1989). It is not uncommon for
instructors to chastise students for lack of preparedness (Barnett, 1991; Gottwald &
Liskin-Gasparro, 1995). In addition, language instructors have been known to blame
8
colleagues who teach lower-level courses for students’ inadequate readiness for reading
(Barnett, 1991).
Another problem often mentioned by language educators is the unrealistic
expectations placed on learners. In relation to the use of literary texts, research conducted
by Gottwald and Liskin-Gasparro (1995) revealed that part of the gap between instructor
expectations and student performance may lie in the pedagogical materials available for
language instruction. They reviewed six first-year college Spanish textbooks to
determine the number of literary and non-literary texts in each, as well as the authors’
approach to text usage. They found that four texts contained samples of literature, but to a
varying degree (Gottwald & Liskin-Gasparro, 1995). The authors felt that this finding
attests that although most textbook authors are currently attempting to introduce students
early on to samples of literary texts, educators should be cautious as to the
appropriateness of the pedagogic treatments associated with these samples. They state
that “literature is not regarded by the textbook authors as something so significantly
different from the other text types that it requires a completely distinct set of activities.
The implication here is that students should see literary texts as equally approachable as
the nonliterary texts” (Gottwald & Liskin-Gasparro, 1995, p. 51). Potentially the
similarity in pedagogical support offered to students, whether reading literary or nonliterary texts in beginning Spanish, may mislead readers into assuming that all types of
texts should be approached and comprehended in the same manner.
Problems with Literature for L2 Instructions: Summary
Although it can be argued that literary texts have always been esteemed in
departments of foreign languages in the United States, many obstacles remain to their
widespread use in elementary and intermediate classes. Some of these obstacles are
institutional in nature (i.e., historical and pedagogical trends, utilitarian vs. humanistic
expectations, lack of articulation regarding program outcomes and objectives), whereas
9
others relate to the personal experience of the learner or the instructor (e.g., students’
preparedness, belief in usefulness and appropriateness, instructor expectations). It may
still be argued that, introduced appropriately, literary texts have the potential for bridging
the gap not only between language and culture but also between language courses and
literature courses. The question remains as how to make literature more accessible to
lower-level students and therefore an appealing option for language learning not only on
the local, but global level as well.
Increasing the Use of Literature at
Lower Proficiency Levels
Bernhardt (1991b) contends that the understanding of literary texts is perhaps the
most difficult task L2 students face. This occurs, she explains, not only due to readers’
limited linguistic skills but also because of the lack of sociocultural and pragmatic
knowledge that native speakers posses. Beck et al. (1991) calls the situation that results
when readers are asked to read beyond their linguistic and knowledge limits as
comprehension breakdown. Comprehension breakdown involves at least two elements:
reader failure and text failure (Beck et al., 1991). Reader failure can be attributed not only
to inadequate linguistic knowledge, but also to the lack of topical or situational
knowledge or the failure to apply relevant knowledge at the appropriate time. Beck et al.
argue that reader failure can be mitigated through the use of pre-reading activities or
activities that activate reader background knowledge prior to contact with the actual text,
and with increasing students’ linguistic competence. Indeed research studies consistently
report that instructional interventions such as pre-reading activities and schema activation
are effective means of increasing the comprehensibility of texts for less-than-proficient
readers (Anderson & Pearson, 1984; Chen & Donin, 1997; Hauptman 2000; KamhiStein, 2003).
In addition to inadequate linguistic knowledge and lack of topical or situational
knowledge, Bernhardt (1991a) describes other intrapersonal variables that influence
10
whether the reader is able to comprehend the L2 text. These variables include reader
purpose, intention, goal, and comprehension monitoring. Other factors intrinsic to the
reader include personal motivation and anxiety level. All of these intrapersonal variables
have been shown to affect how readers process texts and, ultimately, how the text is
comprehended (Alderson, 2000).
But how might a text fail a reader? According to Beck et al. (1991), text failure
occurs when the text’s structure is so obtuse or unfamiliar that the reader cannot connect
or anticipate language, thus losing the intended meaning. Text failure may happen for
many reasons: when there is a lack of clear connections between ideas and events, when
irrelevant information is presented, or when the concepts are of such high density that
cognitive overload inhibits comprehension (Alderson, 2000). The failure of a text to
assist a reader may also result from ambiguous, distant, or indirect references, or to
failure to provide information that allows the reader to activate an appropriate context for
the text content (Beck et al., 1995). Unlike reader failure, which is related to reader
preparedness, text failure is internal; that is, related to the content and the structure of the
text itself.
In the interest of forestalling comprehension breakdown due to text failure,
various types of text modification and manipulation are used. A common method of
providing additional support to the reader is by glossing; that is, defining unfamiliar
lexical items or phrases. The basic intent of vocabulary glossing is to (a) enhance
comprehension (Nation, 1990), (b) increase vocabulary learning (Holley & King, 1971),
and (c) allow for greater use of authentic texts (Rivers, 1987). Vocabulary glossing,
which is discussed in more detail in the following chapter, is done in a number of ways
and with various decrees of effectiveness.
Textual modification is also used to avert or prevent comprehension breakdown.
The two most common types of text modification are simplification; that is, decreasing
the linguistic complexity of syntactical and lexical items, and elaboration. Elaboration
11
involves offsetting unfamiliar linguistic terms by techniques first identified in native
speaker (NS) to non-native speaker (NNS) interactions. These include the use of
synonyms, restatements, explanations, and rhetorical signaling (Parker & Chaudron,
1987). The idea behind the use of elaborative modification is the preservation of the
original linguistic content in a form that mirrors the interactive modifications made
during NS-NNS oral exchanges when communicative breakdown occurs (Long, 1983a).
Additionally, and perhaps more important, studies comparing unmodified L2 reading
texts against simplified and elaborated versions of the same texts have found that
although both modification techniques increase comprehension, the type and amount of
material comprehended is significantly different. When comprehension is measured via
tasks that require readers to extract basic information from the text (who, what, when,
where, etc.) comprehension measures derived from linguistically simplified texts are
comparable to comprehension measures obtained from texts with elaborative
modification (Brown, 1987; Oh, 2001; Ross et al., 1991:Yano et al., 1994). However,
when text comprehension is assessed by measures that require synthesis of information or
inference beyond the text, students who read elaborative modifications have been
reported to outperform readers of both simplified and unmodified texts (Oh, 2001; Ross
et al., 1991; Yano et al., 1994).
Present Study
The goal of this study is to assess the effect of elaboration modifications on
readers’ comprehension of L2 literary texts. This study is based on the premise that
textual modifications, specifically in the form of elaboration, serve as an effective means
to facilitate comprehensibility without sacrificing the lexical and syntactic complexity
inherent in literary texts, a complexity that is essential for L2 linguistic mastery and
sociocultural understanding.
12
This study’s intent is to contribute to a greater understanding of not only the
amount but also the type of information L2 readers retain and comprehend from reading
elaborated versus unmodified literary texts. It reports in numerical terms on the amount
and type of information readers are able to derive from both text types. Specifically, this
study assesses whether there is a quantitative difference between the comprehension
scores of college-level, fourth-semester students of Spanish who read unmodified
versions of short literary texts and the same students who read these texts in an elaborated
form.
Recall protocols assess the amount and type of text-based or text-explicit
information that readers recall, making it possible to determine what a participant has and
has not understood. Additionally, multiple-choice questions measure the amount of
information that readers can infer after reading the texts. This instrument is intended to
assess understanding at a level beyond reader ability to recall information explicited
stated in the texts. Additionally, incidental vocabulary knowledge is measured through
the use of a word recognition instrument that asks readers to identify specific words and
phrases that appeared in the texts. Finally, readers ability to identify lexical items that are
glossed in Spanish in the original text versions is compared to their ability to identify the
same ideas without the use of the Spanish gloss.
Research Questions
1. Are comprehension scores of participants reading elaborated L2 literary texts
different from scores of the readers reading unmodified versions of the same texts?
1.1 Do readers of elaborated literary texts and readers of the unmodified versions
of the same texts recall from these texts propositions that differ quantitatively and
qualitatively?
1.2 Do readers of elaborated and unmodified versions of the same literary texts
draw different inferences from the texts?
13
1.3 Do readers of elaborated and unmodified versions of the same literary texts
differ in their ability to understand words that appear in the original unmodified versions
of these texts?
1. 4 Do readers of the unmodified versions of literary texts identify lexical items
glossed in Spanish in the text margins differently from readers of the same texts
presented without the gloss in an elaborated version?
To answer these questions, this study follows the standard approach to classroombased research by measuring one group’s performance against another’s but, in addition,
it examines each group’s, and at times each individual’s performance against its own. In
other words, between-group and within-group comparisons are used to ascertain the
effectiveness of text modification within as well as across groups.
Scope and Limitations of this Study
The participants in this study are fourth-semester students of Spanish at a large
research university in the Midwestern United States. All participants are native speakers
of English. It may be inferred that these subjects have at least average reading skills in
English, as they are NSs in college who have had at least 12 years of formal schooling.
Therefore, because of the shared common native language of the participants and because
the study is being carried out in a classroom environment where the language is being
taught as a foreign language, the results from this study is be applicable only to American
adult learners of cognate languages; that is, languages with similar orthographic and
lexical characteristics to English such as French, German, and Portuguese.
14
CHAPTER II
REVIEW OF THE LITERATURE
Introduction
In order to determine how literary texts might be more broadly utilized in early L2
instruction, it is important to describe what is already known about how readers interact
with and interpret texts as well as identify areas still in need of investigation. That is the
intent of this chapter. Beginning with a broad view of what it means to read and working
towards the specific areas of concern to this study, an argument is made that literary texts
could be used to a greater extent in early L2 instruction once they are modified in accord
with the cognitive needs of lower-proficiency readers.
Organization of the Chapter
This chapter is organized as follows. First, what it means to read is described
through a cognitive learning model that serves throughout the remainder of the study as
the framework by which the experience of reading is understood. Similarities and
differences between reading in L1 and in L2 are described, highlighting aspects that
influence language processing and comprehension.
Next text- and reader-based factors that lead to comprehension breakdown for
non-proficient L2 readers are described. Particular attention is paid to research dealing
with the influence of text modifications on reading and reading comprehension. Textbased reading difficulties are presented in relation to literary texts, and approaches to text
modification to increase reader accessibility are described.
A distinction is then drawn between two types of text modification: linguistic
simplification and interactional or conversational elaboration. Research findings from
studies employing simplified and elaborated texts are compared and features that
distinguish these two modification methods are described, particularly in relation to
15
cognitive learning theory. A case is made for the superiority of elaborative modifications
over linguistic simplification.
At this point incidental vocabulary retention and acquisition through reading is
described with an emphasis on what it is and how it is enhanced and measured. This leads
to a brief description of methods for assessing L2 reading processing and comprehension.
The chapter concludes with questions still to be answered regarding the potential
relationship between the use of unmodified literary texts in early L2 instruction as
compared to those same texts with elaborative modifications.
Reading: Cognitive Process
Generally speaking, cognitive learning theory, the framework though which
reading is examined in this study, is based on the work of psychologists and
psycholinguists who believe that language learning is similar to other learned skills, only
in this case, the skill to be mastered is extremely complex. According to cognitive theory,
language learning is a skill comprised of a number of subskills, each requiring the
formation of internal representations that serve to regulate and guide performance
(McLaughlin, 1987). These internal representations are based on the specific language
system being learned and include procedures for selecting appropriate vocabulary,
grammatical rules and pragmatic conventions necessary for accurate language
comprehension and production. Although criticized by some for failure to adequately
describe what the mental representations of the language learner may look like (Mitchell
& Myles, 1998), at the heart of cognitive learning theory is the belief that practice and
later integration of basic subskills are required to increase language competence
(McLaughlin, 1987). Through practice, mental representations are restructured and
modified and performance is improved as subskills become automatized. The notions of
automatization and restructuring are central to the understanding of cognitive learning
theory (McLaughlin, 1987; Shriffrin & Schneider, 1977).
16
McLaughlin (1987) contends that the acquisition of any skill involves assessment
and coordination of information from a multitude of perceptual, cognitive and social
domains. Every language-learning task necessitates the integration of a number of
different subskills, each requiring practice in order to become routine. Therefore, a task
that initially taxes processing capacity, through practice, ultimately requires much less
energy to process. In the language of cognitive learning there are two ways in which
information is processed: either a task requires a relatively large amount of processing
capacity, or it proceeds automatically, demanding little processing energy (Shriffrin &
Schneider, 1977). By definition, the former type of processing is controlled; that is,
involves a lot of attentional control on the part of the learner, while the latter type of
processing is automatic (McLaughlin, 1987). Shriffrin and Schneider (1977) explain that
controlled processing is not a learned response but rather a temporary activation of
mental operations in a dichotomous way. Automatic processing on the other hand,
involves the activation of certain processes held in memory that are activated every time
the appropriate inputs are present. For language learning, the initial stages of learning any
skill, such as reading, involve controlled processing or the slow development of certain
subskills and the gradual elimination of error. After a time, subskills become automatic,
leading to increased language competence and improved performance (Shriffrin &
Schneider, 1977). It is argued in this study, however, that language learning thorough
reading is enhanced when a certain amount of controlled processing remains. In other
words, in order to maximize the type of language recognition that promotes language
learning through reading, a balance must be achieved and maintained between learner
reliance on automatic and controlled processes.
This study uses cognitive language learning theory, specifically in regard to
reading, to examine how the automatic versus controlled subskills of L2 reading are
affected by modifying the reading text that learners are required to process. But before
looking specifically at variables that influence the processing of L2 reading texts, L1
17
reading factors that affect text processing and comprehensibility are explored. In
particular, research findings that compare and contrast L1 and L2 reading are described.
Reading: The Relation Between L1 and L2 Reading
Aspects of language performance where the distinction between automatic and
controlled processing is especially salient can be seen when one compares the L1 and L2
reading processes. According to McLaughlin (1987), successful L2 readers interact
actively with the text by adding, deleting, and substituting words where appropriate.
Readers use available cues to determine the most direct path to meaning, drawing on
prior conceptual and linguistic competence to predict what might plausibly come next.
But unlike the case of L1 readers, automatization of the processes necessary for effective
and efficient reading is hard for L2 readers to achieve. This was demonstrated by a study
conducted by McLeod and McLaughlin (1986). They reported that even advanced ESL
students did not seem able to utilize semantic and syntactic cues well. Although the
participants’ increased syntactic and semantic competences should have enabled them to
make nearly twice as many accurate predictions as the beginning L2 readers, they did not
apply this competence to their reading. McLeod and McLaughlin maintain that even at
the advanced level, L2 readers have not yet reached the point in reading performance
where restructuring; that is, automatic processing, occurs involuntarily. Even advanced
L2 readers must rely on deliberate decoding strategies, a slow and labor-intensive
process.
For cognitive L2 learning theorists, readers’ limitations are partially explained by
another important feature of cognitive learning theory: the limited processing capacity of
humans. To function effectively, humans develop ways to organize information. Some
tasks require more attention (controlled processing); others that have been practiced
require less. According to McLaughlin (1987), “the development of any complex
cognitive skill involves building up a set of well-learned, automatic procedures so that
18
controlled processes will be freed for new tasks” (p. 136). In reference to L2 reading, the
goal is to increase the number of subskills the reader is able to process automatically
while limiting the number of subskills requiring the reader’s attentional resources. To
read comprehensibly in an L2, the number of subskills requiring controlled processing
must be limited so as not to exceed processing capacity limitations: a condition
counterproductive to successful reading.
Reading: Similarities Between L1 and L2 Reading
As previously stated, cognitive learning theories describe language learning as a
skill comprised of a number of subskills based on a specific language system. After new
information is practiced and learned, the learner integrates that information into
preexisting internal mental representations through restructuring. But how does this
conceptual understanding of the reading process accommodate an L2 learner who already
possesses a native language system? In what ways does the process of learning a L2
intersect or overlap with L1 processing? How do they differ? As a substantial portion of
the theoretical bases for L2 reading theory is derived from L1 reading theory, it seems
appropriate to briefly describe research findings, particularly those germane to this study,
that describe similarities and the differences between L1 and L2 reading.
There are various stable factors in regard to the reading process, whether
performed in the L1 or the L2. Presented below are some of these elements, many of
which are described in detail by Bernhardt (2000). Similarities between the two reading
processes can be categorized as (a) rate and characteristics of language processing that
develop with increased fluency, (b) influence of reader knowledge on the reading process
and text comprehension, (c) types of text features that do and do not promote
comprehensibility, and (d) how good readers process text. Each is discussed briefly.
19
Similarities: Rate and Fluency of Reading
Two important similarities between L1 and L2 reading processes relate to fluency.
First and most obvious, studies indicate that in both the L1 and the L2, the speed of
processing increases as fluency increases (Brown & Hayes, 1985; Favreau et al., 1980;
Favreau & Segalowitz, 1982). Studies also indicate that with increased fluency, readers
are not only able to process text faster but also comprehend it more easily. Second,
research studies also reveal that language processing is directly related to the reader’s
phonological ability to recognize words (Brown & Hayes, 1985; Everson, 1998; Koda,
1987). Without the phonological link to word recognition, whether in the L1 or the L2,
reading with comprehension is impossible. As this present study investigates L2 readers
with L1 literacy in a cognate language, the phonological barrier to word recognition is
unlikely to present as great an obstacle to reading process and product as would be
expected if the same readers read in a non-cognate language.
Similarities: Reader Knowledge
A second area of overlap between L1 and L2 reading research involves reader
knowledge; specifically, that the knowledge a reader brings to the text influences
comprehension (Alderson & Urquhart, 1988; Carrell, 1987; Parry, 1987, 1996). From this
strand of research, studies with the greatest bearing on this present study relate to those
where researchers have concluded that the manipulation of content, whether the text be in
the L1 or the L2, can lead to differences in comprehension (Alderson & Urquhart, 1988;
Carrell, 1987; Parry, 1987, 1996). Seemingly trivial changes in content result in
surprisingly varied results on reading comprehension measures (Chihara et al., 1989).
Similarities: Text Features
In addition to research that focuses on issues involving the knowledge that a
reader brings to the text, some research dealing with text-based features in L1 and L2
reading has produced similar findings. One factor of paramount importance to this study
20
is that syntactic complexity does not necessarily predict text difficulty (Barnett, 1986;
Ross et al., 1991; Yano et al., 1995). Specifically, in regard to L2 reading research, it has
been found that as long as the text is coherent and the lexicon comprehensible, syntactic
complexity does not increase text difficulty (Oh, 2001; Parker & Chaudron, 1987; Ross et
al., 1991; Yano et al., 1994).
Similarities: Text Processing
It has been found that good readers, both in L1 and L2, are able to keep the
meaning of a text in mind as they read and are able to read word groupings, which
facilitates comprehension (Fransson, 1984; Hosenfeld, 1977; Shohamy, 1982). As
reading speed and fluency increase, fewer attentional resources are required to process
text, and more information is held in long-term memory. For both L1 and L2 readers, the
result is that the short-term memory is freed to keep text meaning in mind, thereby
increasing comprehension. These similarities between L1 and L2 reading processes are in
line with cognitive learning theory, specifically the aspect that predicts that practice leads
to automatization and that with automatization, reading becomes more efficient and more
effective.
Similarities: Summary
A summary of the similarities between L1 and L2 reading research includes the
findings that indicate that as readers progress in proficiency, the speed at which they
recognize words increases, enabling them to process text more holistically (Brown &
Hayes, 1985; Favreau & Segalowitz, 1982; Favreau et al., 1980). In addition, able
readers, through increased phonetic awareness, learn to be more strategic in word
recognition and develop an understanding of when to process componentially (i.e., wordby-word) and when to process holistically (Bernhardt, 1991b; Brown & Hayes, 1985;
Everson, 1998; Koda, 1987). Other findings indicate that the knowledge that a reader
brings to a text influences comprehension and even a small change in contextual
21
information can have a big influence on a reader’s comprehension (Alderson & Urquhart,
1988; Carrell, 1987; Chihara et al., 1989; Parry, 1987, 1996). Studies also reveal that
syntactic complexity does not predict text difficulty (Barnett, 1986; Ross et al., 1991;
Yano et al., 1995) and that good readers keep the meaning of the text in mind as they
read, which in turn increases comprehension (Fransson, 1984; Hosenfeld, 1977;
Shohamy, 1982).
Reading: Dissimilarities Between L1 and L2 Reading
A number of important features differentiate L1 reading from L2 reading. These
dissimilarities can be categorized as (a) variability in access to phonetic code, (b) L1
syntactic and semantic interference, (c) speed of language processing, (d) lexical storage
and recognition issues, (e) affective and motivational factors, and (f) learner maturation
factors including the fact that most L2 readers already possess a fully developed L1
literacy.
Dissimilarities: Variability in Access to Phonetic Code
First, unlike L1 readers, most L2 readers do not have a fully developed
phonological system in the L2 when they begin to read (Barnett, 1989). Therefore, L1
reading models that depend on the reader’s ability to decode text via phonologic symbols
or internal speech cannot be directly applied to L2 reading. Even at advanced proficiency
levels, phonetic decoding remains problematic for L2 readers (Segalowitz et al., 1991).
Segalowitz and his colleagues showed that L2 learners do not read as easily or as quickly
in their L2, even at advanced levels. The researchers attributed the lack of fluency to
poorer processing skills; that is, difficulty in encoding text into phonological symbols,
which ultimately slows basic word recognition. Again, in the present study, which
compares cognate languages, phonetic decoding presents less of a problem than it would
with non-cognates languages.
22
Dissimilarities: L1 Syntactic and Semantic Interference
Problems with L2 phonetic decoding, therefore, are at least partially attributable
to L1 interference. Perkins (1983) contends that L2 readers’ inability to construct
appropriate meaning from a text is also related to reader reliance on L1 syntactic and
semantic inferences to guide L2 processing and comprehension. In cognitive learning
terms, readers expect L2 passages to follow syntactic and semantic patterns similar to the
patterns they see in their L1. Processing constraints come into play when reading in the
L2 due to the high percentage of controlled processes necessary to restructure mental
representations set to meet L1 expectations (Mitchell & Myles, 1998). With such a high
percentage of processes requiring controlled attention, little memory is available for the
cognitive problem solving needed to comprehend text unfamiliar text. Clarke (1980)
refers to this phenomenon as the Short Circuit Hypothesis, in which “limited control over
the language ‘short circuits’ the good reader’s system causing him/her to revert to poor
reader strategies when confronted with a difficult or confusing task in the second
language” (p. 206). One of the tenets of this phenomenon, also refers to the Linguistic
Threshold Hypothesis, is that in order to read in a L2, a level of L2 linguistic ability first
must be achieved and that strong L1 reading skills cannot help readers compensate for
lack of linguistic knowledge (Bernhardt & Kamil, 1995). Stated in another way, a
sufficient number of processes must be automatized in the L2 to free up learners’
attentional resources to deal with the controlled processes needed for processing and
comprehension. Automatized L1 reading skills cannot compensate for a lack of
automatized processes in the L2.
Dissimilarities: Language Processing Speed
For the above-mentioned reasons and others, reading speed varies greatly between
the L1 and L2. A typical fluent L1 reader reads most material at 200 to 300 words a
minute (Grabe, 1999). Comprehension problems may result if, for linguistic reasons, a
23
reader is forced to read at slower speeds, which is typical for L2 readers (Carpenter et al.,
1994). According to Carpenter et al., L2 readers are slow because working memory
capacity is used ineffectively while waiting to assemble clausal information. Their
conclusion is similar to that of other studies that indicate that good readers keep the
meaning of the text in mind as they read (Fransson, 1984; Hosenfeld, 1977; Shohamy,
1982). Again it would seem that L2 readers’ processing constraints are responsible for
slower reading speeds.
Dissimilarities: Lexical Storage Issues
A fourth important difference between L1 and L2 reading is vocabulary
recognition. The lexicon of an educated native reader in the United States has been
estimated to range from 17,000 to 20,000 words (Goulden et al., 1990; Rott, 1999). Rott
(1999) estimates that students in the United States who are fulfilling their college
language requirement have a vocabulary of 3,000 to 5,000 words in their L2. In addition,
L1 students at most grade levels read material in which they know 99% of the words on a
given page (Carver, 1994). In L2 contexts, minimal word knowledge for fluent reading
has been estimated at 95% on any given page, although frequently readers know far
fewer than 95% of the L2 lexicon as they read (Laufer, 1989). This lack of familiarity
with a higher percentage of words leads L2 readers to react differently, both affectively
and motivationally, to reading in the L2 when compared with how they experience
reading in L1.
Dissimilarities: Affective and Motivational Factors
Grabe (1999) contends that it is an atypical experience for a L1 reader in a
postsecondary environment to be asked to read material whose lexical familiarity is far
beyond the level at which he or she generally reads. But rather than atypical, for L2
readers this experience is the norm. As college-level students rarely must cope with such
a high level of processing demands when they read in their L1, the potential to develop a
24
negative motivational respond to L2 reading exists. Grabe adds that research has shown
that L1 readers who consistently encounter difficult reading material on a regular basis
tend to abandon academic reading altogether and he suggests that repeated exposure to
difficult reading material may lead to the same outcome for L2 readers.
Dissimilarities: Learner Maturation Factors
Finally, although beginning L1 readers are normally young children, most L2
readers already possess a fully developed L1 literary, which they use simultaneously with
their newly emerging L2 literacy (Reder, 1994). Alderson (1984) notes that due to this
preexisting and distinct literacy, any appropriate research into L2 reading has to probe the
extent and nature of the reading ability those readers possess in their L1. He asserts,
however, that few studies acknowledge this important factor. Bernhardt (1991b) agrees,
insisting that amount and type of literacy experience that L2 learners have in their L1
affect L2 reading ability. She also stresses that L2 reading research must recognize that
L2 reading is a new and different type of literacy from the readers’ L1 literacy. Different
L1 literacy practices affect L2 reading abilities and strategies as well as the role that the
new L2 literary plays (Hudson, 1998).
Dissimilarities: Summary
In summary, reading research has identified a number of factors differentiating L1
from L2 reading. A summary of these findings include that L2 readers lack a fully
developed phonetic system in the L2 and that their pre-existing phonetic system interferes
with L2 reading (Barnett, 1989; Segalowitz et al., 1991). In addition, readers’ L1
interferes with the ability to construct and assign appropriate meaning to the L2 text
(Barnett, 1989; Perkins, 1983). Another difference is that even for proficient L2 readers,
reading speed between the L1 and the L2 is not comparable (Segalowitz et al., 1991;
Favreau & Segalowitz, 1982). A fourth factor differentiating L1 from L2 reading is that
the number of recognizable vocabulary words on any given page of text is not
25
comparable between the L1 and the L2 (Carver, 1994; Laufer, 1989; Nagy, 1987; Rott,
1999). Additionally, the affective and motivational responses to L2 reading differ from
the responses to L1 reading (Grabe, 1999). Finally, L2 readers, as adults, already have a
preexisting literacy as well as a variety of experiences with literary texts that affect the
L2 reading experience (Alderson, 1984; Bernhardt, 1991b; Reder, 1994).
It is important to note that, based on a review of multiple research studies
conducted by Bernhardt and Kamil (1995), there is considerable consistency in the
amount of variance in L2 reading scores that can be accounted for by readers’ L1
literacy: approximately 20 percent. Therefore in addition to the influence of readers’ L1,
multiple other factors affect how and to what extent L2 readers are able to process
reading input. The Bernhardt and Kamil data indicate that although L1 reading is an
important variable in L2 reading achievement, “in those studies that are able to account
for L2 proficiency, this construct seems to be a substantially more powerful predictor of
L2 reading ability” (1995, p. 30). For this reason, a variety of other factors, particularly
those that are reader based, need to be taken into account when examining how and to
what extent readers process and understand L2 reading texts.
Reading Factors That Contribute to L2 Reading Difficulty
As described in chapter I, a framework that can be helpful to the understanding of
why L2 readers do or do not have trouble processing and comprehending text can be
taken from the work of Beck et al. (1991) with L1 readers. It was their contention that
comprehension breakdown could be attributed to two major factors: reader failure and
text failure. For the purpose of this study, research that describes aspects of both of these
factors is described but with greater emphasis placed on the focus of this study, text
failure. The following discussion highlights elements of text-based factors that can lead
to comprehension breakdown. They include (a) text syntax, (b) text organization, (c)
vocabulary difficulty, and (d) text length. Following this discussion, a brief review of
26
reader-based factors is described, including (a) background knowledge, (b) cultural
expectations, and (c) L2 proficiency. Summaries of text-based and reader-based factors
that contribute to reading comprehension and comprehension breakdown are also
included.
Text-Based Factors: Syntax
It would be irresponsible to discuss text-based reading factors without
acknowledging that both the process and the outcome of reading are multifaceted and
cannot be reduced to words or, in the case of syntax, to the relationship between words.
Nonetheless, in an effort to shed light on specific reading variables that are problematic
for L2 readers, attention is focused on L2 reading studies that have attempted to isolate
syntax, specifically the effect of syntactic complexity, on student comprehension of text
in L2. In addition, it should be stated that the following studies deal with syntax
manipulation within reading texts. Studies that examine readers’ recollections and
responses to isolated sentence pairs outside of a broader reading text are not be described
as “performing syntactic operations outside of a discourse context and using these same
patterns productively within a discourse context are different processes” (Bernhardt,
1991a, p. 34).
Alderson (2000) contends that despite the fact that shorter sentences tend to be
syntactically simpler than longer sentences, considerable research shows that it is not the
length of a sentence that facilitates comprehension, but rather what is communicated by
the sentence. Research conducted by Blau (1982) supports Alderson’s assertion. The
motivation behind Blau was to determine the effect of syntactic complexity on two
groups of ESL students through the use of 18 reading passages created with three levels
of syntactic complexity. Although vocabulary and content of these passages were held
constant, sentence structure varied. Version 1 consisted of short, simple sentences;
version 2, of complex sentences with clues to underlying relationships left intact; and
27
version 3, of complex sentences without clues. Blau’s (1982) research suggested that the
least complex sentences presented an obstacle to comprehension as the speech contained
within them was “choppy and unnatural . . . [and] the relationships and meaning revealed
by the formation of complex sentences were apparently lost” (p. 525). Blau believed that
the complex sentences presented a comprehensibility advantage over syntactically
simplified sentences because the relationships between phrases and sentences in text that
contained more complex syntax were explicitly stated. She also noted that the most
syntactically simple sentences were the least likely to be encountered in a real-life setting.
Her findings revealed not only higher comprehensibility in texts with complex syntax but
also that students perceived that the versions that contained more complex structures
were easier to read.
Blau felt that it was the underlying relationships within and between sentences
that accounted for students’ superior performance, and not the content of vocabulary.
Barry and Lazarte (1995) conducted a similar study that also used passages with varying
degrees of syntactic complexity. Although the researchers modified the syntactic
complexity of L2 texts, their main focus was on the effect of an increased number of
embedded clauses on student recall scores. At the lowest level (level I), the texts were
comprised of only the most essential ideas in single-clause sentences. As texts were
modified to make them syntactically more complex, the researchers added clauses that
contained non-essential information. The level II texts contained one additional clause
and, at level III, two additional clauses were added, each clause containing non-essential
information. The research subjects were high school L2 Spanish learners. Results of this
study indicated that as the level of complexity increased, the amount of essential
information recalled decreased, although at all complexity levels readers recalled
essential information better than non-essential information. The researchers asserted that
students were not picking up the non-essential information contained in the embedded
28
clauses for later recall. In addition, and more important, the presence of non-essential
information was interfering with readers’ ability to recall essential information.
Barry and Lazarte suggested a reason for their findings, which is aligned to
cognitive learning theory. They proposed that sentence structures that contain nonessential information put a strain on the processing capacity of short-term memory, thus
diminishing readers’ ability to recall text (Kintsch & van Dijk, 1978 as cited in Barry &
Lazarte, 1995, p. 491). They also contended, as do cognitive theorists, that humans, as
limited-capacity processors, have limitations to their attentional resources. Barry and
Lazarte’s findings also supported the notion that it is not so much the length or
complexity of a text that determines its difficulty as much as the comprehensional
support provided by the text to the reader. Although difficult syntax may inhibit
comprehension and recall, simplified syntax does not guarantee that the intended message
conveyed by the text will be easier to comprehend unless attention is paid to the content
of the sentences within the text.
Ulijn and Strothers (1990) also studied the effect of syntactic simplifications on a
computer science journal article versus the unmodified version of the same article. Their
research subjects were students of English for Science and Technology (EST) at the
college level. Forty-eight of the subjects were L1 English and 48 were L2 English
learners, all of whom read four sets of English texts, two unmodified and two
syntactically simplified. Their study determined that the complexity of the syntax did not
significantly affect the level of reading comprehension for either expert or novice readers.
The researchers suggested that “syntactic simplification of an EST text is not a real
simplification” and that “teachers might give priority to other conceptual ways of
rewriting texts” (Ulijn & Strothers, 1990, p. 38). They added that although they believed
that a syntactic analysis of a text was not necessary when assessing its appropriateness for
L2 readers, “a complete conceptual and lexical analysis may be” (Ulijn & Strothers,
1990, p. 38).
29
Alderson (2000) warned that the process of making a text less syntactically
complex could have the effect of either distorting the message or increasing difficulties
involving other text features. This assertion was echoed by Yano et al. (1994) when they
advocated for maintaining target language syntax whenever possible in reading passages
because “despite greater length . . . and the use of subordination, a single multiclausal
sentence that explicitly links two propositions can be easier to process, as well as be a
more realistic model of language use than a series of shorter, linguistically simpler
sentences” (p. 191). Bernhardt (1991b) voiced concern about the reading of simplified L2
syntax by adults for an entirely different reason; that is, that texts should be tailored to
adult readers. She stated that “adult learners are more capable of reading adult texts and
should be expected to do so. Syntactically simplified prose or simplistic topics should not
be tolerated by students or by teachers” (Bernhardt, 1991b, p. 227).
Research suggests, therefore, that syntactic simplification is an insufficient means
of increasing text comprehensibility unless other textual elements are also taken into
consideration. In addition, it appears that complex syntax alone is not detrimental to
comprehension as long as the information contained within the sentences is germane to
the topic, presented logically and contributes to overall text comprehension. Although
Barry & Lazarte (1995) demonstrated that the addition of non-essential information
interfered with comprehension, Blau (1982) concluded that it was the lack of surface
signaling, lost with syntactic simplification, which prevented problems for readers, not
only in relation to comprehensibility but also in regard to the perceived text difficulty.
Additional research conducted by Shook (1977) supports both of these conclusions.
Text-Based Factors: Text Organization
Text organization is another text-based factor that contributes to or detracts from
reading comprehension for L2 learners. For the purpose of this study, the terms text
organization and text coherence are used interchangeably in reference to how paragraphs
30
relate to each other and how relationships between ideas are or are not explicitly stated or
expressed in the text. Along with surface-level features such as syntactic and lexical
elements, text organization has long been an object of study for reading researchers in
both L1 and L2.
L1 reading research has demonstrated that revisions that increasing the structural
and explanatory coherence of texts result in substantial increases in recall (Beck et al.,
1991, 1995; Britton & Gulgoz, 1991). Lack of text coherence can inefficiently utilize
attentional resources. If there exists argument overlap between new input and information
already stored in the short-term memory, the new input is accepted and integrated. If not,
a resource-consuming search of all previously processed information is made, resulting in
an increased use of controlled processes (Kintsch & van Dijk, 1978).
When comparing types of revisions that contribute to improved comprehension,
Beck et al. (1991) suggested that revisions that explicitly connect causes to events and
events to consequences through clear signaling were of greatest benefit to
comprehension. Without such signaling, the reader must generate the relationships
between ideas, whereas signaling reduces the load placed on working memory during text
processing (Meyer, 2003). The research of Beck et al. (1991) also determined that
explicit signaling of information throughout the text had the potential for increasing
readers’ comprehension of the text and that, similar to the findings of Barry & Lazarte
(1995), the inclusion of information not directly related to the topic interferes with
comprehension. McNamara et al. (1996) described two of the beneficial effects of greater
text coherence as (a) increased comprehension and (b) decreased amount of active
language processing taking place during reading. Through research, McNamara et al.
(1996) and Kintsch (1994) found that L1 readers who knew little about the domain of the
text benefited the most from coherent texts. They argued that a poorly written text or a
text that assumed a certain amount of background knowledge forced the reader to engage
in compensatory processing and to infer unstated relations in the text (McNamara et al.,
31
1996). Although a less coherent text could potentially stimulate high-knowledge readers
to more actively engage in text processing, McNamara et al. contended that lowknowledge readers derive the most benefit from fully coherent texts.
Urquhart (1984) explored the effects of chronological and spatial order in text
presentation and showed that for both L1 and L2 English readers, texts organized
according to the sequence of events were read faster and were easier to understand than
texts whose temporal sequencing was disturbed. Other studies in L2 reading have
reported similar findings. Horiba (1993) studied text organization, specifically the role of
causal reasoning in L2 narrative comprehension. Her study involved 21 college-aged L2
Japanese learners with L1 English backgrounds. Study results indicated that L2 readers
had difficulty detecting the higher-order causal structures in texts, a process Horiba
referred to as “a central component of L1 readers’ mental representation of the text
information” (1993, p. 49). Horiba noted that as L2 readers’ skills improved, their ability
to detect causal relations increased, but she warned that even advanced L2 readers needed
to process the entire text at least once before they were able to uncover causal
relationships.
Kobayaski (2002) also researched text coherence with L2 readers, specifically the
effects of text organization on reading comprehension as measured by cloze, limited
production response questions, and summary writing. Her subjects were 754 Japanese
ESL university students. Her study revealed that text organization had a significant
impact on students’ performance. Interestingly, she concluded that although clearly
structured texts resulted in better comprehension for more-proficient students, text
structure made little difference in the performance of less-proficient students. Kobayaski
speculated that the reason for this disparity was that the lower-proficiency readers had not
yet reached a level where they were able to identify the text structure much less use the
clearer structure to their advantage, a finding in line with the Linguistic Threshold
Hypothesis. She concluded that “when the level of proficiency is low, the learners have
32
difficulty at a decoding level and consequently can rarely go beyond sentence level
meaning or literal understanding” (Kobayaski, 2002, p. 211). Although McNamara et
al.’s (1996) findings regarding the greater benefit of coherent texts to lower-proficiency
readers seems to contradict Kobayaski’s conclusions, the discrepancy may be attributable
to the definition of “low proficiency,” which for McNamara et al. meant lowerproficiency adult L1 readers, whereas Kobayaski was working with college-level L2
readers of English who were already fully literate in Japanese.
Kobayaski’s findings support those of Horiba (1993), whose research also found
that as competence increased, L2 readers became more adept at detecting causal and
enabling relationships in texts. As Horiba explained, the basic benefit of increased
coherence is that the reader is not only able to decode ideas and events, but also the
relationships between those ideas and events, thus creating a mental representation
facilitating comprehension and recall. Horiba contends that events and ideas that are
causally related are remembered better than those that are not and that highly connected
ideas and events are more memorable, rated as more important, and retrieved faster than
events and ideas with few connections. Alderson (2000) concurs, asserting that the types
of texts written to expose the reasoning that connects causes to events and events to
consequences are easier to understand than texts that fail to make causal sequences clear.
The self-reporting of L2 readers in Hammadou Sullivan (2001) reveals that the most
common signal that readers believed help their comprehension were the words beginning
a new paragraph. Other readers reported that the first word in the sentence helped clarify
temporal sequencing (Beck et al., 1991). Clearly, unambiguous and straightforward text
organization with clear signaling is a factor that should be considered when determining
which text-based features detract from or contribute to a text’s comprehensibility level
for L2 readers.
33
Text-Based Factor: Level of Vocabulary Density or
Difficulty
At least as important to reading comprehension as syntax and text organization
are the lexical features of a text. In this section, studies that focus on the influence of the
presence or absence of higher- and lower-frequency vocabulary on reading
comprehension are reviewed, working from the basic assumption that people who know
more word meanings comprehend text better and that the lexical knowledge of lessproficient L2 readers is inferior to that of more-proficient readers (see Graves, 1986).
Initially, research is presented that relates to the influence of text vocabulary on reading
comprehension in the L1, and this section concludes with research that examines the
importance of vocabulary usage for L2 reading comprehension.
Stahl et al. (1991) studied the effect of vocabulary knowledge and its relation to
text comprehension for 159 English L1 high school students. Students read a magazine
article and, later, comprehension was measured by a written recall protocol. Results
indicated that difficulty in the level of vocabulary tended to affect the number of
propositions, or idea units, recalled, whereas student prior knowledge of the text content
affected which units the readers recalled. The researchers concluded that vocabulary
difficulty had its most major effect on the recall of text details, and strong background
knowledge did not compensate for lack of lexical knowledge. The researchers also
suggested that text difficulty should be reconceptualized in terms of two separable
aspects, one involving support for processing of individual lexical components as in
phrases or sentences, and the other involving support for processes that assist readers
with the overall idea or gist of the passage.
In an attempt to measure the contribution of vocabulary knowledge to L1 text
difficulty, Carver (1994) took a different approach. He conducted two studies, one with
elementary school children and the other with graduate students, where the relationship
between the relative difficulty of a passage was compared to the number of unknown
34
words identified by the reader. Passage difficulty was determined by the difference
between a measure of the reading ability of the student in grade-equivalent units and a
measure of the difficulty level of the material in grade-equivalent units. Carver concluded
that despite the differences in age between the two research populations, findings were
generally consistent: (a) when the material was considered relatively easy, then close to
0% of the words were unknown, (b) when the material was relatively hard, then 2% or
more were unknown, and (c) when the difficulty of the material was closely matched to
the ability of the reader; that is, that the measure of the students’ ability matched that of
the materials’ intended target grade level, then around 1% of the basic words were
unknown. According to Carver, if reading as an instructional activity is intended to
improve students’ reading level, then texts should be matched as closely as possible to
students’ current reading level; that is, so that 1% of the words are unknown to the reader
(Carver, 1990 as cited in Carver, 1994, p. 435).
The vocabulary used in reading texts has consistently been shown to have an
effect on L2 reading comprehension as well, and in many ways research on L1 reading
and L2 reading converge. Laufer (1989) attempted to measure the relationship between
the number of words understood by 100 college-level ESL readers of academic texts with
the quality of their text comprehension. Research results indicated that to adequately
comprehend the texts in this study, readers needed to be familiar with at least 95% of the
lexical items. Laufer (1989) stated the following:
This does not mean that a person cannot understand a text when
the lexical coverage is lower than that. Other factors, like
grammatical clues, text organization, subject-matter familiarity
may facilitate comprehension . . . [but] in the majority of the cases,
when the lexical coverage was below 95%, comprehension was
impaired in spite of the other facilitating factors that might have
affected reading. (p. 319)
Research by Liu & Nation (1985) also concluded that L2 readers needed to know
approximately 95% of the words in a text to gain adequate comprehension and to be able
to guess unknown words from context.
35
Stanovich (1991) affirmed the importance of vocabulary in reading texts when he
stated that “while it is possible for adequate word recognition skill to be accompanied by
poor comprehension abilities, the converse virtually never occurs” (p. 418). In many
ways it appears that lexical knowledge is the key to reading comprehension, whether that
reading be in the L1 or the L2. Although Laufer (1989) stated that adequate grammatical
clues and text organization could not compensate for lack of lexical knowledge, others
feel that the opposite is possible. Alderson (2000) contends that knowledge of the lexis,
as well as more general and specific content knowledge, might well compensate for lack
of linguistic knowledge. This again points to the pivotal role knowledge of the lexis holds
for L2 readers. Although both difficult vocabulary and low topic familiarity can reduce
comprehensibility, easier vocabulary can increase comprehension of unfamiliar topics but
familiar topics do not ease the processing constraints imposed by difficult vocabulary
(Alderson, 2000). Indeed, Bossers (1992) argues that L2 vocabulary usage, much more
than syntax, could be directly connected to students’ success or failure as L2 readers.
Additionally, having to struggle with a reading because of unknown vocabulary not only
negatively affects comprehension but also takes the pleasure out of reading (Alderson,
2000).
One more important factor to discuss regarding vocabulary knowledge is what is
meant, precisely, by knowing a word. Bernhardt (1991a) contends that for a lowproficiency L2 reader, an unfamiliar word is reduced to one concept, and that the concept
is generally derived from a L1 translation. She explains that the mental representations of
a given word or concept are of varying degrees of usefulness and relativeness for L2
readers. Bernhardt (2003) argues that “the reader may ‘know’ the vocabulary item
breakfast but may not have a relevant or meaningful semantic field attached to it” (p.
113). Quinn (2003) supports Bernhardt by stating that when it comes to interpreting a
sign, such as vocabulary item, “it depends on who put the sign up, with which readers in
mind, and what those readers are assumed to know” (pp. 40-41). Quinn also argues that
36
frequently in L2 instruction “a word’s meaning is the [L1] gloss that occurs alongside it
in a bilingual vocabulary list . . . [and] this is about as far as a word meaning goes. It
often seems to come down to a bilingual pairing” (p. 44). What it means to know L2
vocabulary for successful reading comprehension is a topic addressed below.
Text-Based Factor: Text Length
Currently, research on the effects of text length for L2 readers offers conflicting
results. Although some researchers have obtained findings that lead them to advocate for
the use of longer texts and multiple texts in serial form for L2 students (Krashen, 1988;
Maxim, 2002), research conducted by Leow (1997) concluded that shorter versions of
texts are easier for students to comprehend. Even though both factions agree that long
and short reading texts are processed differently, Leow (1997) contends that shorter texts
facilitate comprehension because they reduce the processing demands required of the L2
reader.
Leow (1997) derived this conclusion from his research that compared student
comprehension of an unmodified 631-word Spanish magazine article with
comprehension of three modified version of the same article. Whereas version 1 was the
original versions taken from the magazine, version 2 maintained the text length but
highlighted one particular grammar point. The last two versions were shortened to 384
words, and only version 4 contained the grammatical enhancements found in version 2.
Leow concluded that students understood the shorter texts significantly better than the
longer text based on a task of reproducing 20 pieces of information contained in all
article versions.
In accordance with the procedures used to measure reader understanding, the
subjects who read the shortened text did indeed produce more correct responses (Leow,
1997). A question that should be considered, however, is how the construct of text
comprehension was operationalized in this study. If the only measure of reader
37
comprehension involved elicitation of pieces of information that the readers could recall
directly from the text, then Leow is correct in asserting that these shorter texts were
indeed more comprehensible to research participants. If, however, the intent of the
comprehension instruments was to measure how readers mentally assemble texts as they
read or to assess the extent to which readers holistically understand the author’s intent or
the global text message, then retrieving pieces of information drawn directly from the text
was inadequate.
Many L2 reading researchers believe that the longer the text, the greater the
opportunity for readers to activate metacognitive strategies that will assist their
processing and comprehension, provided that the text is comprehensible (Day &
Bamford, 1998; Kintsch & van Dijk, 1987; Maxim, 2002). Indeed, some argue that short
(i.e., 200-350 words) texts have the opposite effect on L2 readers; that is, their lack of
redundant features, rapid breaks in structure, and change of topic or focus impede rather
than facilitate comprehension (Kintsch & van Dijk, 1978; Swaffar, 1991). Longer texts,
on the other hand, allow for greater topic clarification because readers have more time
and opportunity to identify and refine areas of misunderstanding.
Other researchers argue for the use of longer texts in L2 reading, contending that
the advantages of using longer texts outweigh the disadvantages (Grabe, 1991; Martino &
Block, 1992; Swaffar & Bacon, 1993). Martino & Block (1993) contend that longer texts
encourage readers to use a wider variety of strategic reading processes as, with these
texts, they have the time and opportunity to recognize the author’s presentational style,
leaving resources available to focus on the broader message rather than on individual
lexical or syntactic items. Additionally, students reading longer texts seem to be able to
overcome some of their proficiency limitations by drawing on recurring vocabulary,
characters, and themes (Maxim, 2002). Grabe (1991) sums up the advantages of reading
longer texts by stating that they “build vocabulary and structural awareness, develop
38
automaticity, enhance background knowledge, improve comprehension skills, and
promote confidence and motivation” (p. 396).
Although the present study, motivated by logistic constraints, is conducted with
unmodified texts of less that 600 words, attempts have been made to modify three of
these texts through structural modifications, such as redundancy and topic signaling,
modification features that naturally occur with more frequency in longer texts. Such
modifications, however, cannot compensate for the lost of quantity and richness of input
to which readers of longer texts are exposed.
Text-Based Factors: Summary
Of the text-based reading factors, type of L2 vocabulary is by far the most
influential for text processing and comprehension (Bossers, 1992; Stanovich, 1991). L2
readers cannot compensate for lack of lexical knowledge through other text-based
sources (Alderson, 2000; Laufer, 1989; Liu & Nation, 1985). Additionally, vocabulary
learning in context has been showed to be an ineffectual form of word identification and
therefore is of limited usefulness when employed to facilitate comprehension
(Alderson, 2000; Carmine et al., 1984; Deighton, 1959; Haynes, 1993). For adequate
comprehension, L2 readers need to know approximately 95% of the words in a text
(Laufer, 1989; Liu & Nation, 1985). Lack of vocabulary knowledge not only results in
poorer text comprehension but also changes how readers approach a text, as their
motivation to read and the enjoyment they receive from reading is decreased (Alderson,
2000; Day & Bamford, 2002; Grabe, 1999). In addition, lack of adequate vocabulary
affects the amount of information recalled as well as the details contained in the recalled
information (Stahl et al., 1991). On the other hand, knowledge of the lexis might
compensate for lack of linguistic knowledge (Alderson, 2000). Finally, attention should
be pay to two issues: (a) the clear definition of what it means to know a word and (b) how
unfamiliar words are defined or made known to L2 readers.
39
Other text-based reading factors reviewed include syntax, text organization, and
text length. Research findings reveal that clear relationship signaling within, between,
and among sentences is more important for text processing and comprehension than intersentential constituent ordering or sentence length, with the first words in a sentence and
the initial words in a paragraph being of paramount importance (Alderson, 2000; Beck et
al., 1991; Blau, 1982; Hammadou Sullivan, 2001; McNamara et al. 1996; Meyer, 2003;
Yano et al., 1994). Greater sentence complexity becomes detrimental to reading only
when the information added is non-essential, thus distracting from, rather than adding to,
comprehensibility (Barry & Lazarte, 1995; Beck et al., 1991). Moreover, longer reading
texts have been shown to facilitate both the way readers process and comprehend L2 texts
(Grabe, 1991; Kintsch & van Dijk, 1978; Martino & Block, 1992; Maxim, 2002; Swaffar,
1985; Swaffar & Bacon, 1993). For those researchers who compared sentences of varying
length, sentence length that most closely resembled that found in naturally occurring L2
written and spoken discourse was preferred (Alderson, 2000; Bernhardt, 1991b; Blau,
1982; Ulijn & Strothers, 1990).
Reader-Based Factors: Background Knowledge/Schemata
Despite the fact that this study targets text-based factors that contribute to
comprehension problems for L2 readers, there are a number of reader-based factors
worthy of brief review due to their impact on the process and product of reading. Two of
these factors, which are highly related, are background knowledge and reader schemata.
Alderson (2000) defines schemata as the “interlocking mental structures
representing readers’ knowledge” (p. 33). He contends that as readers process text, they
integrate the new text-based information into their preexisting schemata. The preexisting
schemata, therefore, influence not only what readers recognize and interpret from the
text, but also how they store the incoming information. Background knowledge, on the
other hand, contains scripts or frames for common events into which new knowledge is
40
slotted (Alderson, 2000). For the purpose of this discussion, the terms background
knowledge and schemata are used interchangeably.
For many years, researchers have recognized that the readers’ preexisting
knowledge influences not only what they understand and remember, but also how they
process the text. Swaffar (1991) argues that the use of passages that contain familiar
components or subject matter eases readers’ cognitive load. This occurs because L2
readers can anticipate textual ideas rather than rely exclusively on language to convey
these ideas (Carrell, 1984, 1985; Lee, 1986a). Swaffar (1991) asserts that when focal
attention and macroprocessing; that is, global processing, deal with familiar schemata,
well-trained cognitive pathways are activated, freeing up memory to deal with unfamiliar
language or text details. Indeed, in research conducted by Hammadou (2000), results
revealed that readers’ background knowledge was the most powerful factor affecting
their ability to recall L2 texts.
As stated previously, the current research study deals with text-based factors
influencing L2 reading and not those factors that a reader brings to the text. Background
knowledge and schemata have been touched upon only in recognition of their potential
influence on readers, an influence that is difficult to measure quantitatively. Another
feature highly related to how readers process and understand text, yet elusive and difficult
to define, are cultural expectations and bias.
Reader-Based Factors: Cultural Expectation/Cultural Bias
L2 readers approach texts with cultural biases of which they are unlikely to be
aware. Shook (1997) argues that this native-culture-based set of expectations and
experiences contain readers’ values, assumptions, and norms as well as personal
variations. The reason that this topic is appropriate to this discussion is that researchers
have been able to identify readers’ cultural expectations by noting how even minor
changes to text elements can result in a significant change in students’ comprehension
41
and overall performance. Chihara et al. (1989) demonstrated that culturally determined
expectations could be encoded in “what would seem to be the most trivially simple
elements of surface form such as names and places” (p. 144). Although this finding may
have little bearing on this present research, it is worth keeping in mind that language
samples that appear to be straightforward in either style or message, nonetheless are
imbued with underlying cultural biases and expectations.
Quinn (2003) warns against assuming that readers share the culturally specific
knowledge that a text presumes. He cautions that “the foreign reader in particular needs
to know not only the words, constructions and their written representations, but all the
larger practices and conventions that motivate the deployment of such linguistic
resources and make any text the instance of a genre” (p. 38). Alderson (2000) adds that
texts that follow recognized conventions and appeal to commonly held assumptions are
easier to process, but only for those who happen to acknowledge the conventions or share
the assumptions. As such conventions and assumptions are largely subconscious and
virtually impossible to fully articulate, much research is still needed in this area. The
attempt here is simply to acknowledge their potential impact on the process and product
of reading.
Reader-Based Factors: L2 Proficiency
The last reader-based factor to be discussed, with direct implications on how well
a learner is able to process and comprehend a text, is the reader’s level of L2 language
proficiency. It would seem self-evident that the greater the reader’s language skills, the
stronger his or her reading ability, whether in the L1 or the L2. Some researchers have
speculated, however, that at least in the L2, the correlation is not so straightforward.
Bernhardt (1991b) hypothesizes that not all factors related to L2 reading follow a weakerto-stronger continuum in accord with the increased reading proficiency of the reader.
According to Bernhardt, syntactic errors seem to increase, at least temporarily, as L2
42
reading proficiency increases. She speculates that this occurs as learners become greater
risk takers and therefore misuse and misunderstand complex syntactic forms as they read.
Although Bernhardt concedes that this hypothesis is in need of empirical support, the idea
that increased language proficiency may be temporarily detrimental to aspects of the L2
reading process is intriguing and worthy of future investigation.
Reader-Based Factors: Summary
Three of the primary reader-based factors that are influential for L2 reading
processing and comprehension are (a) background knowledge or schemata, (b) cultural
expectations or bias, and (c) the level of L2 proficiency. Researches have long
acknowledged the important and, at times, pivotal role that reader knowledge brings to
bear on how a text is processed and understood (Alderson, 2000; Carrell, 1984;
Hammadou, 2000; Lee, 1986a; Swaffar, 1991). Readers also bring to their reading of a
text cultural expectations and biases that are, for the most part, unconscious and difficult
to identify (Chihara et al., 1989; Quinn, 2003; Shook, 1997). Of importance to the present
study is the finding that even seemingly insignificant text elements can serve to either
strengthen or diminish how a text is understood and recalled. Recognized conventions
that employ commonly held cultural assumptions are easier for students to process
(Alderson, 2000) but no two readers share identical assumptions or an identical concept
of culture (Quinn, 2003). Finally, although it can be generally assumed that as language
proficiency increases, language processing and comprehension increase proportionately,
researchers should be open to the possibility that not all factors associated with L2
reading develop simultaneously and uniformly (Bernhardt, 1991b).
Why Literary Texts?
As described in chapter I, there are many reasons to use literary texts as a primary
source of reading material in early L2 learning. These reasons include the cultural,
linguistic, and educational input inherent in literary texts. In addition, literary texts are an
43
excellent means of access into the commonly held notions, beliefs, and ideas that are
reflections of the L1 community.
Value of Literary Texts: Cultural
Expressions of culture are communicated both explicitly and implicitly through
the literature of a people. Bernhardt (1991b) goes so far as to say that for L2 learners,
literary texts “provide the most significant source of cultural materials” (p. 227). These
texts may either serve as a means of direct access to culturally specific concepts or
assumptions or as a springboard for L1 cultural comparison or awareness. For that
reason, literary texts function as appropriate language input sources that may lead to
greater appreciation and awareness of aspects of both the L1 and the L2 cultures.
Value of Literary Texts: Linguistic
Linguistically, the language contained within literary texts is a genuine sample of
language used to communicate ideas in a form that has been recognized for its excellence
of expression. Shook (1997) contends that the value of literary texts as vehicles for
developing language skills for beginning learners is often overlooked. They are
appropriate, he stated, because “authors of literary texts will use new turns of phrase,
rhyme, metaphor, simile, word order, author’s voice and/or other striking form-meaning
combination while expressing their own perspectives or the shared values and
assumptions of the culture” (p. 235). Additionally, a broad range of literary styles and
genres are available for integration into a language course, avoiding the possibility that
learners may be exposed only to linguistic styles or perspectives of one or two textbook
authors.
Value of Literary Texts: Educational
Literary texts are also of educational value. Intellectually challenging content,
such as that found in literary readings, may serve to stimulate student interest in a way
44
that texts constructed to illustrate thematic or grammatical concepts may not. Literary
texts not only have intrinsic merit but also represent the types of texts to which collegelevel students have been traditionally exposed in their L1. Additionally, earlier exposure
to literary texts prevents readers from having to make an abrupt transition between text
types as they progress through their L2 studies. Through the use of literary texts in lowerlevel language courses, curricula can also be devised that holistically unite instruction
and methodology across all levels of instruction. But although some educators call for
increased use of literature for early language learners (Barnett, 1991; Bernhardt, 1991b;
Rice 1991; Shook, 1997), the use of literary texts for low-proficiency students is limited
in scope.
Reasons Literary Texts Present a Problem in L2 Instruction
In addition to the arguments presented in chapter I highlighting why literary texts
present a problem for L2 instruction, some educators would argue that another reason that
literary text are avoided is because instructors expect the student to rise to the level of the
text rather than adapt the text to the proficiency level of the student. In her article entitled
“Teaching Literature or Teaching Students?” Bernhardt (1995) speculates as to how the
face of L2 instruction might change if educators adjusted the focus of their instruction.
According to Bernhardt, “if we are in the business of teaching students rather than
teaching texts, then perhaps we can slow the revolving door out of upper-level classes by
tailoring instruction to the appropriate level” (1995, p. 6). Unfortunately, in lowerdivision courses, most students are simply not proficient enough to cope with literary
texts without additional linguistic support. Bernhardt (1991b) asserts that the reason that
literary texts are so difficult for students is that when readers approach these texts, the
only resources at their disposal are their limited linguistic skills and none of the implicit
knowledge that the native-speaking group possesses. Because literary texts are so
lexically and syntactically complex and may be conceptually unfamiliar, cognitive
45
overload can occur. Unprepared students, therefore, are left with the burden of reading a
text that taxes their cognitive abilities to such an extent that the reading process becomes
overwhelming and onerous, which reduces comprehension.
Making L2 Literary Texts Easier to Read
From the previous discussion, it is clear that beginning L2 readers lack the
requisite linguistic and cultural knowledge implicitly assumed by authors of literary texts
intended for L1 readers. For this reason, literary texts are so cognitively challenging that
L2 readers must struggle to read them, resulting in difficult text processing and poor
comprehension. Frequently, reading becomes such a cognitive burden that much of the
pleasure that otherwise may have been derived from reading is lost (Anderson, 2000). If
L2 readers, as limited-capacity processors, cannot process all the information to which
they are exposed at one time, and it is this cognitive overload that constrains the process
and product of reading, than how can texts be modified to ease cognitive demands, thus
decreasing processing constraints and, in turn, increasing reading comprehension and
pleasure?
Text Modification: Available Choices
Evidence suggests that text modifications intended to enhance comprehensibility
for lower-proficiency readers are useful, no matter the form or type (Blau, 1992; Brown,
1987; Long, 1985; Ross et al., 1991; Tsang, 1987). Unmodified texts, on the other hand,
provide neither the lexical support nor the clues to the subtle connections between
phrases and ideas needed to facilitate processing of unknown language (Ross et al., 1991;
Yano et al., 1994). A comprehensible text would reduce the controlled or attentional
processing demands and therefore increase the number of subskills available for
automatic processing. This would thus free up short-term memory so that the reader
could keep in mind the meaning of what is read. But how, exactly, might a text be
modified in order to reduce controlled processing demands?
46
Text Modification: Linguistic versus
Conversational/Interactional
A review of relevant literature reveals two primary types of text modification used
to render texts more comprehensible to limited-proficiency readers. They are (a)
linguistic modifications, in the form of simplification; and (b) conversational or
interactive modifications, in the form of elaboration. The following discussion describes
the features and genesis of linguistic modifications, specifically simplification. This
description is followed by research findings indicating the positive and negative effects of
linguistic modifications. The conversational or interactional approach to text modification
is described in a similar manner. It is argued that although linguistic modifications are
structurally simplier, text modifications in the form of elaboration are cognitively less
taxing, resulting in L2 texts that are more easy to process and comprehend for lowproficiency readers.
Linguistic Modification: Genesis/Roots
Linguistic modification in the form of simplification is the dominant approach to
text modification used in commercially published reading material in L2 settings (Long
& Ross, 1993). The practice of simplifying a text to make the language more
comprehensible for NNSs (non-native speakers) is of long standing, and many of the
simplification techniques currently in use can be traced back to Ferguson’s (1971) work
with foreigner talk. Ferguson categorized various language features that he observed
when NNSs engaged in conversations with NSs (native speakers). He coined the term
foreigner talk (FT) to describe what he observed as a “conventionalized register of
simplified speech . . . used by speakers of a language to outsiders who are felt to have a
very limited command of the language or no knowledge of it at all” (p. 143). Among the
most common features of FT that Ferguson observed were the use of shorter utterances
and speech that was syntactically and propositionally less complex (Long, 1983a). In
47
addition, he described the range of vocabulary used in NS-NNS speech as being more
restricted; idioms and low-frequency lexical items seemed to be avoided.
Linguistic Modification: Simplification
Following Ferguson’s model describing the types of linguistic simplification that
NSs make to accommodate NNSs, textual simplifications also focus on limiting
vocabulary and syntactically complex language. The most common forms of linguistic
simplification include the use of shorter utterances, simpler syntax, simpler lexis, deletion
of unnecessary morphological inflections, and maintenance of canonical word order (Oh,
2001). Simplification of L2 reading materials results in shorter sentences, avoidance of
idioms, deletion or rephrasing of complex structures and use of low-frequency
vocabulary (Long & Ross, 1993). Simplified sentences also contain transparent
structures, low numbers of S-nodes, and the use of surfaced or nonpronominal subjects
(Tweissi, 1998). These and other simplification modifications are made to L2 reading
texts for the purpose of making the text simpler and thus more accessible to lowproficiency readers. Uses of these types of modification techniques are justifiable on
cognitive grounds as well. If fewer attentional, or controlled processes are required to
decode a literary text’s linguistic elements then, logically, more resources are available to
the L2 reader for discourse-level comprehension. Indeed, various research studies
indicate that simplified texts seem to be better understood by L2 readers than unmodified
versions of the same texts.
Simplification: Positive Research Findings
Various studies indicate that, compared with reading L2 texts in their unmodified
forms, simplified texts are generally better understood by readers (Blau, 1982; Brown,
1987; Parker & Chaudron, 1987; Ross et al., 1991; Tsang, 1987; Yano et al., 1994). Ross
et al. (1991), for example, found that of the 483 Japanese ESL students who read
passages that were either in unmodified or linguistically simplified form, those who read
48
the simplified passages scored significantly higher on a 30-item multiple-choice
comprehension test than those who read the unmodified versions.
Tweissi (1998) investigated whether variations in type and amount of linguistic
simplification resulted in differences in the level of message comprehensibility. In this
study, 200 Omani ESL learners were asked to read five different versions of a text.
Tweissi had readers read an unmodified passage (version 1); in addition, he modified the
baseline passage four ways. Modified passages included a highly simplified version
(version 2) where linguistic items were replaced by “simpler alternatives” (Tweissi,
1998, p. 195). In version 3, half of the linguistically challenging items from version 1
were simplified and in version 4, only syntactic simplifications were made. Finally,
version 5 received only lexical simplifications. Tweissi’s results revealed that
simplification in general had a positive effect on reading comprehension although it was
the type of simplification, rather than the amount, that was most important to readers.
In Tweissi’s study, comprehensibility measures indicated statistically significant
differences between the mean scores of the subjects who read the authentic text and all of
the other four versions. Tweissi stated that “this study has shown that authentic texts are
not always the best to give to students who are not ready for them, that simplified texts
may facilitate L2 learners’ reading comprehension, and that the use of lexically
simplified texts is more justifiable than the use of other types of text” (1998, p. 201). He
concluded by asserting that the major premise that motivated his study, that the simpler
the text, the more comprehensible it is to L2 learners, is unwarranted and that “when the
amount of simplification is brought into play, it is more likely the less the better as it is
the type rather than the amount of simplification that may have a higher impact on
reading comprehension” (1998, p. 201).
49
Simplification: Negative Research Findings
Despite simplification’s seemingly facilitative effect on reading comprehension,
few researchers other than Tweissi (1998) advocate its use. Researchers such as Yano et
al. (1994) contend that linguistically simplified input may negatively affect language
acquisition. They argue that “linguistically simplified texts can be self-defeating to the
extent that the purpose of a reading lesson is not the comprehension of a particular text
but the learning of a language in which text is written and the development of
transferable, non-text-specific reading skills” (p. 191). Additionally, Ross et al. (1991)
argue that readers of simplified texts are denied access to authentic models of the L2 with
vocabulary and lexical items that they eventually need to learn.
Other researchers contend that modifications that simplify text not only affect
linguistic elements but frequently the content of the text as well (Brown, 1987; Swaffar,
1985). Swaffar (1985) argues that simplified texts alter the authorial cues because they
are “culturally and linguistically sanitized” (p. 17). This sanitizing may strip away much
of the text’s detail and richness, thus inhibiting readers from keeping the text’s message
in mind as they read. Bernhardt (1984) states that while “simplified texts and materials
may indeed be more teachable than holistic texts, they do not seem to focus on the actual
comprehension process” (p. 329). Honeywell (1977) agrees, asserting that simplified
reading passages may induce readers to develop reading strategies that are inappropriate
for unmodified reading material. A result of readers’ inappropriate use of reading
strategies may result in another phenomenon seen with readers of simplified texts; that is,
an inability to comprehend text meanings that extend beyond the literal text message.
An issue that continually arises when simplified L2 texts are compared to
unmodified versions of authentic texts is that readers of simplified texts are denied
participation in much of the experience of reading a literary text. Researchers suggest that
readers of simplified texts do not develop an appreciation of the text that goes beyond a
basic comprehension of the who, what, when, and where of the text message. In other
50
words, readers of simplified texts are equipped to provide or recall discrete, explicit data
derived from the text but are unable to tie elements together or infer meaning beyond the
text itself. Yano et al. (1994) conclude that simplified texts provide so much less context
that the reader is unable to form a deeper pragmatic understanding of the text, a condition
that the authors deemed necessary for inferring the implications and consequences of the
text. They state that “simplified texts do not enhance inferential comprehension because
they strip away the richness in detail and connections that help a reader to perceive
implicational links” (p. 214). Despite research studies that indicate that readers of
linguistically simplified texts are frequently able to outscore readers of unmodified texts,
the impoverished reading environment resulting from simplification prevents those same
readers from demonstrating an understanding of the text that extends beyond the words
written on the page (Blau, 1982; Brown, 1987; Oh, 2001; Parker & Chaudron, 1987;
Yano et al., 1994).
Conversational Modification: Genesis/Roots
Limitations to the traditional approach to text modification through simplification
are becoming more widely recognized, but this awareness has of yet had little impact on
L2 reading or L2 reading texts. Negative aspects of linguistic modifications have led
researchers to look for alternative viable means to modify texts, as most L2 educators feel
the need to provide some sort of text-based support to limited-proficiency readers. One
area of research that is beginning to shed some light on a different approach to text
modification also has its genesis in studies examining foreigner talk. Rather than focus
exclusively on how NSs simplify speech when addressing NNSs, conversational or
interactional language modifications concentrate attention on other types of language
adjustments used to make speech more comprehensible (Long, 1983a, 1983b). For the
remainder of this discussion, the terms conversational modification and interactional
modification are used interchangeably in reference to various language strategies used in
51
oral exchanges by NSs with NNSs with the intention of making language more
comprehensible. Here, conversational or interactional language modifications refer to
those adjustments that affect the pragmatic structure of FT discourse. Although such
modifications may include simplification of linguistic content, adjustments other than
those of a linguistic nature are also included in this definition. Additionally, even though
these modifications were identified orally, similar types of modifications have been found
to be applicable for use with reading texts.
Ross et al. (1991) documented a number of these conversational language
adjustments (for a review, see Long, 1983a; Larsen-Freeman & Long, 1991). Some of the
speech modifications made by NSs in an attempt to increase NNS comprehension include
more abrupt topic shifts, more use of questions, more repetitions of various kinds, and a
higher frequency of comprehension checks. Additionally, NSs addressing NNSs use a
slower rate of delivery, more careful articulation, stress on key words with pauses before
and after them, and fewer contractions. Morphological and syntactic changes include use
of fewer words and clauses per utterance, preference for canonical word order, retention
of usually deleted optional constituents, and overt marking of grammatical relationships
(Ross et al., 1991). Other changes include fewer idiomatic expressions and the use of full
noun phrases over pronouns.
Although it is possible to argue that some of these changes, such as fewer words
and clauses per utterance, are in fact forms of linguistic simplification, what makes
conversational modifications different from simplification is that multiple techniques are
employed simultaneously rather than restrict the focus to linguistic forms. Indeed studies
derived form FT find that it is the use of multiple adjustments that seems to facilitate
comprehension and, with the exception of rate of delivery, single adjustments are
insufficient to increase comprehension (Ross et al., 1991; Yano et al., 1994).
Various proposals have been advanced to explain why interactional adjustments
are superior to linguistic modifications for L2 comprehension, whether aural or written.
52
One proposal suggests that interactional modifications take a broader approach to L2
comprehension by viewing language processing as more than an exercise in decoding
linguistic elements. In an oral NS-NNS exchange, language is frequently simplified but
the causes of communication breakdown are not necessarily assumed to be the result of
complex structures or unfamiliar vocabulary. NSs employ a wide variety of techniques to
get their message across. What results from their modifications is more speech and
language that employs a wider, rather than a restrictive, array of linguistics and pragmatic
devices. Long (1980, 1983b) contends that in NS-NNS speech, conversational
adjustments are more frequent and pervasive than linguistic adjustments and sometimes
occur when the latter do not. The question is, then, to what extent what is known about
NS-NNS conversational adjustments can or should be applied to text modification for L2
readers.
Conversational Modification: Elaboration
As previously stated, although there is no strict dichotomy between what is
described as linguistic simplification and conversational elaboration, there are differences
between the two approaches, specifically in regard to the conceptualization of what it
means to process and comprehend input. Whereas the aim of linguistic simplification is
to get the message communicated to the listener or reader in the most straightforward and
uncomplicated way possible, elaborative modifications provide a variety of discourse
techniques that potentially appeal to a number of language learning processes. Linguistic
modification produces a “structurally simplified form” of language, whereas the intent of
elaborative modifications is to produce “cognitively simplified forms” (Parker &
Chaudron, 1987, p. 108). Elaborative modifications are thought to be less cognitively
demanding because they mimic input strategies that speakers employ when attempting to
transmit a message. In other words, the intent of elaborated speech or an elaborated text
53
is to appeal to as many cognitive language processes as possible, a technique that NSs
employ instinctively when attempting to make themselves understood to NNSs.
Elaboration: Procedure
A number of researchers have taken what is known about NS-NNS speech
patterns and applied that knowledge to written texts through text elaboration (Brown,
1985, 1987; Oh, 2002; Parker & Chaudron, 1987; Ross et al., 1991; Yano et el., 1994).
The ultimate goal of text elaboration is to ease the cognitive processing burden so that
more of the text meaning is decoded and thus understood. Comprehension is increased
because fewer attentional or controlled processes are needed to decipher the message
during the reading process. The emphasis is on the construction of a conceptual
representation of the text rather than the decoding of linguistic text elements. Ideally, the
L2 reader is supported to the extent that he or she is able to keep the text message in mind
during the reading process as textual elements support cognitive processing.
Theoretically, the reader expends less energy decoding the text so that more cognitive
resources are available to attend to the global text message. The result is a more fluid
mental discourse without the choppiness or unnaturalness that sometimes results once a
text has been simplified.
The fundamental goal of text elaboration is to create a text that assists a reader to
connect pieces of information in order to develop a coherent mental text representation.
This is done by a variety of elaboration techniques. For the purpose of this discussion, the
features of elaboration for L2 reading are categorized in the following, though not
mutually exclusive, groupings. They are (a) message maintenance, (b) message
clarification, (c) text organization with signaling, and (d) repetition and redundancy.
Elaboration Feature: Message Maintenance
Message maintenance refers to the preservation of the original complexity, both
syntactic and lexical, found in the primary text source prior to modification (Ross et al.,
54
1991; Yano et al., 1994). Studies in foreigner talk (FT) have found that NSs frequently do
not simplify syntax or lexis when addressing NNSs, unless they are speaking to NNSs
with considerably low L2 proficiency (Long, 1980). Indeed, Long’s research found that
some NSs used more complex speech with NNSs than they did when addressing
members of the NS control group. The complexity of language found in FT is preserved
through elaborative text modifications. Additionally, L2 reading research empirically
supports maintenance of unmodified reading texts, suggesting that syntactic and lexical
simplification is unnecessary for greater text comprehension (Alderson, 2000; Blau,
1982; Ulijn & Strothers, 1990).
Elaboration Feature: Message Clarification
Message clarification is another feature of FT that has been adapted as an
elaborative modification strategy. In oral exchanges, message clarification is frequently
expressed through body language or comprehension questions (Long, 1983a). With
written text, however, as it is difficult to determine when comprehension breakdown
occurs, texts are clarified assuming that the reader will benefit from additional message
elucidation. Message clarification techniques include greater topic saliency and
explicitness and the use of topic-comment, rather than subject-predicate constructions
(Ross et al., 1991). In addition, phrases are restated or paraphrased and there is greater
syntactic signaling through the use of such techniques as relative or complementary
clause markings (Parker & Chaudron, 1987). Subject fronting and the addition of
clarifying phrases are used as well (Yano et al., 1994).
Elaboration Feature: Text Organization with Signaling
Other techniques employed when texts are elaborated to focus readers’ attention
on text organization through clear signaling, another feature found in FT (Long, 1983a).
The intent of these types of textual revisions is to explain or provide motivation for
actions or reactions and to explicitly link events and events to consequences (Beck et al.,
55
1991). Much of this is done through the use of conjunctions and temporal or causal
markers or through retention of full noun phrases rather than pronouns (Alderson, 2000;
Ross et al., 1991; Yano et al., 1994). Both the modifications that serve to clarify text
message and those that increase text organization have been empirically demonstrated to
support reading comprehension in both the L1 and L2 (Alderson, 2000; Beck et al., 1991,
1995; Britton & Gulgoz, 1991; Horiba, 1993; Kobayaski, 2002; McNamara et al., 1996;
Urquhart, 1994).
Elaboration Feature: Repetition and Redundancy
Another clarification technique found in NS-NNS oral exchanges and adapted
within text elaboration is the use of redundancy and repetition. Included within this
category of techniques are paraphrase, synonym use, definitions of low-frequency
vocabulary and other practices used to support readers’ lack of familiarity with lexical
items (Oh, 2001; Parker & Chaudron, 1987; Ross et al., 1991). Parker and Chaudron
(1987) contend that modifications that support lexical understanding are the most critical
for limited-proficiency readers. Owing to the essential role that the lexis plays in both L1
and L2 reading comprehension and processing, techniques that assist readers in
understandiing words and phrases are of fundamental importance.
In summary, although the goal of conversational modifications through
elaboration and of linguistic modifications through simplification is essentially the same;
that is, to make L2 reading texts easier to process and comprehend, there are differences
between the two approaches. Viewing language comprehension and learning from an
approach that is more process than product oriented, elaborative modifications are
employed in an attempt to influence the number and kind of language processes and
strategies used by the reader. Elaborative modifications techniques can be classified as
message maintenance, message clarification, text organization with signaling, and
repetition and redundancy. Although limited in number, four research studies are
56
presented below that demonstrate the potential benefits of text elaborations for L2
readers.
Elaboration: Research Findings
A limited number of second language researchers have compared the use of
elaborated L2 reading texts with the same texts in either unmodified or simplified form.
What follows is a description of these studies. They are presented chronologically. Each
subsequent study builds on findings obtained from earlier text elaboration studies
expanding the definitional construct and applicability of this form of text modification.
All of the studies support the position that both the product and process of reading are
different when texts are elaborated rather than simplified. In addition, the studies serve to
support the view that elaborated texts are superior to both linguistically simplified and
unmodified reading texts for language learners with low L2 proficiency.
Elaboration Research Findings:
Parker and Chaudron (1987)
Chronologically, Brown (1985) was one of the first researchers to compare an
elaborated style of text modification with unmodified reading texts in an ESL setting.
Brown (1985) found, as did Tsang (1987) who replicated Brown’s study on a
homogeneous Chinese L1 population, that texts with increased redundancy through such
strategies as paraphrases and synonyms were as successful in promoting comprehension
as syntactically simplified texts, specially for lower-level learners (see Parker &
Chaudron, 1987). Parker and Chaudron (1987) coined the term elaboration in reference
to the types of aural and written text modifications that include not only redundancy and
repetition but also clearer signaling of thematic structure (p. 108). Their research
incorporated Long’s findings regarding oral speech modifications (1983a, 1983b, 1980),
testing whether alterations in thematic structure and redundancy would lead to greater
reading comprehension. Parker and Chaudron studied 43 college-level ESL students
57
whose L1 was Chinese, Korean, or Japanese. Two versions of a reading passage were
developed. Passage A “retained all the redundancy and thematic structure that occurs
naturally” and passage B “had all redundancies eliminated and all the thematic structures
reduced to canonical word order” (p. 118). Text comprehension was measured by means
of cloze tests.
Although, statistically, no difference in cloze scores was found between readers of
the elaborated passage and those of the non-elaborated passage, the researchers
contended that the higher reliability and greater correlation of the cloze test scores with
other measures rendered the elaborated text the preferred passage for use in a ESL
classroom. According to the researchers, their study “suggests that elaborated input,
which does not hinder comprehension and appears to be more natural, may be a better
choice than non-elaborated input in language classrooms” (Parker & Chaudron, 1987, p.
122). The researchers argued that if text developers are inclined to present the most
native-like L2 input, they should modify the input in the direction of elaborative rather
than syntactic simplifications because such texts allow for more native-like complexity
and are at least as successful in promoting comprehension.
Elaboration Research Findings:
Brown (1987)
In 1987 Brown expanded on his earlier research (1985) by comparing an
unmodified text with simplified and elaborated versions of the original. His subjects were
30 EFL students in a secondary school setting in Taiwan. Reading comprehension was
checked by means of a 20-item multiple-choice test “stressing recognition rather than
inference” (p. 56). Results revealed that students reading both modified versions scored
significantly better than those who read the unmodified text. There was no significant
difference between the scores for the two modified versions. Interestingly, both modified
versions were at least 33% longer than the original version, with the elaborated by far the
longest. Despite students being given a 50-minute time limit, the greater text length did
58
not present a problem. More importantly, although no statistical difference was noted in
comprehension scores between the students who read the simplified version versus the
elaborated version, a readability estimate determined that the elaborated text was four
grade levels above the simplified text. That same year Brown replicated this study with a
different sample of 30 students and reached similar conclusions (Brown, 1987, pp. 5859).
Brown’s research indicated that modifications used to simplify vocabulary and
decrease sentence length are less important for readers’ comprehension than expansive
interactive modifications such as redundancy, expansion, and clarification. He asserted
that “comprehension is less dependent on linguistic structure than with the level of
information made available to the reader and the frequency with which the reader
encounters the information” (1987, p. 53). Thus, the text can remains at a high level of
linguistic competence but by exploiting interactional text strategies, such as redundancy
or definitions, the reader is offered more opportunity to process information and thus
more chances for successful comprehension.
Elaboration Research Findings:
Ross et al. (1991)
Whereas Brown (1985, 1987) and Parker and Chaudron (1987) connected the
rationale for elaborative modifications to research on aural comprehension, Ross et al.
(1991) use a model of cognitive language learning to explain why elaborated texts are
more comprehensible to readers. Additionally, their research is the first to test readers’
comprehension beyond the level of text-based informational recall. In this study, 13
versions of reading passages were presented to 483 EFL Japanese college students in one
of three forms: native baseline, simplified, and elaborated. After modification, elaborated
texts were an average of six grade levels higher in readability, twice as complex, and
50% longer than the simplified versions. As with Brown (1987) passage length was
59
especially important as, due to time constraints, students who read the elaborated texts
had to read more difficult material more rapidly.
Also as with Brown (1987), 30 multiple-choice items were used to check
comprehension but unlike Brown, the items not only required students to replicate or
reproduce information found in the text but also required students to connect and
synthesize information as well make deductions and inferences about the implications of
the text. Results indicated that the students who read both modified versions of the
baseline passage scored higher than those who read the unmodified version but only the
scores of those reading the simplified version were statistically superior. However, when
looking specifically at test items that required students to synthesize or infer information
from the text, the readers of the elaborated texts scored higher than the readers of both the
simplified and the baseline versions.
Interestingly, Ross et al. (1991) observed that elaborated texts both “increase the
general processing burden” and were “cognitively simpler” than original versions (pp.
24-25). Although it appeared that the readers of the elaborated versions were faced with a
greater processing burden due to text length and complexity, the elaborated texts were in
fact cognitively simpler, due to the processing support provided to readers. For this
reason, readers were able to synthesize and infer information from the elaborated texts.
Ross et al. (1991) described the process as follows:
Elaboration provides the reader with a “second look” at key
terms and concepts consequently increasing the chance that
inference about them can be stimulated in the reading process. In
contrast, unmodified and simplified texts provide less context for
stimulating the deeper pragmatic linkage necessary for inferring
the consequences of passage meanings. Unmodified texts probably
fail because concepts are obscured by the structural and lexical
detail. Simplified texts probably fail because they strip away the
richness of detain helpful for a reader to perceive a texts’
implications (pp. 24-25).
As argued by Ross et al., elaboration seems to serve two purposes: improve
comprehension and provide learners with the raw data they need for language
60
development in the form of access to unknown linguistic items. These researchers
contend that if the purpose of L2 reading material is to provide opportunities to process
texts at a deeper level, elaboration should be considered as it jointly triggers the processes
of understanding language from the context and content from the language itself.
Elaboration Research Findings:
Oh (2001)
Oh (2001) investigated the effects of text simplification and elaboration on 180
Korean secondary school students’ EFL reading comprehension. Six English reading
passages were presented to students in one of three forms: (a) baseline, (b) simplified, or
(c) elaborated. Reading comprehension was assessed by an 18-item multiple-choice test
that assessed general, specific, and inferential comprehension. Students were also divided
into high and low proficiency levels.
Oh’s findings revealed that regardless of proficiency level, the students who read
the elaborated passages performed significantly better than those who read the baseline
passages. As with Brown (1987) and Ross et al. (1991), this finding was especially
interesting because readers of elaborated texts had to read more text in a restricted period
of time. Overall comprehension of the passages significantly improved among students at
both proficiency levels as a result of elaboration, but those with lower proficiency seemed
to benefit the most. Although high-proficiency students’ total comprehension scores were
the highest when reading the simplified texts, low-proficiency students scored better on
all measures when reading texts with elaborative modifications. In addition, inferential
comprehension was superior for readers at both proficiency levels when reading the
elaborated texts. Oh found that both high- and low-proficiency students perceived that
they understood more information when reading the elaborated texts in comparison to the
baseline texts.
61
Present Study: Rationale for Exclusion of Simplified Text
The present study compared student comprehension of unmodified literary L2
texts with how they process and comprehend the same texts in an elaborated form.
Simplified texts were not considered, as the integrity of literary texts is difficult to
maintain in a simplified form. The text-based characteristics that render literary texts as a
valuable form of L2 reading material are lost with simplification. Additionally, studies
indicate that simplified texts do not promote or engage the same type and amount of
language learning strategies as elaborated texts and therefore do a poorer job in preparing
students for processing of subsequent language input. As simplified texts are not genuine
representations of authentic language discourse, they do not represent the type of reading
passages that learner are likely to encounter outside of the L2 classroom. Although
readers of simplified texts tend to comprehend more of the material they read than
readers of either unmodified or elaborated versions, these findings are deceptive due to
the nature of the comprehension measures used.
Incidental Vocabulary Acquisition Through Reading
Before describing how comprehension of unmodified literary texts was compared
to learners’ comprehension of elaborated versions of the same texts, characteristics of
incidental vocabulary acquisition are described. There are several reasons for this
discussion, first and most important because incidental vocabulary acquisition was
assessed in this study. Second, as previously discussed in reference to the effects of textbased factors on reading, of all of these factors, L2 vocabulary knowledge has been found
to be the most influential for text processing and comprehension (Alderson, 2000;
Bossers, 1992). But even though the influence of vocabulary knowledge on text
comprehension is well documented, the effect of reading on the acquisition of previously
unknown words is controversial. Many researchers believe that reading for
comprehension is not the primary means of increasing L2 lexicon (Alderson, 2000;
62
Carmine et al., 1984; Deighton, 1959; Haynes, 1993). Others, however, argue that a
substantial proportion of the vocabulary acquired by L2 learners after the first few
thousand most common words is accomplished through reading (Huckin & Coady, 1999)
and that reading must be considered as a potential source of lexical exposure and
development (Hulstijn et al., 1996; Paribakht & Wesche, 1996, 1999; Rott, 1999). For
these reasons, incidental vocabulary acquisition through reading is discussed and
measured. The primary focus of this discussion involves four elements; (a) the
operationalization of the construct; that is, what incidental vocabulary acquisition means;
(b) a description of the text-based factors that influence lexical learning; (c) an
explanation of how L2 text modifications may support vocabulary acquisition and,
finally, (d) how incidental vocabulary acquisition is measured.
Incidental Vocabulary Acquisition: Definition
The hypothesis that both L1 and L2 readers learn at least some vocabulary as a
result of reading is not new, nor is the assertion that some of this learning is done
incidentally. This vocabulary learning is said to be incidental because it occurs as a byproduct of other cognitive exercises, generally those involving comprehension (Gass,
1999). Although Gass argues that there is no surefire way to know whether a lexical item
has in fact been acquired incidentally (1999, p. 337), studies have shown that students
can, at times, correctly guess the meanings of words while reading, words that they
previously identified as unknown to them (Hulstijn, 1993; Paribakht & Wesche, 1996,
1997). This process is considered especially important for the acquisition of less
frequently encountered lexical items; that is, words that are likely a part of a native
speaker’s written rather than oral lexicon (Coady, 1993; Ellis, 1994).
But what exactly constitutes the acquisition of a lexical item? Some would argue
that a word is not truly acquired, whether intentionally or incidentally, until the learner
demonstrates its comprehension not only immediately after encountering it; that is,
63
through immediate recall, but also after a period of time (Mondria & Wit-De Boer, 1991).
Although a discussion of what it means to acquire versus merely recall a word and its
meaning goes beyond the scope of this study, suffice it to say that the incidental
acquisition continuum could potentially span from first exposure to delayed, repetitive,
productive use. For the purpose of this study incidental vocabulary learning is used
interchangeably with intentional vocabulary acquisition and will be defined by a
learner’s ability to correctly produce an acceptable equivalent of the target word or
phrase in his or her L1. What is acceptable is defined further when assessment procedures
are discussed, as are other issues germane to this study such as the categorization of the
specific lexical items to which study participants are exposed.
As pointed out by Huckin and Coady (1999), incidental vocabulary learning
cannot be entirely incidental as the reader must pay at least some attention to the words to
be able later to identify them. Much of the vocabulary gain demonstrated through reading
is believed to be a result of guessing on the part of the L2 reader based on his or her
reliance on contextual clues. Contextual support, however, is not the only factor that
promotes incidental vocabulary acquisition. Lexical inferencing also “involves making
informed guesses as to the meaning of a word in light of all available linguistic cues in
combination with the learner’s general knowledge of the world, her awareness of context
and her relevant linguistic knowledge” (Haastrup, 1991, p. 40). As this study deals
primarily with text features that influence language acquisition, learner-centered
variables, although acknowledged, are not addressed.
Incidental Vocabulary Acquisition: Text-Based Factors
That Influence Lexical Learning
What are the text-based factors that influence L2 vocabulary learning while
reading? Aside from reader factors, linguistic elements are variables associated with the
unknown word in question or those associated with the surrounding text. Classifications
of word factors include elements such as (a) part of speech (i.e., noun, verb, etc.), (b)
64
degree of concreteness of abstractness, (c) transparency of word structure, and (d) degree
of correspondence between meaning of the word and that of the same word in the L1
(Mondria & Wit-De Boer, 1991, pp. 252-253). In addition to the degree of
correspondence between L1 and L2 word meanings, another variable is whether the same
word concept exists in the reader’s L1. Additionally, how relevant the reader considers
the comprehension of any given word to the understanding of the text to be is likely to
determine how much attention he or she invests in guessing its meaning (Hulstijn et al.,
1996).
Factors associated with the surrounding text include, most basically, how many
unfamiliar words are present and how well the reader is able to comprehend the text
message. Coady (1997) argues that a good reader can guess the meanings of some
unfamiliar words in a text but to do so, must also know most of the other words. Huckin
and Coady (1999) agree, contending that recognition on sight of most of the surrounding
words in a text has long been considered a prerequisite to correct guessing of word
meaning in context.
Another factor that influences the likelihood that a reader will acquire new
vocabulary while reading relates to how well the reader is able to understand the text as a
whole. As with other forms of lexical support, the overall difficulty of the text must be
balanced; that is, it cannot be too easy or too difficult. The perception of the text’s overall
difficulty has been shown to “strongly influence” whether a reader attempts to infer the
meanings of unknown words (Paribakht & Wesche, 1999, p. 210). If the text is too easy,
with only a few unknown lexical items, comprehension of the text message is frequently
possible even if the reader chooses to ignore unknown words. On the other hand, if the
text’s difficulty results in high levels of frustration, readers may give up without even
attempting to decode the main ideas, much less individual lexical items. Readability
issues come into play here. A rather obvious way to increase text-based support for
incidental vocabulary acquisition is to have learners read texts that are written at a
65
manageable level of difficultly. Attempts have been made to develop indexes and
formulas that estimate text readability (e.g., Klare, 1984) but, as previously stated in the
discussion involving text elaboration versus simplification, these indexes do not always
reflect the actual experience of the readers (Brown, 1997; Ross et al., 1991) and, for that
reason, are not considered here.
Two other text-based features that have been found to support L2 incidental
vocabulary learning include the frequency in which an unknown word appears and how
much the word is supported by redundant textual clues such as those provided by
paraphrase or the use of a synonym or antonym. Frequency of occurrence fosters
incidental vocabulary acquisition because the reappearance of a word reinforces the
form-meaning connection associated with that word (Hulstijn et al., 1996). Moreover, as
words reappear in different contexts, or through repetition and paraphrase, the formmeaning connection is not only reinforced but also strengthened and expanded, both
semantically and syntactically.
The above-mentioned text-based features are thought to support lexical
acquisition because they influence the type of cognitive processing in which the reader
engages. These influences occur because the manipulation of text features results in
varying levels of reader attentiveness to the unknown words or phrases. But what degree
of attention is necessary for vocabulary learning to take place and, more important, how
is the reader’s attention best directed to the words that he or she is the most likely to
acquire incidentally? Working under the assumption that as a limited-capacity processor
the learner has finite cognitive resources and those resources are already engaged in the
task of passage comprehension, how might the conditions for incidental vocabulary
acquisition be optimized without compromising the reader’s ability to understand the
text?
66
Incidental Vocabulary Acquisition: Text Modification
In the previous section factors that optimize incidental vocabulary acquisition
when reading for meaning were described. They include features inherent to the words
themselves and factors associated with the context in which unknown words appear. The
ultimate goal when selecting or modifying texts to support incidental vocabulary
acquisition, therefore, is to provide readers with a comprehensible text with ample
contextual clues to facilitate word recognition. As previously discussed, this process is
best described as a balancing act. A text that provides too little contextual support may
lead readers to ignore unknown words or incorrectly infer their meanings (Hulstijn et al.,
1996) or, as pointed out by Nation and Coady, “the very redundancy or richness of
information in a given context which enables a reader to guess an unknown word
successfully, could also predict that that same reader is less likely to learn the word
because he or she was able to comprehend the text without knowing the word” (1988, p.
101). The optimal text for vocabulary learning, therefore, would be one that challenges
rather than overwhelms the reader and provides a sufficient yet not excessive number of
linguistic clues. Under these conditions, the reader becomes aware of the importance of
certain words and therefore attends to them, but without expending undue energy that
might distracts from the main reader objective, that of message comprehension.
Elaborative modifications, as previously described, seem to be a viable text modification
option for incidental vocabulary acquisition. This is because the two main objectives of
this type of text modification, those of increased linguistic support and enhanced
comprehensibility, not only serve to clarify the primary text message but also provide
additional contextual clues to make lexical items more salient.
Incidental Vocabulary Acquisition: Marginal Glosses
Before turning to the question of how incidental vocabulary acquisition is
assessed, one factor that has been shown to increase not only text comprehension but
67
also, more important, vocabulary acquisition, is discussed; that is, marginal glosses.
Studies indicate that marginal glosses that provide word definitions in either the reader’s
L1 or L2 enhance incidental vocabulary learning (Hulstijn, 1992; Jacobs et al., 1994;
Watanabe, 1992). When word meanings are provided, L1 definitions appear to have more
effect than L2 definitions (Krantz, 1991; Scherfer, 1993). But the effect such definitions
have on incidental vocabulary acquisition is not clear. Reading research in this area has
shown that in relation to vocabulary learning, meaning inferred yields higher retention
rates than meaning given (Hulstijn, 1992; Mondria & Wit-de Boer, 1991; Watanabe,
1992). This finding harkens back to the notion that minimal cognitive effort is
disadvantageous for learners. Moon (1997) also points out that glossed lexical items tend
to communicate an uncomplicated one-to-one meaning correspondence with the
definitions provided in the margins. According to Moon, words in context cannot be
reduced to their glossed definitions. This is because as soon as words are part of a text,
they form “meaningful” and “inseparable” units with each other (p. 43), units difficult to
appreciate if these they are reduced to their L1 equivalents. Perhaps a preferable
alternative to marginal glosses, therefore, would be textual elaboration where all lexical
items are left to be understood and, perhaps, learned within the confines of a naturally
occurring syntactic and semantic environment.
Measuring Incidental Vocabulary Learning
As previously mentioned, it is virtually impossible to claim that a vocabulary item
has been wholly acquired through incidental exposure. Moreover, other than through the
use of nonsense words, (i.e., Pulido, 2003; Urano, 2000) it is equally difficult to prove
that the meaning of a word has been wholly acquired through any one reading
experience. To do so, one must account for every previous exposure to the word, whether
in or out of the language learning setting. Additionally, as vocabulary acquisition is a
system in flux, any measure of vocabulary learning is just that: a single measure of an
68
instance of learner performance and not necessarily a reflection of language competence.
But despite these limitations, attempts are made to measure whether L2 learners can and
do acquire new vocabulary incidentally.
In studies in which incidental vocabulary acquisition is measured, a number of
factors should be taken into account. As previously mentioned, a good measurement
procedure first insures that the words are indeed unknown to the learners. This is
generally done thorough the use of nonsense words; that is, neologisms created following
language-specific linguistic parameters and constraints (Urano, 2000). In addition, as the
primary objective should be text comprehension and not vocabulary acquisition, words
targeted for later assessment need to be perceived by the readers as relevant, intriguing,
or even annoying, thus warranting further attention (Hulstijn et al., 1996). If the meaning
of the unknown word seems to the reader to be unnecessary for general understanding,
that word will more likely be ignored so that greater attention can be paid to words
perceived to be more important for overall text comprehension (Gass, 1999).
Additionally, and as previously mentioned, the overall difficulty of the text should
be taken into consideration, as should the number and type of encounters that the reader
will have with the selected lexical items. Research in L1 reading has shown that the
probability of gaining receptive word knowledge of an unfamiliar word during reading
ranges from 0.08 to 0.22 when that knowledge is measured directly after exposure (Nagy
et al., 1985; Shu et al., 1995). Other researchers contend, however, that as few as two
encounters with unfamiliar words during reading can significantly affect a reader’s
vocabulary growth (Rott, 1999). Much of this variation in research findings may well
stem from how vocabulary knowledge and growth are conceptualized and, more
important, how they are operationalized in studies measuring vocabulary acquisition.
To a great extent the discrepancy in research results appears to relate to when, in
reference to reading time, students are assessed on their vocabulary knowledge, as well as
by the form that the assessment takes. In other words, whether an assessment instrument
69
measures a reader’s receptive or productive vocabulary knowledge will have an effect on
how well the learner performs, as will whether the reader is tested immediately after
reading or after an extended period of time. Although the full relationship between
receptive and productive vocabulary knowledge is unclear, it is generally believed that
receptive knowledge is the initial step toward productive knowledge on a word
knowledge continuum (Melka, 1997; Rott, 1999). Also, when vocabulary testing is
conducted immediately after reading, scores are significantly higher than at any other
time (Rott, 1999). In summary, in order to appropriately assess incidental vocabulary
acquisition while reading, one should take into account the following variables: (a)
previous exposure to unfamiliar words, (b) the difficulty of the level of the text in which
the unknown words appear (c) the perceived relevance of words, (d) the number and
extent of encounters with the words while reading, and (e) when and how vocabulary
knowledge is assessed.
Assessing Reading Comprehension
In the remainder of this chapter the methods used to assess how readers interact
with and comprehend texts, in a more general sense, are discussed. This section is
organized as follows: First, general assumptions related to the assessment of reading are
presented. For clarification purposes, two sets of terms are defined: (a) the difference
between text-explicit and text-implicit assessment measures and (b) what is meant by text
remembering versus text learning. Next the most common assessment methods used in
L2 reading comprehension are discussed. The methods featured here include (a) multiplechoice questions/selected responses, (b) recall/summary protocols, and (c) shortanswer/short constructed response questions. Pros and cons of each of these methods are
included. Finally, various implications associated with the implementation of L2 reading
assessment measures are discussed, including procedures for measuring acquisition of
incidental vocabulary.
70
Assessing Reading Comprehension: General Assumptions
Researchers generally agree that there is no perfect measure of reading
comprehension. The results of every assessment task provide only a partial picture of
what the reader is experiencing or knows. In addition, it can be argued that the
assessment tasks that are selected reflect the test developer’s view of reading as much, if
not more, than does the reader’s experience with the text. That said, there are certain
practices that can be followed to provide a more complete and reliable representation of
the process and product of reading.
First, as every assessment measure provides only a partial perspective, multiple
means of assessment are preferable to single reading measurements (Alderson, 2000;
Bernhardt, 1991b; Wolf, 1993). Multiple measures help assure that the testing
instruments are adequately measuring the construct under evaluation. A successful
battery of assessment measures is, therefore, integrative in nature, examining not only the
extent to which a text actually communicates a message, but also the reader’s experience
with the text. In other words, assessment of reading comprehension should target both the
process and product of reading. This entails using assessment instruments that provide
insight into how readers cope with texts while at the same time provide quantifiable data
for comparison and contrast (Bernhardt, 1991b).
Wolf (1993) warns, however, that different assessment tasks frequently yield
different results, and that these results that are not necessarily comparable. For this
reason, care must be taken not only with the development of assessment measures but
also with their interpretation. To assist test developers in creating valid objective
measures of reading comprehension, Wolf (1993) suggests that developers keep in mind
several criteria. First, tasks should be passage dependent and items should test
information from different levels of the passage, such as main ideas and implicit
information. This means that test takers should not be able to answer test questions
without having first read the text. Also, tasks that elicit text-explicit as well as text-
71
implicit information should be included as they represent a broad picture of text
comprehension. Additionally, a test taker should not be able to lift answers from the text
without reading the passage. This means that salient information such as dates and times
that are easily identifiable should be avoided as test items. Finally, items should assess
what students have understood and not what they already knew about the topic. In other
words, the background and literary knowledge that a reader possesses should be taken
into account when scoring test results. Although difficult to achieve in every instance,
such recommendations provide a guide to assist test designers to create valid and reliable
assessments of reading comprehension.
Definition of Terms
Certain terms associated with the assessment of reading are described below.
These terms are presented here due to their importance to latter discussions and relate to
the measurement of learner reading process and product. The terms are (a) text-explicit
versus text-implicit measures, and (b) memory of a text versus learning from a text.
Text-Explicit versus Text-Implicit Assessment Measures
An important distinction to recognize when preparing instruments that measure
reading comprehension is whether the task is designed to measure text-explicit or textimplicit information. Text-explicit measures are those that target information that is
directly found in the text. Alderson (2000) contends that questions intended to measure
textually explicit information are those where both the information that formed the
question and the correct answer are found in the same sentence, and include who, what,
when and where question types.
Text-implicit measures, on the other hand, require readers either to combine
information across sentences or paragraphs or to reason beyond the text by synthesizing
information. In order to infer, readers must develop an understanding that, although
motivated by information in the text, is at least partially based on prior experience or
72
knowledge. According to Hammadou Sullivan (2001), to infer readers must use
generalizations of typical events and apply what they know about such events to the text
at hand. In addition to generalizing from typical events, readers also use their knowledge
of literary styles and genres when inferring, a knowledge that cannot be derived without
repeated exposure to different types of texts.
Memory For a Text versus Learning From a Text
There is an important reason to emphasis the distinction between text-explicit and
text-implicit reading tasks, particularly in relation to what it means to comprehend a text.
It has been argued that assessment instruments that exclusively measure text-explicit
knowledge are, in actuality, tests of memory and that inferential measures are needed in
other to assess a deeper level of text comprehension (Kintsch, 1994). Kintsch calls the
difference between these two levels of comprehension memory for a text versus learning
from a text.
According to Kintsch, memory for a text means that a reader can reproduce part
of the text in some form, more or less verbatim and more or less completely. Learning
from a text, on the other hand, implies that a reader is able to use the information
provided by the text in ways other than replication. Kintsch states that “the distinction
made here between learning and memory is a matter of the criteria used to define
learning: Learning requires deep understanding of the subject matter, so that the
information acquired can be used productively in novel environments: for mere memory,
as assessed by reproduction of the text, a more shallow understanding suffices” (1994, p.
294). In other words, there are many ways in which a text can be comprehended and, for
that reason, different assessment tasks are necessary to elicit distinct representations of
reader knowledge. To infer meaning from a text requires not only the comprehension of
the text’s surface features but additionally, a deeper, internalized level of understanding.
This type of understanding allows the reader to arrive at novel findings by incorporating
73
text-based information with the reader’s preexisting knowledge. This internalization is
achieved through combining and integrating textually explicit information with reader
assumptions or generalizations.
Assessing Reading Comprehension: Most Common
Methods Used in L2 Reading
The primary purpose of any assessment task is to provide relevant information for
making decisions about individuals or groups of individuals. Ostensibly, dimensions of
language ability can be inferred from reader responses. But although some tasks elucidate
certain dimensions of language ability, these same tasks may overlook other dimensions.
All assessment measures have limitations and no measure is capable of fulfilling every
assessment goal. Yet, in L2 reading, certain methods are used more frequently than
others, most likely because they are familiar to test takers and developers and because
they are convenient and efficient (Alderson, 2000). Included in the list of most common
methods are (a) multiple-choice/selected response questions, (b) gap-completion
exercises, (c) recall/summary protocols, and (d) short-answer/short constructed response
questions. A brief description of multiple-choice, recall protocols, and short constructed
responses is included here, including the pros and cons of all three, as each is employed
in this study.
Multiple-Choice Questions/Selected Response
By far, the most common method currently used to assess L2 reading
comprehension is multiple-choice (MC) questions or as they are sometimes called,
selected response questions (Alderson, 2000; Bachman & Palmer, 1996; Perkins, 1998).
Typically, in MC questions students are expected to provide the “best answer” or the
“correct answer” among the choices given (Bachman & Palmer, 1996, pp. 202-203). MC
questions are adaptable to a wide range of testing settings, easy to score, and provide
74
insights on strategies used by L2 readers. But despite the prolific use and popularity of
MC questions, this assessment method is not without drawbacks.
One problem with MC questions is that well-constructed questions that target the
intended language ability or abilities are difficult to develop. Bernhardt (1991b) contends
that even formally and professionally developed multiple-choice tests “fall into the
passage independency category” (see Wolf, 1993, p. 474). That is, test takers are able to
successfully answer these questions without having read the text. Another criticism is that
some MC questions do not require careful reading of the passage to answer the questions.
In these instances test takers are able to rely on the recognition of a few key words or on
cues from the other questions to respond successfully (Wolf, 1993). In addition, some
argue that the ability to answer MC questions is a separate skill that can be learned,
independent from the language ability being tested (Alderson, 2000). Here, students note
the inter-relatedness of questions and employ a general understanding of MC test formats
to do well on tests (Bernhardt, 1983a).
Students tested with MC questions also have a statistical advantage over those
tested with most other measures. This is illustrated by the fact that on a MC test with four
responses per question, the probability exists that the test taker will correctly answer one
in every four questions just by chance. Although most standardized tests use a scoring
procedure that corrects for guessing (Liskin-Gasparro, personal communication, April 27,
2004), the majority of texts used to measure L2 reading comprehension have not been
standardized. Finally, research findings have indicated that MC tests scores are generally
higher than those on open-ended questions, which in turn are higher than those on gapcompletion activities, which leads one to doubt that such tests are accurate reflections of
the language construct in question (Wolf, 1993). But despite these shortcomings, the
majority of language test developers have historically preferred MC questions above all
other testing methods (Alderson, 2000).
75
Recall/ Summary Protocols
Assessment instruments included within the category of recall/summary protocols
are referred to by a number of names, including free-recall tests and immediate-recall
protocols. Although the aforementioned testing methods are not completely synonymous,
they do share a number of similar characteristics. Within this section, for the purpose of
simplicity, all recall assessment tasks are referred to as recall protocols. With recall
protocols, students are asked to read a text, put it aside, and either summarize what they
have read or write down as much as they can remember about the text. Generally, recall
protocols are preformed immediately after reading although, depending on the intent of
the assessment, recall protocols can also be delayed.
Many factors distinguish recall protocols from other assessment measures,
including that students respond to these tasks in their L1. This is done because it is
believed that a test of L2 reading comprehension should not be confused with a test of L2
language production. Reading comprehension may be masked by the readers’ inability to
express themselves in the L2. Indeed, it has been suggested that even at very advanced
levels, learner’s ability to demonstrate what they understand suffers when assessed in
their L2 rather than their L1 (Bachman, 1990).
Many L2 reading researchers contend that recall protocols provide one of the
purest measure of comprehension, because neither the test questions nor the test
developer interfere with the quality and quantity of response (Alderson, 2000; Bernhardt,
1983a, 1991b). It is also claimed that these protocols provide a partial picture of the
reader’s learning processes, as they reveal information about how information is stored
and organized as well as the type of retrieval strategies the reader employs when
reconstructing the text (Bernhardt, 1983a). The recall protocol does not allow students to
guess their way through a text. In addition, recall protocols demand that readers
understand a text well enough to be able to recall it in a more or less coherent and logical
76
manner, thus allowing misunderstanding and gaps in comprehension to surface (Heinz,
1993), a feature not found in most assessment instruments.
But as with all assessment instruments, recall protocols are not without problems.
Hammadou Sullivan (2001) suggests that with recall protocols, successful readers are
able to mask or delete what they are unclear about. It has also been argued that recall
protocols are more a measure of short-term memory and recollection than a reflection of
true understanding, a reading distinction made by Kintsch (1994). His studies indicate
that texts memorized verbatim are subject to more rapid forgetting than texts stored for
meaning (Kintsch, 1974), thus calling into question whether the practice of verbatim
reading recall should be encouraged. Additionally, others contend that recall protocols
can only provide a reflection of information that is text-explicit and therefore are of
limited use when inferential knowledge is required of the text taker. Research conducted
by Hammadou Sullivan (2001) support this conclusion. In her study that measured L2
readers’ inferencing ability via a recall protocol, she found that such recall protocols
produced only limited insight into readers’ ability to infer. Recall protocols have also
been criticized for encouraging readers to treat all information as equally important rather
than place more emphasis on essential information and main ideas (Riley & Lee, 1996).
Finally, the scoring of recall protocols is generally based on a predetermined set of
propositions, or meaning units, identified by the test developer. The classification and
evaluation of these propositional components is time consuming and of questionable
validity and reliability.
Short Constructed Response Questions
A measure of reading comprehension that falls somewhere between the specific
responses of MC and gap-completion and the open-endedness of recall protocols is the
short-answer/short constructed response task, referred to here as short constructed
response questions. Here, test takers are asked questions that require either brief
77
responses, as in the form of yes/no or true/false, to those requiring a few sentences. As
with MC, short constructed responses should be worded so that all possible answers are
foreseeable, although it is difficult to predict every potential response. But this same
shortcoming, that of unpredictability, can also be viewed as a strength, particularly when
short constructed responses are used. Such questions elicit unanticipated responses that
may result in greater insights for test developers. Hammadou Sullivan (2001) found this
to be the case with her use of short constructed responses to test L2 reading
comprehension. She found that these questions provided the push that readers needed to
display what they had not fully understood. Also, readers shared more elaborative
inferences when prompted by short constructed responses than they had with recall
protocols. Additionally, the influence of readers’ prior knowledge was evidenced to a
greater extent through the use of short constructed responses than through recall
protocols.
On the negative side, as is the case with MC items, short constructed responses, if
not properly written, can allow test takers to respond by simply matching words used in
the questions to those in the passage. Also, if poorly worded, these types of questions test
only isolated facts and details and do not promote synthesis or elaboration of material
(Wolf, 1993). Bernhardt (1991b) points out that short constructed responses place limits
on the range and type of possible responses and can influence test takers’ comprehension
of the passage. Also, in answering short constructed responses, readers sometimes
discover clues to the content of the text that, otherwise, they may not have fully
understood, making the wording of these questions of paramount importance
(Hammadou Sullivan, 2001).
Assessing Reading Comprehension: Implications
As stated at the beginning of this section, there is no single ideal assessment
method to measure L2 reading comprehension. The selection of assessment method
78
depends on many factors, not the least of which is what one intends to measure and what
is done with and inferred from the assessment findings. Even with the best of preparation
and intentions, no single measurement procedure is able to capture all of the facets of any
one language components much less all the components that comprise language ability.
Moreover, of the methods reviewed here, all of have their strengths and weakness, with
some better suited for certain testing situations than others. What is clear is that to get the
most comprehensive and accurate picture possible, tests should be selected whose results
most closely reflect the language ability or abilities under consideration. In addition and
when possible, a variety of testing methods should be used and results should be
compared for common and divergent findings or features. Additionally, assessment
instruments should attempt to capture language comprehension thorough the broadest
means possible. For that reason, it is best to use methods that test readers’ inferential as
well as text-explicit knowledge as well as provide both subjective and objective
reflections of the reading experience.
Chapter II Summary: What is Known and Unknown
Expanding upon the premise set forth in chapter I that literary texts deserve a
more prominent role in early L2 instruction, in this chapter, information was presented to
suggest why this premise might be valid. Through a cognitive model of language learning
many of the similarities and dissimilarities that readers experience when reading in their
L1 and L2 were explored and traced back to the processing demands placed upon learners
as they interact with texts. As limited-capacity processors with finite cognitive resources,
it was suggested that readers experience comprehension breakdown when the number of
controlled processes required for full text engagement exceeds processing capability or
when these processes are used minimally or ineffectively. A reader may use excessive,
insufficient, or ineffectual attentional resources for a variety of reasons, some internal to
79
the reader and to a large extent out of the educator’s or the researcher’s control, but many
others external and therefore modifiable.
Following Beck et al. (1991), comprehension breakdown was attributes to either
reader failure or text failure. Although issues that result in reader failure cannot be
ignored and to a certain extent are measurable and manipulable, ways in which a text
fails, or at the very least poorly supports a reader, are of paramount concern to this study.
Texts fail L2 readers for a variety of reasons, including lack of coherence and
organization, difficulty of language, or because they include information that is
irrelevant, too simplistic, or that distracts from the text message. For L2 readers of
cognate languages, linguistic complexity and text length are not the main obstacles to
reading processing and comprehension.
The primary method of text modification used to increase message accessibility to
non-proficient readers has been linguistic simplification. Although demonstrated to
increase reader comprehension of items explicitly stated within the text, simplified texts
do not promote nor engage the same type and amount of language learning processes as
does the alternative form of modification, elaboration. This is demonstrated through
learners’ lack of ability to infer text meaning after reading simplified texts. Furthermore,
although simplified texts are structurally simpler, elaborated texts appear to make more
effective use of cognitive resources because they activate a wider and more diverse
variety of language processing strategies that serve to facilitate comprehension.
Additionally, elaborative modifications provide the type of support needed for incidental
vocabulary acquisition to occur.
Much, however, is still left unanswered. The majority of research conducted to
this point dealing with text modification has either compared unmodified texts with
simplified texts or simplified texts with elaborated texts. To date no studies have
compared comprehension of unmodified L2 reading texts with comprehension of
elaborated versions of the same texts. Additionally, no study has compared
80
comprehension of elaborated literary texts against the same texts in unmodified form.
Moreover, none of the studies that have incorporated the use of elaborated texts have
modified those texts in accordance with empirical data obtained from student readers.
Instead, elaborations have been added arbitrarily at places in the texts where the
modifiers believed students were likely to encounter problems. Finally, whereas more
recent studies have incorporated both text-implicit and text-explicit measures of
comprehension, no research has attempted to measure text recall, inferential text-implicit
information, and the incidental vocabulary learning by readers of unmodified and
elaborated versions of literary texts. All three assessment techniques were employed in
this study. The procedures used to conduct this study are explained in chapter III.
81
CHAPTER III
METHOD
Introduction
In the previous chapter, focus turned to some of the text-based features
demonstrated to contribute to comprehension breakdown for limited-proficiency L2
readers. It was argued that conversational elaboration techniques engage a wider range of
cognitive learning processes in the lower-proficiency reader than do linguistic
modifications. For this reason, readers of elaborated texts seem to comprehend what they
read on a deeper level than do readers of simplified texts, as demonstrated by their ability
to infer meaning within and beyond the text.
But although various studies have compared readers’ comprehension of
elaborated and simplified texts, no study has compared how these readers process and
understand the same literary texts in elaborated and unmodified form. In addition,
acquisition of incidental vocabulary has yet to be measured when comparing unmodified
reading texts against the same texts in an elaborated form. Keeping in mind the call
within the profession to increase the use of literary texts in early L2 instruction (Barnett,
1991; Kramsch 1987 & 1993; Rice, 1991; Schulz, 1981; Shanahan, 1997), the potential
implications of the use of elaborated literary texts are worth considering. Are literary
texts with elaborative modifications easier for L2 readers to process than the same texts
in unmodified form? Although research suggests that such elaborated texts engage,
cognitively speaking, a different array of language processing mechanisms than do
unaltered texts and thus are more comprehensible than linguistically simplified texts, is
the same true when comparing elaborated literary texts with unmodified literary texts?
Do readers comprehend these texts differently? Is their acquisition of incidental
vocabulary in any way altered? And what of the reading process? How does the
experience of reading an elaborated literary text compare to the experience of reading the
82
same text in an unmodified form? A description of the methodological steps employed in
an attempt to answer the questions posed above is presented in this chapter.
Organization of the Chapter
This chapter is organized as follows. First the design of the study is presented.
Next the operationalization of the study is introduced including a description of the study
participants. At this point, focus turns to the reading texts used in the study and, more
specifically, the procedure used to make elaborative modifications. This explanation is
followed by a description of the data collection procedure and the three testing methods
used to measure students’ reading comprehension. Each of the data collection instruments
is described using the following criteria: (a) rationale for selection, (b) preparation and
piloting, and (c) method of scoring.
Study Design: General Description
The intent of this study was to gather, access, and compare data collected from L2
learners who had read elaborated and unmodified literary texts. Specifically, the study
was designed to measure whether there is a quantitative and/or qualitative difference
between what readers of elaborated versus unmodified literary texts (a) recall from the
texts, (b) infer from the texts, and (c) demonstrate vocabulary knowledge in relation to
lexical items taken from the texts.
Study Design: Operationalization
Study Design: Participants
The participants in this study were undergraduate students enrolled in the fourthsemester Spanish course at a large research university in the Midwestern United States
during the fall 2004 semester. Students in this course have varied backgrounds and
experiences with the Spanish language and with Spanish language instruction. Some
students are placed into this course upon admission to the university based on their scores
83
on a locally developed placement test, whereas others have taken the prerequisite collegelevel Spanish language courses either at this university or at another tertiary institution.
Although, chronologically, this course may be the second, third, or fourth consecutive
college Spanish class for some, others are returning to Spanish instruction after an
absence of one or more years. Although some of the students plan on minoring or
majoring in Spanish, the majority take this course to complete their general education
requirement in foreign language and do not plan to continue studying Spanish.
For classification purposes, it is assumed that the students in this course have
completed the equivalent of three semesters of college Spanish language study.
Therefore, following Lee (1990), these participants are classified as late-beginners or
early-intermediate language learners with 300 hours of Spanish instruction.1
Study Design: Reading Texts
Each participant read two texts, one in elaborated form and one in unmodified
form, from a potential text pool of six texts. The six texts were comprised of two versions
of each of three readings taken from the intermediate Spanish textbook currently used in
this course, Conexiones: Comunicación y cultura 2nd edition (2002). Reading texts were
selected from the fourth-semester textbook because they represent the type and style of
literary readings currently used in L2 instructional settings. In addition, all three texts
were similar in narrative style and did not violate formal story schemata (Carrell, 1981).
Moreover, the texts were approximately the same length, were similar in format and
presentation, and treated topics that correspond to three of the chapter themes from this
course: (a) financial matters, (b) leisure time activities, and (c) the future.
1 FL learners who have 0 to 300 hours of classroom instruction at the college-level are
classified as beginners. Learners with 301 to 600 hours of classroom exposure are classified as
intermediates. Students at this university receive 80 hours of classroom instruction per semester.
84
As previously mentioned, background knowledge has been shown to be an
important variable in relation to text comprehension (Alderson, 2000; Carrell, 1984;
Hammadou, 2000; Lee, 1986a; Swaffar, 1991). It was not the intent of this study,
however, to measure the effect of participants’ background knowledge on
comprehension. Nevertheless, it was important to select readings about which most
readers had similar knowledge or experience. Two of the three readings meet that
criterion as their themes involve newspaper reading as a leisure activity and the potential
events surrounding the end of the world. It was assumed that most students have
experience with reading newspapers and lack experience of an apocalyptic nature.
Participants’ amount of experience and background knowledge in relation to the third
theme, playing the lottery, is varied. This variation is accounted for when study results
are interpreted.
The selected texts were all relatively short, ranging from 371-588 words in
unmodified form. Elaboration has been shown to increase text length as much as 50%
(Ross et al., 1991), which is indeed the case here, as the elaborated versions ranged from
531-946 words, an average length increase of 65%. Despite this increase, the texts were
still representative of the text types that L2 learners read in an instructed setting (Swaffar,
1991) and were of an appropriate length to be read and recalled within a 50-minutes class
period. Readers of the elaborated versions were, nonetheless, at a disadvantage, as they
were expected to complete the readings and assessment procedures in the same amount of
time as the students who read the unmodified texts. This disadvantage could potentially
be offset, however, because texts of greater length are, by their nature, more redundant
with more clarifying features (Kintsch & van Dijk, 1987; Maxim, 2002; Swaffar, 1991)
with elaborative modifications contributing to their redundancy.
85
Reading Texts: Elaborative Modification Preparation
Procedures guiding the use of elaborative modifications in previous studies have
been primarily guided by researchers’ intuition regarding where and how the
comprehensibility of the text might become problematic for the readers rather than by
empirical evidence. Phrases such as “highlighting main ideas,” “adding cohesive ties,”
and “clarifying the structure” have been used when describing how elaborative
techniques have been implemented (Brown, 1987; Oh, 2001; Parker & Chaudron, 1987;
Ross et al., 1991). These descriptions, however, do not explain why certain revisions
were made and others were not. For this reason, in this study elaborative modifications
were made in accordance with empirical data collected from student readers when
engaged in the process of reading the three texts in question.
In the spring 2004, six intact classes of fourth-semester Spanish participated in
think-aloud protocols (Pressley & Afflerbach, 1995) while reading the three unmodified
text versions. These think-aloud protocols were recorded on audiotape. Students were
asked to verbalize what they were thinking about as they read and not to be concerned
about the coherence of their thoughts. The purpose of these think-aloud protocols was to
identify types and locations of comprehension breakdown. The think-aloud protocols also
revealed how students approached the texts, the kinds of the reading strategies they used,
and the places where glossed lexical items did or did not aid comprehension.
Initially, each student’s verbal comments were analyzed separately. Eight two
students participated in the think-aloud protocols; three classes read one of the
unmodified text versions with marginal glosses provided by the textbook authors and
three other classes read the same texts with the glosses removed. The motivation behind
removing the glosses was to see the impact that the glosses had on students’
comprehension. A secondary motivation was to determine to what extent students used
the marginal glosses and how helpful glossing was to overall comprehension.
86
As student participation during the think-aloud protocols was voluntary, 14
students chose either not to participate or the quality of their participation was such that
their comments were of limited use (e.g., reading the passage aloud in Spanish without
additional comments; commenting on first few lines of text only). The data from those 14
students were excluded, resulting in 68 think-aloud protocols that were compiled and
analyzed.
Each class read a different text version (i.e., 3 texts: a glossed and unglossed
version of each) and six separate lists of problem areas were created. In order to qualify
as a problem, at least half of the readers of the text needed to express confusion or
misunderstanding at that specific point in the text. As students used different strategies to
approach their respective texts, care had to be taken to avoid second-guessing readers’
intentions. The issue of second-guessing was especially challenging when identifying
problem areas in the protocols of the lowest-proficiency students and with the most
difficult texts. When comprehension became too difficult, weaker readers, or those
reading the more difficult texts, had a tendency to give up. For this reason, the thinkaloud protocols did not provide equally reliable or comprehensive information for all
readers. Consequentially, the determination of areas of comprehension difficult was
primarily obtained from data generated by the strongest readers and, for that reason, is of
limited use as a representative sample of all participating readers. Once the protocols
were individually analyzed, confusion problems were compiled and categorized by
class/text version.
When all six lists had been created, three master lists were compiled, one for each
text. These master lists included not only the locations and types of problems verbalized
by the students, but also specific examples of students’ misunderstandings. From these
master lists, misunderstanding categories were then created as well as a list of the most
problematic words or phrases in each reading. Problematic words and phrases were also
weighted and the terms deemed most essential for comprehension were identified and
87
later targeted for elaborative support. An item was considered essential for
comprehension when the majority of readers who expressed confusion at that point were
unable to decipher the general meaning of the sentence or phrase without the aid of that
word or phrase. Patterns of confusion were also identified in all three texts. (See
Appendix A for combined types and examples of confusion patterns for the three literary
texts.)
Reading Texts: Elaborative Modification Procedure
Using the students’ confusion patterns and lists of problematic lexical items taken
from the think-aloud protocols, the three literary texts were modified according to the
elaborative modification features outlined in chapter II. As described in that chapter, the
fundamental goal of the elaborative modification was to create texts that would assist the
readers to connect pieces of information in order to develop a coherent mental text
representation. The elaboration techniques used to help connect these pieces of
information are (a) message maintenance, (b) message clarification, (c) text organization
with signaling, and (d) repetition and redundancy. Additionally, other non-linguistic
alterations were made to the texts dealing primarily with the removal of glossed
vocabulary and attempts to replicate the texts’ titles and illustrations. In this section, each
of these modification techniques is described.
One of the most important considerations when modifying the texts was
maintenance of the text message; that is, the preservation of the complexity, both
syntactic and lexical, found in the original text. Although the lexical complexity was
maintained in all three texts, on occasion, syntactic order was changed either by means of
subject fronting or by explicitly adding or relocating pronoun referents. Occasionally,
sentences with multiple clauses were broken into two sentences but, more frequently,
sentence length was increased by adding redundancy features, additional clauses,
88
paraphrases, resulting in a greater number of sentences; that is, sentences that were more
rather than less complex.
In addition to message maintenance, message clarification techniques were
employed. This procedure was used at points in the text where readers had trouble
connecting an action to a specific agent or had difficulty relating something mentioned
previously in the text to subsequent references to the same topic. Message maintenance
was also preserved through the use of temporal markers and the addition of relative and
complementary clauses.
Another technique, that of signaling clarification, was used to focus readers’
attention on text features intended as information organizers. Much of this was done
through the use of conjunctions or through the inclusion of full noun phrases in place of
pronouns. In one text, signaling clarification was the motivation behind a formatting
change that converted a grouping of elements originally described in narrative form
within a single paragraph into a list in which items were identified numerically (i.e., 1, 2,
3). This clarification was made in response to readers’ inability to understand that the
data contained within the original paragraph were separate, quantifiable elements all
pertaining to the same list of items.
The texts were also elaborated through use of repetition and redundancy, a
modification technique that includes paraphrases, synonyms, and definitions of lowfrequency vocabulary. This technique was used frequently owing to the critical role
lexical understanding has been found to play in L2 reading, especially for lowerproficiency readers (Bernhardt, 1991a).
In addition to linguistic modifications, alterations were made to other text
features, specifically in response to information gathered from the think-aloud protocols.
For instance, it was decided that the titles and illustrations that accompanied each text be
maintained in all text versions. This decision was reached after realizing the extent to
which readers relied on the graphics and titles as advance organizer to get a sense of the
89
overall text message. Although the preservation of the illustrations may have influenced
students’ overall comprehension, this variable was not crucial to study results as the same
illustrations were present in both the unmodified and elaborated text versions.
Another variable identified as influential to comprehension was text glossing. In
the case of these three unmodified texts, some lexical items were glossed by translation
into English, whereas others were defined or explained in Spanish. In the unmodified
versions, individual words, phrases, and sentences were glossed with bullets indicating
that a definition was provided in the margin (see Appendix B for copies of the
unmodified texts as scanned from the textbook). During the think-aloud protocols it was
determined that few students consistently and successfully incorporated the glossed terms
into their overall understanding of the readings. Despite this, there was no doubt that
readers did rely on the glosses, specifically for re-orientation purposes. Additionally, as
the intent of this study was to compare the effects of elaboration with the type of reading
texts commonly used by intermediate-level L2 students, it was determined that the
glosses should be maintained in the unmodified text versions.
Because two goals of textual elaboration are to clarify the information in the text
and in the language of the text, the decision was made to remove the marginal glosses in
the elaborated versions. There was, however, one exception to this rule that should be
noted. Two Spanish terms, both with basically the same meaning, were glossed in each of
the three texts four times. These terms were es decir and o sea ‘that is.’ In the elaborated
versions, these terms were frequently used to introduce a relative clause for clarification
purposes. Ideally, readers familiar with elaborative modification would be familiar with
these phrases, but in this study, it could not be assumed that learners understood their
significance. As it was felt that their role in signaling an upcoming linguistic clarification
could prove pivotal to overall text understanding, these phrases were glossed in English
in the text margins in all three elaborated versions.
90
The last non-linguistic modification feature taken into consideration when the
texts were elaborated was the format in which the texts were presented to students. For
the first day of data collection it was decided that students reading the unmodified text
versions would read the texts directly out of their textbooks. The textbooks were used so
that readers might approximate, as nearly as possible, their typical reading experience.
An unanticipated problem arose, however, when the researcher noted that a number of
students were reading from used textbooks whose previous owners had partially or
wholly translated the reading text. Indeed, on one occasion a participant was seen reading
the student-annotated translation to other research participants during the study.
The problem of annotated texts was rectified on subsequent data collection days
when students were given the unmodified texts as scanned copies from the textbook. This
also avoided the potential use of the glossary provided in the back of the textbook. In
addition, students were instructed not to use online or paper dictionaries, so as to
eliminate the potential variability that dictionary use might have on study results.
Participants reading elaborated versions had no translation recourse and those reading
unmodified versions had only the marginal glosses.
After the first data collection day, all student participants were given paper copies
of the texts. As described, the unmodified versions were scanned from the textbook and
the elaborated versions were formatted to resemble the unmodified textbook readings.
Students were given print copies of the texts rather than asked to read online, not only to
replicate the typical reading experience, but also because in previous studies students had
expressed dissatisfaction with online reading (Heinz, 1993). The elaborated texts were
formatted to resemble the unmodified textbook versions by matching font size and type
and by scanning the titles and color illustrations from the textbook. Copies of the
elaborated versions of all three texts are presented in Appendix C. Although, the font size
in the appendix appears larger than that of the elaborated version, during the study, the
size and appearance of both texts were closely matched.
91
Study Design: Data Collection Procedure
Working during regularly scheduled class periods, intact Intermediate Spanish II
classes were randomly asked to read either an unmodified or elaborated version of a
reading text. Altogether, 13 sections of Intermediate Spanish II participated in this study.
The unmodified versions of all three texts are required reading in this fourth-semester
course. To avoid the possibility that students came to class having already read the texts,
participants were instructed to read and recall the texts during the class period on the day
before the texts were assigned for homework. Each section read two texts, one on each of
two days. Knowing that on day two, students might expect to read another text, on that
day participants were asked not only if they had read the text prior to class but, if so, how
much time they had spent reading it. Students were told that an honest response to this
inquire would not negatively effect their performance in the course. Few students
indicated having seen the text and, of those who did, the majority stated that they had
spent only two or three minutes looking it over.
There were three data collection days in all, although each section of students
participated on only two days. On data collection days, participants were asked to report
to a computer lab rather than their regularly scheduled classroom. Once in the lab they
were asked to log onto the computers and to access Web Course Tool (WebCT), an
online course management tool currently available to all students on the university server.
Students were then told that they would be reading a text taken from their textbooks, an
assignment originally scheduled as homework for the upcoming day’s class. Prior to
reading, students were told that they would be tested online after completing the reading.
They were also told that they would be expected to recall information about the reading,
in English, and for that reason, they should read the text carefully and as many times as
necessary to understand its meaning. The reason that they were informed of the recall
assignment prior to reading was because, according to Lee (1986b), students’ recalls are
significantly improved when they were given prior notification. Students were also told
92
that they should raise their hands when they had finished reading, which would notify the
researcher to remove the text copies. Students were instructed that, once the texts were
removed, they would need to click on an icon associated with that particular reading on
the computer screen and follow the testing instructions given to them. Finally, students
were told to let the researcher know if they had any questions or problems either prior to
or during the reading process.
After instructions were completed and questions answered, each student
participant was given a copy of either an unmodified or elaborated version of one of the
reading text. Although the particular version was assigned randomly, certain criteria were
followed. First, each group of students read one text in unmodified form on one day and a
different text in elaborated form on the subsequent data collection day. Also, as five out
of eight instructors in this course taught two sections of participating students, on any
given data collection day, one section from each instructor read an unmodified version
and the other class read an elaborated version. This was done to offset the potential
influence that instructors might have on test results (Bachman & Palmer, 1996). The
number of students that could be accommodated in the computer lab at any given hour
also determined section selection. It was important that all students read the same version
at the same time due to variation in text length to guard against the possibility that
students reading the shorter text might influence reading time of those reading longer
texts or vice versa. Finally, equality in numbers was another determining factor related to
the selection of participant numbers. Ultimately, 10 classes read the Fiebre de lotto, 5 in
unmodified and 5 in elaborated form, and 8 classes each read Tiempo libre and
Apocalipsis, 4 in unmodified and 4 in elaborated form respectively
Table 3.1. 3.2 and 3.3 indicate the number of study participants per reading and
per assessment instrument. As can be noted in the table, a number of individual results
for each assessment instrument were not captured, most likely due to students not saving
answers. This hypothesis was confirmed by the automatic timing feature in the software,
93
which indicated that participants invested considerable time on items that were
subsequently recorded as missing data. When data were collected from each reader a
second time and participants were more familiar with the assessment technique, a higher
number of data sets were saved for later evaluation. Despite the loss of some data, the
percentage of responses not saved for later evaluation was minimal: .038% of the recall
protocol results and .033% of the MC and vocabulary results.
Table 3.1. Missing Participant Numbers per Section per Reading and per Assessment
Instrument: Fiebre de lotto
Text
Section
N
FL
A
12
-1
0
FL
E
16
0
FL
F
16
FL
L
FL
I
Totals
Recalls MC Vocab
Text
Section
N Recalls MC
Vocab
-1
FL*
B
14
-1
0
-1
0
0
FL*
C
16
0
0
0
0
-2
-2
FL*
G
16
0
-1
0
17
0
-3
-2
FL*
H
14
0
-2
0
13
0
0
0
FL*
M
15
-5
0
0
74
73
69
69
Totals
75
69
72
74
FL= Fiebre de lotto (unmodified): FL*= Fiebre de lotto (elaborated)
Table 3.2. Missing Participant Numbers per Section per Reading and per Assessment
Instrument: Tiempo libre
Text
Section
N
TL
D
13
-1
0
TL
H
15
-2
TL
J
18
TL
M
Totals
Recalls MC Vocab
Text
Section
N
-1
TL*
E
18
-1
0
-1
0
-1
TL*
F
17
0
0
-1
-1
0
-1
TL*
K
14
-1
0
0
14
0
0
-1
TL*
L
16
0
-1
-1
60
56
60
56
Totals
63
63
62
60
TL= Tiempo libre (unmodified): TL* = Tiempo libre (elaborated)
Recalls MC
Vocab
94
Table 3.3. Missing Participant Numbers per Section per Reading and per Assessment
Instrument: Apocalipsis
Text
Section
N
AP
B
15
-2
-1
AP
C
16
0
AP
G
18
AP
K
Totals
Recalls MC Vocab
Text
Section
N
Recalls MC Vocab
0
AP*
A
12
0
0
0
0
0
AP*
D
11
0
-1
0
0
0
0
AP*
I
15
0
0
0
16
0
0
0
AP*
J
18
0
0
-1
65
63
64
65
Totals
56
56
55
55
AP= Apocalipsis (unmodified): AP* = Apocalipsis (elaborated
Assessing Reading: Testing Methods Used in this Study
To capture a broad representation of the process and product of reading, three
different assessment instruments were used as described in chapter II. Each assessment
task was designed to assess a different aspect of reading. One task targeted overt
comprehension; that is, text-explicit information, whereas another elicited text-implicit
data. The last assessment instrument targeted students’ vocabulary comprehension and,
by extension, their incidental vocabulary learning. All three measurement techniques
obtained results that were quantifiable and, therefore, amenable to comparison and
contrast. As multiple means of assessment are preferable to a single means (Wolf, 1993),
three separate tasks are employed: (a) recall protocols, (b) multiple-choice questions, and
(c) short constructed responses. These three assessment instruments are representative of
some of the most commonly used L2 reading assessment tasks (Alderson, 2000) and were
piloted in summer 2004 prior to use in the study.
Testing Methods: Assessment Instrument #1
Recall Protocols for Text-Explicit Information:
Rationale for Use
As previously described in chapter II, many L2 reading researchers believe that
recall protocols provide one of the purest measure of comprehension because neither the
95
test questions nor the test developer interfere with the quality and quantity of response
(Alderson, 2000; Berkemeyer, 1989; Bernhardt, 1983a, 1991b). It has also been asserted
that recall protocols provide a partial picture of how readers experience the text, revealing
how information is stored, organized, synthesized and reproduced (Bernhardt, 1983a).
According to Berkemeyer (1989), “the immediate recall protocol demands that the reader
comprehend the text well enough to be able to recall it in a coherent and logical manner .
. . [which] allows misunderstandings and gaps in comprehension to surface” (p. 131).
Because recall protocols reveal information about the interconnectedness of information
that readers store in their short-term memory, they can be appropriately used to expose
the propositional-level representations that are linked to pre-existing knowledge, and are
therefore available for retrieval (Kintsch, 1974). Recall protocols can be seen, therefore,
as the operationalization of comprehension. As noted by Bernhardt (1991b), “the use of
immediate recall protocols for operationalization of comprehension is generally not
assailed” (p. 218). For these reasons, recall protocols have become one of the preferred
methods of reading assessment and are appropriate for use in this study.
Apart from the access to reader processing that recall data provide, in this study
recall protocols were used as a means of assessing reader comprehension for a variety of
other reasons. If, according to cognitive learning theory, readers’ processing capacity
determines the amount and type of input that can be dealt with at a given time, then the
text features that are the most salient and processable are those most likely to be recalled
(Kintsch & van Dijk, 1978). With a large-scale study such as this one, propositional
recall patterns among readers can be identified, classified, and described. Furthermore,
the influence or lack of influence on participants’ recall of the information contained in
the unmodified texts can be identified before and after elaborative modifications have
been instituted. Although Kintsch (1994) might argue that recall protocols constitute
remembering a text rather than learning from a text, it can also be maintained that not
even remembering can take place without comprehension (Berkemeyer, 1989).
96
Particularly when used as part of a battery of assessment measures, a recall protocol can
provide valuable data not accessible through other means of assessment. Both Bernhardt
(1983b) and Lee (1986b) demonstrated that a recall protocol is capable of producing
quantitative results that can be subjected to various forms of statistical analysis.
Additionally, Bernhardt (1983a) showed that the same type of data can be qualitatively
examined to gain insight into aspects of reading that are more process than product
oriented.
Unfortunately, the standard procedures involved in developing scoring protocols
and templates for recall protocols are cumbersome and time consuming (Bernhardt,
1991b). Developing scoring templates with well-known protocol assessment systems
such as that in Meyer (1985) has “traditionally taken between 25 and 50 hours per 250word text” (Bernhardt, 1991b, p. 202). Scoring individual student recall protocols is time
consuming as well. Bernhardt estimates that each protocol, depending on its length, takes
one-half to one hour to score. This time investment is not practical or realistic for a study
of this size. For this reason, following Bernhardt (1991b) and Johnson (1970), weightedpausal units were used to measure students’ ability to recall texts. The procedure is
explained below.
Recall Protocols: Preparation of Texts for Scoring
In order to score students’ recalls, the unmodified reading texts were prepared to
use for recall comparisons. First, a professional translator translated the three readings
from Spanish to English. Translation was necessary because student recall protocols were
written in English (Lee, 1986b). After translation, the texts were broken down into
propositional units. In accord with Bernhardt (1991b), simple propositions, determined
by pausal units or breath groups, were used in this study. Bernhardt states that “a pausal
unit is one that has a pause on each end of it during normally paced oral reading” (1991b,
p. 208). Pausal propositional units were created by asking two expert L1 English readers
97
to read the translated texts to themselves, marking the places where they paused as they
read (see Bernhardt, 1991b, p. 209). After these readers had marked the texts according to
naturally occurring reading pauses, each text was re-evaluated by a third expert L1
English reader and by the researcher. It seemed apparent during re-evaluation that the
identification of naturally occurring pauses, although an appropriate starting point for text
breakdown, was inadequate when the goal of pausal template creation was ultimately the
assessment of the quantity and quality of reader recall. The following example from the
translated text Fiebre de lotto illustrates this point.
Initially, the two expert readers consistently marked pausal units at syntactic text
breaks such as at the beginning of prepositional phrases. This point can be illustrated in
how the experts divided the first translated sentence of Fiebre de lotto: The 160
employees/of the Savings and Loan Bank/agreed to spend/all of their savings/on the
combined purchase/of half a million dollars/in lottery numbers/from the Florida lottery . .
. . When the pausal units were later re-evaluated, the appropriateness of always using the
beginning of a prepositional phrase to indicate a pause was called into question.
Although in most cases, pausal breaks at the beginning of prepositional phrases were
deemed appropriate, there were exceptions. This is illustrated by the phrase “lifted their
arms above them,” which appears later in the same reading. Here it was determined that
the prepositional phrase “above them” did not add enough new information to the idea
being expressed to warrant being identified as a separate pausal unit. In other words, the
pausal units that were ultimately created were intended to serve the instrument’s function,
that of measuring students’ comprehension of text information. For this reason the pausal
units were created with both semantic as well as syntactic concerns in mind. (See
Appendix D for division of texts into pausal units.)
Later, weighted propositional units were identified following Bernhardt (1991b).
Two native L1 English speakers weighted these units by indicating which pausal units
they considered to be more- or less-essential to text comprehension. These readers were
98
asked to cross out idea units of lesser importance to the overall meaning of the text
(Bernhardt, 1991b, pp. 208-217). Approximately half of the pausal units were
subsequently removed, resulting in a division between more-essential and less-essential
idea units. All told, 52% of the idea units were determined to be more essential and 48%
less essential. When student recalls were later compared against the pausal unit rubrics
prepared by the expert readers, propositional weighing was considered in addition to
length of recall.
Recall Protocols: Scoring
Recall protocol data were collected from 207 students altogether with most
students provided vocabulary data twice. Although written recalls were collected from all
students, for reasons of practicality, only a subset was scored. Out of a total of 379
recalls, 36 recalls were scored. These 36 were selected for scoring so that a representative
sample of strong, moderate, and weak recalls could be scored for each text version (See
procedure for stratified random sampling in McCallon & McClaran, 1974, pp. 9-12.).
These recalls were evaluated simultaneously by five raters to insure consistency
(Bachman & Palmer, 1996, pp. 221-222). The raters consisted of expert L1 English
readers who had previously taught or were currently teaching college-level Spanish. The
36 represented for each of the six reading text versions the two longest, the two shortest,
and the two whose text length most approximated the average, henceforth referred to as
the most average, as determined through word count. During simultaneous scoring, none
of the readers knew whether the recalls that they were scoring were produced by readers
of the unmodified or the elaborated versions. In addition, raters did not know which ideas
units had been determined to be more or less essential.
To help assure reliability, following Bachman and Palmer (1996), simultaneous
rater training and recall scoring was conducted. Rater training consisted of (a) reading the
texts in translation and discussing the rating rubrics one by one prior to scoring, (b)
99
practice rating recalls not considered part of the subset, (c) comparing application of the
rubric, and (d) discussing discrepancies and variation among raters (Bachman & Palmer
1996). Initially, raters practiced by scoring recalls separately and then comparing their
results. For the operational scoring, it was decided to use an approach that would be
comparable, as well as more efficient. One reader (i.e., the researcher) read the recalls
aloud, pausing at intervals to allow the raters time to identify idea units. When raters
disagreed, discussion ensued, at times resulting in agreement and, at other times,
disagreement. As agreement was not necessary, the raters resumed scoring after these
brief discussions. Due to these discussions, the resulting Pearson correlation coefficient
among raters was r (5)=.99.
One week after the group rating session, nine recalls out of the original 36 were
re-scored by the same raters to ensure intra-rater reliability (Deville, personal
communication, January 14, 2005). Through this method a correlation between idea unit
totals from the first scoring was compared to idea unit totals from the second scoring and
demonstrated a Pearson correlation coefficient of r (5)=-.84 for intra-rater reliability. A
negative correlation was obtained because raters identified fewer idea units when
working on their own. Nonetheless, correlations between identified ideas units among
raters per recall protocol remained strong when recalls protocols were scored
individually.
Figure 3.1 illustrates raters’ identification of total idea units during simultaneous
rating and then after one week. Overall fewer idea units were identified by all raters when
working on their own but the number of idea units found after one week did not vary
significantly among raters.
100
Figure 3.1. Raters’ Preliminary and Subsequent Identification of Idea Units
Raters' Preliminary and Subsequent Identification of Idea Units
Score
300
Pre
Sub
250
200
1
2
3
4
5
Pre
261
265
264
263
265
Sub
251
233
227
250
222
Rater
Testing Methods: Assessment Instrument #2
Multiple-Choice Questions for Text-Implicit Information:
Rationale for Selection/Use
Although scores derived from recall protocols provide some insight into the
product and process of reading, they are of limited value as indicators of what Kintsch
(1994) calls “learning from a text” (p. 294). As described in chapter II, Kintsch (1994)
defines two levels of text comprehension, with text learning representing a more
profound internalization of understanding than that which can be demonstrate through a
reader’s ability to recall a text. According to Kintsch, readers demonstrated learning from
a text through their ability to synthesize information holistically and reach conclusions
not explicitly stated in the text. Recall protocols can demonstrate only what readers have
decoded for comprehension and remembered for reproduction. To determine whether a
101
reader has learned from a text, additional assessment measures are necessary because,
according to Kintsch, it is only through a reader’s ability to infer new knowledge that text
learning is revealed (1994).
Inferential knowledge can be defined as the ability to answer a question related to
meaning not directly stated in the text (Alderson, 2000). Previous work by L2 reading
researchers has resulting in the categorization of two types of inferences: (a) within-text
inferences that synthesize or summarize ideas found in the text, and (b) elaborative
inferences that combine concepts from the text with readers’ prior knowledge (Barry &
Lazarte, 1998; Hammadou Sullivan, 2001). As previously stated, a certain amount of
background knowledge is assumed, no matter what the testing method. In this study, in
an attempt to decrease the influence of reader background knowledge, the inferential
questions were created to target within-text knowledge rather than elaborative inferential
knowledge.
Despite certain drawbacks mentioned in chapter II, readers’ inferential knowledge
was measured by means of multiple-choice (MC) questions. The justification for the use
of this method is based on the fact that many negative features associated with MC
questions could be controlled for, given that students have to synthesize their knowledge
rather than recall facts. As the answers are not explicitly stated in the text, the objection
to the use of MC questions because they do not require a careful reading of the text were
mitigated. The same can be said for the presence of keyword triggers that evoke
responses that do not require comprehension. Lifting of keywords directly from the texts
was not likely to occur in this study because participants read in Spanish and were tested
in English and because no Spanish/English cognates were used in the creation of the MC
questions. Also, they did not have access to the texts while responding to the MC
questions.
Nonetheless, care was taken to assure that the inference questions were not
interdependent and that responses measure what the readers understand based on the text
102
and not on prior knowledge. Although research has shown that MC questions act as an
information source providing readers with options, including appropriate conclusions,
which they might have never considered (Kembo, 2001), the questions were intentionally
created with plausible options as distracters; that is, choices worthy of consideration but
ultimately not supported in the texts themselves.
Multiple-Choice Questions: Preparation and Piloting
Only three MC assessment tasks were created, one for each of the three texts. This
was done so that essential information present in both text versions could be directly
compared. As inferencing from reading passages involves many different components of
language ability, MC questions were created that asked participants to provide the “best
answer” from a list of credible options rather than the “correct answer” (Bachman &
Palmer, 1996, pp. 202-203). Moreover, as with the recall protocols, English was the
language of assessment for these questions to avoid comprehension complications that
might result from language unfamiliarity. Test items were written in simple language and
piloted on readers of similar L2 proficiency. Following Fecteau (1999), the MC
inferencing questions in this study targeted (a) narrative perspective or point of view, (b)
the predominant story theme, (c) the mood of the characters or narrator, and (d) the
overall tone of the person telling the story. Unlike Fecteau’s questions, each question in
this instrument had four rather than three possible responses to decrease the probability
that participants could randomly select the correct response and to increase the
discrimination potential of each question (Bachman, 1990). In addition, and again
following Fecteau, the four inference questions were presented in order of assumed level
of difficulty. Familiarity with narrative viewpoints was deemed the easiest concept
followed in order by themes, character or narrator mood and story tone. This order was
based on the assumption that previous instruction probably addressed types of narration,
themes and characters more often than tone or authors’ aims (Fecteau, 1999). The order
103
and the wording of each question in the three MC assessment instruments were similar,
although not identical. The placement of correct responses (i.e., the a, b, c, or d position)
was assigned randomly by the computer and varied from text to text.
During the summer 2004, all three assessment instruments were piloted with 30
students enrolled in fourth-semester Spanish classes. Three sections of students
participated by reading the unmodified versions of all three texts and completing the
three assessment instruments that accompanied each text. For the MC questions during
piloting, five rather than four possible responses were given to students so that the
response that distracted the fewest students could be later eliminated. Care was taken to
assure that the stronger students were able to recognize the correct responses, with strong
students defined as those who most frequently provided the correct answers on all three
assessment days.
During piloting, each of the 12 MC items was analyzed by means of a
discrimination index2 using discrimination values cut off level of .40; a level considered
to indicate high discriminatory power (Ebel, 1954). All four MC questions for Fiebre de
lotto and Tiempo libre had a discrimination range from .40-.83 with the exception of one
item, which was changed prior to the research study. Only one of the MC questions for
the Apocalipsis reading, however, had a discrimination index higher than .40. In addition,
on two occasions, a distracter drew too many strong students away from the correct
response and, therefore, these distracters were changed prior to data collection.
An additional problem arose when the students who had distinguished themselves
as strong students on the first two reading evaluation days faltered almost to the same
extent as the weak students on day 3. It was determined that this was likely attributable to
2 An item discrimination index indicates the extent to which success on an item
corresponds to success on the overall test. According to Kelly et al., “the discrimination index is
computed from equal-sized high and love scoring groups on the test. Subtract the number of
successes by the low group on the item from the number of successes by the high group and
divide this difference by the size of the group” (www.rasch.org/rmt/rmt163a.htm).
104
the difficulty level of the Apocalipsis text, an issue about which students frequently
complained during the think-aloud protocols. After piloting, three of the total 12 MC
questions were adjusted, and one was changed completely to increase the discriminatory
power of the instrument.
Multiple-Choice Questions: Data Collection Procedure
During the primary research study, MC questions were presented to the
participants online immediately after completion of the recall protocol. As with the recall
protocols, participants did not have access to the texts when responding to these
questions. Research by Johnson (1984) indicated that on questions central to text
comprehension, students’ performance improved when they did not refer back to the text.
He hypothesized that when the text was available, the question-answer process involves
search-and-match strategies rather than “actual comprehension” (as cited in Alderson,
2000, p. 107). Moreover, findings by Alderson (2000) indicate that central questions are
easier to understand with the text removed, whereas peripheral, or non-essential questions
are easier when readers have access to the text to search and match. He also contends that
texts should be removed in testing situations where students’ knowledge of the main idea
is assessed, as is the case here (p. 107).
As with the recall protocol, participants were instructed to take as much time as
they wished to respond to the MC questions. Because the questions and response options
were written in a simple and straightforward way, most participants required only a few
minutes to complete this task. Each question appeared by itself on the screen but could be
skipped and revisited later. Participants were asked to make an informed guess if they
were unsure of the answer to assure that all questions were seen by the participants
(Bachman & Palmer, 1996). (See Appendix E for multiple-choice questions as they
appeared to readers on WebCT.)
105
Multiple-Choice Questions: Scoring
Multiple-choice data were collected from 206 students although, as with the other
assessment instruments, most students provided data twice. As with the recall protocols,
scoring of the MC questions was computerized. The computer-generated results were
analyzed descriptively and inferentially to provide summary statements about the data
and to determine whether results could be generalized to this population of L2 readers. Of
particular interest when interpreting the results were data that directly compared the
inferential ability of the readers of the same text in its elaborated versus its unmodified
form.
Testing Methods: Assessment Instrument #3
Short Constructed Responses for Vocabulary Knowledge:
Rationale for Selection/ Use
Following Knight (1994) and Pulido (2003), readers were asked to supply the
meanings of lexical items found in the texts in their L1 (English) via short constructed
responses. Short constructed responses given in the L1 are particularly valuable for
assessing L2 word knowledge as readers are pushed to demonstrate what they do and do
not comprehend lexically (Hammadou Sullivan, 2001). Negative aspects of short
constructed responses, previously described in chapter II, were considered when the
questions were constructed. They include (a) poorly written questions, (b) readers’
matching of questions with direct quotations taken from text passages and, (c) correctresponse clues revealed through the questions themselves. These negative aspects were
not an issue here due to the nature of question development, a procedure described in the
preceding section. The criticism, however, that these types of questions measure only
knowledge of isolated facts and details and do not promote synthesis or elaboration of
material (Wolf, 1993) is valid. Synthesis and elaboration of textual information,
however, had been addressed and assessed in the two other assessment instruments.
106
Short Constructed Responses: Preparation and Piloting
As with the two prior instruments, lexical comprehension through short
constructed responses was piloted in the summer 2004. This instrument, however, was
developed late in the piloting period when the data garnered through an open-ended
paragraph format on the first two data collection days proved too unwieldy to quantify.
During the last day of the three-day piloting period, however, lexical knowledge through
short constructed responses was included and results were analyzed.
During this pilot study, 16 lexical items deemed by the researcher to be of
moderate difficulty were targeted for assessment. The selection of these lexical items was
not of paramount importance as the emphasis here was on the successful functioning of
the assessment tool more so than on the results obtained. Readers were asked to provide
short definitions in English for 21 Spanish words or phrases and to provide five
additional definitions, also in English, for English cognates of Spanish terms found in the
reading. Results revealed that only one Spanish word, será ‘will be,’ was correctly
identified by more than 50% of the students (64%). Of the 16 Spanish items tested, 6
were completely unfamiliar to all students and 5 others recognized by fewer than 20% of
the readers. The five English cognates received a range of 10 to zero correct answers,
respectively, with “to languish” being identified correctly by the largest number of
readers and “capricious” incorrectly identified by all students.
But as per the previous discussion regarding MC item discrimination, care must
be taken when interpreting these results as students were asked to identify lexical items
after reading what proved to be the most challenging text: Apocalipsis. As previously
mentioned, as well as demonstrated by the think-aloud protocols, readers of texts pitched
beyond their comprehension level seem to become disheartened and have a tendency to
give up. As this third and final pilot reading resulted in the lowest comprehension results
on all measures, the results obtained from the lexical assessment instrument cannot be
107
viewed as typical of those that would have been obtained should this instrument have
been used with the two previous reading texts.
Again, the primary motivation behind the use of the short constructed response
technique during pilot testing was not so much the scores achieved by readers as the
testing of the online short constructed response technique for the subsequent research
project. Through piloting it was determined that to increase lexical comprehension, the
target words would be presented in context rather than in isolation. Therefore, during the
research study, the lexical items were presented in short phrases, similar and sometimes
identical to those that students had just seen in the texts. This was done as too little
contextual support has been found to result in readers who have more difficulty with
word identification (Bernhardt, 1991a; Haastrup, 1991; Hulstijn et al., 1996). The pilot
study also revealed a need for clear, concise instructions outlining reader expectations as
well as a model question to demonstrate appropriate response types. Therefore, these two
features were added prior to testing.
To create the word lists for the primary research study, the problematic lexical
item lists, previously created from data obtained from the think-aloud protocols, were
used. Initially, these word lists were coded according to the percentage of think-aloud
students who struggled with the word or concept. The words or phrases deemed the most
problematic were those that were correctly understood by fewer than 25% of readers.
Next, of the three texts, the one with the fewest problematic words, Tiempo libre with 34
items, was used to determine the number of lexical items that would be tested for each of
the three lexical assessment instruments. This was done in order to maintain consistency
in instrument length. As with the recall protocols and the MC questions, the three lexical
assessment instruments were identical for readers of both the unmodified and elaborated
versions of the reading texts.
108
Lexical items for testing were ultimately chosen using the following criteria: (a)
all words glossed in Spanish were included to gauge the effectiveness of elaboration
versus glossing in word identification; (b) all words glossed in English were eliminated to
prevent verbatim lifting of English glosses in response to vocabulary questions; (c) all
words that did not have direct English equivalent were excluded (e.g., complex relative
pronouns such as el cual), although idiomatic expressions were maintained; (d) all
problematic words that were repeated thorough elaborative modifications were excluded
as repetition has been shown to enhance comprehension; and of the remaining lexical
items; and (e) all words demonstrated to be the most problematic during the think-aloud
protocols were chosen. Table 3.4 illustrates the resulting categorical breakdown of
targeted items.
Table 3.4. Categorical Breakdown of Targeted Lexical Items
Fiebre de lotto
Tiempo libre
Apocalipsis
Nouns (5)
Noun (5)
Idioms (5)
Verbs in preterit (3)
Verbs in preterit (5)
Nouns (4)
Idioms (3)
Idioms (3)
Verbs in future (4)
Adverbs (2)
Adjectives (2)
Adjectives (3)
Verbs in infinitive (2)
Verbs in infinitive (1)
Verbs in infinitive (1)
Verb in preterit with indirect
object pronoun (1)
Verb in present reflective (1)
Verb in present subjunctive with
direct object pronoun (1)
Adjectives (1)
Adverbs (1)
Preposition (1)
Note: Of these lexical items, those glossed into Spanish in the margin included six from
Fiebre de lotto, three from Tiempo libre, and two from Apocalipsis.
Care was also taken to avoid including lexical items that had been defined
through use of synonym or paraphrase in the elaborated versions, so as not to unfairly
109
bias results. Eighteen items were selected from Tiempo libre and the same number of
items was later used for the creation of the other two test instruments. Prior to the
selection of lexical items, participants were not tested to gauge their familiarity with the
words. Familiarity was not tested as all sections of Intermediate Spanish II, with the
exception of the night classes, participated in this study. For that reason, it can be
assumed that these students represent the entire population and, as such, their knowledge
of the selected items is normally distributed.
Short Constructed Responses: Data Collection
As with the other two assessment procedures, testing for vocabulary knowledge
occurred online immediately after completion of the multiple-choice (MC) questions.
Also, as with the other assessment instruments, participants did not have access to the
texts when responding to these questions and they could use as much time as necessary to
complete the assessment. Participants were instructed to translate into English certain
words or phrases that they would see underlined, which, they were told, were all lexical
items originating from the texts they had just read. They were also instructed to be aware
that some of the expressions were idioms, meaning that they could not be translated word
for word. Furthermore, they were told to guess the meaning of words that they were
unsure of, as there would be no penalty for incorrect responses. Finally, they were
reminded to translate only the underlined word or phrase and not the entire sentence.
For clarification purposes, a sample question, presented with an appropriate
response already provided, was included for each of the three texts. Each of these sample
questions was taken from the first line of the text the participants had just read for the
sake of familiarity. As with the MC questions, these questions were presented one at a
time and followed the chronological order in which they appeared in the story. Although
participants saw only one question at a time, they had the option to skip questions and
revisit them later. For the sake of contextualization, each underlined lexical item was
110
presented within a short sentences or phrase taken from the unmodified versions of the
texts. (See Appendix F for lists of the tested lexical item.)
Short Constructed Responses: Scoring
At the conclusion of the research study, the results from the short constructed
response section were downloaded onto an Excel spreadsheet from WebCT for ease of
scoring. Vocabulary data were collected from 204 students although, as with the other
assessment instruments, most students provided vocabulary data twice.
A three-point scale was used to score the participants’ short constructed
responses. Answers that were translated into English correctly received two points,
answers that were incorrectly translated received no points and those deserving of partial
credit received one point. All responses were scored by the primary research, but answers
receiving partial credit were re-scored by a L1 Spanish rater. Initially, 174 studentgenerated responses were given partial credit by the researcher. Of these, the L1 Spanish
rater determined that 85 were worthy of partial credit, 2 worthy of full credit, and the rest
not worthy of any credit. Scoring adjustments were made in accord with the
determinations made by this second L1 Spanish rater.
Chapter III Summary
In this chapter the general description and operationalization of the study were
presented, including a description of the participants, a description of how the reading
texts were selected and modified, and an explanation for the data collection procedure. In
addition, each of the testing methods used in the study were described, including the
rationale for their selection and use, their preparation and piloting, and the methods of
scoring. The results of the research are presented in chapter IV.
111
CHAPTER IV
RESULTS
Introduction
To answer the research questions presented in chapter I regarding whether
comprehension scores of participants reading elaborated L2 literary texts were different
from the scores of readers reading unmodified versions of these texts, research results are
presented in this chapter in the following manner. First the results of the recall protocol
data are given in response to research question 1.1; that is, whether readers of elaborated
versus unmodified literary texts recall propositions from these texts that differ
quantitatively and qualitatively. Differences in propositional recall were measured
quantitatively by (a) word count per version, (b) general estimated equations (GEEs) for
the effect of the overall influence of elaboration on response length, (c) word count per
version per reading text, (d) idea units per version, and (e) idea unit per version per
reading text. Differences in propositional recall are also measured qualitatively by
looking at the percentage of more-essential versus less-essential idea units per text
version. Here version refers to the unmodified versus the elaborated versions, and
reading text refers to Fiebre de lotto, Tiempo libre, or Apocalipsis.
Next, in response to research question 1.2 regarding whether readers of elaborated
versus unmodified versions of literary texts draw different inferences from the texts, the
following results are presented. Readers’ inferencing ability was measured by their
number of correct responses to the multiple-choice (MC) questions. The MC responses
were measured (a) per text version, (b) by means of GEEs to measure for the effect of
modification, (c) per text version per reading text, (d) per question type, (e) per question
type per text version and, finally, (f) per question type per text version per reading text.
Here question type refers to the four categories of elicited responses: (a) the narrator’s
112
point of view, (b) the narrator’s tone, (c) the mood of the character or characters, and (d)
the predominant theme of the story.
In response to research question 1.3, regarding whether readers of elaborated
versus unmodified literary texts differ in their ability to understand words from those
texts, the following results are presented. Readers’ vocabulary knowledge was measured
by the number of correct responses (a) per text version, (b) by means of GEEs to measure
for the effect of modification, and (c) per version per reading text.
Finally, in response to research question 1.4 regarding whether readers of the
unmodified versions of literary texts identify lexical items glossed in Spanish in the texts’
margins differently from readers of the same texts presented in an elaborated, unglossed
version, vocabulary knowledge was measured once again. Here, readers’ vocabulary
knowledge of glossed versus unglossed terms was measured according to the number of
correct responses per version and per reading text.
Results of Recall of Text-Based Information
Number of Words per Text Version
In response to the first facet of the first research question, regarding whether
readers of unmodified versus elaborated versions of literary texts recall different
propositions quantitatively from these texts, word counts were tabulated. Specifically, the
question under consideration was whether students, as a group, were able to recall more
information from the texts they read in the unmodified version or in the elaborated
version. Results revealed that the total recall length for all recalls produced by readers of
unmodified texts was 25,892 words, whereas readers of elaborated texts produced recalls
with a combined word length of 30,029 words. The mean recall length for the unmodified
versions was 135 words and the mean recall length for the elaborated versions was 161
words. In other words, the recalls of the students who read the elaborated texts contained
113
16% more words than those produced by the readers of the unmodified versions. (See
table 4.1.)
Table 4.1. Total Word Count per Text Version
Text Version
N
Total Words
Mean Words
Unmodified
192
25,892
134.85
Elaborated
186
30,029
161.45
General Estimated Equations for Effect of Elaboration
General estimated equations (GEEs) were used on word count totals as, due to the
study design, these counts were neither independent nor repeated or paired. (See table
4.2.) This statistical method is used to study longitudinal data that involve the repeated
measures of individuals over time, which is the case here. General estimated equations
were used, specifically, because data were collected twice from each reader but no reader
produced data from all three reading texts, hence the data collection for individuals was
not consecutive 3. General estimation equations provided an estimate of the correlation
between each individual’s responses and offered an estimate of the effect’s variance.
Here, the specific effect in question is the overall influence of elaboration on the total
number of words. The recall data were analyzed by GEEs to test for the effect of text on
recall length, as well as modification on the individuals’ recalls between texts.
3 For more information on the use and implications for generalized estimated
equations, consult the University of Manitoba website, in particular:
http://www.umanitoba.ca/centres/mchp/concept/thesaurus/thesaurus_G.html#GEE
114
A reliability estimate was computed based on the data from all three data
collection days. A significance level of .05 was used. Here, there is a significant effect for
text type and a significant effect for elaboration on word count (p value<.0001).
Table 4.2. General Estimated Equations for Difference in Word Count Between Text
Versions
Score Statistics for Type 3 General Estimated Equation Analysis
Variable
df
Chi-square
Pr > Chi-square
Text
2
36.34
<.0001
Modification
1
18.20
<.0001
Number of Words per Text Version per Reading Text
Although recall word length was generally higher for readers of the elaborated
text versions, readers of certain texts produced recalls whose disparity in length was
greater than for other texts. (See table 4.3.) The readers of the two versions of Apocalipsis
produced the greatest disparity in recall length with a mean recall length for the
unmodified version of 99 words, in contrast to 135 words in the elaborated version, a
difference of almost 28%. The readers of the unmodified version of Fiebre de lotto
produced recalls with a mean length of 141 words, whereas the readers of the elaborated
version produced recalls with a mean length of 175 words. These figures represent a
difference of 19%. Readers of the unmodified version of Tiempo libre produced more
words than the readers of the elaborated version with a mean length of 167 words in
comparison to a mean length of 160 words, respectively, representing a difference of 4%.
In addition to the difference in word count, standard deviation and standard error
differences were tabulated and are also presented in table 4.3. These numbers indicate
that the greatest statistical disparities occurred between the two versions of Apocalipsis
and, to a lesser extent, between the versions of Fiebre de lotto. There was considerable
115
less variability in word count totals produced by the readers of the two versions of
Tiempo libre.
Table 4.3. Word Count Totals per Text Version per Reading Text
Text Version per
Reading Text
N
Total
Words
Mean
Score
Standard
Deviation
Standard
Error
Percent
Difference
FL (unmodified)
73
10,327
141.47
72.56
8.49
FL* (elaborated)
69
12,083
175.12
79.72
9.59
↑ 19%
TL (unmodified)
56
9,376
167.21
61.50
8.21
↑ 4%
TL* (elaborated)
63
10,314
160.30
60.74
7.65
AP (unmodified)
63
6,189
98.24
33.75
4.25
AP* (elaborated)
56
7,632
136.29
71.61
9.57
↑ 28%
Next, the percent difference between word count per version was analyzed using
independent samples-t test. Tables 4.4, 4.5 and 4.6 illustrate the results obtained through
analyses of the three reading texts.
Table 4.4. Independent Samples t-Test for Word Count Totals: Fiebre de lotto
Levene's Test
for Equality of
Variances
Equal
Variances
Assumed
t-test for Equality of Means
F
Sig.
T
df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
1.928
.167
-2.670
140
.008
-34.13
12.78
116
An inspection of the two groups indicates that the average recall length for
readers of the unmodified version of Fiebre de lotto at 141 words is significantly lower
than the average recall length of 175 words produced by the readers of the elaborated
version at the alpha =.05 level (t=2.67, df=140, p<.008).
Table 4.5. Independent Samples t-Test for Word Count Totals: Tiempo libre
Levene's Test
for Equality of
Variances
Equal
Variances
Assumed
t-test for Equality of Means
F
Sig.
T
Df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
.301
.584
.616
117
.539
6.91
11.22
An inspection of the findings from the two groups who read Tiempo libre
indicates that the average recall length for readers of the unmodified version at 167 words
is not significantly higher than the average recall length of 160 words from the readers of
the elaborated version at the alpha =.05 level (t=.616, df=117, p<.539).
The results obtained from the use of the computer program Statistical Program for
the Social Sciences (SPSS), indicated that the variances were significantly different for
the Apocalipsis results as demonstrated by Levene’s test for the assumption of equal
variances as seen in table 4.6. (Consult Morgan et al., 2004, for further information.)
Therefore, using the results obtained, assuming inequality of variances, the word count
total for readers of the unmodified version at 98 words is significantly lower than the
word count total at 136 words from the readers of the elaborated version for Apocalipsis
at the alpha =.05 level (t=-3.48, df=76, p<.001).
117
Table 4.6. Independent Samples t-Test for Word Count Totals: Apocalipsis
Levene's Test
for Equality of
Variances
Equal
Variances
Assumed
t-test for Equality of Means
F
Sig.
t
df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
17.56
.000
-3.614
117
.000
-36.44
10.68
-3.480
76.23
.001
-36.44
10.47
Equal
Variances Not
Assumed
Word count total, however, can be viewed in another way. Table 4.7 shows word
count totals in relation to the length of texts that the students read. Here, the length of
each text is shown as well as the number of sentences per version, the mean sentence
length, and the mean recall word count. Finally, the length of the recall per text version is
shown by a percentage, in comparison to the length of the text that was read.
Table 4.7. Recall Length Compared to Text and Sentence Length
Text Version per
Reading Text
Text
Word
Count
Total
Number of
Sentences in
Reading
Mean
Sentence
Length in
Words
Mean
Recall
Length
Recall Length
as Compared
to Sentence
Length
FL (unmodified)
587
28
20.27
141.47
24%
FL* (elaborated)
957
30
31.47
175.12
18%
TL (unmodified)
371
20
18.86
167.21
45%
TL* (elaborated)
531
19
26.75
160.30
11%
AP (unmodified)
378
27
14.03
98.24
26%
AP* (elaborated)
615
24
22.68
136.29
22%
118
Of interest is the disparity seen with the text Tiempo libre between the recall
lengths of the two versions when compared to the lengths of the text versions the students
read. The possible implications of these results are further discussed in chapter V.
Total Idea Units per Text Version (Representative Sample)
When comparing the number of idea units recalled by students who read
unmodified texts with the number of idea units recalled by students who read elaborated
versions, only the representative sample of 36 recalls was considered. Idea units were
determined by calculating a mean score from the raters’ totals. As described in chapter
III, the recalls selected to represent the sample consisted of the two longest, two shortest,
and the two most average length recalls of all the recalls produced per version as
determined by word count. When looking at the 18 readers of the unmodified versions,
the total number of idea units produced was 439, whereas the 18 readers of the elaborated
versions produced a total of 593 idea units. These figures indicate that the number of
recalled idea units produced by readers of the elaborated versions was 26% greater than
the number produced by readers of the unmodified versions. (See table 4.8.)
Table 4.8. Idea Unit Totals per Text Version (Representative Sample)
Text Version
N
Idea Units
Mean
Standard
Deviation
Standard Error
Mean
Unmodified
18
439
24.38
17.73
17.73
Elaborated
18
593
32.77
21.37
21.37
Percent
Difference
↑26%
Next, the difference in the number of idea units was analyzed by means of an
independent samples t-test. The results are presented in table 4.9. The independent
samples t-test revealed that the mean difference in number of idea units recalled by the
readers of the elaborated versions is not significantly different from the total number
119
recalled by readers of the unmodified versions at the alpha =.05 level (t=-1.28, df=34,
p<.209).
Table 4.9. Independent Samples-t Test, Total Idea Units per Text Version
(Representative Sample)
Levene's Test for Equality of
Variances
Equal
Variance
Assumed
t-test for Equality of Means
F
Sig,
T
df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
.719
.403
-1.281
34
.209
-8.35
6.55
Idea Units per Reading Text (Representative Sample)
In addition to calculating the number of idea units per text version, idea units per
reading text were also calculated. The descriptive statistics presented in table 4.10 show
mean scores, standard deviation, and standard error of the mean per text.
Table 4.10. Descriptive Statistics for Idea Units per Text (Representative Sample)
Text
N
Mean
Std. Deviation
Std. Error
Fiebre de lotto
12
28.58
25.06
7.23
Tiempo libre
12
31.58
17.72
5.11
Apocalipsis
12
27.17
21.72
6.27
Total
36
29.11
21.16
3.53
The idea units recall per reading by the representative sample were also analyzed by
means of an analysis of variance (ANOVA) to see if there was a significant difference
120
between any two of the readings. (See table 4.11.) The ANOVA revealed no significant
difference between texts at F (2,33)=.129, p=.879.
Table 4.11. ANOVA Results for Idea Units per Reading Text (Representative Sample)
Sum of Squares
df
Mean Squares
Between Groups
122.06
2
Within Groups
15555.50
33
62.02
Totals
15677.56
35
471.38
F
Sig.
.129
.879
Idea Unit Means per Text Version per Reading Text
(Representative Sample)
Again using the representative sample recalls, idea unit totals were calculated
according to reading text, as seen in figure 4.1. Here again, readers of the elaborated
versions consistently recalled more idea units than the readers of the unmodified
versions, although there was considerable variability among text versions and readings.
Although the numbers of idea units recalled from readers of the two versions of Tiempo
libre were similar, the idea unit recalled from readers of Fiebre de lotto and Apocalipsis
showed more variability.
Idea unit means per text version per reading text were also analyzed by means of
an independent samples t-test to assess whether the difference between the average
number of idea units recalled per text version was statistically significant. None of the
three reading texts produced differences that were statistically significant at the alpha
=.05 level (Fiebre de lotto: t=-.392, df=10, p<.703; Tiempo libre: t= -.499, df=10,
p< .629; Apocalipsis: t=-1.049, df=10, p<.319).
121
Figure 4.1. Idea Unit Means per Text Version per Reading Text (Representative Sample)
Mean Idea Unit per Text Version per Reading Text
80
60
Idea Unit Means 40
20
0
FL(s)
Unmodified
Elaborated
FL(s)
4.5
12.5
FL(a)
FL(a)
12
24.5
FL(l)
TL(s)
FL(l)
51
65
TL(a)
TL(l) AP(s)
Reading Text
TL(s)
8
12.5
TL(a)
35
35
AP(a)
TL(l)
46.5
52.5
AP(l)
AP(s)
7
9.5
AP(a)
14
25
AP(l)
33.5
66.5
Note: (s) refers to the shortest recalls, (a) to the most average, and (l) to the longest
Notably, there is a considerable difference between the mean number of idea units
recalled by readers of certain text versions. Most conspicuous is the mean number of idea
units recalled by readers of the unmodified version of Fiebre de lotto, 4.5, in contrast to
the elaborated version at 12.5. The difference, however notable, must be viewed in light
of the fact that the means were obtained from a small sample of two recalls. For this
reason, these scores were not subjected to further analysis.
When the mean number of idea units was viewed according to the text-length
criteria employed to select the representative sample, the readers of the shortest
unmodified versions recalled a total mean of 6.5 idea units, in comparison to a mean of
11.5 idea units for the elaborated versions. In a similar fashion, the readers of the most
average length unmodified versions mean recall was 20 idea units per text in comparison
to a mean of 28 idea units for the readers of the elaborated versions. Finally, the mean
122
number of idea units per recall for the readers of the longest unmodified versions was 44
idea units whereas the mean number of idea units per recall for readers of the longest
elaborated versions was 61. At no time was the mean number of units recalled greater for
the unmodified versions than that of their elaborated-version counterparts, although
readers of the most average elaborated and unmodified versions recalled, on average, the
same number of idea units at 35. (See figure 4.2.)
Figure 4.2. Mean Number of Idea Units per Text Version per Short, Average and Long
Recall (Representative Sample)
Mean Idea Units per Text Version per Recall Length
80
60
Idea Unit Mean
40
20
0
Short
Average
Long
Recall Length
Unmodified Idea Unit
Averages
Elaborated Idea Unit
Averages
Short
Average
Long
6.5
20
44
11.5
28
61
The average numbers of idea units recalled among reading texts can also be
viewed as percentages. Readers of the elaborated versions recalled an average of 46%
more idea units after reading the Fiebre de lotto text, an average of 16% more idea units
after reading the Tiempo libre text, and an average of 40% more idea units after reading
123
the Apocalipsis text, all in comparison to those who read unmodified versions of the
respective texts.
Essential Idea Unit per Text Version
(Representative Sample)
As described in chapter III, idea units were also categorized as those considered
more or less essential to overall text comprehension. Following this categorization, the
total number of idea units recalled in relation to the total number of essential idea units
recalled was compared between text versions. In table 4.12 the number of idea units from
the representative sample of 36 recalls was compared to the predetermined percentage of
more-essential idea units in relation to total idea units for all three texts: 52%. Of the idea
units recalled by readers of the unmodified texts versions, 66.8% were more essential to
text meaning. In contrast, the idea units recalled by readers of the elaborated versions
contained 63.5% more-essential units.
Table 4.12. More-Essential versus Less-Essential Recalled Idea Units per Text Version
(Representative Sample)
Text Version
Percent of More-Essential Units
Unmodified
66.8%
Elaborated
63.5%
All Three Texts in Unmodified Form
52.3%
The mean difference between these two percentages was then analyzed by means
of an independence samples t-test. The results, as displayed in table 4.13 show no
significant difference between the groups at the alpha =.05 level (t=-.274, df=16, p<.788).
124
Table 4.13.Independent Samples t-test for More-Essential versus Less-Essential Recalled
Idea Units per Text Version (Representative Sample)
Levene's Test
for Equality of
Variances
Equal
Variance
Assumed
t-test for Equality of Means
F
Sig,
T
Df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
.291
.597
-.274
16
.788
-1.889
6.90
Results of Multiple-Choice Inferential Information
Correct Responses per Text Version
In response to the second part of the research question regarding whether readers
of elaborated versus unmodified literary texts draw different inferences from the texts, a
number of research results are presented. The initial results report the number of correct
multiple-choice (MC) responses produced by readers of the unmodified versus elaborated
versions of all three reading texts. Altogether, students who read unmodified versions of
these texts produced 447 correct answers out of a total of 772 potential correct responses.
Conversely, readers of the elaborated texts versions produced 464 correct responses out
of a total of 756 potential correct responses. (See table 4.14.) Viewed as percentages,
readers of the unmodified versions correctly responded correctly 58% of the time,
whereas readers of the elaborated versions responded correctly 61% of the time.
Table 4.14. Correct Multiple-Choice Responses per Text Version
Text Version
N
Correct
Reponses
Potentially Correct
Responses
Percentage
Correct
Unmodified
193
447
772
58%
Elaborated
189
464
756
61%
125
These figures can also be describes on a 4-point score, as the value of each correct
answer equaled 1 point: Out of a potential score of 4.00 points, readers of the unmodified
versions obtained an average score of 2.34 points, whereas readers of the elaborated
versions obtained an average score of 2.45 points.
General Estimated Equations for Effect of Elaboration
As with total word count per version, general estimated equations (GEEs) were
used on the MC data as, viewed across texts, these data were neither independent nor
repeated or paired. (See table 4.15.)
A reliability estimate was computed based on the data from all three data
collection days. Here, there is a significant effect for text type (p <.0001), but elaboration
did not result in a significant effect for MC scores (p <.3109).
Table 4.15. General Estimated Equations for Difference in MC Results Between Text
Versions
Score Statistics for Type 3 General Estimated Equation Analysis
Variable
df
Chi-square
Pr > Chi-square
Text
2
18.28
0.0001
Modification
1
1.03
0.3109
Correct Responses per Text Version per Reading Text
The above-mentioned scores were also analyzed for each reading text. Here,
results indicate that there was little difference in scores obtained by readers of the
unmodified versions when compared to readers of the elaborated versions of each text.
Using the mean total of correct responses per text version per reading text to compare
these differences, the scores of students who read Tiempo libre and those who read
Apocalipsis were almost identical. The descriptive statistics for the MC responses per text
126
version per reading text are presented in table 4.16 and show that the readers of both
versions of Tiempo libre obtained the highest scores; the lowest scores were from readers
of the unmodified version of the Fiebre de lotto text.
Table 4.16. Multiple-Choice Totals per Text Version per Reading Text
Text Version per
Reading Text
N
Total
Correct
Mean
(4-point scale)
Standard
Deviation
Standard
Error
FL (unmodified)
69
139
2.03
1.071
.129
FL* (elaborated)
72
170
2.36
1.066
.126
TL (unmodified)
60
165
2.75
1.068
.138
TL* (elaborated)
62
171
2.76
.987
.125
AP (unmodified)
64
142
2.23
.842
.106
AP* (elaborated)
55
123
2.24
.838
.113
Another way to look at these same results is to view them on 4-point scale. This is
demonstrated figure 4.3. Here it can be seen that the readers of Tiempo libre obtained a
mean score of 2.75 points after reading the unmodified version and a mean score of 2.76
points after reading the elaborated version of a total of 4 possible points. The mean scores
produced by readers of the Apocalipsis text versions were similar as well with the readers
of the unmodified version obtaining a mean of 2.23 out of 4.00 compared with readers of
the elaborated version whose mean score was 2.24. There was, however, more disparity
in scores obtained from readers of the two versions of Fiebre de lotto. Here, readers of
the unmodified version obtained a mean score of 2.03 out of 4.00 whereas the readers of
the elaborated version scored a mean of 2.36. Although the readers of the elaborated
versions consistently out-scored those reading the unmodified versions, at times the
difference was minimal, as is demonstrated in the graphic.
127
Figure 4.3. Correct Multiple-Choice Responses per Text Version per Reading Text
Correct Multiple Choice Responses per Text Version per Reading Text
4
3.5
3
Unmodified
Elaborated
2.5
Possible Points 2
1.5
1
0.5
Elaborated
Unmodified
0
FL
TL
AP
Reading Texts
To further examine the disparity in responses to MC questions obtained by
readers of the two versions of Fiebre de lotto, these data were analyzed through the use
of an independent samples t-test to test for the statistical significance of the difference.
Here, only the disparity in results between the readers of the two versions of Fiebre de
lotto were tested as the difference in results was greatest from readers of these two text
versions. The analysis is presented in table 4.17. Results reveal that the mean difference
between correct responses to the MC questions for Fiebre de lotto approaches but does
not reach significance at the alpha =.05 level (t=-1.846, df=139, p<.067).
128
Table 4.17. Independent Samples t-Test for MC Responses: Fiebre de lotto
Levene's Test
for Equality of
Variances
Equal
Variance
Assumed
t-test for Equality of Means
F
Sig,
T
df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
.99
.321
-1.846
139
.067
-.33
.18
Correct Responses per Question Type
Another result of interest to this study is the number of correct responses per
question type. Here, as previously described, question type refers to the four categories of
questions: (a) the narrator’s point of view, (b) the predominant theme of the story, (c) the
mood of the character or characters, and (d) the narrator’s tone. Specifically, results are
intended to reveal which of the four question types students found the most and the least
difficult as determined by their percentage of correct responses. Students’ scores indicate
that questions addressing narrator point of view were the easiest to answer, followed by
questions regarding narrator tone, the stories’ predominant theme and, finally, the mood
of the characters or characters. On average, students were able to correctly identify the
narrator’s viewpoint 73% of the time, the story’s predominant theme 66% of the time, the
mood of the characters 47% of the time, and narrator’s tone 53% of the time. Table 4.18
indicates these percentages.
Table 4.18. Average Correct Responses to Multiple-Choice Questions per Question Type
Narrator’s
Viewpoint
Story’s Predominant
Theme
Mood of
Character(s)
Narrator’s Tone
73%
66%
47%
53%
129
Correct Responses per Question Type per Text Version
Correct responses to MC questions were also analyzed by question type per text
version. Table 4.19 shows these results viewed as percentages, with the higher percentage
per version displayed in bold type. The percent difference between the text versions is
presented as well.
Table 4.19. Average Correct Responses to Multiple-choice Questions per Question Type
per Text Version
Question Type
Percent Correct
Unmodified
Percent Correct
Elaborated
Percent Difference
Narrator’s Viewpoint
72
74
2%
Story’s Predominate Theme
62
69
7%
Mood of Character(s)
50
45
5%
Narrator’s Tone
49
57
8%
Reviewing these results, one sees that no matter the text version, the questions
dealing with the narrator’s viewpoint were the easiest for students, followed by the
questions addressing the story’s theme. Of particular interest are the questions that
targeted both the mood of the character(s) and the narrator’s tone. Here readers of the
unmodified versions had less trouble identifying the mood of the characters but more
difficulty identifying the narrator’s tone. Furthermore, questions dealing with narrator’s
tone and the story’s predominant theme created the greatest disparity in respondents’
results between the two groups of readers, with the readers of the elaborated versions outscoring the readers of the unmodified versions by 7% and 8%, respectively.
130
Correct Responses per Question Type per Text Version per
Reading Text
Percentages of correct responses per question type were also viewed not only
depending on text version but also according to reading text. These results are presented
in figures 4.4, 4.5, and 4.6.
Readers of both versions of Tiempo libre out-performed the readers of both
versions of Fiebre de lotto and Apocalipsis, with the exception of the question concerning
narrator’s tone in Apocalipsis. On that question, readers of both versions of Apocalipsis
performed better than the readers on any other question no matter the version or text. The
greatest discrepancy in scores between text versions occurred with answers obtained from
readers of Fiebre de lotto, specifically those questions dealing with the narrator’s tone
and story theme. Here the readers of the elaborated version out-scored the readers of the
unmodified version by 24% and 16%, respectively. Readers of the elaborated version of
Fiebre de lotto, however, were not consistently more accurate than their counterparts on
all questions. In fact, at no time did any one group of readers of any one text version
consistently out-score the readers of the corresponding version on all four questions,
although readers of three of the elaborated versions out-performed the readers of the
corresponding unmodified version when answering three out of four questions.
Statistically speaking, however, there was no difference between the accuracy of
performance of any group of readings on any MC question whether reading an
unmodified and an elaborated text version.
Issues of question difficulty also come into play when interpreting these results.
As can be noted in figure 4.6, the discriminatory strength of the MC questions,
particularly the questions used after the Apocalipsis reading, are questionable.
131
Figure 4.4. Mean Number of Correct Responses for MC Questions per Question Type:
Fiebre de lotto
Percentage Correct Responses MC
Fiebre de lotto
100
90
80
68
70
69
71
64
60
Percentage 50
55
45
40
Unmodified
Elaborated
38 28
30
20
10
Elaborated
0
Viewpoint
Unmodified
Tone
Mood
Theme
Figure 4.5. Mean Number of Correct Responses for MC Questions per Question Type:
Tiempo libre
Percentage Correct Responses MC
Tiempo libre
100
90
79
80
78
74
70
74
72
60
58
Percentage 50
67
48
Unmodified
40
Elaborated
30
20
10
Elaborated
0
Viewpoint
Tone
Unmodified
Mood
Theme
132
Figure 4.6. Mean Number of Correct Responses for MC Questions per Question Type:
Apocalipsis
Percentage Correct Responses MC
Apocalipsis
100
91
90
75
80
84
73
70
60
Percentage 50
Unmodified
40
39
30
33
Elaborated
25
20
25
10
Elaborated
0
Viewpoint
Tone
Unmodified
Mood
Theme
Results of Vocabulary Knowledge
Correct Responses per Text Version
The results presented here are in response to the third part of the research question
that asks whether readers of elaborated and unmodified literary texts, respectively, differ
in their ability to understand words in the texts. Specifically, the question under
consideration is how well the readers of the unmodified text versions were able to
identify the meaning of the tested lexical items when compared to the readers of the
elaborated versions of the same texts.
Initially, to answer this question, the total scores of all readers of the three
unmodified versions were compared to the total scores of all readers of the three
elaborated versions. As discussed in chapter III, correct responses were awarded 2 points,
incorrect responses received no points and those responses meriting partial credit
133
received 1 point. These scores were then divided by the potential score; that is, the score
that would have awarded assuming every questions was answered correctly. These results
are presented as percentages in table 4.20.
If every student reading the three unmodified versions had answered each
vocabulary question correctly, the potential score for all sections was 6,840 points. In
reality, readers of the unmodified versions scored a total of 1,736 points, representing an
accuracy rate of 25.4%. The potential total score of all students reading the elaborated
versions was 6,804 points. In reality, these readers scored a total of 2,122 points
representing an accuracy rate of 31.2% accuracy.
Table 4.20. Correct Vocabulary Responses per Text Version
Text Version
N
Total
Points
Total Potentially
Points
Percentage
Correct
Unmodified
190
1,736
6840
25%
Elaborated
189
2,122
6804
31%
Percent
Difference
↑6
General Estimated Equations for Effect of Elaboration
Next, correct vocabulary responses per version were then analyzed by means of
general estimated equations (GEEs) to test the overall effect of elaboration on vocabulary
knowledge. (See table 4.21.) The recall data were analyzed by GEEs to test for the effect
of text as well as modification on the individuals’ results across texts.
A reliability estimate was computed based on the data from all three data
collection days. Here, there is a significant effect for text type (p <.0001) and a
significant effect for elaboration on vocabulary scores (p <.0001).
134
Table 4.21. General Estimated Equations for Difference in Vocabulary Knowledge
Between Text Versions
Score Statistics for Type 3 General Estimated Equation Analysis
Variable
df
Chi-square
Pr > Chi-square
Text
2
35.25
<.0001
Modification
1
21.87
<.0001
Correct Response per Text Version per Reading Text
To determine the percentage of correct responses per text version per reading text,
the accuracy rate previously described was further analyzed. The figures in table 4.22
were obtained by taking the total number of points earned by readers divided by total
possible points. In every instance the readers of the elaborated versions out-preformed the
readers of the unmodified versions. The greatest disparity occurred between the readers
of the two versions of Apocalipsis, where the readers of the elaborated version out-scored
the readers of the unmodified version by 13%. Readers of the elaborated version of
Fiebre de lotto earned scores that were 5% higher than readers of the unmodified version
of the same text, and readers of the elaborated version of Tiempo libre received scores
that were 1% higher than the readers of the unmodified version.
Table 4.22. Vocabulary Scores per Text Version per Reading Text (In Percent)
Text Version per
Reading
N
Correct Scores/
Potential Scores
Percentage
Correct
FL (unmodified)
69
646/2484
26%
FL* (elaborated)
74
828/2664
31%
TL (unmodified)
56
417/2016
21%
TL* (elaborated)
60
470/2160
22%
AP (unmodified)
65
673/2340
29%
AP* (elaborated)
55
824/1980
42%
Percentage
Difference
↑ 5%
↑ 1%
↑ 13%
135
Table 4.23 shows this information in a different form. Here, vocabulary scores are
presented as mean number of correct responses, including the standard deviation and
standard error results.
Table 4.23. Vocabulary Scores per Text Version per Reading Text
Text Version per
Reading Text
N
Mean
(36 points possible)
Standard
Deviation
Standard Error
FL (unmodified)
69
9.39
6.02
.724
FL* (elaborated)
72
11.18
6.45
.749
TL (unmodified)
56
7.44
6.94
.928
TL* (elaborated)
60
7.83
5.49
.709
AP (unmodified)
65
10.35
6.89
.855
AP* (elaborated)
55
14.98
7.42
1.001
To further analyze the disparity in results obtained from vocabulary results
between the two text versions of each text, independent samples t-tests were conducted
for each of the readings. Results are presented in tables 4.24, 4.25, and 4.26.
Table 4.24. Independent Samples t-test for Vocabulary Results per Text Version: Fiebre
de lotto
Levene's Test for
Equality of Variances
Equal
Variance
Assumed
t-test for Equality of Means
F
Sig,
t
df
Sig.
(2-tailed)
.797
.373
-1.720
141
.088
Mean Difference Std. Error Difference
-1.797
1.045
136
Results reveal that the mean difference between the vocabulary responses given
by readers of the elaborated version of Fiebre de lotto approaches, but does not reach, a
level of significant different when compared to the responses given by readers of the
unmodified version of the same text at the alpha =.05 level (t=-1.72, df=141, p<.088).
Table 4.25. Independent Samples t-test for Vocabulary Results per Text Version:
Tiempo libre
Levene's Test
for Equality of
Variances
Equal Variance
Assumed
t-test for Equality of Means
F
Sig.
t
df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
1.063
.305
-.334
114
.739
-.3869
1.158
Results reveal that the mean difference between the vocabulary responses given
by readers of the elaborated version of Tiempo libre are not significantly different than
the responses given by readers of the unmodified version of the same text at the alpha
=.05 level (t=-.334, df=114, p<.739).
Table 4.26. Independent Samples t-test for Vocabulary Results per Text Version:
Apocalipsis
Levene's Test for Equality
of Variances
Equal
Variance
Assumed
t-test for Equality of Means
F
Sig.
t
df
Sig.
(2-tailed)
Mean
Difference
Std. Error
Difference
.819
.367
-3.538
118
.001
-4.628
1.307
137
Here, results reveal that the mean difference between the vocabulary responses
given by readers of the elaborated version of Apocalipsis is significantly different from
the responses given by readers of the unmodified version of the same text at the alpha
=.05 level (t=-3.538, df=118, p<.001).
Correct Response for Glossed Items per Text Version
Finally, in response to research question 1.4 regarding whether readers of the
unmodified versions of literary texts identify lexical items glossed in Spanish in the texts’
margins differently from readers of the same texts in which the same lexical items were
presented in an elaborated, unglossed form, responses to glossed items were measured.
When evaluating the performance of readers of both text versions in relation to their
correct responses to glossed vocabulary items, item selection was limited and included
only those words glossed in Spanish in the margins of the unmodified text versions. For
this reason, readers’ scores from 11 lexical items out of 54 total items were considered
here. Table 4.27 illustrates these results. Averaging between text versions, the readers of
the unmodified texts correctly identified the items that were glossed in Spanish in the
margins 29% of the time. In contrast, the readers of the elaborated versions, without
access to the Spanish gloss in the margin, correctly identified these items 31% of the
time. On average, the items glossed in Spanish in the unmodified version were correctly
identified more often in the elaborated version but this did not hold true for every glossed
lexical item. In fact, 7 out of the 11 total lexical items were more frequently identified
correctly by readers of the unmodified versions but by a smaller degree than those readers
of the elaborated versions. Higher percentages for lexical items per version are indicated
by bold type.
138
Table 4.27. Correct Response for Glossed Items per Text Version and per Reading Text
Reading Text
Glossed Items
Fiebre de lotto
quedarían en la calle
28%
51%
23%
velitas
16%
15%
1%
pestes
38%
36%
2%
ni en las curvas
4%
1%
3%
rictus
27%
36%
9%
importe
29%
31%
3%
hormigas
68%
55%
7%
Me flaquearon
22%
14%
8%
se les antoje
46%
45%
1%
harán caso omiso
29%
32%
3%
forcejeo
19%
17%
2%
Tiempo libre
Apocalipsis
Percent Correct Percent Correct
(Unmodified)
(Elaborated)
Percent
Difference
Speaking generally, the readers of the elaborated versions performed as well as
the readers of the unmodified versions with access to the marginal glosses, although
viewed item by item, the direct influence is not as clear. The interpretation of these
findings is addressed in chapter V, as well as limitations and implications.
Summary Chapter IV
In this chapter, research results were presented in response to the four research
questions regarding the effect of textual elaboration on readers’ ability to recall
information from the text, make inferences beyond the text, and identity vocabulary
words, both glossed and unglossed, that appear in the unmodified texts versions. In
chapter V the results are described in more detail and include the implications for practice
and further research. Study limitations are also discussed.
139
CHAPTER V
DISCUSSION
Introduction
The purpose of this study was to analyze textual elaboration as a modification
option for literary texts in an intermediate Spanish language course. It was suggested that
literary texts might obtain a more predominant place in second language instruction,
especially for readers with lower levels of language proficiency, if learners were supplied
with additional tools to facilitate comprehension. Elaboration, it was suggested, might be
a tool that could ease the type of comprehension difficulty that occurs due to what Beck
et al. (1991) referred to as text failure; that is, the difficulties readers encounter when the
text’s structure is inaccessible or unfamiliar to them. Here, readers’ response to textual
elaboration was tested and measured.
In this chapter, summaries of the research results are presented sequentially
following the order of the research questions. Each summary includes a discussion of the
theoretical and practical implications of the relevant results. This discussion is followed
by a description of study limitations as well as recommendations for future research
possibilities incorporating texts that have been modified elaboratively.
Summary of Results: Recall Protocol Quantitative
The first facet of the research question investigated here asks whether L2 learners’
comprehension scores differ after reading a literary text in an elaborated version
compared to the scores of readers of the same text in an unmodified form. A recall
protocol was used to measure comprehension. This question was divided to access both
quantitative and qualitative differences in readers’ responses. Quantitatively, the amount
of information that the readers of the elaborated versions were able to recall proved
significantly greater than the amount of information that the readers of the unmodified
versions were able to recall. Moreover, it should be noted that the readers of the
140
elaborated versions read longer texts and produced longer recalls than the readers of the
unmodified versions within the same 50-minute class period.
It should not be assumed from these results, however, that more information
necessarily means that the writers of these protocols displayed better or qualitatively
superior comprehension. In fact, Hammadou Sullivan (2001) noted that during the
written recall protocol task that she conducted, at times the readers who were the most
successful at retaining the largest number of ideas units did so with the fewest words. She
argued that precision of writing or conciseness of thought may be related to these readers’
ability to recall information.
But, as previously mentioned, readers cannot recall information that they have not
first understood. In other words, the immediate recall protocol requires that the reader
comprehend the text well enough to be able to recall it in a coherent and logical manner
(Berkemeyer, 1989). Thus, at least on the most basic level, an argument can be made,
provided that the readers of the elaborated texts were not more redundant writers than the
students who read the unmodified versions, that the recalls produced by readers of the
elaborated texts contained more information than was contained in the recalls produced
by the readers of the unmodified versions. A question of interest, therefore, is why
readers of the elaborated texts recalled more information.
Working with L1 reading texts, Beck et al. (1991) and Britton and Gulgoz (1991)
found that readers of more coherent texts--an elaboration strategy implemented here-produced longer recalls than those who read less coherent texts. Horiba (1993) came to a
similar conclusion. She found that when texts were altered so that events and ideas were
causally related, the readers remembered them better in comparison to texts that had not
been altered. It can be argued, therefore, that the elaborated versions of the texts became
more memorable for readers and that the information they contained was retrieved faster
and more completely.
141
Additionally, it could be argued that the increased length of the elaborated texts
may have contributed to the greater length of learner recalls. Through this line of
reasoning, it may be argued that readers recalled more because they read more. But the
text information added through elaborative modifications did not add any new
information to the text message. So as long as the recalls did not mirror the elaborated
versions through use of clarification and redundancy, techniques of little use for students
narrating in their L1, it does not follow that the longer recalls were the result of text
length.
As discussed in chapter II, many L2 reading researchers believe that the longer
the text, the greater the opportunity for readers to activate metacognitive strategies that
will assist their processing and comprehension (Day & Bamford, 1998; Kintsch & van
Dijk, 1987; Maxim, 2002). This opportunity offered by longer texts may help explain
why readers of the elaborated versions were able to out-perform their unmodified-text
counterparts in regard to the amount of information recalled. Readers not only read the
information once but, because of restatments, clarifications, and redundancy, they were
given opportunities to process the same or similar information a second time.
It was the position of Parker and Chaudron (1987) that elaborative text
modifications produced “cognitively simplified forms” (p.108) given that the ultimate
goal of elaboration is to ease the processing burden so that more text meaning can be
decoded and thus understood. The present study supports their conclusion. The amount
that readers were able to comprehend increased because fewer attentional or controlled
processes were needed to decipher the message during the reading process and, therefore,
the readers’ energy could focus on the construction of a conceptual representation of the
text rather than on the linguistic decoding of text elements. Although the information that
readers recalled would be likely classified by Kintsch (1994) as information remembered
rather than learned, in terms quantity of response, elaborative modifications positively
affected readers’ ability to retrieve and reproduce the information they had just read.
142
Before turning to the quality of information that was recalled, the variation in
recall length between versions and among texts should be noted. Without a doubt, the
greatest difference in recall length involved those recalls from readers of the two versions
of Apocalipsis. During the think-aloud protocols, students commented on the difficulty
level of this text to a greater degree than the difficulty level of the other two texts. Recall
length disparity was also notable between the two versions of Fiebre de lotto, but the
length of recalls produced by readers of the two versions of Tiempo libre, arguably the
easiest text from a linguistic perspective, was identical.
One could argue that because of the relative linguistic and thematic simplicity of
Tiempo libre (i.e., written in first person, present tense, deals with newspaper reading) the
number of words recalled from the two versions was similar because most readers
comprehended the story. The simplicity of the story obviated the need for elaborative
alterations because the readers understood well enough without modification. In support
of this hypothesis, one need only look at the disparity between recall length in relation to
text length, which was presented in table 4.7. The mean length of the recalls produced by
readers of the two versions of Fiebre de lotto and Apocalipsis was about 22% of the
length of the respective version and text. These means change, however, in the case of
Tiempo libre. Readers of the unmodified version produced recalls that were 45% as long
as the unmodified text version. In contrast, the mean length of the recalls produced by the
readers of the elaborated version was 11% as long as the elaborated text version. But
readers of the two versions of Tiempo libre produced recalls of essentially the same
length; that is, recalls of approximately 164 words. This occurrence contrasts with the
results from the other two reading texts, in which there was a correlation between text
length and recall length. With Tiempo libre, it could be argued, when judging from recall
length alone, that the effort involved in elaborating the text did not pay off in students’
recall results, which suggests that elaborative modifications should be used judiciously,
143
as elaborative modifications have more impact on students with texts that are more
difficult.
Cognitively speaking, elaboration, although not having a detrimental effect on
comprehension, did nothing to enhance the controlled processes necessary to maximize
comprehension. Readers’ attentional resources were only minimally engaged in the task
of comprehension because the text was comprehensible without the need to activate
multiple cognitive strategies or controlled thought processes. The extra language added
during elaboration did not help readers to understand the general text message to any
greater degree and, as will be discussed in the sections dealing with incidental
information and vocabulary, did not help readers on various other levels either.
Summary of Results: Recall Protocol Qualitative
The qualitative results described here were based not only on the quality but also
on the quantity of idea units recalled. As described in chapter III, the idea units were
identified through naturally occurring reading pauses and were weighted in relation to
their importance to the overall text message. It should be noted at the outset of this
section that the small sample size used to arrive at the qualitative results puts severe
limitations on the implications that can be inferred from the findings. With that caveat in
mind, statistically speaking, there was no difference in the type or number of idea units
recalled from readers of the unmodified versions versus the readers of the elaborated
versions of the three texts. Although the readers of the elaborated versions recalled, in
toto, 593 idea units compared to 439 total idea units by readers of the unmodified
versions, this difference is attributable, at least in part, to the fact that the recalls from
students reading the elaborated versions were, on average, longer. For this reason, it is
logical to assume that longer recalls would consist of more idea units.
Recall length should also be kept in mind when looking at the total number of
idea units per reading. Here there was no statistical difference among texts when
144
comparing the mean number of idea units produced per reading, although the readers of
Tiempo libre, on average, recalled more total idea units after reading the two versions
with the least variation in recall length among all recall scores (see table 4.10). This
finding indicates that the readers found this text to be more comprehensible than the other
two, with or without elaborative modifications.
Another finding from this research relates to the role of syntactic complexity in
reading comprehension. As was demonstrated in table 4.7, readers of the elaborated
versions of the three texts identified as many idea units as did readers of the unmodified
versions of the texts, even though the sentences in the elaborated versions contained, on
average, nine more words per sentence. This result supports the findings of various other
researchers, who contend that syntactic complexity does not necessarily predict L2 text
difficulty as long as the text is coherent and the lexicon comprehensible (Barnett, 1986;
Oh, 2001; Parker & Chaudron, 1987; Ross et al., 1991; Yano et al., 1995).
The number of idea units was also analyzed according to the recall length criteria
used for selection of the representative sample. Despite the disparity between the mean
number of idea units recalled by readers of various texts and versions, the differences
were not significant. It should be noted, however, that these averages were calculated
from a sample of only two readers per group; that is, the long, most average, and short
recall groups.
In regard to the weighting of idea units, as expected, all readers in the
representative sample recalled a greater percentage of more-essential idea units than lessessential idea units (see Bernhardt, 1991b; Johnson, 1970). There was no difference,
however, in the ratio of more-essential to less-essential idea units per text version or per
text version per reading. Specifically, readers of the unmodified versions recalled a
higher percentage of more-essential idea units (66.8%) than did readers of the elaborated
versions (63.5%). Although not statistically significant, readers of the unmodified
versions did produce a greater percentage of more-essential idea unit, which may indicate
145
that they retained more information of greater importance to overall text comprehension.
On the other hand, the greater percentage may suggest that the readers of the elaborated
versions consistently recalled at least as many more-essential idea units as their
counterparts because they recalled more idea units overall, so the percentage of moreessential idea units was lower. As it was beyond the scope of this study to match specific
idea units among recalls, it is difficult to interpret these results without additional
analysis.
Summary of Results: Multiple-Choice Questions
In response to the second facet of the research question regarding whether readers
of elaborated versus unmodified literary texts draw different inferences from the texts,
results indicated that there was no difference in inferencing ability between the
participants who read the unmodified versions and those who read the elaborated
versions. Overall, the readers of the unmodified versions correctly responded 58% of the
time to the MC questions, whereas the readers of the elaborated versions responded
correctly 61% of the time. When viewed according to reading text, those reading Tiempo
libre, no matter the version, out-scored readers of both of the other texts, which supports
the assertion that this text was the easiest for readers to understand. Viewed as
percentages, the scores from readers of the two versions of Tiempo libre were, for all
practical purposes, identical, as were the scores from readers of the two versions of
Apocalipsis. The only variation in scores occurred between the versions of Fiebre de
lotto. Although the mean scores from the readers of the elaborated version were higher
(i.e., 2.36 versus 2.06 for the unmodified version), the difference was not statistically
significant.
When looking at MC responses per questions type, the hierarchy for question
difficulty followed Fecteau (1999), except that readers found that questions that required
the identification of the mood of the character(s) more challenging than those asking
146
them to identify the narrator’s tone. Interestingly, the questions dealing with the story’s
predominant themes were answered correctly more often by the readers of the elaborated
versions, supporting the notion that elaborative changes help readers keep the main idea
or storyline of the text in mind (Parker & Chaudron, 1987). Additionally, the readers of
the elaborated versions were more likely to respond correctly to the questions asking
them to identify the word that best fit the tone of the person telling the story. The only
question type where readers of the unmodified versions out-performed their counterparts
were those questions that dealt with the characters’ moods.
Even though readers of the elaborated versions, on average, out-scored the readers
of the unmodified versions on three out of the four question types and on three out of the
four questions per reading, the differences between the groups were not statistically
significant. All readers appear to have been unable to access the higher-order subskills
required to make the connections necessary to infer meaning beyond the text
(McLaughin, 1987). This helps to explain why, when looking at the accuracy rate for the
MC questions from all readers tested, accuracy did not reach 60%.
Cognitively speaking, therefore, it can be concluded that the elaborative
modifications did not prove robust enough to influence the ability of readers to infer
meaning beyond the literal text-based information presented to them. It can be argued,
however, that the elaborative modifications did not diminish inferencing performance,
given that the readers of elaborated versions out-scored their counterparts on all three
texts, albeit at times to a minor degree.
Summary of Results: Incidental Vocabulary Learning
Perhaps the most interesting findings from the study relate to the third facet of the
research question, which asks whether L2 learners’ vocabulary comprehension scores
after reading elaborated literary texts differ from scores of readers of the same texts in
unmodified form. Statistically speaking, vocabulary scores from readers of the elaborated
147
versions were significantly higher than the scores of readers of the unmodified versions.
Although these vocabulary results, like the results produced from the recall protocols and
the MC measures, did not differ as much for the readers of Tiempo libre, across texts
elaborative modifications did influence vocabulary results. Readers of the elaborated
versions out-performed the readers of the unmodified versions with the greatest disparity
in results occurring between readers of the two versions of Apocalipsis (a 13%
difference), followed by the readers of Fiebre de lotto (5%) and, finally, by the readers of
Tiempo libre (1%).
Taking into account these differences, generally speaking, most readers correctly
identified only a fraction of the vocabulary items tested, ranging from a high score of
42% from the readers of the elaborated version of Apocalipsis to a low score of 21% from
the readers of the unmodified version of Tiempo libre. Of particular interest were the
vocabulary results from readers of the two versions of Tiempo libre, especially
considering the superior performance of these readers on all of the other assessment
measures. In regard to vocabulary, the readers of Tiempo libre understood fewer of the
targeted words than did the readers of the other two texts. This finding adds support to
Paribakht and Wesche (1999), who contend that the likelihood that a reader will acquire
new vocabulary while reading relates to how well he or she is able to understand the text.
The optimal text for vocabulary learning, according to Paribakht and Wesche, would be
one that challenges but does not overwhelm the reader. Under these conditions, the reader
becomes aware of the importance of certain words and therefore attends to them, but
without expending undue attentional energy that might distract from the main reading
objective, that of message comprehension. If a text is too easy for optimal vocabulary
learning, which was likely the case with Tiempo libre, comprehension of the text message
is frequently possible even if the reader chooses to ignore unknown words.
Once again, this argument regarding text difficulty relates to the readers’ ability to
utilize their attentional or controlled processes most effectively and efficiently. Results
148
obtained from readers of the Tiempo libre text versions can be used to exemplify how
readers allocate cognitive resources. Despite the variation in text length between the two
versions of Tiempo libre (unmodified: 371words; elaborated: 531 words), the recall
lengths, the number of idea units recalled, and the number of correct MC responses all
point to better text message comprehension by the readers of both version of Tiempo libre
in contrast to readers of the other two texts. When looking at the identification of
vocabulary words, however, the readers of Tiempo libre correctly identified only 21.5%
of the vocabulary, whereas readers of Fiebre de lotto were able to identify 28.5% of the
vocabulary and the readers of Apocalipsis, arguably the most difficult text, correctly
identified 35.5% of the vocabulary items. These results can be explained by the amount
of attention that readers paid to the unknown lexical items. The ease of text
comprehensibility allowed readers of Tiempo libre to pay less attention to unknown
words because the text message was understandable without the need to attend to the
unfamiliar words and phrases.
The primary goal of this study, however, was to see whether elaborative
modifications increased a readers’ text comprehension. Viewed in this light, incidental
vocabulary acquisition was a by-product of reading for meaning. Although it is
heartening to see that the readers of Apocalipsis, especially in the elaborated form,
identified more vocabulary than the readers of the other two texts, the recalls from these
same readers were the shortest and these students were out-scored on the MC questions
by all readers except those who read the unmodified version of Fiebre de lotto. How
readers allocate their attentional resources, whether primarily to the task of message
maintenance or to the task of vocabulary comprehension, is worthy of consideration
regardless of whether the texts were modified or not, as readers must process texts with a
finite supply of resources. It is argued here that, assuming an appropriate level of
difficulty, readers of texts with elaborative modifications recall more information from
the texts and attend to, and thus be able to identify, more lexical items that appear in the
149
texts, than do readers of unmodified literary texts. These effects do not occur to the same
extent, however, if texts do not present a sufficient cognitive challenge to readers.
One point regarding attentional or controlled resources related to the vocabulary
results needs to be clarified. The instructions given to participants on the data collection
days included that they would be asked to recall as much information about the text as
possible after they were finished reading. Students were not told that they would be
assessed on their vocabulary knowledge. The instructions no doubt influenced how
readers approached the texts. Readers of both versions of Apocalipsis, and readers of both
versions of Fiebre de lotto to a lesser extent, paid closer attention to the vocabulary
because it was necessary to understand certain words in order to comprehend the texts
well enough to recall them later. How students approached the texts might have been
altogether different if pre-reading instructions had been different.
Again, although students were not consciously allocating more resources to
vocabulary understanding for the more challenging texts, attention to specific lexical
items was necessary for text comprehension. Likely the reading behavior of those who
read the unmodified form of Apocalipsis was motivated by the need to later recall the
text, but problems with message comprehension as demonstrated by their shorter recall
length and poorer performance in comparison to their counterparts on the MC questions,
hindered their ability to utilize the context for identification or clarification of unknown
words. This lack of message comprehension was mitigated through elaboration as
reflected in recall length and vocabulary scores. These results support researchers who
argue that lexical items are best acquired incidentally when they appear within a context
whose message is at least minimally comprehensible (Coady, 1997; Hulstijn et al., 1996).
Another vocabulary finding of interest is that readers of the elaborated versions
were able to identify more vocabulary that appeared in the readings even though the
lexical items that were chosen for testing had been selected to avoid words that were
repeated or restated thorough elaborative modifications. Words that would be repeated
150
subsequently in the text were excluded because repetition has been shown to enhance
comprehension (Hulstijn et al., 1996) potentially biasing study results in favor of the
elaborated texts. One wonders, however, whether the results of the vocabulary
assessment would have been different had the items selected for readers’ identification
been those that had been targeted for repetition through elaboration.
Summary of Results: Influence of Spanish Glossing
The final facet of the research dealt with the influence of marginal glossing in
Spanish. Specifically, the question asks whether readers of the unmodified versions of the
literary texts identify lexical items glossed in Spanish in the text margins differently from
readers of the same texts presented without a marginal gloss but with elaborative
modification. The results are described below.
The information included in this section refers to readers’ responses to questions
involving the 11 lexical items that were glossed in Spanish in the margins of the
unmodified text versions. The Spanish glosses were removed in the elaborated versions,
as were the English glosses, although both types of glossing were maintained in the
unmodified versions. The idea behind the removal of the glosses was to measure the
effect of glossing on readers’ recollection of word meanings immediately after reading
the texts. The primary question asked here was whether elaboration was a strong enough
variable to influence readers’ comprehension of these items to a different degree than the
degree provided through the use of the marginal definitions in Spanish.
In this section, no definitive answer is given to this question due to the limited
number and the variability of the vocabulary items. The conclusions given here are
suggestive, at best, and warrant further study. The glossed items that produced the largest
discrepancy in results are discussed but the conclusions reached are provisional and are
not intended to be generalizable to other populations.
151
Generally speaking, the findings involving glossing were mixed and mostly
related to the type of gloss provided for the students in the unmodified versions and to the
type of contextual support supplied to readers through elaborative changes. Although the
readers of the elaborated versions correctly identified the selected terms more frequently
(31%) than did the readers who had access to the marginal glosses (29%), this was not
true for every, or even most, lexical items.
The lexical phrase most influenced by elaborative modifications came from the
Fiebre de lotto text. This phrase, quedarían en las calles ‘would be left in the street,’ was
modified in the elaborated version by the addition of the words sin trabajo ‘without
work.’ With the phrase changed, 23% more readers recalled its meaning correctly
compared with readers of the unmodified text in which the phrase was glossed as
perderían sus trabajos ‘would lose their jobs’ in the text’s margin. Speculatively,
potential reasons why the readers’ comprehension improved with the inclusion of the
prepositional phrase within the sentence include (a) the familiarity of the words contained
in the modification and (b) the ease of access, meaning that readers did not have to divert
attention away from the text in order to understand the meaning of the sentence. Readers
of the elaborated version were able to comprehend the meaning of the sentence using
fewer attentional resources; that is, they expended less energy.
Another word that resulted in a large disparity in readers’ results appeared near
the end of the Fiebre de lotto text. The noun rictus ‘sneer’ or ‘grimace’ was glossed in
Spanish as sonrisa ‘smile’ in the unmodified version. Although sonrisa is a term familiar
to many Spanish students at this level, its use here proved problematic. Again, a
speculative but plausible explanation for the disparity in results is that the students
reading the glossed version had difficulty reconciling the notion of smiling at a point in
the text where the narrative described the characters’ reaction to losing the lottery.
Through elaboration, the phrase de sorpresa ‘of surprise’ was added to rictus. With the
addition of this change, a greater percentage of students (9%) were able to more closely
152
interpret the meaning of the phase with definitions such as ‘shocked expression’ even
without the aid of the gloss.
On many occasions, however, the influence of the Spanish gloss was as strong, if
not stronger, than the effect of elaboration. This was especially the case for readers of
Tiempo libre. These readers produced the poorest results of all three texts when incidental
vocabulary learning was tested, but the readers of the unmodified version out-scored
those reading the elaborated version when it came to identifying items glossed in
Spanish. The phrase me flaquearon ‘weakend’ in reference to the narrator’s legs was
glossed in the margin as se debilitaron, also meaning ‘weakened.’ In the elaborative
version the phrase se debilitaron was removed from the margin and inserted directly into
the text but this change did not increase readers’ ability to correctly provide the
definition. On the contrary, those reading with the marginal gloss out-scored their
counterparts by 8%. This occurred in the same text, again, with the word hormigas ‘ants,’
which was glossed as insectos ‘insects’ in the margin. During elaboration, hormigas was
followed by o de otros insectos similares ‘or from other similar insects,’ yet adding the
information directly to the text did not result in increased word recognition. Again, those
reading the unmodified version out-scored their counterparts by 7%.
The interpretation of what occurred with the Tiempo libre text is only speculative
but it could again be tied to the distribution of attentional resources. It has already been
shown that readers found this text the easiest to comprehend of the three. In addition,
these same readers scored the lowest on the all vocabulary items, which points to the fact
that they did not need to recognize these terms in order to understand the text. But these
particular items were bulleted and glossed, thereby drawing readers’ attention to the
marginal glosses. If, as seems to be the case with Fiebre de lotto and Apocalipsis,
readers’ cognitive processes were taxed because of comprehension concerns, not enough
energy was expended to commit the word meanings to memory even when the words
were glossed. With Tiempo libre, however, comprehension processes were not taxed to
153
the same extent. Therefore, the definitions in the gloss were committed at least to readers’
short-term memory as demonstrated by later recognition. The same lexical items were not
made salient to the readers of the elaborated version of Tiempo libre; they were just other
words included within the narrative, pointing again to the influence that text difficulty
exerts on vocabulary learning.
Overall Findings
The study’s findings indicate that elaborative modifications to literary texts
influence the amount and kind of information that intermediate-level Spanish learners
comprehend about the texts. Limited-proficiency L2 readers of short literary texts that
have been modified elaboratively can recall more information about the texts and are able
to identify more vocabulary that appeared within the texts. They cannot, however, infer
meaning from these texts to a greater degree than can the readers of unmodified versions
of the same texts. Moreover, text difficulty plays a pivotal role in relation to the
effectiveness of elaborative modifications. In term of comprehension, students who read
unmodified texts that are relatively easy to comprehend benefit least from elaborative
modifications.
Study Limitation
The primary limitation of this study is in relation to the results obtained from the
recall protocols. At the inception of study design, it was hoped that all of the student
recalls (n=379) would be analyzed by the type and the number of idea units which they
contained. Results obtained from such a large sample would have been more powerful
than those attained from the representative sample (n=36) evaluated here. The small
number of recalls subjected to analysis limits the inferential ability of this study.
154
Pedagogical Implications
The results of this study suggest that, with elaborative modifications, literary texts
could have an expanded role in L2 instruction for lower-proficiency students. It is
important, however, for students to read texts that are sufficiently, but not excessively,
challenging if they are to gain the optimal benefit from this form of modification.
Benefits to readers include a greater understanding of text content and a higher decree of
incidental vocabulary learning. Another advantage associated with the use of elaborated
texts is that students read, and thus are exposed to, a greater amount of comprehensible
input. In addition, elaboration negates the need for marginal glosses so that readers can
continue to read in the L2 without the need to resort to translation to understand word
meanings. The absence of marginal glosses also serves to keep readers’ attention within
the body of the text, which results in readers being able to use their processing resources
on comprehension of text content. Disadvantages to the use of elaboration include
identifying where and how to elaborate a text, the time involved in text modification, and
the difficulty involved in finding and matching texts to students’ proficiency level.
Recommendations for Future Research
This study, as most do, leaves more questions unanswered than answered,
opening up multiple avenues for further investigation. These results touched upon, but
did not investigate in detail, how readers use their limited cognitive and attentional
resources when reading literary texts in a second language. Knowing that readers have
limited processing resources at their disposal, future research could combine the use of
elaboration with work on input enhancement (Leow, 1997; Sharwood Smith, 1994) and
noticing (Schmidt, 1995) to assess whether the beneficial effects of elaboration could be
optimized by making readers aware of the elaborative modifications. Future research
could assess whether students attend to the elaborative additions differently when they
have been made aware that the modifications were included as aids to comprehension.
155
Making modifications salient to readers through changes in font size or text formatting
would also serve to clearly distinguish between the authors’ words and those of the
person modifying the text.
Conclusion
The findings of this study are but one piece of a much larger puzzle that attempts
to explain how lower-proficiency L2 readers process and understand literary readings. It
is hoped that these results contribute to an overall understanding of the experience and
result of reading literary texts so that these texts can be used earlier and more frequently
for L2 instruction.
156
APPENDIX A
COMBINED TYPES AND EXAMPLES OF
CONFUSION PATTERNS
FOR LITERARY TEXTS
157
Combined Types and Examples of Confusion Patterns for Three Literary Texts
General Type of
Problem
Fiebre de lotto
Tiempo libre
Apocalipsis
1) Difficulty in
determining agency
Examples: … a los
innumerables santos de
su presencia para que
seleccionara uno de los
números or ‘to the
innumerable saints for
their presence so they
would select one of the
numbers’ Readers not
sure who was to select
one of the numbers; El
fin de semana se
convirtió en una tortura
or ‘The weekend
became (a) torture’
Reader interpretations
include “they converted
the weekend” and “he
converted the
weekend”.
Example: Entró mi
esposa, me levantó del
suelo…or ‘My wife
entered and she picked
me up.’ Only two out of
19 readers realized that
the wife lifted the
husband from the
ground. Most believed
that the husband got up
once his wife entered.
Example: No hablamos
or ‘We won’t talk’
interpreted as “you can
talk to them,” “they
would talk,” “talk of,”
“will not talk,” “no
speak,” “we do not
have to speak,” “they
will talk,” “no talking
about,” “we don’t
speak.” Only one
correct translation out
of 10 readings.
2) Difficulty making
transition into
sentence or phrase
Examples: que la
fortuna los había
atropellado. El corazón
les latía
aceleradamente or ‘that
fortune had trampled
over them. Their hearts
beat rapidly.’ These
two sentences are
presented consecutively
within the same
paragraph. The
transition between these
two sentences presented
a difficulty for most
readers. Additionally,
unfamiliarity with the
meaning of the phrase
para tal efecto or ‘to
this end/for this reason’
which begins second
paragraph of reading,
problematic.
Example: … y me puse
a leer la primera
página. Luego de
enterarme… or ‘…and
started to read the front
(first) page. After I
found out…’ Starting
the sentence with the
word luego, which can
mean ‘then, later, soon,
etc.’ as well as here
meaning ‘afterwards’ or
‘after’ resulted in
confusion for most
students
Example: …grandes
aparatos electrónicos:
provocarán catástrofes.
Pero hasta el bisturí se
deslizará or . “…large
electronic appliances:
they will induce
catastrophes. Even the
scalpel …” is
interpreted as “… will
provoke catastrophes.
But later…”
158
3) Difficulty
identifying direct
object pronoun
referent
Examples: Lo salvaron
otros mensajeros or
‘Other messengers
saved him’ readers
unable to identify to
whom the lo is
referring; La fortuna los
había atropellado or
‘fortune had trampled
over them’ Again
unable to identify to
whom the los is
referring; no los iban a
ver or ‘wouldn’t see
them’ in reference to
the bank clientele not
understood by readers
Example: no me ha
importado
ensuciármelos or ‘It’s
never mattered that I’ve
dirtied them (fingers)’
pronouns not correctly
translated by any reader
4) Difficulty
connecting
information
presented earlier in
the text with
information
presented later
Examples: In the third
paragraph, the author
presents three “rules”
that need to be followed
by the bank employees.
Readers are not
identifying the list that
follows as representing
the “rules”.
Additionally, when
talking later of one of
those rules regarding
not speaking to others
about the lottery luego
de abrir el sobre con
los datos or ‘until
opening the envelop
with the information,’
readers do not seem to
relate the datos as the
information regarding
lottery results.
Example: In the first
sentence of the text, the
protagonist complains
of staining his fingers
with ink from the
newspaper. No further
mention is made of this
until sentence eight,
where the protagonist
states that he finds his
fingers “grimier than
usual”. Even for those
few readers who
correctly understood the
meaning of both
sentences, most had
difficulty making the
connection between the
“stain” in sentence one
and the “grim” in
sentence eight.
5) Difficulty
identifying indirect
object pronoun
referent
Examples: Este señor
les cobró una cantidad
exagerada or ‘This man
charged them an
exaggerate amount.’
Unable to identify to
whom the les is
referring; also speaking
of the bank employees,
lo poco que les
quedaba or ‘the little
that they had left’ not
understood; el corazón
les latía
aceleradamente or
‘their hearts beat
rapidly’ the les not
understood; le echaron
Examples: Los
aniquilarán las cosas
or ‘Things will
annihilate them’; donde
nunca lo encuentre or
‘where it will never be
found’ ; cortarán los
dedos que los or ‘will
cut the fingers that hold
them.’ All pronouns
misunderstood by
readers.
159
understood; el corazón
les latía
aceleradamente or
‘their hearts beat
rapidly’ the les not
understood; le echaron
la culpa or ‘they
blamed it (the traffic)’,
again, the le not
understood
6) Difficulty
identifying relative
pronoun antecedent
Example: un experto.
. . el cual. . . Readers
do not associate el
cual with its
antecedent un experto
7) Difficulty
determining time
frame of text
Example: Many readers
do not recognize that
the text is written in the
future tense. Most
readers interpreted for
será or ‘will be’ as “is”,
or even “was”.
Lexical Problems
Fiebre de lotto
Tiempo libre
Apocalipsis
1) Inability to
comprehend
meaning of word or
phrase
Example: Although
despedidos or ‘fired’ is
an essential word in
order to understand
how the story ends,
only one reader
understood its meaning.
Example: The words
mancha and the verb
manchar meaning
‘stain’ and ‘to stain.’
are pivotal to the
comprehension of this
reading yet not glossed
(or understood) except
by one reader out of 19.
Examples: máquina or
‘machine’ correctly
understood by less than
half the readers and of
import to
comprehension of
reading.
Desalojar or ‘replace’
interpreted as
“dislodge,” “destroy,”
“dis-instill,”
“destruction,”
“disappear” and
“distinguish”
The term el hombre in
reference to ‘man’ in
the generalized sense is
not understood by many
readers with one stating
“Who is this man that
they keep talking
about?”
160
2) Direct translation
of idioms
Examples: salarios de
hambre or
‘terrible/low/poor
salaries’ as “salaries of
the hungry” or “salaries
of men”; dar el primer
paso or ‘take the first
step’ as “give the first
pass”
Example: con tal de
estar al día or ‘as long
as I am up to date’ as
“with to be to the day”;
me di cuenta or ‘I
realized’ only
understood by one of
the 19 readers and
translated into many
expressions such as ‘he
gave me.’
Example: por sí solas
or ‘by themselves’
poorly understood and
interpreted literally by
one reader as “because
yes only”; poner en
marcha or ‘to start’ also
poorly understood.
Translations include
“put in march”
3) Applying one
meaning of a word
when another is
appropriate in this
context
Example: quedarían en
la calle or ‘would be
left in the street (let
go)’ as “fit in the
street”; que dicha
entidad maneja or ‘that
this organization
handles’ as “that the
entity drives”; sentir
gritos or ‘notice shouts’
as “feel shouts”; sobre
as in ‘envelop’ as
“over”; manzana as in
‘block’ as “apple" and
retirarse as in ‘leave’ as
“retire.”
Example: se trataba de
or ‘dealt with or
involved” as “he tried”
Example: encender la
televisiónor ‘turning on
the televisión’
interpreted as “burning
the television”;
seguirán brillando or
‘will continue or keep
shining’ as ‘will follow
shining’
4) Word or phrase
translated as
cognate or as
similar English
word
Examples: acordaron
or ‘they agreed’ as
“according”; datos or
‘information’ as
“dates”; intervenir or
‘to intervene’ as
“interview”; vistieron
or ‘they wore or dressed
in’ as “visitors”,
“visited” or “visits”
Examples: importe as
“important”;
“blanqueador” as
“blanket”.
Examples: sitios or
‘places’ interpreted as
“cities”; disponible or
‘available’ as
“disposable”; conducir
or ‘to drive’ as
“conduct”; entre or
‘between’ as “enter”;
plagadas or ‘full of’ as
“plagued”; trabarán or
‘will jam’ as various
forms of the word
trabajo or “work”;
montantes or ‘drawer
frames’ as “mountains”;
sobrevendrá or ‘will
come about’ as
“overselling”
5) Mistaking one
word for a similar
word
Example: manzana or
‘block’ for mañana or
‘morning/tomorrow’;
hambre or ‘hunger’ as
hombre for “man”;
pozo or ‘jackpot’ for
poco or “little/few”
Example: mostrarse or
‘to show’ as molestar
or “to bother”, esconder
or ‘to hide” as
encontrar or “to find.”
Also suelo or ‘floor’
understood as cielo or
“sky”; luego or ‘later’
understood as lago or
“lake.”
161
pozo or ‘jackpot’ for
poco or “little/few”
Also suelo or ‘floor’
understood as cielo or
“sky”; luego or ‘later’
understood as lago or
“lake.”
Ignored or misused glosses
The glosses included with the texts are constructed in such a manner that the words or phrases glossed
into English are in italics. When Spanish synonyms are provided in the margin, italics are not used. Few
readers appear to have noted the italics/no italics distinction. Additionally, other issues presented below
relate to how the glosses were misused or ignored by readers.
Fiebre de lotto
The word regadas is glossed in English as ‘spread’ in the phrase Las enormes carteleras regadas a lo
largo y a lo ancho del estado or “The enormous billboards spread far and wide across the country.” One
reader translated this to mean ‘the jackpot had a large spread.’ The phrase a granel or ‘in
abundance/great quantity’ is glossed but in such a way that it is ignored by all readers. The word rictus is
glossed as sonrisa or ‘smile’ a cause of confusion for many students as they are unable to understand
why the employees would smile if they had lost the lottery. A more appropriate translation would have
been ‘sneer’ or ‘bitter smile.’
Tiempo libre
Me costó hilar is problematic even with gloss, which is in English and states ‘to reflect on (literary, to
spin)’. Translations included ‘reflect the idea,’ ‘reflect on job,’ ‘it cost my job.’ ‘I charged job to reflect.’
and ‘it cost work to spin.’
Two students reading this text noted that switching between English and Spanish in the gloss was
confusing. At one point in the text the word flaquearon is glossed to se debilitaron but only two students
were able to correctly interpret the glossed meaning as “to grow weak” or “to give out."
Apocalipsis
The word volubles is glossed as “unstable” in the phrase mostrarse arrogantes, despóticas, volubles but,
when interpreting the gloss, reader applies the glossed definition of the word despóticas instead of the
definition to volubles: Wrong glossed definition applied to se les antoje as well as for se desmandarán.
Harán caso omiso also glossed into Spanish equivalent but gloss not understood by any reader; sume and
restará glossed into English but glossed definitions improperly incorporated by most readers.
Definition of English cognates unknown to readers. Examples: patetismo glossed as ‘pathos’ but various
readers state that they do not know what ‘pathos. means in English. Additionally, despóticas as in
‘despotic’ not understood by many nor caprichoso as in ‘capricious.’
162
APPENDIX B
UNMODIFIED LITERARY TEXTS
163
Zayas-Bazán, E., S. Bacon, et al. Conexiones: Comunicación y cultura, 2nd ed. ©2002.
Reprinted with permission Pearson Education, Inc. Upper Saddle River, NJ
164
(Continued on following page)
165
Zayas-Bazán, E., S. Bacon, et al. Conexiones: Comunicación y cultura, 2nd ed. ©2002.
Reprinted with permission Pearson Education, Inc. Upper Saddle River, NJ
166
Zayas-Bazán, E., S. Bacon, et al. Conexiones: Comunicación y cultura, 2nd ed. ©2002.
Reprinted with permission Pearson Education, Inc. Upper Saddle River, NJ
167
APPENDIX C
ELABORATED LITERARY TEXTS
168
Fiebre de lotto
Los 160 trabajadores del Banco de Ahorros y Préstamos acordaron, es decir°,
that is
todos se pusieron de acuerdo, en gastar todos sus ahorros para comprar conjuntamente,
entre todos, medio millón de dólares en números de la lotería de la Florida. El pozo acumulado
de esta lotería subía, minuto a minuto, a una cantidad de dinero extravagante. Las enormes
carteleras regadas a lo largo y a lo ancho del estado que anunciaban los premios hacían
ascender la cifra hasta llegar a los 100 millones. Los empleados ponían todas sus esperanzas
y planes para el futuro en el premio gordo para combatir así los rumores que circulaban en
el banco de que en pocas semanas el banco y los empleados iban a ser absorbidos por el
Banco Interamericano y posiblemente ellos quedarían en la calle sin trabajo.
Para tal efecto, los empleados contrataron y después pagaron a un experto en
combinaciones numéricas, el cual era un hombre que había sido expulsado de la Lotería
Estatal de la Florida por negociar con los secretos que dicha entidad del estado maneja en
cuestiones de sorteos y premios.
Este señor les cobró a los empleados del banco una cantidad exagerada que no se
pudo revelar a nadie por aquellos de los impuestos del estado de la Florida. Antes de mandar
al mensajero a comprar los números, por escrito los empleados acordaron tres reglas o normas
que debían cumplirse por todos exactamente al pie de la letra para evitar estropear o arruinar
la suerte del grupo.
1) Ningún empleado podía comprar por su propia cuenta más boletos de la lotería;
2) No se podía hablar con nadie acerca de lo mismo, o sea°, de la lotería, hasta
that is
el lunes siguiente a las ocho de la mañana, luego o en el momento de abrir un sobre
con los datos que cada empleado encontraría en su escritorio;
3) Todos los empleados tenían que dedicar una oración y encender velitas en la iglesia y en
su casa a los innumerables santos de su presencia para que algún santo seleccionara
uno de los números comprados por ellos.
169
Una fila de personas que deseaba comprar boletos le daba vueltas a la manzana en
la calle. Había tantas personas en esta fila que la gente que esperaba le armó una trifulca y un
escándalo al mensajero por demorarse obteniendo los números y así impedir que los otros
compraran boletos. Otros mensajeros de otras entidades y compañías salvaron a este mensajero
de esta trifulca. Estos otros mensajeros estaban haciendo la misma diligencia para distintas
personas que también deseaban comprar boletos.
A medida que pasaba la semana, la atención de los empleados iba desmejorando
y empeorando progresivamente hasta llegar a la completa ineficiencia del jueves y viernes.
En estos días, los empleados atendieron con tal desgana y apatía a los clientes que muchos
clientes optaron por retirarse del banco maldiciendo, gritando malas palabras mientras
salían del banco.
El viernes antes de la selección de los números de la lotería, los empleados hicieron
una fiesta de despedida porque creían que no iban a tener que trabajar más y muchos de ellos
empeñaron y vendieron lo poco que les quedaba de sus posesiones para comprar muchas
bebidas y comidas a granel. La fiesta de despedida terminó en una bacanal, o sea°,
that is
una orgía, como las fiestas de final de año. La policía tuvo que intervenir porque la
mayoría de las personas que asistieron a la fiesta salió a la calle a gritar pestes e insultos
contra el banco, protestando por los salarios de hambre que los jefes del banco les pagaban
y también, protestando por tener que contar dinero a montones que no era de ellos. Pero
ahora los jefe del banco no los iban a ver en ninguna parte ni en las curvas porque después de
ganar la lotería los empleados se iban a dar la gran vida como se la daban ahora los dueños del
banco.
Ese fin de semana se convirtió en una tortura para los empleados del banco.
Ninguno de ellos se atrevió a violar el pacto de las tres reglas por temor a echar a
perder o arruinar la suerte del grupo. Nadie quería cargar con la culpa ni tener la
responsabilidad de seguir viviendo y arrastrando una vida mediocre y sin sentido.
El lunes todos los empleados del banco se vistieron con sus mejores ropas porque ellos
no querían demostrar que eran unos miserables que la fortuna y la suerte los habían atropellado
como un auto atropella a un perro en la calle. Cuando los empleados llegaron el lunes al banco,
el corazón de cada uno de ellos les latía aceleradamente. Hasta los empleados que siempre
llegaban con retraso, ese día se levantaron con tiempo para evitar el tráfico; el mismo tráfico al
que ellos le echaban la culpa de sus demoras cuando llegaban normalmente tarde al trabajo.
El sobre que contenía los resultados de la lotería estaba sobre la mesa o el escritorio
de cada uno de los empleados. La emoción de cada empleado los paralizó. Nadie se atrevía
a dar el primer paso. Poco a poco por todo el banco se empezaron a sentir gritos, desmayos,
llantos. Varios empleados caían fulminados agarrándose el pecho como si tuvieran un
ataque al corazón. Varios de ellos elevaban los brazos a lo alto mientras decían «¡Dios mío!»
«Dios mío!» Al ver los ojos inconmensurablemente abiertos de otros, y un rictus de
sorpresa, es decir°, una expresión de incredibilidad en los demás, lentamente los
that is
últimos empleados abrieron el sobre para enterarse y ser informados que habían perdido
su trabajo porque ellos habían sidos despedidos, y que el pozo acumulado de la lotería para la
próxima semana sería de 200 millones de dólares.
170
Tiempo libre
Todas las mañanas compro el periódico y todas las mañanas, al leerlo, me mancho
los dedos de tinta, una tinta de color negro que sale de las páginas del periódico. Nunca
me ha importado ensuciarme los dedos con tal de estar al día, o sea°, bien informado en
that is
relación con las noticias del día. Pero esta mañana sentí un gran malestar como una
that is
incomodidad o inquietud apenas, es decir°, tan pronto como toqué el periódico.
Creí que solamente se trataba de uno de mis acostumbrados mareos, porque sufro de
that is
vértigo a veces. Pagué el importe, o sea°, el costo del diario y regresé a mi casa. Mi esposa
había salido de compras. Me acomodé en mi sillón favorito, encendí un cigarro y me puse
a leer la primera página. Luego de enterarme –al leer la noticia en el periódico- de que un
jet se había desplomado en un accidente muy violento, yo volví a sentirme mal; vi mis dedos
y los encontré más tiznados, más sucios de tinta que de costumbre. Con un dolor de cabeza
terrible, fui al baño, me lavé las manos con toda calma y, ya ahora más tranquilo, regresé al sillón.
Cuando iba a tomar mi cigarro, descubrí que una mancha negra de tinta cubría mis dedos.
De inmediato retorné al baño, me tallé las manos con zacate, piedra pómez y, finalmente, me
lavé con blanqueador; pero todo el intento de lavarme las manos fue inútil, porque la mancha
de tinta creció y me invadió hasta los codos y cubrió la mitad de los brazos. Ahora, más preocupado
que molesto, yo llamé al doctor y él me recomendó que lo mejor era que tomara unas vacaciones,
o que durmiera. En el momento en que yo hablaba por teléfono, me di cuenta de que, en realidad,
no se trataba de una mancha típica que viene de la tinta negra del periódico, sino de un número
infinito de letras pequeñísimas, apeñuscadas y agrupadas, como una inquieta multitud de
hormigas negras o de otros insectos similares. Después, yo llamé a las oficinas del periódico
para elevar mi rotunda protesta; me contestó una voz de mujer, y esta mujer solamente me
insultó y me trató de loco. Cuando yo colgué el teléfono, las letritas habían avanzado ya hasta
171
mi cintura, es decir°, hasta la mitad de mi cuerpo. Asustado, corrí hacia la puerta de
that is
entrada de mi casa; pero, antes de poder abrirla, perdí mucha fuerza y se me debilitaron y
me flaquearon las piernas, y yo caí estrepitosamente con mucho ruido al suelo. Cuando yo
estaba en el suelo, tirado bocarriba descubrí, que además de la gran cantidad de letras
hormiga, como insectos, que ahora ocupaban todo mi cuerpo, había también una que
otra fotografía. Así yo estuve durante varias horas hasta que escuché que abrían la puerta.
Me costó un gran trabajo mental llegar a la conclusión e hilar la idea, pero al fin pensé
que había llegado mi salvación. En este momento mi esposa entró, ella me levantó del suelo,
me cargó bajo el brazo, se acomodó en mi sillón favorito, y ella me hojeó despreocupadamente.
Como si yo fuera el periódico del día, ella se puso a leerme.
172
Apocalipsis
En el futuro, el fin de la humanidad, el Apocalipsis, no será esa fantasmagoría ideada por
San Juan en los Salmos de la Biblia. Ni ángeles con trompetas, ni monstruos, ni batallas,
ni combates en el cielo y en la tierra. El fin de la humanidad será lento, gradual, sin ruido,
con mucho silencio, sin patetismo o emoción excesiva: una agonía progresiva. Los hombres
se extinguirán uno a uno. Las cosas aniquilarán y destruirán a los hombres, a través de la
rebelión de las cosas, la resistencia, la desobediencia de las cosas. Las cosas- después de
desalojar y tomar el lugar de los animales y de las plantas e instalarse en todos los sitios y
lugares y ocupar todo el espacio disponible- comenzarán a mostrarse arrogantes, despóticas
y tiránicas, inestables y volubles, de humor caprichoso e inconstante. Su funcionamiento
no se ajustará ni se adaptará a las instrucciones de sus manuales. Las cosas modificarán por
sí solas sus mecanismos sin la intervención de los humanos. Luego las cosas funcionarán
cuando se les antoje o, es decir°, sólo cuando deseen funcionar. Por último las cosas mecánicas
se insubordinarán, se declararán en franca rebeldía, se desmandarán y desobedecerán a los
seres humanos y a sus reglas, y harán caso omiso de las órdenes del hombre. El hombre
querrá que una máquina o un aparato mecánico sume, o sea°, haga un simple cálculo
aritmético, pero en vez de sumar los números, la máquina restará; quitando en vez de
incorporando números. El hombre intentará poner en marcha o, es decir°, encender un
motor, y el motor se negará y no querrá funcionar. Operaciones simples y cotidianas de
todos los días como encender la televisión o conducir un automóvil se convertirán en
maniobras y manipulaciones complicadísimas, costosas y caras, plagadas, o sea°, llenas
de sorpresas y de riesgos. Y no sólo las máquinas mecánicas y los motores se amotinarán
y se declararán en rebeldía: también los simples objetos. El hombre no podrá sostener
ningún objeto entre las manos porque se le escapará, se le caerá al suelo o al piso y este
simple objeto se esconderá en un rincón del cuarto donde él nunca lo encuentre.
Las cerraduras que se usan para guardar posesiones bajo llave se trabarán obstaculizando
el acceso y será imposible abrirlas. Los cajones donde se guardan ropa y utensilios
se aferrarán y se mantendrán en sus montantes y nadie logrará abrirlos. Modestas y sencillas
tijeras mantendrán las puntas del pico tenazmente apretadas y será imposible abrirlas.
that is
that is
that is
that is
173
Y los cuchillos y tenedores, en lugar de cortar la comida, cortarán los dedos de las
personas que los manejen. No vamos a hablar de los relojes: señalarán cualquier
hora, no la hora correcta. No hablamos tampoco de los grandes aparatos electrónicos:
provocarán catástrofes. Hasta el bisturí, el cuchillo que usan los médicos para operar
a sus pacientes, se deslizará y cortará donde quiera, sin que los cirujanos que hagan
las operaciones puedan impedirlo. Este bisturí cortará hacia cualquier parte, y el enfermo
morirá con sus órganos desgarrados, con su cuerpo abierto por este cuchillo que cortara
donde no debe cortar. La humanidad languidecerá con poca energía y fuerza entre las
cosas hostiles, indóciles, desobedientes, subversivas y rebeldes. El constante forcejeo
con las cosas irá minando y consumiendo las fuerzas humanas. Y el exterminio de la raza
de los hombres sobrevendrá y ocurrirá a consecuencia del triunfo de las cosas. Cuando
el último hombre desaparezca de este mundo, las cosas frías, bruñidas y brillantes,
relucientes y lustrosas, duras, metálicas, sordas, mudas, insensibles e indiferentes,
seguirán brillando bajo la luz del sol, y bajo la luz de la luna, por toda la eternidad.
174
APPENDIX D
DIVISION OF TEXTS INTO PAUSAL UNITS
175
Fiebre de lotto (Lottery Fever)
Idea/Pausal Units: 157 total/ 82 essential (in bold type)
1
The 160 employees
48
on their own;
2
of the Savings and Loan Bank
49
no one could speak with anyone else
3
agreed to spend
50
about this
4
all of their savings
51
until the following Monday
5
on the combined purchase
52
at eight in the morning,
6
of half a million dollars
53
after opening the envelope
7
in lottery numbers
54
with the information
8
from the Florida lottery
55
that each would find
9
whose accumulated jackpot
56
on his or her desk;
10
was growing
57
everyone had to say a prayer
11
minute by minute.
58
and light candles
12
The enormous billboards
59
to the innumerable saints
13
spread far and wide
60
for their presence
14
across the country
61
so they would select
15
helped the figure grow
62
one of their purchased numbers.
16
to 100 million.
63
One of the lines
17
They put all of their hope
64
that wrapped around the block
18
in the grand prize
65
raised a row
19
in order to combat the rumors
66
with the messenger
20
that in a few weeks
67
for taking such a long time
21
they would be absorbed
68
to buy the numbers.
22
by the Inter American Bank
69
He was saved
23
and would possibly be let go.
70
by other messengers
24
To this end
71
from other entities
25
they hired
72
that were on the same errand.
26
an expert
73
As the week passed,
27
in numerical combinations
74
their attention got progressively
worse
28
who had been fired
75
until arriving at the complete
inefficiency
29
by the state lottery
76
of Thursday and Friday.
30
for trading in secrets
77
On these days
31
in matters of raffles
78
they attended their clients
176
32
this organization handles.
79
with such disinterest
33
This gentleman
80
that many chose to leave
34
charged them
81
cursing.
35
an exorbitant amount
82
On Friday
36
that couldn’t be disclosed
83
they had a going away party
37
for tax purposes.
84
and many pawned
38
Before sending the messenger
85
what little they had left
39
to buy the tickets,
86
to buy
40
they agreed
87
vast amounts
41
in writing
88
of drinks and food.
42
on some rules
89
The party ended
43
that they would follow
90
in a bacchanalia
44
down to the last detail
91
like on New Year’s Eve.
45
to avoid ruining their luck.
92
The police had to intervene
46
No one could buy
93
because the majority
47
an additional ticket
94
went out to the street
95
to curse the bank,
127
in order to avoid the traffic
96
protesting
128
they always blamed
97
the terrible salaries
129
for their tardiness.
98
it paid them
130
The envelope
99
for counting mountains of money
131
was on the table.
100
that wasn’t theirs,
132
Emotion
101
and now
133
paralyzed them.
102
they wouldn’t see them at all,
134
No one dared
103
anywhere,
135
take the first step.
104
because they were going to be the ones
136
Little by little
105
living the high life
137
they began to notice
106
like the owners of the bank.
138
shouts,
107
That weekend was torture.
139
fainting,
108
No one dared
140
sobs.
109
violate the pact
141
Several fell over
110
for fear of causing
142
dead,
111
the loss of the group’s luck.
143
hands clutching their chests.
112
No one wanted
144
Several lifted their arms above them
113
to carry the blame
145
while saying
177
114
of continuing on
146
“My God! My God!”
115
with a mediocre
147
Seeing the bulging eyes of some
116
and meaningless life.
148
and the deathly surprise of others,
117
On Monday
148
the last ones
118
they all dressed in their best clothes.
149
opened their envelopes
119
They didn’t want to let on
150
slowly
120
that they were wretched people
151
to discover
121
trampled on
152
that they had been fired
122
by fortune.
153
and that the jackpot
123
Their hearts beat rapidly.
154
that would carry over
124
Even the ones who always arrived late,
155
to the next week
125
that day
156
would be worth
126
woke up on time
157
200 million dollars.
178
Tiempo libre (Free Time)
Idea/Pausal Units: 95 total/ 55 essential (in bold type)
1
Every morning
49
and he suggested
2
I buy the paper,
50
that the best remedy
3
and every morning,
51
would be a vacation
4
while reading it,
52
or sleep.
5
I stain my fingers
53
While I was talking on the phone,
6
with ink.
54
I realized that it wasn’t really a stain
at all,
7
It’s never mattered
55
rather an infinite number
8
that I’ve dirtied them
56
of miniscule letters
9
as long as I was up to date
57
crowded together,
10
with the news.
58
like a restless multitude
11
But this morning
59
of black ants.
12
I felt extremely uneasy
60
Afterward,
13
as soon as I touched the newspaper.
61
I called the office of the newspaper
14
I thought it was only one of my
usual dizzy spells.
62
in order to raise my emphatic
protest;
15
I paid the cost of the daily
63
a woman’s voice answered,
16
and went back home.
64
and she insulted me
17
My wife had gone shopping.
65
and treated me like a crazy man.
18
I settled into my favorite armchair,
66
By the time I hung up,
19
lit a cigarette
67
the little letters
20
and started to read
68
had advanced to my waist.
21
the front page.
69
Frightened,
22
After I found out a jet had dropped out
of the sky,
70
I ran toward
23
I started to feel bad again;
71
the front door;
24
I caught a glimpse of my fingers
72
but before I could open it,
25
and found them grimier than usual.
73
my legs gave way
26
With a terrible headache,
74
and I fell down
27
I went to the bathroom,
75
noisily.
28
washed my hands
76
Flat on my back
29
calmly
77
I discovered that,
30
and, again composed,
78
in addition to the great quantity of
ant-like letters
179
31
I went back to my chair.
79
that now occupied all of my body,
32
When I was about to pick up my
cigarette,
80
there was also a photograph here
and there.
33
I discovered
81
I was like this for several hours
34
that a black stain
82
until I heard the door open.
35
covered my fingers.
83
It took me a while
36
I returned to the bathroom
84
to put the idea together but
37
immediately.
85
I finally came to the conclusion
38
I worked
86
that my salvation had arrived.
39
with a dishcloth,
87
My wife entered,
40
a pumice stone
88
picked me up from the floor,
41
and finally I washed myself
89
carried me
42
with bleach,
90
under her arm,
43
but the attempt was useless
91
got comfortable
44
because the stain grew
92
in my favorite chair,
45
and assaulted me
93
leafed through me
46
up to my elbows.
94
nonchalantly
47
Now, more worried than irritated,
95
and began to read.
48
I called my doctor
180
Apocalipsis (Apocalypse)
Idea/Pausal Units: 126 total/ 61 essential (in bold type)
1
The end of mankind
47
will ignore
2
will not be a phantasmagoria
48
the orders
3
of St. John of the Psalms.
49
of man.
4
Nor angles
50
Man will want one machine
5
with trumpets,
51
to add
6
nor monsters,
52
and the machine will subtract.
7
nor battles
53
Man will try
8
in heaven
54
to start a motor
9
or on earth.
55
and the motor will refuse.
10
The end of humanity
56
Simple everyday operations
11
will be slow,
57
like turning on a TV or
12
regular,
58
driving a car
13
noiseless,
59
will become
14
without poignancy:
60
complex
15
a gradual agony.
61
maneuvers,
16
Men will expire
62
expensive,
17
one by one.
63
plagued with surprises
18
Things will annihilate them,
64
and risks.
19
the rebellion of things,
65
And not only will the machines
20
the resistance,
66
and motors
21
the disobedience of things.
67
rise up,
22
Things,
68
but also the simplest objects.
23
once they replace
69
Man will be unable to hold on to any
object
24
the animals
70
because it will escape,
25
and plants
71
it will fall
26
and install themselves
72
to the floor;
27
everywhere
73
it will hide
28
and occupy
74
in a corner
29
all available
75
where it will never be found.
30
space,
76
The locks will jam.
31
will begin to display themselves as
77
The drawers will seize
32
arrogant,
78
on their hangers/glides
181
33
despotic,
79
and no one will be able to open them.
34
unpredictable,
80
Modest
35
of capricious
81
scissors
36
humor.
82
will keep their blades
37
Their function will not fit
83
stubbornly tight.
38
with that in the instruction
manuals.
84
And knives
39
They will modify their own
workings
85
and forks,
40
by themselves.
86
instead of cutting food,
41
Later,
87
will cut the fingers
42
they will work only when they feel
like it.
88
that hold them.
43
Finally
89
We won’t talk about watches:
44
they will rise up,
90
they will mark the hour they wish.
45
they will declare themselves in frank
defiance,
91
We won’t talk about
46
they will get out of hand,
92
large
93
electronic
110
And the extermination
94
appliances:
111
of the race of man
95
they will induce catastrophes.
112
will be a direct consequence
96
Even the scalpel
113
of the triumph of things.
97
will slip
114
When the last man disappears,
98
and the surgeons
115
cold,
99
will be unable to stop it
116
polished,
100
from going any which way,
117
glittering,
101
and the ailing man will die
118
hard,
102
with his organs in tatters.
119
metallic,
103
Humanity will languish
120
deaf,
104
between hostile,
121
mute,
105
disobedient
122
insensitive
106
and subversive
123
things will keep shining
107
things.
124
under the light of the sun,
108
The continuous struggle with things
125
the light of the moon,
109
will undermine their strength.
126
for all eternity.
182
APPENDIX E
MULTIPLE-CHOICE QUESTIONS
183
Multiple-Choice Questions:
Fiebre de lotto
Directions: Complete each sentence by selecting the response that corresponds the
most closely to your understanding of the text. Indicate only one letter. If you are unsure
of an answer, make as informed a guess as possible based on what you do know. You
may skip over a question and return to it later.
1.) From whose point of view is this story told?
a. Someone outside the story
b. One of the bank employees
c. Someone inside the story who is not involved in the events
d. Unable to determine from the story
2.) Which phrase best describes the story’s predominant theme?
a. Nothing ventured, nothing gained.
b. Money is the root of all evil.
c. Hard work is its own reward.
d. Don’t put all your eggs in one basket.
3.) In the beginning of the story, which word best describes the mood of the
characters?
a. Eager
b. Hopeful
c. Excited
d. Worried
4.) What word best describes the tone of the person telling this story?
a. Celebratory
b. Humorous
c. Nonjudgmental
d. Embittered
184
Multiple-Choice Questions:
Tiempo libre
Directions: Complete each sentence by selecting the response that corresponds the
most closely to your understanding of the text. Indicate only one letter. If you are unsure
of an answer, make as informed a guess as possible based on what you do know. You
may skip over a question and return to it later.
1.) From whose point of view is the story told?
a. Someone outside the story
b. The main character
c. The main character's wife
d. An unnamed character in the story
2.) Which phrase best describes the story's predominant theme?
a. Things aren't always as they appear.
b. You can't judge a book by its cover.
c. It's true if you believe it to be true.
d. Better safe than sorry.
3.) In the beginning of the story, which word best describes the mood of the
character?
a. Bored
b. Calm
c. Resigned
d. Curious
4.) Which word best describes the tone of the person telling this story?
a. Ironic
b. Sarcastic
c. Puzzled
d. Angry
185
Multiple-Choice Questions:
Apocalipsis
Directions: Complete each sentence by selecting the response that corresponds the
most closely to your understanding of the text. Indicate only one letter. If you are unsure
of an answer, make as informed a guess as possible based on what you do know. You
may skip over a question and return to it later.
1.) According to the narrator's perspective, the events in this story
a. Will take place
b. Might take place
c. Did take place
d. Unable to determine from the story
e. Could take place
2.) Which phrase best describes the story's predominant theme?
a. Possessions are taking over our lives.
b. If you do not take care of what you have now, you might not have it
enjoy in the future.
c. It is through creating, not obtaining, that life has purpose.
d. It is never too late to change things for the better.
3.) In the beginning of the story, which word best describes how the narrator
depicts the events?
a. Disorganized
b. Unexpected
c. Chaotic
d. Precautionary
4.) Which word best describes the tone of the person telling this story?
a. Cynical
b. Detached
c. Surprised
d. Ironic
186
Appendix F
Assessed Lexical Items
187
Assessed Lexical Items
Fiebre de lotto
Example:
(You see)
Los 160 trabajadores del Banco de Ahorros y Préstamos….
(You write)
workers or employees
1. Los 160 trabajadores del Banco de Ahorros y Préstamos acordaron gastar todos sus ahorros.
2. Los trabajadores gastaron todos sus ahorros para comprar conjuntamente….
3. Había rumores que posiblemente los trabajadores quedarían en la calle.
4. Para tal efecto, contrataron a un experto en combinaciones numéricas.
5. Este señor les cobró una cantidad exagerada.
6. Los empleados acordaron reglas que debían cumplir.
7. Debían cumplir las reglas para evitar estropear la suerte.
8. No se podía hablar con nadie acerca de lo mismo.
9. Todos tenían que encender velitas.
10. Una fila le daba vuelta a la manzana.
11. A medida que pasaba la semana, la atención iba desmejorando.
12. La mayoría salió a la calle a gritar pestes.
13. Ellos no los iban a ver ni en las curves.
14. Ninguno se atrevió a violar el pacto.
15. No se atrevieron a violar el pacto por temor a echar a perder la suerte del grupo.
16. Todos, al ver los ojos abiertos de otros y un rictus de sorpresa en los demás…
17. Los últimos abrieron el sobre para enterarse…
18. Ellos habían sido despedidos.
188
Assessed Lexical Items
Tiempo libre
Example:
(You see)
Todas las mañanas compro el periódico
(You write)
I buy or I purchase
1. Al leer el periódico me mancho los dedos….
2. Nunca me ha importado ensuciármelos con tal de estar al día en las noticias.
3. Pero esta mañana yo sentí un gran malestar apenas toqué el periódico.
4. Yo pagué el importe del diario.
5. Luego de enterarme de que un jet se había desplomado….
6. De inmediato, retorné al baño y me tallé con piedra pómez.
7. La mancha creció.
8. La mancha invadió hasta los codos.
9. Cuando yo hablaba por teléfono, me di cuenta…..
10. En realidad, no se trataba de una mancha.
11. …un número de letras como una inquieta multitud de hormigas negras…
12. …un número de letras como una inquieta multitud de hormigas negras…
13. Cuando colgué, las letras habían avanzadas.
14. Las letras habían avanzado hasta mi cintura.
15. Asustado, yo corrí hacia la puerta.
16. Me flaquearon las piernas y yo caí.
17. Cuando mi esposa entró, me levantó del suelo.
18. Mi esposa me cargó bajo el brazo.
Assessed Lexical Items
189
Apocalipsis
Example:
(You see)
El fin de humanidad no será esa fantasmagoría ideada…
(You write)
The end of humanity or mankind
1. Ni ángeles con trompetas, ni monstruos, ni batallas…
2. Las cosas aniquilarán a los hombres.
3. Las cosas, después de desalojar a los animales, …
4. Las cosas comenzarán a mostrarse arrogantes y de humor caprichoso.
5. Las cosas modificarán por sí solas.
6. Las cosas funcionarán cuando se les antoje.
7. Las cosas harán caso omiso de las órdenes del hombre.
8. El hombre intentará poner en marcha un motor.
9. Operaciones simples y cotidianas como encender la televisión…
10. Las cerraduras se trabarán.
11. Los cuchillos y tenedores cortarán los dedos que los manejen.
12. Los relojes señalarán cualquier hora.
13. El bisturí se deslizará sin que los cirujanos puedan impedirlo.
14. El bisturí cortará hacia cualquier parte.
15. La humanidad languidecerá entre las cosas.
16. El constante forcejeo con las cosas irá minando las fuerzas.
17. Las cosas frías metálicas e insensibles ….
18. Las cosas seguirán brillando por toda la eternidad.
190
REFERENCES
Alderson, J. C. (1979). "The Cloze Procedure and Proficiency in English as a Foreign
Language." TESOL Quarterly 13: 219-227.
Alderson, J. C. (1984). Reading in a Foreign Language. New York, Longman.
Alderson, J. C. (2000). Assessing Reading. Cambridge, Cambridge University Press.
Alderson, J. C. and A. H. Urquhart (1988). This Test is Unfair: I'm Not an Economist.
Interactive Approaches to Second Language Reading. P. C. Carrell, J. Devine and D.
E. Eskey. Cambridge, Cambridge University Press: 168-182.
Anderson, R. C. and P. D. Pearson (1984). A Schema-Theoretic View of Basic Processes
in Reading Comprehension. Handbook of Reading Research. P. D. Pearson.
Longman, NY: 255-291.
Bachman, L. F. (1982). "The Trait Structure of Cloze Test Scores." TESOL Quarterly 16:
61-70.
Bachman, L. F. (1990). Fundamental Considerations in Language Testing. Oxford,
Oxford University Press.
Bachman, L. F. and A. S. Palmer (1996). Language Testing in Practice. Oxford, Oxford
University Press.
Barnett, M. (1986). "Syntactic and Lexical/Semantic Skills in Foreign Language
Reading: Importance and Interaction." Modern Language Journal 70: 343-349.
Barnett, M. (1989). More Than Meets the Eye: Foreign Language Learner Reading:
Theory and Practice. Englewood Cliffs, NJ, Prentice-Hall.
Barnett, M. (1991). "Language and Literature: False Dichotomies, Real Allies." ADFL
Bulletin 22(3): 7-11.
Barry, S. and A. Lazarte (1995). "Embedded Clause Effect on Recall: Does High Prior
Knowledge on Content Domain Overcome Syntactic Complexity in Students of
Spanish?" Modern Language Journal 79: 491-503.
Barry, S. and A. Lazarte (1998). "Evidence for Mental Models: How Do Prior
Knowledge, Syntactic Complexity, and Reading Topic Affect Inference Generation in
a Recall Task for Nonnative Readers of Spanish?" Modern Language Journal 82:
176-193.
Beck, I. L., M. G. McKeown, et al. (1991). "Revising Social Studies Text From a Text
Processing Perspective: Evidence of Improved Comprehensibility." International
Reading Association 26: 251-276.
Berkemeyer, V. C. (1989). "Recall Protocol Data: Some Classroom Implications." Die
Unterrichtspraxis 21(3): 131-137.
191
Bernhardt, E. B. (1983a). "Testing Foreign Language Reading Comprehension: The
Immediate Recall Protocol." Unterrichtspraxis 16(1): 27-33.
Bernhardt, E. B. (1983b). "Three Approaches to Reading Comprehension in Intermediate
German." Modern Language Journal 67: 111-115.
Bernhardt, E. B. (1984). "Towards an Information Processing Perspective in Foreign
Language Reading." Modern Language Journal 68: 322-331.
Bernhardt, E. B. (1991a). A Psycholinguistic Perspective on Second Language Literacy.
Reading in Two Languages. J. H. Hulstijn and J. F. Malter. Amsterdam, Free
University Press. 7: 31-44.
Bernhardt, E. B. (1991b). Reading Development in a Second Language: Theoretical,
Empirical and Classroom Perspectives. Norwood, NJ, Ablex Publishing Corporation.
Bernhardt, E. B. (1995). "Teaching Literature or Teaching Students?" ADFL Bulletin
26(2): 5-6.
Bernhardt, E. B. (2000). Second-Language Reading as a Case Study of Reading
Scholarship in the 20th Century. Handbook of Reading Research III. M. K. Kamil, P.
H. Mosenthal, P. D. Pearson and R. Barr. Mahwah, NJ, Erlbaum: 791-811.
Bernhardt, E. B. (2003). "Challenges to Reading Research for a Multilingual World."
Reading Research Quarterly 38(1): 112-117.
Bernhardt, E. B. and M. L. Kamil (1995). "Interpreting Relationships Between L1 and L2
Reading. Consolidating Linguistic Threshold and the Linguistic Interdependence
Hypotheses." Applied Linguistic 61(1): 15-34.
Blau, E. K. (1982). "The Effects on Syntax on Readability for ESL Students in Puerto
Rico." TESOL Quarterly 16: 517-528.
Bossers, B. (1992). Reading in Two Languages. Unpublished PhD Thesis. Amsterdam,
Vrije University.
Britton, B. K. and S. Gulgoz (1991). "Using Kintsch's Model to Improve Instructional
Text: Effects of Inference Calls on Recall and Cognitive Structures." Journal of
Educational Psychology 83: 329-345.
Brown, R. (1985). A Comparison of the Comprehensibility of Modified and Unmodified
ESL Reading Material. ESL. Manoa, HI, University of Hawaii, Manoa.
Brown, R. (1987). "A Comparison of the Comprehensibility of Modified and Unmodified
Reading Materials for ESL." University of Hawaii Working Papers in ESL 6(1): 4979.
Brown, T. L. and M. Hayes (1985). Literacy Background and Reading Development in a
Second Language. The Development of Reading Skills. T. H. Carr. San Francisco,
Jossey-Bass: 19-34.
Carnine, D., E. J. Kameenui, et al. (1984). "Utilization of Contextual Information in
Determining the Meaning of Unfamiliar Words." Reading Research Quarterly 19(2):
188-204.
192
Carpenter, P., A. Miyake, et al. (1994). Working Memory Constraints in Comprehension:
Evidence from Individual Differences, Aphasia, and Aging. Handbook of
Psycholinguistics. M. A. Gernsbacher. San Diego, CA, Erlbaum.
Carrell, P. (1984). "The Effects of Rhetorical Organization on ESL Readers." TESOL
Quarterly 18: 441-469.
Carrell, P. (1985). "Facilitating ESL Reading by Teaching Text Structure." TESOL
Quarterly 19: 727-752.
Carrell, P. L. (1981). Culture-Specific Schemata in L2 Comprehension. Ninth Illinois
TESOL/BE Annual Convention: The First Midwest TESOL Conference, Illinois.
Carrell, P. L. (1987). "Content and Formal Schemata in ESL Reading." TESOL Quarterly
2(3): 461-481.
Carver, R. P. (1990). Reading Rate: A Review of Research and Theory. New York,
Academic Press.
Carver, R. P. (1994). "Percentage of Unknown Vocabulary Words in Text as a Function
of the Relative Difficulty of the Text: Implications for Instruction." Journal of
Reading Behavior 26(4): 413-437.
Chaudron, C. (1983). "Simplification of Input: Topic Reinstatements and their Effects on
L2 Learners' Recognition and Recall." TESOL Quarterly 17(3): 437-458.
Chen, Q. and J. Donin (1997). "Discourse Processing of First and Second Language
Biology Texts: Effects of Language Proficiency and Domain-Specific Knowledge."
Modern Language Journal 81: 209-227.
Chihara, T., T. Sakurai, et al. (1989). "Background and Culture as Factors in EFL
Reading Comprehension." Language Testing 6(2): 143-151.
Clarke, M. A. (1980). "The Short Circuit Hypothesis of ESL Reading: Or When
Language Competence Interferes with Performance." Modern Language Journal 64:
203-209.
Coady, J. (1993). Research in ESL/EFL Vocabulary Acquisition: Putting it in Context.
Second Language Reading and Vocabulary Learning. T. Huckin and J. Coady.
Norwood, NJ, Ablex Publishing Corporation: 3-23.
Coady, J. (1997). L2 Vocabulary Acquisition: A Synthesis of the Research. Second
Language Vocabulary Acquisition. J. Coady and T. Huckin. New York, Cambridge
University Press: 273-290.
Davis, J. (1992). "Reading Literature in the Foreign Language: The
Comprehension/Response Connection." The French Review 65(3): 359-367.
Day, R. R. and J. Bamford (1998). Extensive Reading in the Second Language
Classroom. Cambridge, UK, Press Syndicate of the University of Cambridge.
Deighton, L. (1959). Vocabulary Development in the Classroom. New York, Bureau of
Publications, Teachers College, Columbia University.
193
Deville, C. (2005). Meeting. M. O'Donnell. Iowa City.
Ebel, R. (1954). "Procedures for the Analysis of Classroom Tests." Educational and
Psychological Measurement 14: 352-364.
Ellis, N. (1994). Implicit and Explicit Learning of Languages. London, Academic Press.
Everson, M. E. (1998). "Word Recognition Among Learners of Chinese as a Foreign
Language: Investigating the Relationship Between Naming and Knowing." Modern
Language Journal 82: 194-204.
Favreau, M. and N. S. Segalowitz (1982). "Second Language Reading in Fluent
Bilinguals." Applied Psycholinguistics 3: 329-341.
Favreau, M., M. K. Komoda, et al. (1980). "Second Language Reading: Implications of
the Word Superiority Effect in Skilled Bilinguals." Canadian Journal of Psychology
34: 370-380.
Fecteau, M. L. (1999). "First and Second Language Reading Comprehension of Literacy
Texts." Modern Language Journal 83: 475-493.
Ferguson, C. (1971). Absence of Copula and the Notion of Simplicity: A Study of
Normal Speech, Baby Talk, Foreign Talk, and Pidgins. Pidginization and Creolization
of Languages. D. Hymes. Cambridge, Cambridge University Press: 141-150.
Fransson, A. (1984). Cramming or Understanding? Effects of Intrinsic and Extrinsic
Motivation on Approach to Learning and Test Performance. Reading in a Foreign
Language. J. C. Alderson and A. H. Urquhart. London, Longman: 86-115.
Gass, S. (1999). "Discussion: Incidental Vocabulary Learning." Studies in Second
Language Acquisition 21(2): 319-333.
Goodman, K. S. (1973). Psychological Universals in the Reading Process.
Psycholinguistics and Reading. F. Smith. New York, NY, Holt, Rinehart and
Winston: 21-27.
Goodman, K. S. (2003). Psychological Universals in the Reading Process. On the
Revolution of Reading: The Selected Writings of Kenneth S. Goodman. A. D.
Flurkey and X. Jingguo. Portsmouth, NH, Heinemann.
Gottwald, E. and J. E. Liskin-Gasparro (1995). Approaches to Literature in First-Year
College Spanish: Surveying the Textbooks. Creating Opportunities for Excellence
through Language. E. Spinelli. Lincolnwood, IL, National Textbook Company: 4155.
Goulden, R., P. Nation, et al. (1990). "How Large Can a Receptive Vocabulary Be?"
Applied Linguistic 11: 341-351.
Grabe, W. (1991). "Current Developments in Second Language Reading Research."
TESOL Quarterly 35: 375-406.
194
Grabe, W. (1999). Developments in Reading Research and Their Implications for
Computer-Adaptive Reading Assessment. Issues in Computer-Adaptive Testing of
Reading Proficiency. M. Chalhoub-Deville. Cambridge, UK, Cambridge University
Press.
Graves, M. F. (1986). Vocabulary Learning and Instruction. Review of Educational
Research. E. Z. Rothkopt and L. C. Ehri. Washington DC, American Educational
Research Association. 13: 49-89.
Haastrup, K. (1991). Lexical Inferencing Procedures or Talking About Words. Tübingen,
Gunter Narr.
Hammadou, J. A. (2000). "The Impact of Analogy and Content Knowledge on Reading
Comprehension." Modern Language Journal 75: 27-38.
Hammadou Sullivan, J. A. (2001). Advanced Foreign Language Readers' Inferencing.
Literacy and the Second Language Learner. J. A. Hammadou Sullivan. Greenwich,
Connecticut, IAP Publishing: 217-237.
Harper, S. N. (1988). "Strategies for Teaching Literature at the Undergraduate Level."
Modern Language Journal 72: 402-408.
Hauptman, P. C. (2000). "Some Hypotheses on the Nature of Difficulty and Ease in
Second Language Reading: An Application of Schema Theory." Foreign Language
Annals 33(6): 622-631.
Haynes, M. (1993). Patterns and Perils of Guessing in Second Language Reading. Second
Language Reading and Vocabulary Learning. T. Huckin, M. Haynes and J. Coady.
Norwood, NJ, Ablex Publishing Corporation: 46-64.
Heinz, P. J. (1993). Towards Enhanced, Authentic Second Language Reading
Comprehension Assessment, Research, and Theory Building: The Development and
Analysis of an Automated Recall Protocol Scoring System. Unpublished Doctoral
Dissertation Columbus, Ohio, Ohio State University.
Holley, F. M. and J. K. King (1971). "Vocabulary Glosses in Foreign Language Reading
Materials." Language Learning 21: 213-119.
Honeywell, J. (1977). "Simplification." TESOL Quarterly 11: 431-440.
Horiba, Y. (1993). "The Role of Causal Reasoning and Language Competence in
Narrative Comprehension." Studies in Second Language Acquisition 15: 49-81.
Hosenfeld, C. (1977). "A Preliminary Investigation of the Reading Strategies of
Successful and Nonsuccessful Second Language Readers." System 5: 110-123.
Huckin, T. and J. Coady (1999). "Incidental Vocabulary Acquisition in a Second
Language." Studies in Second Language Acquisition 21(2): 181-193.
Hudson, T. (1998). "Theoretical Perspectives on Reading." Annual Review of Applied
Linguistics 18: 43-60.
195
Hulstijn, J. H. (1992). Retention of Referred and Given Word Meanings: Experiments in
Incidental Vocabulary Learning. Vocabulary and Applied Linguistics. P. J. L. Arnaud
and H. Bejoint. London, Macmillan: 113-125.
Hulstijn, J. H. (1993). "When Do Foreign-Language Readers Look Up the Meaning of
Unfamiliar Words?" Modern Language Journal 77: 139-147.
Hulstijn, J., M. Hollander, et al. (1996). "Incidental Vocabulary Learning by Advanced
Foreign Language Students: The Influence of Marginal Glosses, Dictionary Use, and
Reoccurrence of Unknown Words." Modern Language Journal 80: 327-339.
Jacob, G. M., P. Dufon, et al. (1994). "L1 and L2 Vocabulary Glosses in L2 Reading
Passages: Their Effectiveness for Increasing Comprehension and Vocabulary
Knowledge." Journal of Research in Reading 17: 19-28.
Johnson, P. (1984). "Prior Knowledge and Reading Comprehension Text Bias." Reading
Research Quarterly XIX(2): 219-139.
Johnson, R. E. (1970). "Recall of Prose as a Function of the Structural Importance of the
Linguistic Units." Journal of Verbal Learning and Verbal Behavior 9: 12-20.
Jonz, J. (1991). "Cloze Items Types and Second Language Comprehension." Language
Testing 8: 1-22.
Kamhi-Stein, L. D. (2003). "Reading in Two Languages: How Attitudes Toward Home
Language and Beliefs About Reading Affect the Behaviors or "Underprepared" L2
College Readers." TESOL Quarterly 37(1): 35-72.
Kembo, J. A. (2001). "Testing of Inference Behaviour in a Second Language."
International Journal of Bilingual Education and Bilingualism 4(2): 77-96.
Kintsch, W. (1974). The Representation of Meaning in Memory. Hillsdale, NJ, Erlbaum.
Kintsch, W. (1994). "Text Comprehension, Memory and Learning." American
Psychologist 49(4): 294-303.
Kintsch, W. and T. A. van Dijk (1978). "Towards a Model of Text Comprehension and
Production." Psychological Review 85(5): 363-394.
Kintsch, W., D. Welsch, et al. (1990). "Sentence Memory: A Theoretical Analysis."
Journal Of Memory and Language 29: 133-159.
Klare, G. R. (1984). Readability. Handbook of Reading Research. P. D. Pearson. New
York, Longman: 681-744.
Kobayaski, M. (2002). "Method Effects on Reading Comprehension Test Performance:
Text Organization and Response Format." Language Testing 19(2): 193-220.
Koda, K. (1987). Cognitive Strategy Transfer in Second Language Reading. Research in
Reading in ESL. J. Devine, P. L. Carrell and D. E. Eskey. Washington, DC, TESOL:
124-144.
Kramsch, C. (1985). "Literary Texts in the Classroom: A Discourse." Modern Language
Journal 69: 356-366.
196
Kramsch, C. (1993). Context and Culture in Language Teaching. Oxford, United
Kingdom, Oxford University Press.
Kramsch, C. E. (1987). "Special Issue on the Teaching of Literature." Die
Unterrichtspraxis 2: 159-160.
Krantz, G. (1991). Learning Vocabulary in a Foreign Language: A Study of Reading
Strategies. Goteborg, Sweden, Acta Universitatis Gothburgensis.
Krashen, S. D. (1988). Do We Learn to Read by Reading? The Relationship Between
Free Reading and Reading Ability. Linguistics in Context: Connecting Observation
and Understanding. D. Tannen. Norwood, NJ, Ablex Publishing Corporation: 32-38.
Lado, R. (1986). "Analysis of Native Speaker Performance on a Cloze Test." Language
Testing 3: 130-146.
Ladousse, G. P. (2001). "Using Literature in the Language Classroom: Whys and
Wherefores." English Teacher: An International Journal 5(1): 27-34.
Larsen-Freeman, D. and M. H. Long (1991). An Introduction to Second Language
Acquisition. London, Longman.
Laufer, B. (1989). What Percentage of Text-Lexis is Essential for Comprehension?
Special Language: From Human Thinking to Thinking Machines. C. Lauren and M.
Nordman. Clevedon, England, Multilingual Matters: 316-336.
Lee, J. F. (1986a). "Background Knowledge and L2 Reading." Modern Language Journal
70: 350-354.
Lee, J. F. (1986b). "On the Use of Recall Task to Measure L2 Reading Comprehension."
Studies in Second Language Acquisition 8: 201-211.
Lee, J. F. (1990). A Review of Empirical Comparisons of Non-Native Reading Behaviors
Across Stages of Language Development. Variability in Second Language
Acquisition. H. Burmeister and P. L. Rounds. Eugene, OR, University of Oregon
Press. II: 453-472.
Leow, R. P. (1997). "The Effects of Input Enhancement and Text Length in Adult L2
Readers’ Comprehension and Intake of Second Language Acquisition." Applied
Language Learning 8(2): 151-182.
Liu, N. and I. S. P. Nation (1985). "Factors Affecting Guessing Vocabulary in Context."
RELC Journal 16(1): 33-42.
Long, M. H. (1980). Input, Interaction, and Second Language Acquisition. Unpublished
Doctoral Dissertation. Los Angeles, University of California at Los Angeles.
Long, M. H. (1983a). "Linguistic and Conversational Adjustments to Non-native
Speakers." Studies in Second Language Acquisition 5: 177-193.
Long, M. H. (1983b). "Native Speaker/Non-Native Speaker Conversation and the
Negotiation of Comprehensible Input." Applied Linguistics 4: 126-141.
197
Long, M. H. (1985). Input and Second Language Acquisition Theory. Input in Second
Language Acquisition. S. Gass and C. Madden. Rowley, MA, Newbury House: 377393.
Long, M. H. and S. Ross (1993). Modifications that Preserve Language and Content.
Simplification: Theory and Application. M. Tickoo. Singapore. Anthology Series 31:
29-52.
Martino, M. and E. Block (1993). "Let Them Read Books." College ESL 2(2): 12-20.
Maxim, H. H. (2002). "A Study into the Feasibility and Effects of Reading Extended
Authentic Discourse in the Beginning Language Classroom." Modern Language
Journal 86: 20-35.
McCallon, E. and R. McClaran (1974). Determining an Appropriate Sampling Method.
Austin, Texas, Learning Concepts.
McGuigan, H. E. (1979). Difficulty Levels of Modern German Literary Prose vis-a-vis
the Reader: An Empirical Investigation Using the Cloze Procedure with American
Undergraduate and Graduate Students of German and Native German University
Students, University of Minnesota.
McLaughlin, B. (1987). Theories of Second Language Learning. London, Edward
Arnold.
McLeod, B. and B. McLaughlin (1986). "Restructuring or Automaticity? Reading in a
Second Language." Language Learning 36: 109-123.
McNamara, D. S., E. Kintsch, et al. (1996). "Are Good Texts Always Better? Interactions
of Text Coherence, Background Knowledge, and Levels of Understanding in
Learning from Text." Cognition and Instruction 14(1): 1-43.
Melka, F. (1997). Receptive versus Productive Aspects of Vocabulary. Vocabulary
Acquisition, Description and Pedagogy. N. Schmitt and M. McCarthy. New York,
Cambridge University Press: 84-102.
Merriam, S. B. (2001). Qualitative Research and Case Study Applications in Education.
San Francisco, Jossey-Bass.
Meyer, B. (1985). Prose Analysis: Purposes, Procedures and Problems. Analyzing and
Understanding Expository Text. B. K. Britton and J. Black. Hillsdale, NJ, Erlbaum:
11-64 and 269-304.
Meyer, B. (2003). "Text Coherence and Readability." Topics in Language Disorders 23:
204-211.
Mitchell, R. and F. Myles (1998). Second Language Learning Theories. London, Arnold
Publishers.
Mondria, J. and M. Wit-De Boer (1991). "The Effects of Contextual Richness on the
Guessability and the Retention of Words in a Foreign Language." Applied Linguistics
12(3): 249-265.
198
Morgan, G., A., Leech N. L., et al. (2004).SPSS for Introductory Statistics: Use and
Interpreation 2nd ed., Mahwah, NJ, Erlbaum.
Moon, R. (1997). Vocabulary Connections: Multi-word Items in English. Vocabulary:
Description, Acquisition, and Pedagogy. N. Schmitt and M. McCarthy. New York,
Cambridge University Press: 40-63.
Nagy, W. E. (1987). "Learning Word Meanings From Context During Normal Reading."
American Educational Research Journal 24(2): 237-270.
Nation, I. S. P. (1990). Teaching and Learning Vocabulary. Rowley, MA, Newbury
House.
Nation, I. S. P. and J. Coady (1988). Vocabulary and Reading. Vocabulary and Language
Teaching. R. Carter and M. McCarthy. London, Longman: 97-110.
Oh, S. Y. (2001). "Two Types of Input Modification and EFL Reading Comprehension:
Simplification versus Elaboration." TESOL Quarterly 35(1): 69-96.
Pak, J. (1986). The Effect of Vocabulary Glossing on ESL Reading Comprehension.
Unpublished Manuscript. Manoa, University of Hawai’i.
Paribakht, T. and M. Wesche (1996). "Enhancing Vocabulary Acquisition Through
Reading: A Hierarchy of Text-Related Exercises Types." The Canadian Modern
Language Review 52: 154-178.
Paribakht, T. and M. Wesche (1997). Vocabulary Enhancement Activities and Reading
for Meaning in Second Language Vocabulary Acquisition. Second Language
Vocabulary Acquisition. J. Coady and T. Huckins. New York, Cambridge University
Press: 174-200.
Paribakht, T. and M. Wesche (1999). "Reading and "Incidental" L2 Vocabulary
Acquisition." Studies in Second Language Acquisition 21: 195-224.
Parker, K. and C. Chaudron (1987). "The Effects of Linguistic Simplification and
Elaborative Modifications on L2 Comprehension." University of Hawai’i Working
Papers in ESL 6: 107-133.
Parry, K. (1996). "Cultural, Literacy and L2 Reading." TESOL Quarterly 30(4): 665-692.
Parry, K. (1987). Reading in a Second Culture. Research in Reading as English as a
Second Language. J. Devine, P. L. Carrell and D. E. Eskey. Washington, DC,
TESOL: 59-70.
Patrikis, P. C. (2002). "The Return of the Text." ADFL Bulletin 34(1): 51-53.
Perkins, K. (1983). "Semantic Constructivity in ESL Reading Comprehension." TESOL
Quarterly 17: 19-27.
Perkins, K. (1998). "Assessing Reading." Annual Review of Applied Linguistics 18: 208218.
Pressley, M. and P. Afflerbach (1995). Verbal Protocols of Reading: The Nature of
Constructively Responsive Reading. Hillsdale, NJ, Erlbaum.
199
Pulido, D. (2003). "Modeling the Role of Second Language Proficiency and Topic
Familiarity in Second Language Incidental Vocabulary Acquisition Through
Reading." Language Learning 53(2): 233-284.
Quinn, C. J. (2003). Taking It From the Top: The Growth and Care of Genres. Acts of
Reading: Exploring Connections in Pedagogy of Japanese. H. Nara and M. Noda.
Honolulu, University of Hawai’i Press: 38-60.
Reder, S. (1994). Practice-Engagement Theory: A Sociocultural Approach to Literary
Across Cultures. Literary Across Languages and Cultures. B. M. Ferdman, R. M.
Weber and A. G. Ramirez. Albany, NY, State University of New York Press: 33-74.
Rice, D. B. (1991). "Language Proficiency and Textual Theory: How the Twain Might
Meet." ADFL Bulletin 22(3): 12-15.
Richards, J. C. and T. S. Rodgers (2001). Approaches and Methods in Language
Teaching. Cambridge, UK, Cambridge University Press.
Riley, G. L. and J. F. Lee (1996). "A Comparison of Recall and Summary Protocols as
Measures of Second Language Reading Comprehension." Language Testing 13: 173189.
Rivers, W. M. (1987). Interaction as the Key to Language Learning for Communication.
Interactive Language Teaching. W. M. Rivers. Cambridge, UK, Cambridge
University Press: 3-16.
Ross, S., M. H. Long, et al. (1991). "Simplification or Elaboration? The Effects of Two
Types of Text Modifications on Foreign Language Reading Comprehension."
University of Hawaii Working Papers in ESL 10: 1-32.
Rott, S. (1999). "The Effect of Exposure Frequency on Intermediate Language Learners'
Incidental Vocabulary Acquisition and Retention through Reading." Studies in
Second Language Acquisition 21(4): 589-619.
Rudolph, F. (1962). The American College and University: A History. Athens, GA,
University of Georgia Press.
Scherfer, P. (1993). "Indirect L2-Vocabulary Learning." Linguistics 31: 1141-1153.
Schiffrin, R. M. and W. Schneider (1977). "Controlled and Automatic Human
Information Processing: Perceptual Learning, Automatic, Attending, and a General
Theory." Psychological Review 84: 127-190.
Schmidt, R. (1995). Attention and Awareness in Foreign Language Learning. Honolulu,
HI, University of Hawai’i Press.
Schulz, R. A. (1981). "Literature and Readability: Bridging the Gap in Foreign Language
Reading." Modern Language Journal 65: 43-53.
Segalowitz, N., C. Poulsen, et al. (1991). Lower Level Components of Reading Skill in
Higher Level Bilinguals: Implications for Reading Instruction. Reading in Two
Languages, AILA Review. J. H. Hulstijn and J. F. Matter. Amsterdam, Free
University Press. 8: 15-30.
200
Shanahan, D. (1997). "Articulating the Relationship Between Language, Literature, and
Culture: Toward a New Agenda for Foreign Language Teaching and Research."
Modern Language Journal 81: 164-174.
Sharwood Smith, M. (1994). Second Language Learning: Theoretical Foundations.
London, Longman Publishing.
Shohamy, E. (1982). "Affective Considerations in Language Testing." Modern Language
Journal 66: 13-17.
Shook, D. (1997). "Identifying and Overcoming Possible Mismatches in the Beginning
Reader-literacy Text Interaction." Hispania 80: 234-23.
Shook, R. (1977). "Discourse Structure in Reading." TESL Reporter 10(2,3, & 4).
Shu, H., R. C. Anderson, et al. (1995). "Incidental Learning of Word Meanings While
Reading: A Chinese and American Cross-cultural Study." Reading Research
Quarterly 30: 76-95.
Stahl, S. A., V. Chou Hare, et al. (1991). "Defining the Role of Prior Knowledge and
Vocabulary in Reading Comprehension: The Retiring of Number 41." Journal of
Reading Behavior 23(4): 487-508.
Stanovich, K. J. (1991). Word Recognition: Changing Perspectives. Handbook of
Reading Research. R. Barr, M. Kamil, P. Mosenthal and D. Pearson. New York,
Longman. II: 418-452.
Storme, J. A. and H. J. Siskin (1989). Developing Extensive Reading Skills: The
Transition to Literature. Defining the Essentials for the Foreign Language Classroom.
D. McAlpine. Lincolnwood, IL, National Textbook Company: 24-39.
Swaffar, J. and S. Bacon (1993). Reading and Listening Comprehension: Perspectives on
Research and Implications for Practice. Research in Language Learning: Principles,
Processes, and Prospects. A. Omaggio-Hadley. Lincolnwood, IL, National Textbook
Company: 124-155.
Swaffar, J. K. (1985). "Reading Authentic Texts in a Foreign Language: A Cognitive
Model." Modern Language Journal 69: 15-34.
Swaffar, J. K. (1991). Language Learning is More Than Learning a Language:
Rethinking Reading and Writing Tasks in Textbooks for Beginning Language Study.
Foreign Language Acquisition Research and the Classroom. B. Freed. Lexington MA,
Heath: 252-279.
Tsang, W-K. (1987). Text Modification in ESL Reading Comprehension. ESL. Honolulu,
HI, University of Hawai’i.
Tweissi, A. I. (1998). "The Effects of the Amount and Type of Simplification on Foreign
Language Reading." Reading in a Foreign Language 11(2): 191-204.
Ulijn, J. M. and J. B. Strother (1990). "The Effect of Syntactic Simplification on Reading
EST Texts as L1 and L2." Journal of Reading Research 13(1): 38-54.
201
Urano, K. (2000). Lexical Simplification and Elaboration: Sentence Comprehension and
Incidental Vocabulary Acquisition. ESL. Honoulu, Hawai'i, University of Hawai'i.
Urquhart, A. H. (1984). The Effect of Rhetorical Ordering on Readability. Reading in a
Foreign Language. J. C. Alderson and A. H. Urquhart. London, Longman: 160-180.
Watanabe, Y. (1992). Incidental Learning of Vocabulary: Retention of Inferred Meanings
versus Given Meanings, Unpublished Master's Thesis. Manoa, HI, University of
Hawai'i.
Wolf, D. F. (1993). "A Comparison of Assessment Tasks Used to Measure Foreign
Language Reading Comprehension." Modern Language Journal 77: 473-489.
Yano, Y., M. Long, et al. (1994). "The Effects of Simplified and Elaborated Texts on
Foreign Language Reading Comprehension." Language Learning 44(2): 189-219.
Zayas-Bazán, E., S. Bacon, et al. (2002). Conexiones: Comunicacion y cultura 2ed.
Upper Saddle River, NJ, Pearson Publishing.