as a PDF - CiteSeerX

Input Enhancement and L2 Question Formation
LYDIA WHITE, NINA SPADA
McGill University
Concordia University
In this study, we investigate the extent to which form-focused instruction and
corrective feedback (i.e. 'input enhancement'), provided within a primarily
communicative program, contribute to learners' accuracy in question formation. Over a two-week period, three experimental classes of beginner level
francophone ESL learners (aged 10-12 years) were exposed to a variety of input
enhancement activities on question formation. Their performance on paperand-pencil tasks and an oral communication task was assessed on a pre-post
test basis and compared with an uninstructed control group. The results
indicate that instruction contributed to syntactic accuracy and that learners
who were exposed to the input enhancement activities significantly outperformed the uninstructed learners. These results are interpreted as evidence
that input enhancement can bring about genuine changes in learners' interlanguage systems.
1. INTRODUCTION
Many current approaches to second language (L2) teaching adopt the position
that instruction should focus on the provision of rich and varied comprehensible input and provide opportunities for learners to use language in spontaneous, meaningful interaction. In addition, it is often argued that it is
superfluous or even counter-productive to expose L2 learners to instruction
that focuses on grammatical form. This position is justified (implicitly or
explicitly) by the assumption that L2 learners proceed in a similar fashion to
first language (LI) learners (Ellis 1984; Krashen 1985). The crucial input in
LI acquisition is positive evidence, utterances which give the learner unconscious knowledge of what the language allows. LI acquirers master their
mother tongue on the basis of such input, attaining complete fluency and
accuracy. If the same is true in L2 acquisition, then there is no role for formfocused instruction.
There is no doubt that L2 learners can achieve considerable success in
contexts where they are simply exposed to meaningful, naturalistic input.
However, such input alone does not necessarily lead learners to high degrees of
accuracy or high levels of development in the L2 (Harley and Swain 1984;
Swain 1985). Some researchers argue that learners' accuracy can be improved
by providing a variety of 'consciousness raising' activities (Sharwood Smith
1981; Rutherford 1987), which aim to draw the learner's attention to properties
Applied Linguistics, Vol. 12, No. 4 © Oxford University Press 1991
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
PATSY M. LIGHTBOWN, LEILA RANTA
L. WHITE, N. SPADA, P. LIGHTBOWN, AND L. RANTA
417
2. INPUT ENHANCEMENT AND INPUT
It is worth speculating on the potential beneficial effects of input enhancement
in L2 acquisition. What can it achieve that is not achieved by input provided in
meaningful interaction? There are a number of ways in which input enhancement may supplement naturalistic positive evidence.
First, it is quite possible that the L2 learner, even though exposed to certain
structures, will fail to perceive them in naturalistic input. In other words, the
input does not become intake (Corder 1967). By drawing the learner's attention
explicitly to formal properties of the L2, form-focused instruction provides
a more salient kind of positive evidence, which may help to sensitize the
learner to aspects of the L2 which would otherwise pass unnoticed. For
example, English questions and negatives are sometimes formed with the
auxiliary verb do, as in (1):
1 a. Does Mary like John?
b. Mary does not like John.
Such sentence types are likely to be quite frequent in the positive input to L2
learners, and yet they may fail to perceive the presence of do, filtering it out
because the interlanguage grammar cannot analyse it as yet or simply because of
its lack of salience in the natural stream of speech. Input enhancement, then, can
be used to draw the learner's attention to properties of the input that might
otherwise be missed.
Second, input enhancement may be used to help learners 'unlearn' incorrect
analyses of the L2 by supplying negative evidence, that is, information about
forms which are not possible in the target language. For instance, a learner who
is trying to work out the past tense of sing may form the initial hypothesis that it
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
of the L2 by focusing on the form or grammar of the L2. Recently, Sharwood
Smith (1991) has proposed that the term 'consciousness raising' be replaced by
'input enhancement'. He argues that one can know only that aspects of the input
have been highlighted in some way; it is impossible to tell whether the learner's
consciousness has been raised.
Recently a number of experimental studies have investigated the effects of
form-focused instruction on the L2 development of specific structures: English
relative clauses (Gass 1982; Eckman etal. 1988; Doughty in press) possessive
determiners (Zobl 1985); German word order (Pienemann 1984); the French
determiner system (Tomasello and Herron 1988, 1989); French conditionals
(Day and Shapson 1991) and past tense forms (Harley 1989). These studies
have found a beneficial effect for such instructional intervention, at least in the
short term. Furthermore, our own research, both observational and experimental, has indicated the effectiveness of input enhancement in the context of
communicative language teaching, in both the short term (White 1991a, 1991b)
and the long term (Lightbown 1991). The research reported in the present
paper is a further test of the hypothesis that input enhancement can facilitate L2
acquisition.
418
INPUT ENHANCEMENT
2 a.
b.
c.
d.
e.
f.
John drinks his coffee quickly.
Carefully Mary opened the door.
Jane has often visited Paris.
Bill sometimes eats pizza.
The children walk slowly to school.
*Elizabeth takes usually the bus.
If a learner assumes the possibility of sentences like 2f, negative evidence would
be necessary for disconfirmation in the absence of a single pre-empting
alternative.1 In this case, input enhancement may supply crucial negative
evidence. Although negative evidence is assumed to play a minimal,role in LI
acquisition, it may be more important in L2 acquisition if learners make
incorrect generalizations (often influenced by perceived similarity to their LI)
that cannot be disconfirmed on the basis of positive input alone.2
In programs that focus on communicative language teaching, it is intended
that primarily naturalistic positive evidence be provided in the classroom.
However, learners in classrooms where much of the input comes from other
learners will inevitably be exposed to incorrect 'positive evidence'. There is no
reason to assume that learners are capable of filtering out the 'poor' input and
utilizing only the 'good' input for second language acquisition. Teaching
programs that include a focus on form and some error correction provide more
emphasis on what is possible in the L2, as well as explicit indications of what is
not possible.
3. QUESTION FORMATION IN ENGLISH AND FRENCH
In this paper, we look at French-speaking learners of English, with respect to the
potential effectiveness of form-focused instruction and correction for the
acquisition of English question formation.
English questions presumably occur frequently in the input to L2 learners,
especially in the classroom context. English yes/no questions involve subject-
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
is singed (on the basis of other English past tense forms in the input). The
learner uses this form, then notices that in fact people say sang, and adopts this
instead. The change from singed to sang can be motivated purely by positive
evidence, provided that learners realize that there is only one possible past tense
form, so that encountering an alternative form will 'pre-empt' the original
incorrect analysis (Pinker 1984; Rutherford 1989). However, another
possibility is that L2 learners do not use positive evidence to pre-empt in this
way, treating singed and sang as two acceptable past tense forms. In that case,
input enhancement can be used to point out to the learner that one form must
replace the other.
Other cases arise where there appears to be no positive evidence to pre-empt
incorrect analyses. For example, adverb placement in English is relatively free,
as can be seen in 2a to 2e and yet there is a position from which adverbs are
usually excluded, namely from between the verb and object, as in 2f:
L. WHITE, N. SPADA, P. LIGHTBOWN, AND L. RANTA
419
auxiliary inversion, and require do-support if no other auxiliary is present, as in
3a and 3b:
3 a. Is John playing tennis?
b. Does Mary live in Montreal?
4 What is John playing?
Again, the auxiliary do must be used if no other auxiliary is available, as in 5:
5 Where does Mary live?
French has several ways of forming questions, some of which are similar to
English. Yes/no questions can be formed with the phrase est-ce que attached to
a declarative sentence, as in 6, with subject-verb inversion, as in 7, or, very
commonly, with declarative word order and rising intonation, as in 8:
6 Est-ce que Jean joueau tennis?
7 Joue-t-il au tennis?
8 II joueau tennis?
French differs from English in that main verbs can invert, as can be seen in 7.
There is no equivalent of do -support. Inversion in French is only possible with
pronouns, as in 7. That is, 9 is ungrammatical:
9 *Joue-Jean au tennis?
In spoken French, wh -questions can be formed by leaving the wh -phrase in situ
as in 10, or by movement, as in 11 and 12. Questions which leave subject and
verb in uninverted position (as in 11) and questions with inversion (as in 12) are
both possible. Once again, inversion is optional with pronouns but is not
permitted with noun phrases, as shown by the ungrammatically of 13.3
10 Marie demeure oil?
l l a . Oil est-ce que Marie demeure?
b. Ou c'est qu'elle demeure?
12 Ou demeure-t-elle?
13 *Oii demeure Marie?
French and English, then, differ in a number of respects, as far as question
formation is concerned: inversion is optional with pronoun subjects in French,4
but prohibited with noun phrase subjects; when inversion occurs, main verbs
can invert; wh-in-situ is permitted even in non-echo questions.
Could the francophone learner of English attain syntactic accuracy in English
questions from the L2 input alone? At first glance, it might appear that the
English input reveals that English questions are formed via inversion, using the
auxiliary do in some cases, with wh -phrases in initial positions. In that case, the
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
W/i-questons are formed by movement of a wh word or phrase to initial
position, and they also require inversion of the subject and an auxiliary verb, as
in 4.
420
INPUT ENHANCEMENT
positive input could serve a disconfirming role for incorrect hypotheses about
English question formation, and it might seem superfluous to provide for input
enhancement through form-focused instruction and correction. For example, if
the learner forms a question like 14, positive evidence in the form of questions
like 15 should provide disconfirmation:
14 * What John wants?
15 What does John want?
4. DESCRIPTION OF STUDIES: PHASES 1 AND 2
The experiments reported on here constitute the first two phases of an ongoing
study of the development of English questions. Subjects in both phases were in
intensive English as a Second Language (ELS) programs in the Province of
Quebec, Canada. They were all native speakers of French, in Grade 5 or 6 (1012 years old). Learners enter these programs having had little prior instruction
in English, that is, not more than two hours of instruction per week in the school
year or two preceding Grade 5 or 6. Furthermore, they have very little contact
with English outside the classroom. Our subjects thus represent beginner level
ESL learners.
In these intensive programs, students receive approximately 5 hours of ESL
instruction every day for 5 months of the school year. During their 5 months of
intensive ESL, the learners study English only and do not receive any subject
matter instruction in the L2. In the remaining 5 months, they receive their
regular subject-matter instruction in French. The ESL instruction is communicatively-based and focuses on the provision of opportunities for learners
to use language in meaningful and creative ways. The program is organized
around a variety of topics and themes, and normally little explicit form-focused
instruction is provided (Spada and Lightbown 1989; Lightbown and Spada
1990).
These classes differ from most second language classes that have been
observed in that students ask many questions (cf. Long and Sato 1983; White
and Lightbown 1984). They do so quite spontaneously to get information from
the teacher, from visitors, or from each other. In addition, the teachers tend to
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
However, this presupposes that the L2 learner uses evidence like 15 to preempt 14. On the other hand, if the French-speaking learner starts off with the
hypothesis that English questions, like French questions, are formed in several
different ways, with and without inversion, or with and without wh -movement,
then there is a potential role for form-focused instruction, to provide the
necessary evidence that English question formation is more restricted than
French.5
Furthermore, subject-auxiliary inversion may be a non-salient feature of
English. Even when learners hear many examples of questions with inversion,
they may not perceive the fronted auxiliary. In this case, input enhancement may
serve to draw learners' attention to a feature which is otherwise difficult to
perceive.
L. WHITE, N. SPADA, P. LIGHTBOWN, AND L. RANTA
421
plan activities which require students to ask many questions (guessing games,
show and tell, etc.). However, in our observations which preceded the studies
reported here, it was rare for the students to be corrected on the word-order
aspects of their questions.
4.1 Phase 1
Table 1: Phase 1 correction task: mean
accuracy scores
Croups
Instructed (n = 53)
Uninstructed (n = 76)
Post-test
Mean
SD
8.4
3.58
4.43
2.31
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
The teaching and testing of questions in Phase 1 were originally planned as a
control condition for another study on the effect of instruction on francophone
learners' placement of adverbs (White 1991a, 1991b), as well as a pilot for
Phase 2 of the present study. Two classes (n=53) received form-focused
instruction on question formation, while three other classes (n = 76) received
no instruction on questions but were provided with instruction on another
structure (adverb placement). This instruction was given by their classroom
teachers after students had been in the intensive program for about three
months. Subjects received approximately three hours of explicit instruction on
question formation in the first week of a two week period. There were about two
hours of follow-up activities during the second week and teachers were encouraged to provide corrective feedback as well. Materials concentrated on the
auxiliaries can, be, and do, and the question words what, where, and why.
Throughout these activities, teachers provided explicit instruction in (and
correction of) placement of subjects, auxiliaries, and wh -pronouns.
Subjects were tested on the first school day after the instruction was completed, in order to obtain some preliminary information on the effects of instruction on question formation. The test consisted of 15 written wh -questions,
which were presented either in (partially) scrambled word order (for example,
The teachers do why homework give ?), or without subject auxiliary inversion
(for example, Where your brother is playing baseball?), or in an already correct
order (for example, What is the baby eating?) The subjects' task was to determine whether each question was correctly formed and, if not, to put the words in
a correct order (a task they were familiar with from the activities introduced in
the instructional period).
Mean accuracy scores on the correction task are presented in Table 1.
(Accurate responses consist of those correctly identifying already grammatical
questions, as well as those identifying and correcting ungrammatical questions.)
There were no signifiant differences between the classes assigned to the
422
INPUT ENHANCEMENT
-
Uninstructed
Instructed
Figure I : Phase I correction task: distribution of responses, in percentages
It can be seen that the predominant response of the groups who were not
instructed in question formation was to form questions without inversion:
58 per cent of their responses were of this type, and only 24 per cent of their
responses were fully grammatical questions. In contrast, the instructed groups
gave a significantly lower proportion of questions without inversion (34 per
cent) and a correspondingly higher proportion of correctly formed questions
(55 per cent), suggesting that the instruction had had positive effects on their
accuracy.
4.2 Phase 2
In Phase 2 of the study, four additional classes of intensive program learners
were studied. Three classes (n = 79) received form-focused instruction on
question formation. and one class (n = 29) was instructed on adverb placement,
serving as an uninstructed control group. In addition, native speaker cornparison data were collected from a monolingual English-speaking Grade 6 class
(n = 29) in Victoria, British Columbia, Canada.
Exercises and activities, once again, emphasized sentences using the
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
instructed condition (F(1,51) = 0.62, p
0.44), so their results have been
combined. The same is true of the classes assigned to the uninstructed condition
(F(2,73) = 1.64, p = 0.20). Analysis of variance shows that the difference
between the mean scores of the instructed group and the uninstructed group is
significant (F(1,127) = 64.84, p < 0.00 1).
The distribution of responses is presented in Figure 1. Responses (n = 1,931)
were analysed in terms of four categories: Grammatical (= the sentence was a
fully grammatical question), WhS (= the sentence was grammatical except for
the failure to invert subject and auxiliary), Scrambled (= muddled word order),
and Other (= omissions, incompletes, etc.).
L. WHITE, N. SPADA, P. LIGHTBOWN, AND L. RANTA
423
4.3 Phases 1 and 2 compared
As indicated above, in Phase 1 students were taught questions in order to give
them experience with materials being used in a study of adverb placement, for
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
auxiliaries can, be, and do in the present tense, and the question words what,
where, and why. Subjects received approximately four and a half hours of
explicit instruction on question formation in the first week of a two-week period
after approximately three months in the communicative intensive program.
During the second week, there were about four hours of follow-up activities,
and teachers were encouraged to continue to provide corrective feedback on
students' errors in question formation, both inside and outside the specific
instructional activities. The materials covered a wide range of task types,
including some similar to those which were used in the tests to measure the
subjects' performance.
It is important to note that the teaching materials were developed in a tightly
prescribed manner to ensure that the teachers would teach the activities in the
same order, in the same manner, and in the same time allotment. This was
accomplished by giving each teacher a specially prepared package of instructional materials and specific guidelines for their implementation. The materials
provided everything that was required for the teachers and students to complete
the exercises and activities. This included student hand-outs, pictures, and word
cards for various tasks, as well as classroom posters indicating inversion rules
for question formation. Teachers were audio-recorded while teaching the
materials and they were asked to fill out a questionnaire afterwards to provide
additional information concerning any changes or revisions they made while
working through the materials.6
Subjects were tested by means of two written tasks, and an oral communication task:
(1) a cartoon task in which subjects were presented with 20 cartoons with
scrambled words underneath them. Subjects had to arrange the words in the
cartoon bubble to create a correct question. (See example in Appendix.)
(2) a preference task which consisted of 33 pairs of sentences, of which 20
involved question formation and the rest were distractors. Subjects had to read
each pair and choose between responses indicating whether both were
grammatical, or neither, or whether only one of the two was grammatical. (See
example in Appendix.)
(3) an oral communication task in which students looked at sets of four pictures
and asked questions until they could match one of their pictures with the picture
held by the experimenter. Students' questions were audio-recorded and
transcribed for analysis.
These tests were administered to the subjects on the school day before the
instruction began (Pre-test), on the first school day after the instruction was
completed (Post-test), and again five weeks after instruction (Follow-up). A
long-term follow-up was conducted five months later, results of which will be
reported elsewhere.
424
INPUT ENHANCEMENT
5. PHASE 2 RESULTS
5.1 The cartoon task
Subjects were assigned a score on the cartoon task, reflecting the number of
correctly formed questions (out of 20) that they produced. The mean number of
correctly formed questions for each group is presented in Table 2. Since pretest differences between the three instructed classes proved not to be significant
(F(2,76) = 0.61, p = 0.55), their results have been combined.
Table 2: Phase 2 cartoon
task: mean number of correctly formed questions
Croups
Instructed (n = 79)
Uninstructed (n = 29)
Native speakers (n = 29)
Post-test
Pre-lesl
Mean
SD
Mean
SD
11.32
11.55
3.25
2.86
15.87
12.35
19.21
2.94
3.25
0.9
Follow-up
Mean
SD
16.09
2.78
There is a clear effect for instruction on this task. While there is no significant
difference between the mean scores of the instructed and uninstructed groups
prior to instruction (/"(1,106) = 0.12, p = 0.73), a repeated measures ANOVA
shows a significant interaction between group and time of testing (F(l,105) =
44.71, p < 0.001). The scores of the instructed group improved much more
than those of the uninstructed. This is illustrated in Figure 2. The instructed
group does not, however, achieve native speaker accuracy; there is a significant
difference between the native speakers and both the instructed and the uninstructed groups at the post-test (F(2,134) = 46.13, p < 0.001).
Results from the follow-up test, taken only by the instructed classes, suggest
that the effects of instruction are not lost in the five-week period between the
post-test and the follow-up (see Table 2). There is a significant difference
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
which they were serving as a control group (White 1991a, 1991b). The teaching
and testing of questions was restricted to wh -questions, and there was a single
testing procedure (the correction task). In Phase 2, the main focus of the
research was on the effect of instruction on the development of questions, and
teaching and testing were expanded considerably. Both yes/no and wh -question
formation were taught and tested.
The Phase 1 test was a set of decontextualized sentences. As only one
response was required for each sentence, we could not tell whether learners
would in fact have accepted or produced other formulations of each sentence. In
Phase 2, we sought greater contexrualization (via the cartoon task) and insight
into whether subjects would accept a wider range of question forms (via the
preference task). The oral task was intended to permit investigation of the extent
to which knowledge demonstrated in the paper-and-pencil tasks would extend
to spontaneous communicative language.
L. WHITE. N. SPADA, P. LIGHTBOWN, AND L. RANTA
425
between the mean scores of the instructed group on the three test occasions
(pre-test, post-test, and follow-up) (F(2,77) = 209.84, p < 0.001). Post hoc
Scheffe procedures show that both the post-test and follow-up scores are
significantly higher than the pre-test scores ( p < 0.01), and there is no significant
decline from post-test to follow-up.
The distribution of responses (n = 4,273) according to type is given in
Figure 3. Responses were analysed in the same way as in Phase 1.' Before
instruction, 57 per cent of responses were fully grammatical. This increased to
Instructed: pre-test
El Instructed: post-test
Uninstructed: pre-test
Uninstructed: post-test
Figure 3: Phase 2 cartoon task: distribution of responses, in percentages
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
Figure 2: Phase 2 cartoon task: mean number of correctlyformed questions
426
INPUT ENHANCEMENT
80 per cent immediately after instruction, and there was a corresponding drop
in errors involving failure to invert and scrambled word orders. There was no
such change in the responses of the uninstructed group, suggesting once again
that instruction affects syntactic accuracy.
Table 3: Phase 2 preference task: mean number of correctly judged sentence
pairs
Groups
Instructed (n = 79)
Uninstructed (n = 29)
Native speakers (n = 29)
Pre-test
Post-test
Mean
SD
Mean
SD
8.47
8.17
2.4
2.42
11.67
8.2
17.45
3.13
2.7
1.7
Follow-up
Mean
SD
10.88
3.14
Results from the preference task show trends parallel to those from the
cartoon task, although accuracy on the preference task is somewhat lower than
on the cartoon task.8 At the pre-test there was no significant difference between
the instructed and uninstructed groups (F(l,106) = 0.32, p = 0.57). A repeated
measures ANOVA shows a significant interaction between group and time of
testing (F(l,105) = 23.85, p < 0.001). Only the scores of the instructed group
show improvement, as illlustrated in Figure 4.
As before, there was a significant difference between the native speakers and
both the instructed and uninstructed groups at the post-test (F(2,134) = 81.92,
p < 0.001). As far as the follow-up is concerned, there is a significant difference
between the mean scores of the instructed group on the three test occasions
(F(2,77) = 49.55, p < 0.001) (see Table 3). Post hoc Scheffe tests show that the
difference between the post-test and the follow-up is not significant at the 0.01
level, suggesting that the effects of instruction are lasting.
5.3 The oral data
The oral communication data was analysed in terms of the percentage of wellformed questions, where well-formedness is interpreted in terms of word order,
specifically correct placement of the subject and the auxiliary verb in yes/no
questions, and correct placement of the subject, the auxiliary, and the w/z-word
in wh -questions.9 Due to errors in the administration of the oral task to one of
the instructed groups and to the uninstructed group, results for these groups
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
5.2 The preference task
On the preference task, subjects were assigned a score reflecting the number of
correctly judged sentence pairs (out of the 20 sentence pairs that dealt with
question formation). The mean number of correctly judged sentence pairs is
presented in Table 3. Once again, the results of the instructed classes have been
combined, since there was no significant difference between their mean scores
at the pre-test (F(2,16) = 2.09, p = 0.13).
L. WHITE, N. SPADA, P. LIGHTBOWN. AND L.RANTA
427
could not be interpreted and will not be discussed here. The task has not yet
been administered to a group of native speaker^.'^' The results presented below
are for two of the instructed classes (n = 53). As there is no significant difference
in the mean scores for these two classes (F(1,5 1) = 0.21, p = 0.65) results will
be reported together.
Due to the elicitation procedure, the total number of questions produced by
the subjects was very similar at each test session. Each student asked approximately 17 questions on each test occasion. Table 4 shows the total number of
questions and accurate questions produced."
Learners produced a higher proportion of accurate questions on the post-test
than on the pre-test and there was further improvement on the follow-up. (See
Table 5 for mean percentages of accurate questions produced at each test.)
Analysis of variance shows that these differences are significant (F(2,104) =
27.48, p < 0.001). Post hoc Scheffe tests confirm that there is a significant
difference between pre-test and post-test and between pre-test and follow-up
scores at the 0.01 level, while the difference between post-test and follow-up is
not significant. The improvement in oral accuracy was widely distributed
throughout the subject group, as shown by the fact that 35 students showed
Table 4: Phase 2 oral communication task: number of
questions produced
Accurate questions
Total questions
Pre-test
Post-test
Follow-up
358
905
498
914
551
884
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
Figure 4: Phase Zprejierence task: mean number of correctlyjudged sentence pairs
428
INPUT ENHANCEMENT
Table 5: Phase 2 oral communication task: mean percentage of well-formed
questions
Groups
Instructed (n = 53)
Mean
SD
Mean
SD
Follow-up
Mean
SD
40.66
26.01
55.21
25.37
62.48
Post-test
Pre-test
21.52
6. DISCUSSION
In both phases of this study, instruction that focused on the formation of
questions had an immediate impact on syntactic accuracy. This impact was not
lost, at least during the five-week period preceding the follow-up testing. There
was no decline in performance on the paper-and-pencil tasks; furthermore, oral
performance improved during this time. These results differ from those of a
previous study on the effect of instruction on question formation (Ellis 1984).
In that study, positive examples of questions, as well as explicit correction, were
given over a short period of time. However, there was no explicit form-focused
instruction. In contrast, both form-focused instruction and corrective feedback
were provided in the present study, and instruction lasted for a longer period.
Both the Phase 1 task and the Phase 2 cartoon task were 'unscrambling' tasks.
There was a substantial difference in the results of these two tasks as far as the
proportion of fully grammatical responses is concerned. Students performing
the Phase 2 task before instruction were considerably more accurate than the
comparable uninstructed groups in the Phase 1 study. The explanation for this
seems to lie in the linguistic content of each task, rather than in the subjects'
general ability with English, judging by other test scores. Some of the accuracy
on the Phase 2 cartoon task may be attributable to a small set of stereotypes or
chunk-learned questions, which affected wh -questions starting with what is or
where is, and yes/no questions starting with do you and can I (the latter in the
case of two of the three classes).12 Subjects were highly accurate on sentences
like these, at the level of 87 per cent or more, even at pre-testing.
In the Phase 1 task, all sentence subjects were full noun phrases, not
pronouns. In contrast, the most natural questions for the Phase 2 cartoons
required pronouns rather than full noun phrases, because the referents were
clearly present in the context. The Phase 1 task, with exclusively noun phrase
subjects, led to responses in which failure to invert predominated. The Phase 2
task, on the other hand, with a majority of pronoun subjects, led to responses in
which inversion predominated, even at pre-testing.13 The results that we
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
increases in accuracy of at least 10 per cent from pre-test to follow-up; 15
students remained within 10 per cent of their pre-test scores; and only 3
students declined in accuracy by more than 10 per cent.
These results suggest that instruction had an effect on spontaneous oral tasks
as well as on the paper-and-pencil tasks reported above.
L. WHITE, N. SPADA, P. LIGHTBOWN, AND L. RANTA
429
7. CONCLUSION
In previous research with learners in these communicative meaning-based
programs, we have observed substantial growth in their language development
in the absence of form-focused instruction. Results from the present study
indicate that such learners can also benefit from input enhancement. Some may
dismiss the paper-and-pencil tasks as 'monitored' tasks, and thus less interesting
as an indication of the effects of instruction on interlanguage competence. But
an influence was also observed on students' oral performance in relatively
spontaneous or 'unmonitored1 tasks. We take these changes in oral performance
as evidence that input enhancement can bring about genuine changes in
learners' interlanguage systems.
(Revised version received March 1991)
ACKNOWLEDGEMENTS
This research was conducted with the assistance of a number of research grants from
FCAR and SSHRCC to research teams at McGill and Concordia. We should like to
thank all the teachers who participated, our research assistants, and Bill Rutherford,
Mike Sharwood Smith, Jack Upshur, and two anonymous reviewers for their comments
and advice.
NOTES
1
But see White (1991a, 1991b) for further discussion.
2
On the whole, then, form-focused instruction provides positive evidence, and error
correction provides negative evidence. However, form-focused instruction may also
provide negative evidence when it takes the form of rules which restrict what the language
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
obtained support Zobl's (1979) claim that francophones extend to English the
restriction against inversion in the case of noun phrase subjects and suggest that
the potential role of input enhancement is actually greater in the case of some
question forms than others. The effect of noun phrase versus pronoun subjects
is being investigated in our ongoing research.
The Phase 2 cartoon task suggests that a high degree of word order accuracy
is achieved as a result of teaching, in that the mean number of correctly formed
questions increases from 11.32 to 15.87. However, the cartoon task allows for
only one response to each picture. In this context, the preference task and oral
performance results are particularly interesting: they suggest that subjects
assume that English allows more word orders in question formation than is
actually the case. In the preference task, the incidence of 'both right' responses
to pairs of sentences where one of the sentences is uninverted drops after
instruction, but it does not disappear. In the oral data, cases of inversion and
lack of inversion are found in the same subject. In other words, although the
form-focused instruction increases their knowledge of the correctness of
inversion, this does not necessarily pre-empt non-inversion straight away, nor
other incorrect question forms. It is noteworthy that subjects hardly ever used
wh-in-situ in English, suggesting that this is not perceived as transferable
(Kellerman 1979).
430
INPUT ENHANCEMENT
REFERENCES
Corder, S. P. 196 7. 'The significance of learners' errors.' IRAL 5:161-70.
Day, E. and S. Shapson. 1991. 'Integrating formal and functional approaches in language
teaching in French immersion: An experimental study.' Language Learning 41/1:25—
58.
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
can do. For more detailed discussion of the issue of positive and negative evidence, see
White (1989).
3
There are, in fact, additional ways of forming questions in French. For example, the
subject noun phrase and the pronoun can occur together as in a and b:
a Jean, joue-t-il au tennis?
b Oil Marie demeure-t-elle?
Quebec French has another way of forming yes/no questions, by means of a particle
which is inserted after the finite verb, as in c. Subject and verb are not inverted. The form
of the particle is usually given as -tu, or -ti in some dialects.
c Elle vient-tu avec nous?
4
English also permits some use of non-inversion in questions but the pragmatic
assumptions which are implied are more marked than in French, where questions without
inversion are neutral and of high frequency.
5
In addition, as pointed out by Bill Rutherford (personal communication), there are
other input factors which might prevent pre-emption from taking place. For example,
English includes headless relative clauses, where the wh expression is fronted but
subject-auxiliary inversion does not take place, as in What we're going to do now is work
on our projects. Preliminary analyses of transcripts from teachers involved in our study
show that such sentences do indeed occur.
6
Preliminary analyses of the tapes indicate that the teachers covered the materials and
followed the guidelines. A detailed qualitative and quantitative analysis of the classroom
input and error correction provided by the teachers is currently underway.
7
In Phase 2, the response category 'No inversion' replaces the Phase 1 category 'WhS'.
'No inversion' includes failure to invert subject and auxiliary in both yes/no and whquestions. Subjects' response patterns to yes/no and wh -questions were very similar,
hence these two question types have been collapsed in our analyses.
8
As with the cartoon task, subjects' treatment of yes/no and wh -questions was similar,
so the question types are collapsed.
9
A second analysis of the oral performance data, based on the proposal by Pienemann,
Johnston, and Brindley (1988) for a six-stage sequence in the acquisition of English
questions, will be reported in Lightbown, Ranta, and Spada (1991).
10
These gaps are being filled by ongoing research.
11
Once again, results from yes/no and wh -questions have been collapsed. In general,
learners showed greater accuracy on wh -questions than yes/no questions. It is beyond the
scope of this paper to discuss the implications of this finding.
12
In the Phase 2 study, we found significant variation between the three classes on
questions with the auxiliary can. One class produced significantly more non-inverted can
questions than the others on both the pre- and post-tests. The explanation for this appears
to be that the teacher of this class had stressed (prior to the instruction on question
formation) that may should be used to ask permission, and can only to inquire about
ability. Our test included permission questions with can, and these students failed to invert
subject and auxiliary when the auxiliary was can, although using inversion in other cases.
13
Only four of the twenty sentences in the Phase 2 task had noun phrase subjects.
L. WHITE, N. SPADA, P. LIGHTBOWN, AND L. RANTA
4 31
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
Doughty, C. (in press) 'Second language instruction does make a difference: Evidence
from an empirical study of second language relativization.' Studies in Second Language
Acquisition 13.
Eckman, F. R., L. Bell, and D. Nelson. 1988. 'On the generalization of relative clause
instruction in the acquisition of English as a second language.' Applied Linguistics 9:120.
Ellis, R. 1984. 'Can syntax be taught? A study of the effects of formal instruction on the
acquisition of WH questions by children.' Applied Linguistics 5:138-55.
Gass, S. 1982. 'From theory to practice' in M. Hines and W.Rutherford (eds.): On
TESOL '57. Washington, DC:TESOL.
Harley, B. 1989. 'Functional grammar in French immersion: A classroom experiment.'
Applied Linguistics 10:331-59.
Harley, B. and M. Swain. 1984. 'The interlanguage of immersion students and its
implications for second language teaching' in A. Davies, C. Criper, and A. P. R.
Howatt (eds.): Interlanguage. Edinburgh: Edinburgh University Press.
Kellerman, E. 1979. 'Transfer and non-transfer: Where are we now?' Studies in Second
Language Acquisition 2:37-57.
Krashen, S. 1985. The Input Hypothesis: Issues and Implications. London: Longman.
Lightbown, P. M. 1991. 'What have we here? Some observations of the influence of
instruction on L2 learning' in R. Phillipson, E. Kellerman, L. Selinker, M. Sharwood
Smith, and M. Swain (eds.): Foreign Language Pedagogy Research: A Commemorative
Volume for Claus Feerch. Clevedon: Multilingual Matters.
Lightbown, P. M. and N. Spada. 1990. 'Focus-on-form and corrective feedback in
communicative language teaching: Effects on second language learning.' Studies in
Second Language Acquisition 12:429-48.
Lightbown, P. M., L. Ranta, and N. Spada. 1991. 'The effect of instruction on "1Lformed" questions.' Paper presented at the Second Language Research Forum,
University of Southern California, Los Angeles.
Long, M. and C. Sato. 1983. 'Classroom foreigner talk discourse: Forms and functions of
teachers' questions' in H. Seliger and M. Long (eds.): Classroom-oriented Research in
Second Language Acquisition. Rowley, Mass.: Newbury House.
Pienemann, M. 1984. 'Psychological constraints on the teachability of languages.'
Studies in Second Language Acquisition 6:186-214.
Pienemann, M., M. Johnston, and G. Brindley. 1988. 'Constructing an acquisitionbased procedure for second language assessment.' Studies in Second Language
Acquisition 10:217-44.
Pinker, S. 1984. Language Learnability and Language Development. Cambridge, Mass.:
Harvard University Press.
Rutherford, W. 1987. Second Language Grammar: Learning and Teaching. London:
Longman.
Rutherford, W. 1989. 'Preemption and the learning of L2 grammars.' Studies in Second
Language Acquisition 11:441-57.
Sharwood Smith, M. 1981. 'Consciousness raising and the second language learner.'
Applied Linguistics 2:159-68.
Sharwood Smith, M. 1991. 'Speaking to many minds: On the relevance of different types
of language information for the L2 learner.' Second Language Research 7:2.
Spada, N. and P. M. Lightbown. 1989. 'Intensive ESL programmes in Quebec primary
schools.' TESL Canada Journal 7/1:11-28.
Swain, M. 1985. 'Communicative competence: Some roles of comprehensible input and
INPUT ENHANCEMENT
432
APPENDIX
Example question from the Phase 2 cartoon tahk:
Example question from the Phase 2 prcfercnce task:
a. What i \ your brother doing'.'
h. What l o u r brother i doing'!
Only A
I\
correct
A and B arc correct
Onl? R I \ corrcct
A and R arc ~ncorrect
Don't hnou
Downloaded from http://applij.oxfordjournals.org/ at Penn State University (Paterno Lib) on September 12, 2016
comprehensible output in its development' in S. Gass and C. Madden (eds.): Input in
Second Language Acquisition. Rowley. Mass.: Newbury House.
Tomasello, M. and C. Herron. 1988. 'Down the garden path: Inducing and correcting
overgeneralization errors in the foreign language classroom.' Applied l'svcholinguistics 9237-36.
Tomasello, M. and C. Herron. 1989. 'Feedback for language transfer errors.' Studies in
Second Language Acquisirion 1 I :385-Y 5.
White, J. and P. M. Lightbown. 1984. 'Asking and answering in ESL classes.' Chnudiun
Modern Language Keview 4O:2 28-44.
White, L. 1989. Univer.su1 (;rirmmur and Second Language Acquisition. Amsterdam:
John Benjamins.
White, L. 1Y9la. 'Adverb placement in second language acquisition: some effects of
positive and negative evidence in the classroom.' Second Lat~guageKeseurch 7:1336 1.
White. L. 199lb. 'The vcrb-movement parameter in second language acquisition.'
Language Acquisition 1:337-60.
Zobl, H. 1979. 'Nominal and pronominal interrogation in the speech of adult
francophone ESL learners: Some insights into the workings of transfer.' SPEAQ
Journal 3/1-269-93.
Zobl, H . 1985. 'Grammars in search of input and intake' in S. Ciass and C'. Madden (eds.):
Input it1 Second Language Acquisition. Rowley, Mass.: Newbury House.