Does negotiation of meaning promote reading comprehension? A

Reading Research Quarterly
Vol. 35, No. 3
July/August/September 2000
©2000 International Reading Association
(pp. 426–443)
Does negotiation of meaning promote
reading comprehension? A study of
multilingual primary school classes
Kris Van den Branden
Catholic University, Leuven, Belgium
I
(i.e., manipulated with regard to its level of difficulty prior to the moment when the reader is confronted with it).
The most common form of premodification involves
simplifying the input so that it meets the reader’s supposed level of language proficiency. The complexity of
the written input is downgraded by reducing the number
of words in sentences, reducing the number of embedded sentences, and avoiding advanced syntactic constructions. Vocabulary is simplified by avoiding technical
jargon, abstract terminology, idiomatic expressions, and
words with which the reader is supposed to be unfamiliar. Similarly, topics with which the reader is not acquainted are dismissed. For all their manipulation of form
and content, educational reading materials have often
been criticized for being unnatural and boring models of
language.
In classrooms where reading instruction is used as a
way to promote language proficiency (besides literacy),
such as in second or foreign language classes and in primary school, the simplification of written input may prove
counterproductive. Long (1987) has called this the logical
problem of simplifying input: If structures and lexical
items with which the readers are unfamiliar are removed,
how can they learn language from it? In other words, if
texts are stripped of all that is new and unknown, they
may be useful as input for comprehension only, but not
as input for learning (cf. Faerch & Kasper, 1986).
As an alternative to the simplification of input, other
kinds of manipulations to reading materials have been
n this article a quasi-experimental study is presented
that focused on the effect of negotiation of meaning
on reading comprehension. The study was prompted by the growing attention that has been paid to
negotiation of meaning in empirical studies and theoretical publications on language education and language acquisition. However, the available research on negotiation
of meaning has primarily focused on the oral mode. Few
attempts have been made to study to what extent negotiation of meaning can be useful in reading instruction,
and what actions teachers should take in order for negotiation of meaning to optimally promote students’ reading
comprehension.
Theoretical background
Recent theories of first- and second-language acquisition have stressed the learning potential that is inherent
in comprehension problems (cf. Ellis, 1994; Gallaway &
Richards, 1994). Inferring the meaning of unfamiliar
words or expressions, either through the use of context
or through interaction with an interlocutor, may be the
first step for learners toward acquiring these words or
expressions.
In L1- and L2-reading instruction, however, one can
often observe a tendency to avoid comprehension problems, rather than to exploit them. In many commercially
published reading materials as well as in texts that teachers prepare by themselves, input tends to be premodified
426
ABSTRACTS
Does negotiation of meaning promote reading comprehension?
A study of multilingual primary school classes
During the past decade, a body of empirical evidence has become
available showing that negotiation of meaning promotes the comprehension of oral input. However, to what extent negotiation of
meaning can be useful in reading instruction in order to make written input comprehensible remains an open question. A quasiexperimental study was carried out in 8 multilingual primary schools
in Flanders. The participants were confronted with a difficult text in
4 conditions: unmodified written input, premodified written input,
unmodified written input + oral negotiation with a peer, and unmodified written input + oral negotiation with the rest of the class.
The results of the study show that negotiating the meaning of un-
modified written input led to higher comprehension than premodifying the same input. In turn, meaning negotiation in which the
teacher was involved was superior to peer negotiation. A further
analysis of peer negotiation showed that comprehension scores were
higher for students who had cooperated with a peer of a different
level of language proficiency than for students who had cooperated
with a peer of a similar level of language proficiency. The study
also yields useful guidelines as to the role teachers should play and
the actions they should take when the meaning of written input is
negotiated in the classroom.
¿La negociación del significado promueve la comprensión lectora? Un estudio de aulas
multilingues de escuela primaria
Durante la década pasada, se ha proporcionado un cuerpo de evidencia empírica que muestra que la negociación del significado promueve la comprensión del input oral. Sin embargo, hasta qué punto la negociación del significado puede ser útil en la enseñanza de la
lectura mejorando la comprensión del input escrito, es una cuestión
que permanece abierta. Se llevó a cabo un estudio cuasiexperimental en 8 escuelas primarias multilingües de Flandes. Se
confrontó a los participantes con un texto difícil en 4 condiciones:
input escrito no modificado, input escrito previamente modificado,
input escrito no modificado + negociación oral con un par e input
escrito no modificado + negociación oral con el resto de la clase. Los
resultados del estudio mostraron que la negociación del significado
de un input escrito no modificado resultaba en una mayor comprensión que la modificación previa del mismo input. A su vez, la
negociación del significado en la que estaba involucrado el docente
fue superior a la negociación entre pares. El análisis posterior de la
negociación entre pares mostró que los puntajes en comprensión
fueron más altos para los estudiantes que habían cooperado con un
par de un nivel diferente de dominio lingüístico que para los estudiantes que habían cooperado con un par de un nivel similar de
dominio lingüístico. El estudio también proporciona lineamientos
útiles acerca del rol que deben jugar los docentes y las acciones
que deben realizar cuando se negocia el significado del input escrito en el aula.
Fördert die Auseinandersetzung mit der Begriffsdeutung das Leseverständnis?
Eine Studie über mehrsprachige Erste Klassen
Während des vergangenen Jahrzehnts wurde eine Fülle an empirischem Beweismaterial verfügbar gemacht, welches aufzeigt, daß
die Auseinandersetzung mit Begriffsdeutungen das Verstehen
mündlicher Anweisungen fördert. In welchem Ausmaße jedoch die
Auseinandersetzung mit Begriffsdeutungen beim Lesen von
Anweisungen nützlich sein kann, um schriftliche Anweisungen besser begreiflich zu machen, bleibt eine offene Frage. Eine quasiexperimentelle Studie wurde in 8 mehrsprachigen Grundschulen in
Flandern ausgeführt. Die Teilnehmer wurden mit einem schwierigen Text unter 4 Bedingungen konfrontiert: unveränderte schriftliche
Anweisungen, vorab veränderte Anweisungen, unveränderte Anweisungen + mündliche Auseinandersetzung mit einem Mitschüler,
und nicht modifizierte schriftliche Angaben + mündliche Auseinandersetzungen mit dem Rest der Klasse. Die Resultate der Studie
zeigen, daß die Auseinandersetzung mit der Begriffsdeutung von unveränderten schriftlichen Anweisungen zu höherem Verstehen
führten, als im voraus gemachte auferlegte Veränderungen der gleichen Anweisungen. Andererseits waren die Auseinandersetzungen
mit der Begriffsdeutung mit erweiterter Einbeziehung des Lehrers
den nur aus Mitschülern bestehenden weit überlegen. Eine weitere
Analyse der Auseinandersetzung unter Mitschülern zeigte, daß die
Verständnisbewertung höher bei Schülern war, die mit einem
Schüler differenzierter Stufe im Sprachverständnis zusammenarbeiteten, als bei jenen Schülern mit Mitschüler auf gleicher Stufe im
Sprachverständnis. Die Studie erbrachte auch nützliche Richtlinien,
welche Rolle die Lehrer dabei spielen sollten und der Maßnahmen,
die ergriffen werden sollten, wenn die Bedeutung von schriftlichen
Eingaben im Klassenzimmer vermittelt werden.
427
ABSTRACTS
La négociation du sens améliore-t-elle la compréhension de la lecture ?
Une étude de classes primaires plurilingues
Au cours de la dernière décennie, un ensemble de données empiriques est apparu qui a montré que la négociation du sens favorise
la compréhension de l’input oral. Cependant, la question demeure
de savoir dans quelle mesure la négociation du sens peut être utile
dans l’enseignement de la lecture pour rendre compréhensible l’input écrit. On a entrepris une étude quasi-expérimentale dans 8
écoles primaires multilingues des Flandres. Les participants ont été
confrontés à un texte difficile dans 4 conditions : input écrit non
modifié, input écrit modifié au préalable, input écrit non modifié +
négociation orale avec un pair, et input écrit non modifié + négociation orale avec le reste de la classe. Les résultats de l’étude montrent que la négociation du sens de l‘input non modifié conduit à
une meilleure compréhension que la modification préalable du
même input. Par ailleurs, la négociation du sens dans laquelle le
maître était impliqué a été supérieure à la négociation avec un pair.
Une analyse ultérieure de la négociation avec les pairs a montré
que les scores de compréhension étaient plus élevés pour les élèves
qui avaient coopéré avec un pair ayant un niveau différent de
maîtrise du langage que pour les élèves ayant coopéré avec un pair
ayant un niveau de langage équivalent. L’étude fournit également
des indications utiles quant au rôle que devraient jouer les enseignants et comment agir quand le sens de l’input écrit est négocié
dans la classe.
428
suggested, elaboration being one of the most prominent
(Yano, Long, & Ross, 1994). This latter kind of premodification involves maintaining a sufficient level of lexical
and syntactic complexity, but compensating for this by
clarifying message content and structure (e.g., by removal
of pronouns with unclear antecedents and by highlighting important concepts) and by adding redundancy (e.g.,
through the use of repetition and paraphrase and the retention of full noun phrases). Some empirical evidence is
available that elaborated texts, while keeping the language learning potential intact, are as comprehensible as
simplified texts (Parker & Chaudron, 1987; Yano et al.,
1994).
However, elaborations have to be carefully chosen,
for they may also miss their target. That is, instead of promoting comprehension, elaborations may add to confusion and nonunderstanding. Some empirical evidence
exists on modifications having a negative effect on comprehension (Derwing, 1996; Ehrlich, Avery, & Yorio,
1989). In particular, the use of irrelevant detail and unnecessary repetition, both elaborative modifications, may
lead to confusion rather than comprehension.
Simplification and elaborative modification may further miss their target in that they only work for a part of
the readers they are supposed to support. Many course
designers have a fairly homogeneous target group in
mind when developing reading materials. But, in many
classrooms, especially in multilingual ones, learners may
differ widely in terms of such things as language proficiency, reading proficiency, learning styles, knowledge of
the world, or knowledge of text schemata. As a consequence, text modifications may suit only the needs of a
happy few in a particular classroom; the same text may
be too easy for some students and too difficult for others
in the same classroom. It is up to the teacher, then, to
compensate for this through interaction—for instance,
through the negotiation of meaning.
Negotiation of meaning stands for the joint efforts
that interlocutors make in oral or written interaction to
deal with problems of message comprehensibility (Pica,
1994; Varonis & Gass, 1985). Pica (1994) defined negotiation of meaning as
[T]he modification and restructuring of interaction that occurs when learners and their interlocutors anticipate, perceive, or experience difficulties in message comprehensibility. As they negotiate, they work linguistically to
achieve the needed comprehensibility, whether repeating
a message verbatim, adjusting its syntax, changing its
words, or modifying its form and meaning in a host of
other ways. (p. 494)
Interlocutors may negotiate the meaning of something
one of them said or wrote, as in the excerpt below, or
Negotiation
they may negotiate the meaning of something said or
written by a third party. The latter typically happens in
classrooms, for instance, when teachers and their students discuss the meaning of difficult words in a text
from the coursebook.
English native speaker:
Okay, with a big chimney.
English L2 learner:
What is chimney?
English native speaker:
Chimney is where the smoke
comes out of.
(Pica, Holliday, Lewis, Berducci, & Newman, 1991, p. 345)
Problems of message comprehensibility may be
caused by several factors: (a) by a gap between the listener’s or reader’s level of language proficiency and the
input with which he or she was confronted, (b) by the
fact that the listener or reader was not paying (enough)
attention to the input, and (c) by the fact that the message was not conveyed clearly enough by the speaker or
writer. Meaning negotiations, aimed to restore mutual
comprehension, are not necessarily successful (Van den
Branden, 1997); in fact, they can drag on for quite some
time without any effect, or can result in even bigger confusion and noncomprehension. Moreover, explicit signals
of comprehension, like “hmmm” or “I see,” which are often used to round off a meaning negotiation, should not
be taken at face value. People sometimes feign understanding (Aston, 1986; Hawkins, 1985), and this they do
for different reasons: (a) to avoid looking stupid or being
impolite, (b) to avoid making a bad impression, (c) to
save their interlocutor’s face, or (d) to move on with the
conversation. In general, then, negotiation of meaning always refers to a process and not to a certain outcome.
By enhancing comprehension, negotiation of meaning has been claimed to promote language acquisition
(Golinkoff, 1983; Long, 1985; Pica, 1994; Wells, 1985).
The argument is that, through negotiation of meaning,
language learners can come to comprehend input that
lies above their current level of language proficiency. In
its turn, the comprehension of input, especially input that
the learner at first does not understand, is generally regarded as one of the crucial factors involved in language
acquisition (Krashen, 1985; Pine, 1994; White, 1987).
From a conceptual point of view, the claims on the
positive impact of negotiation of meaning on language
acquisition tie up with a view of (language) learning as
an essentially social phenomenon (Bruner, 1978;
Vygotsky, 1978; Wells, 1985). Following Vygotsky
(1978), a distinction can be drawn between learners’ actual developmental level and their potential developmental level. The actual developmental level defines the limits
of what learners, at any moment in their development,
can do and understand on their own through indepen-
429
dent problem solving. Problems that learners cannot yet
tackle on their own may act as a powerful incentive to
overcome these limits, at least if the problems are dealt
with. Social interaction may be very facilitative in this respect: Through social interaction, especially with more
knowledgeable partners, problems may be not only well
defined and made explicit, such that learners have a
clearer view of the gap between their current knowledge
or skills and the knowledge or skills demanded by the
task, but also solved. Experiencing how new problems
may be solved, or actively collaborating with others in
doing so, may guide learners toward independently solving similar problems in the future. Vygotsky (1978) argued that, in this way, social interaction may act as a
steppingstone and even be a prerequisite for independent developmental achievement: What learners learn to
do today in interaction with more knowledgeable partners, they will internalize and do independently tomorrow. Or, in terms of input comprehension, what learners
comprehend today in interaction (negotiation) with more
proficient partners, they will internalize and comprehend
independently tomorrow.
An extensive corpus of empirical evidence is now
available showing that negotiation of meaning indeed enhances the comprehension of input. This research pertains to both the comprehension of L2 input in
second-language acquisition (cf. Ellis, Tanaka, &
Yamazaki, 1994; Gass & Varonis, 1994; Loschky, 1994;
Pica, 1994; Pica, Doughty, & Young, 1986; Pica, Young, &
Doughty, 1987) and L1 input in first-language acquisition
(Pine, 1994; Wells, 1985). In some of these studies, comprehension of negotiated input was superior to comprehension of premodified versions of the same input.
Though steadily growing, the body of empirical evidence is still restricted in scope. Most of the abovementioned studies have focused exclusively on the
comprehension of oral input. With regard to reading
comprehension, very little empirical evidence substantiating the effect of negotiation of meaning is available.
Some indications can be inferred from the vein of literature that has dealt with discussing texts in the classroom
(cf. Bloome & Green, 1992; Eeds & Wells, 1989), but as
Almasi (1995) rightly pointed out, “many forms of literary
discussion exist and it is unclear how different patterns of
interaction between students and teachers affect students
as they engage with text” (p. 314).
As to the many forms of literary discussion Almasi
(1995) referred to, differences have to do with (among
other things) the following four considerations:
1. What level or aspect of the text is discussed. For instance, discussions may focus on particular linguistic elements of the text, such as key vocabulary and
grammatical constructions; they may highlight the role of
430
extralinguistic elements, such as illustrations; or they may
focus, for example, on the main ideas of the text or on
the reception of the text by different readers.
2. Who is involved in the discussion. One of the
main distinctions in this respect is the one between peer
interactions, in which students are asked to discuss the
text with other students, and teacher-student interactions,
in which the teacher is involved, together with all or
some of the students.
3. What is expected of the participants in the discussion. A wide variety of discussion formats exists, both in
the literature and in the classroom. Students may be expected to read the text aloud, explain the meaning of difficult words, summarize passages they have read, predict
what the text will be about on the basis of the title or the
illustrations, practice particular reading strategies, express
their personal opinion on issues raised in the text, or do
a combination of these. Similarly, the teacher may function as a role model (as in the first stages of reciprocal
teaching) or demonstrate reading strategies or particular
ways of reading aloud. Or the teacher may function
much more in the background, supporting students in
coping with texts themselves, or in explicating their own
interpretation of the text.
4. What possible effects might be expected. Potential
effects of discussions have been formulated, for example,
in terms of enhanced comprehension, enhanced reading
skill, enhanced skill in performing certain reading strategies, and enhanced reading motivation. The available empirical evidence is mixed, sometimes hard to interpret,
and fragmentary. For instance, while Almasi’s (1995) research favored peer-led interactions over teacher-led interactions because teachers tend to dominate text
discussion too much, Brown and Palincsar’s (1989) research on the Reciprocal Teaching Model emphasized the
role of the teacher as a model. The comparison between
these two research studies also yields mixed evidence as
to whether students first need to receive direct instruction
in how to discuss a text (as the Reciprocal Teaching
Model seems to suggest) or not, and as to whether discussion or negotiation has favorable effects on reading
comprehension irrespective of the genre of the text involved. Similarly, while Brown and Palincsar’s (1989) research pointed to the effect of focusing on main ideas in
text comprehension, Reutzel and Hollingsworth (1988)
stressed the impact of focusing on elements at the microlevel (i.e., key vocabulary) in order to enhance global
reading comprehension. In all, then, which aspects of
discussions have what effect on different aspects of reading skill and comprehension remain unclear, while at the
same time text discussion constitutes very current educational practice.
READING RESEARCH QUARTERLY July/August/September 2000 35/3
Apart from the heavy emphasis on oral input, the
body of empirical evidence on the effects of negotiation
of meaning is limited in other respects as well. Most studies that have been conducted in this area have involved
the use of English as a second language. Few other languages or cultures have been involved. In addition, almost all researchers have worked with adult learners. The
effect of negotiation of meaning on input comprehension
by young children has not yet been the subject of any
study. The youngest participants that have been involved
in the available research are adolescents (Ellis et al.,
1994). Moreover, in most studies, the number of subjects
was quite limited. Finally, most of the studies have been
conducted in so-called laboratory settings; learners were
usually carefully selected along criteria such as level of
language proficiency, ethnic background, and gender,
and were often externally motivated (even paid) to participate. In general, it has been taken for granted far too
easily that research results observed in a tightly controlled
experimental design can simply be transferred to the unpredictable, chaotic context of the real-life heterogeneous
language classroom.
As a result, how useful negotiation of meaning can
be in reading instruction remains largely an open question. Is negotiation of meaning just a compensatory device for premodification of reading texts, primarily
addressing students whose level of language proficiency
does not match the difficulty level of the text? Or does
negotiation of meaning constitute a viable alternative to
the premodification of written input, in the sense that
texts have to be only roughly tuned to, or above, the
learners’ level of language proficiency and that input has
to be made comprehensible through negotiation? In both
cases, but especially in the latter, it is essential that empirical evidence be found on when and how classroom negotiation works best for individual learners and on what
actions teachers should take (and avoid) for negotiation
of meaning to promote all learners’ comprehension of
written input.
In view of the foregoing, a quasi-experimental
study was carried out that (a) focused on the effects of
negotiation of meaning on the comprehension of written
input; (b) involved the use of Dutch instead of English;
(c) involved the participation of primary school pupils;
and (d) attempted to link up with real-life classroom conditions, in that the classes involved remained intact so
that learners at different levels of (Dutch) language proficiency were part of the same group and learners were
not externally motivated to participate.
The two main research questions the study aimed at
answering were (a) Does negotiation of meaning promote the comprehension of Dutch written input by primary school pupils and (b) under which conditions does
Negotiation
negotiation of meaning optimally promote the comprehension of written input in the context of the real-life language classroom.
Method
Participants
The study participants were 151 primary school
children. Their ages ranged from 10 to 12 years. There
was an even division of gender: 76 of the 151 pupils
were boys, the other 75 pupils were girls. The children
were spread over the fifth grades of eight Flemish primary schools; throughout the study, the children remained
in their (intact) classes. In Flemish primary schools girls
and boys are in the same class together. The average
number of pupils in a class ranges between 15 and 25.
Children are assigned to grades according to their year of
birth; only pupils who fail to reach a minimum level of
proficiency have to stay down a grade. Flemish primary
school classes are heterogeneous from a socioeconomic
point of view: Pupils of widely differing socioeconomic
backgrounds are in the same schools. During the past 2
decades, schools have also become increasingly heterogeneous from an ethnic point of view following the influx of (mainly Moroccan, Turkish, and Italian) migrant
pupils. These latter pupils do not acquire Dutch, which is
the main medium of instruction, as their mother tongue.
They face the considerable task of receiving an education
in a language that is not their own. This is reflected in
their academic results; on average, nonnative speakers of
Dutch perform relatively more poorly in Flemish education than native speakers. In the sample, 92 pupils (61%)
were native speakers (NSs) of Dutch. The remaining 59
pupils (39%) were nonnative speakers (NNSs) of Dutch,
most of them children of Moroccan immigrants who arrived in Flanders from the 1960s onwards.
The participants’ level of Dutch language proficiency was established prior to the actual experiment by
means of an editing test (see Appendix A). Editing tests
are reading comprehension tests that are constructed by
inserting words (at random) from one text into another.
The participants’ task is to scrap the superfluous words
and in this way to restore the original text. This test revealed that the eight classes involved were highly comparable as far as their mean level of Dutch language
proficiency was concerned. The test further showed that,
in line with this general tendency, the nonnative speakers’ mean level of Dutch language proficiency was statistically significantly lower than the native speakers’
(independent t-test; p < .001).
On the basis of this editing test, the pupils were assigned to one of four Dutch proficiency levels: very high
431
(level A), moderately high (level B), moderately low (level C), and very low (level D). To establish these levels,
the grand mean and the standard deviation of the editing
test were used. Level A, the highest level, included scores
higher than the grand mean plus 1 standard deviation;
level B included scores between the grand mean and the
grand mean plus 1 standard deviation; level C included
scores between the grand mean and the grand mean minus 1 standard deviation; and level D included the scores
lower than level C. As Table 1 shows, the majority of
nonnative speakers of Dutch (72.9%) belonged to levels
C and D while the majority of native speakers (77%) belonged to levels A and B. Still, about a fourth of the nonnative speakers’ Dutch language proficiency was
relatively high (level B), whereas about a fourth of the
native speakers showed a relatively low Dutch language
proficiency (levels C or D).
Materials
The pupils were confronted with a reading task. For
the occasion of the study, the researcher wrote a Dutch
detective story. This genre was chosen after oral interviews with the teachers of the eight classes; the teachers
showed a clear consensus that detective stories were the
children’s favorite when it came to reading. Hence,
chances were that the students would be highly motivated to try and understand the input with which they were
confronted.
The story, called “Harry in het Stormkasteel” (Harry
in the Stormcastle), comprised 12 short chapters, each of
approximately the same length. The plot of the story is
quite simple. Harry, a detective, is told that a treasure is
hidden in the Stormcastle. The 12 chapters report on
Harry’s adventures during his search. The same pattern
recurs on every page. Harry finds himself in one of the
chambers of the castle, confronted with the choice of
where to go next. In all chambers, there are five possibilities, yet only one is completely safe and leads the detective on the right track to the treasure. From the
Table 1
NSs’ and NNSs’ level of Dutch language
proficiency
Level A
Level B
Level C
Level D
Total
19
52
18
3
92
NNSs
0
16
21
22
59
Total
19
68
39
25
151
NSs
Note. A = score higher than the grand mean plus 1 standard deviation on the editing test; B = score between grand mean and level A scores; C = score between level B scores and the grand mean minus 1 standard deviation; D = score lower than
level C scores.
432
information that Harry gathers on the way, the reader can
infer which of the five possibilities is the right one.
It is this question with which the pupils are confronted in a comprehension test that follows each chapter. The comprehension test focused on the students’
comprehension of the main ideas in the chapter. As in a
multiple-choice test, the pupils have to tick the right one
of five possible answers to the question, “What is Harry
to do next?” Five, rather than three, possibilities were offered to diminish the chances of guesswork. The chapter
immediately preceding the test provided Harry (and,
through Harry’s eyes, the reader) with crucial information
on each of these possible answers. To cope with the test,
then, comprehension of the chapter was essential. An example of a part of a chapter of the story and its accompanying comprehension test is provided in Appendix B.
A first version of the chapters and the accompanying tests showed low reliability coefficients (Cronbach’s
alpha, .71). Some of the chapters and the comprehension
test items were revised as a result. The final version
showed much higher reliability both in a pilot study and
in the actual study (Cronbach’s alpha, .91).
Procedure
Four reading conditions were distinguished, two of
which involved some form of negotiation of meaning
while the other two did not.
1. In the unmodified input condition the pupils
were asked to read a chapter of the story in silence and,
after having finished reading this chapter, to solve the accompanying comprehension test. After tackling the test
and giving it to the researcher, the pupils were handed
the next chapter.
2. In the premodified input condition (see
Appendix B for an example) the readers were confronted
with a modified version of the original chapters.
Modifications were made on three levels: (a) vocabulary
was simplified so that approximately 95% of the words
were covered by a basic word list of familiar words for
fifth graders; (b) syntax was simplified by a reduction of
the number of long and embedded sentences; and (c) a
greater proportion of anaphoric references through verbatim repetitions, rather than through the use of other
devices such as pronouns, was introduced. Input modifications, then, included both simplifications (vocabulary
and syntax) and elaborative modifications (the verbatim
anaphoric references).
3. In the collective negotiation condition the pupils
first read an unmodified chapter individually. When all
the pupils had finished reading, they were offered the
opportunity to negotiate about the meaning of unfamiliar
words and phrases with the rest of the class and the
teacher. The teacher did not provide any kind of direct
READING RESEARCH QUARTERLY July/August/September 2000 35/3
instruction on how to negotiate (as in the Reciprocal
Teaching Model), nor did he or she explain unfamiliar input by providing synonyms or paraphrases. Instead, the
teacher incited the pupils to search for the meaning of
unfamiliar input themselves. This choice was deliberately
made in order to obtain empirical data on classroom negotiation of written input that were complementary to
Brown and Palincsar’s (1989) and Almasi’s (1995). As
Brown and Palincsar’s research suggested, the teacher’s
support may have an important impact on the students’
text comprehension, but in view of Almasi’s findings, this
effect may be greatest if the teacher refrains from dominating the discussion too much. So it was up to the
pupils in the collective negotiation condition to decide
which elements of the text would be discussed. The
teacher in the collective negotiation condition did give
hints if necessary and endorsed the correct meaning of
difficult input if it had been formulated by a pupil. Only
if very few or no indicators of noncomprehension arose
did the teacher take the initiative and check the pupils’
comprehension of keywords. The moment this collective
negotiation was closed, the pupils individually tackled
the comprehension test linked to the chapter and then
proceeded with the next chapter. To control the teacher
variable for this condition, the researcher took over this
role in all eight classes. Seated in the back of the classroom, the teacher of each class witnessed the experiment. The interventions of the researcher during this
collective negotiation were based on the flowchart in
Appendix C.
4. In the pair negotiation condition the pupils read
the unmodified version but were instructed, before being
confronted with the comprehension test, to negotiate
with another pupil about the meaning of incomprehensible words and phrases. Again, the pupils themselves decided which words and expressions they would focus on.
For this condition pupils were teamed up with a friend of
theirs, so as to make sure that no one felt inhibited to ask
for clarification or to exhibit noncomprehension of parts
of the text (following Varonis & Gass, 1985). Pupils were
asked to whisper, and desks were set apart as much as
possible to avoid pupils overhearing the negotiations of
other pairs. As soon as the two pupils of a pair agreed
that they could round off their negotiation, they raised
their hands. Each pupil was then given the comprehension test of the chapter at issue. The pupils tackled this
test individually.
The participants were not assigned randomly to one
particular condition, but all of them worked through all
conditions (repeated measures design). The 12 chapters
of the story were divided into four sections of three chapters, each section corresponding with one of the four
above-mentioned conditions. So, all pupils had to read
Negotiation
three chapters in the unmodified input condition, three
chapters in the premodified input condition, and so on.
The order in which the participants worked through the
various conditions was counterbalanced. Whereas one
class read section one (the first three chapters) in the unmodified input condition, another class read the same
section in the premodified input condition, still another
class in the pair negotiation condition, and so on. Since
each chapter could yield the subjects 1 point (i.e., a correct answer to the comprehension test), the maximum
score for each section (condition) was 3, and the maximum score on the whole experiment was 12. Throughout, alpha levels were set at .05.
A preliminary pilot study, involving two fifth grades
that were not in the actual experiment, showed that the
level of difficulty of the four sections was highly comparable. Moreover, the same pilot study showed that the input of all sections lay above the participants’ current level
of language proficiency. This was necessary to be able to
maximally assess the effects of the various conditions on
input comprehension.
The pilot study interactions were audiotaped, both
between pupils in the pair negotiation condition and between the class and the teacher during the collective negotiation condition. The following excerpt typically
illustrates the negotiations between the pupils in the pair
negotiation condition during the pilot study:
A:
Hey, what’s dowsiness? (Note: pupil misreads drowsiness)
B:
Huh?
A:
Dow.... Here (reading aloud) the rhinoceros wading
in the moat of the castle, find itself in a (...) continuous state of do... drowsiness....
B:
Hmmm...I see, uh (coughs), drunk, drunk maybe,
don’t know, I think, maybe, drunk....
A:
How drunk? From drinking the water of the moat?
From drinking whisky maybe (laughter)? Burps
burps....
B:
Yeah, but if he’s drunk, he sees everything double,
and he’s a bad guard. Look here, (reading) as a guard
of the castle it can be ruled out, maybe because he’s
drunk all the time.
A:
Uhu. So so, Harry shouldn’t get on his back?
B:
Huh?
A:
If he gets on his back, he is so drunk that he goes the
wrong way.
B:
Yeah, maybe.
The excerpt shows that, in line with the instructions that
they received, the pupils usually started negotiating the
meaning of a word or expression with which one of
them was unfamiliar. Sometimes they gradually shifted to
433
a negotiation of content (i.e., to a discussion of what
Harry the detective was supposed to do). However, for
the latter question to be solved, unraveling the meaning
of the difficult input was necessary. In other words, negotiating the meaning of key words and phrases was
task-essential.
The pilot study, however, also revealed that some
of the pupils—especially less proficient pupils—were
quite impressed with, and inhibited by, the recording
equipment. When they listened to the recordings, the
pupils’ teachers called the interactional behavior of some
of the pupils “atypical.” This also applied to recordings of
the collective negotiation condition (an excerpt from one
of the collective negotiations in the pilot study is included in Appendix D). Therefore, I decided not to record
the interactions during the actual experiment. Although
much information as to what exactly went on in the negotiations would inevitably be lost, the risk that recording
the interactions could decrease the potential effect of the
negotiations was decisive. Practical considerations further
supported this decision. Recording pupils speaking in a
low voice demands high-quality equipment, and this was
not available for the full period of the experiment.
Furthermore, recording the children in the experiment
would have taken time, in each of the classes, to familiarize them with the recording equipment; however, this
time was lacking. All the same, there was no reason to
believe that the interactions during the actual study
would be any different from those recorded during the
pilot study.
The pilot study further showed that the children, especially those who were not inhibited by the recording
equipment, were quite motivated to perform the task.
The eagerness with which the pupils read the chapters
and tried to come up with the proper advice for Harry
the detective, as well as their praise of the task and the
story after the experiment was rounded off, clearly justified the researcher’s choice for this kind of text.
Table 2
Mean scores and standard deviations on three
comprehension test items in the
four conditions
M
SD
Unmodified input
Premodified input
Collective negotiation
Pair negotiation
0.93
1.26
2.37
1.56
.76
.91
.63
.75
Total
6.12
1.98
434
Results
The results of the comprehension tests revealed a
consistent picture. As Table 2 shows, all classes, irrespective of the order in which they worked through the conditions, showed the lowest mean scores in the unmodified input condition. In this condition, pupils produced
less than one out of three correct responses. Mean scores
for the premodified input condition were better than for
the unmodified input condition, but not as good as for
the two negotiation conditions. With regard to these two
latter conditions, all groups showed the highest mean
scores in the collective negotiation condition. In the pair
negotiation condition the mean number of correct responses amounted to 50% of all responses given, while in
the collective negotiation condition the pupils provided a
correct response in more than two thirds of the comprehension test items.
An analysis of variance with repeated measures
showed a statistically significant main effect for condition
(F = 124.81, df = 3, p < .001). Post hoc analyses were carried out to explore the pairwise comparisons between
the means of the four conditions. The following conclusions can be drawn:
1. The pupils in the study displayed statistically significantly higher comprehension scores when they were
confronted with a premodified version of the text than
when confronted with the unmodified version (p < .001).
So, the input modifications that were selected in this
study, including both simplification and elaboration, did
enhance comprehension.
2. The pupils in the study displayed statistically significantly higher comprehension scores when they were
offered the chance to negotiate on the meaning of difficult words and phrases than when merely exposed to the
unmodified version. Both collective negotiation and pair
negotiation yielded such statistically significant effects
(p < .001).
3. The pupils displayed statistically significantly
higher comprehension scores when they had negotiated
on the meaning of difficult words and phrases than when
working on the premodified version (p < .001). This applies to both negotiation with another pupil (pair negotiation) and negotiation with the teacher and the rest of the
class (collective negotiation).
4. Comparing the two conditions involving negotiation of meaning, the collective negotiation condition
yielded statistically significantly higher comprehension
scores than the pair negotiation condition (p < .001).
To explore whether these overall results applied to
pupils at different levels of Dutch language proficiency, a
repeated measures analysis was conducted with the level
of Dutch language proficiency as the independent vari-
READING RESEARCH QUARTERLY July/August/September 2000 35/3
Table 3
Mean scores and standard deviations of the four levels of language proficiency
in the four conditions
Level D
(n = 25)
M
SD
Level C
(n = 39)
M
SD
Level B
(n = 68)
M
SD
Level A
(n = 19)
M
SD
Unmodified input
0.64
0.76
0.67
0.53
0.98
0.74
1.63
0.76
Premodified input
0.68
0.63
0.87
0.80
1.57
0.87
1.74
0.87
Collective negotiation
2.08
0.57
2.13
0.61
2.54
0.56
2.63
0.68
Pair negotiation
1.32
0.80
1.31
0.83
1.71
0.67
1.84
0.60
able. The interaction of condition by level of language
proficiency was not significant (F = 1.73, df = 9, p = .079).
Table 3 provides more information in this respect.
As can be inferred from Table 3, the pupils of level
B profited most from the premodified input condition. In
the unmodified input condition, the greatest difference
between the comprehension scores of different proficiency levels is situated between levels A and B. The difference between A and B is actually twice as large as the
difference between B and C/D. In other words, for the
unmodified input condition, level A distinguishes itself
from the other three groups. On the other hand, the
break for the premodified input condition is situated between levels A/B and levels C/D.
Planned interaction comparisons showed that the
pupils of level B displayed statistically significantly higher
comprehension scores in the premodified input condition
than in the unmodified input condition (p < .01), whereas
the pupils of the other three levels (A, C, and D) did not.
So, the input modifications in the premodified input condition did promote comprehension of written input but
not for all pupils.
Conversely, in both negotiation conditions the
pupils of all proficiency levels scored statistically significantly higher than in the unmodified input condition. In
the pair negotiation condition, not all pairs did equally
well. As Table 4 shows, pupils belonging to levels C or D
profited more from the negotiation if they were teamed
up with a pupil who belonged to level A or B than with
another pupil of level C or D. T-tests for independent
samples showed this partner effect to be statistically significant (p < .01). Surprisingly, in the same condition
pupils of levels A and B scored better (yet not statistically
significantly better) when they were paired with a less
proficient pupil (of levels C or D) than when cooperating
with another A or B pupil.
For all the positive results of the pair negotiation,
the collective negotiation condition was statistically significantly superior to all other conditions for all proficiency
levels. Moreover, differences between the mean compre-
Negotiation
hension scores of level D and the scores of level A were
smaller in the collective negotiation condition than in any
other condition. These differences were the greatest in
the premodified input condition.
As Table 5 shows, the same observations can be
made when native speakers’ and nonnative speakers’
comprehension scores are compared. Repeated measures
analysis with L1 background as the independent variable
showed that the interaction of condition by L1 background was statistically significant (F = 2.90, df = 3, p =
.035). Post hoc analyses showed that the nonnative
speakers did not statistically significantly profit from the
premodified input condition, whereas the native speakers
did. The only nonnative speakers who scored considerably higher in the premodified input condition than in
the unmodified input condition were the few who belonged to proficiency level B. Within the group of native
speakers, the same tendency was visible.
In contrast, both native speakers and nonnative
speakers scored statistically significantly higher comprehension scores in the two negotiation conditions than in
the unmodified input condition. The difference between
native speakers’ and nonnative speakers’ comprehension
scores was the greatest in the premodified input condition
and the smallest in the collective negotiation condition.
Table 4
Mean scores in pair negotiation condition related to partner’s level of language proficiency
Partner A/B
Partner C/D
Partner effect
Level A/B
1.67
1.86
ns
Level C/D
1.61
1.03
p < .01
Note. For this table and the associated analysis, the four levels of language proficiency were reduced to two levels (A/B versus C/D) to avoid the number of cases
(for some combinations) becoming too small for statistical analyses.
435
Table 5
Mean comprehension scores and standard
deviations of NSs and NNSs in the four
conditions
NS
(n = 92)
M
SD
NNS
(n = 59)
M
SD
Unnmodified input
1.13
0.79
0.61
0.59
Premodified input
1.53
0.91
0.85
0.74
Collective negotiation
2.47
0.62
2.22
0.62
Pair negotiation
1.70
0.77
1.34
0.68
Discussio n
The fact that, in this study, the negotiation conditions had stronger and more uniform effects than the premodified input condition corroborates the findings of
earlier studies in this area. Premodifying a text by a combination of elaboration and simplification had only restricted and selective effects in this study. In fact, the
effect of the premodified input condition took the undesirable shape of a Matthew effect. Only pupils of relatively high language proficiency profited, especially those of
Level B. As a result, the difference between the comprehension scores of the lowest level of language proficiency (level D) and those of the highest level of language
proficiency (level A) was nowhere as large as in the premodified input condition. For the pupils of ability levels
C and D, more input modifications would seem to have
been necessary to yield a level B-like effect. However,
this would have led to a text that no longer holds any
challenge for the more proficient pupils, and that can no
longer promote anyone’s language acquisition for the
lack of unfamiliar input. Furthermore, the fact that level A
did not profit from the premodified input condition in the
same (statistically significant) way as level B suggests
there is also an upper limit to the effect of premodifying
written input on comprehension.
Conversely, both in the pair negotiation and in the
collective negotiation condition a pattern of interaction
was set up in which individual comprehension problems
were allowed, and even pushed, to come to the surface
and to be solved. In these conditions, comprehension
problems were not predicted beforehand or collectively
ascribed to a homogeneous group, as is the case with
premodified input. Comprehension problems were treated following individual learners’ signals of noncomprehension, so that the input could be finely tuned to their
level of proficiency.
436
Ellis et al. (1994) summed up the difference between premodifying and interactionally modifying input
through negotiation as follows: “[I]t is simply difficult to
predict what adjustments are needed in vacuo. The
whole point of interactionally adjusted input is that it occurs in context when learners signal their comprehension
difficulty. It responds to problems rather than predicts
them” (p. 475).
As a result, for the two negotiation conditions, the
comprehension scores of the four proficiency levels not
only were statistically significantly higher, but also lay
closer together than for the premodified and unmodified
input condition. In fact, it was probably as a consequence of this leveling effect of the negotiation conditions that the omnibus interaction of condition by level of
language proficiency did not reach statistical significance.
As for the pair negotiation condition, pupils may be
said to have performed quite well because, due to the
teacher-dominated, lockstep-type of education they usually received, they had very little experience with peer interaction. In other words, for some pupils it may not
have been straightforward to be dependent on a peer for
obtaining crucial information rather than on the teacher.
The fact that pairs that were heterogeneous in terms
of level of language proficiency yielded higher individual
comprehension scores than homogeneous pairs is quite
important with a view to optimizing reading instruction
and language education in general. As far as the less proficient pupils are concerned, the advantage of working
with a peer of another (i.e., a higher) level of language
proficiency, rather than with another less proficient peer,
sounds logical. One might reasonably expect the highly
proficient peer to be able to explain more incomprehensible input to the less proficient pupil than a peer who is
no more proficient. However, the highly proficient pupils
also profited from working with a pupil of another (i.e., a
lower) level of language proficiency. This tendency may
suggest that the efforts highly proficient pupils have to
make to explain input that is (way) above the proficiency
level of their interlocutors bring them profit as well.
Trying to explain a difficult word or sentence to another
may be a way of narrowing down its meaning for oneself.
This latter remark links up with those theories on
language acquisition, and on learning and education in
general, that reserve a strong role for talking, or writing,
as a way of sorting out one’s thoughts—that is, for talking
or writing to learn (Brown & Palincsar, 1989; Day, 1986;
Kuhn, 1989; Scardamalia & Bereiter, 1987; Wells, 1986).
For instance, Brown and Palincsar (1989) stated that conceptual growth “is more likely when one is required to
explain, elaborate, or defend one’s position to others, as
well as to oneself; striving for an explanation often makes
READING RESEARCH QUARTERLY July/August/September 2000 35/3
a learner integrate and elaborate knowledge in new
ways” (p. 395).
Not only the explaining in itself, but also the expectation of having to explain something to another may
have had an impact in the pair negotiation condition. It
might be argued that in heterogeneous dyads the members with the superior language skills felt extra motivated
to comprehend the input with which they were confronted just because there was a less proficient partner who
counted on them.
What the results in the pair negotiation condition
imply, then, is that, both for higher and lower level
pupils, group heterogeneity (in terms of level of language
proficiency) can be an asset rather than a drawback
when it comes to coping with difficult written input. If
the members of a group are on friendly terms, or at the
very least do not feel inhibited to display their noncomprehension, heterogeneous groups may yield intensive
and, for both parties, rewarding processes of negotiation
of meaning and, hence, excellent opportunities to obtain
comprehensible input. Or put in different terms, negotiation of meaning between two language learners can have
beneficial effects on comprehension not only for the interlocutor who indicates a noncomprehension problem,
but also for the one trying to solve this problem.
Such excellent opportunities for coming to grips
with unfamiliar written input also seem to have been offered in the collective negotiation condition. One explanation for the positive results in this condition might lie
in the difficulty of the text. In other words, the assistance
of only one other pupil or of a priori made input modifications may, for many subjects, not have sufficed to solve
all, or the most important, comprehension problems.
During the collective negotiation condition, pupils were
looking for the meaning of unfamiliar input collectively,
with the researcher taking care that the more reticent and
less proficient pupils also participated in the negotiation
process. It is important to note that the teacher never formulated the meaning of a word or phrase himself or herself, but that the unraveling was completely the pupils’
doing, even if this took what in educational circles could
be labeled an unrealistic amount of time. Moreover, the
participation of the researcher guaranteed that, at the end
of a negotiation routine, the pupils arrived at the correct
meaning of a word or phrase. Whenever the pupils came
up with incorrect hypotheses (which they very often did),
the researcher interfered to put them back on the right
track by way of subtle hints or negative feedback. Last
but not least, the researcher prevented the pupils from
giving up searching too soon. It is probably the combination of all these factors that has raised the quality of the
collective negotiation so as to yield such positive results.
Negotiation
The results in the collective negotiation condition,
however, should not be taken as an argument in favor of
any type of teacher-fronted education. One defining characteristic of teacher-fronted education as it has often been
described in the literature on classroom interaction is the
teacher’s prominence, both in monopolizing speaking
turns and in strictly narrowing down what pupils are allowed to say. In contrast to this, during the collective negotiation, the initiative was clearly in the pupils’ hands,
both in signaling comprehension problems and in designing routes to reach a solution for them. In fact, what becomes all the more clear is that for all pupils, but
especially for those with a low level of language proficiency, collective moments constitute an excellent setting
for obtaining comprehensible input only if certain conditions are fulfilled. Much, if not all, depends on the way
teachers handle the interaction and on the opportunities
they are prepared to offer the pupils to contribute in all
relevant and necessary ways to a successful outcome.
A comment should be made on the effects of the
various conditions on the comprehension by native
speakers and nonnative speakers of Dutch. The results of
the experiment indicate that level of language proficiency, rather than first-language background, was the dominating factor in determining comprehension scores. For
regions in the world where some controversy still exists
on whether nonnative speakers of the main medium of
instruction should receive education in classes separate
from native speakers, results like these emphasize that L1
background may not be as determining a criterion in this
issue as some have claimed. The fundamental question is
how language education deals with heterogeneity in
terms of language proficiency. The results of this experiment suggest that trying to exploit heterogeneity may be
a more fruitful approach to the problem than trying to reduce heterogeneity to acceptable proportions by radical
forms of achievement grouping.
Finally, this study is not without its limitations. For
one, only narrative text was involved in the study; generalizations to other genres such as expository text cannot
be readily made. Second, the fact that the researcher delivered the treatment, though justifiable from a methodological point of view, limits the extent to which the
results of the study may be transferable to the real-life
practice of reading education, and may even be a factor
confounding the results. Furthermore, much valuable information undoubtedly was lost because of the decision
not to make any recordings of the negotiations in the experiment. This study, then, clearly calls for further research in which other text types, real teachers, and other
forms of discussion and negotiation are involved.
437
Conclusions
The problem of education is not that language
learners are confronted with incomprehensible input. On
the contrary, comprehension problems, especially those
that arise as a result of a gap between the learner’s current level of language proficiency and the proficiency
needed to comprehend the input with which the learner
is confronted, inherently contain rich potential for further
language development. It is exactly through bridging
these gaps that learning may come about. Course developers, especially those developing reading materials for
the L1- and L2-language classroom, should refrain from
oversimplifying texts and stripping them of their learning
potential. This study provides a forceful indication that
rich models of the target language, rather than impoverished and unnatural models, do language learners the
best service. Clearly, materials and books that were not
specifically created for pedagogical purposes may be
quite valuable in this respect, because these often constitute the richest, most motivating, and the most natural
source of narrative and expository text that is available.
All this, however, implies that teachers working
with rich reading materials have to make sure that the
comprehension problems their students struggle with
come to the surface and are solved. Moreover, the results
of this experiment suggest that negotiation of meaning
has a particularly promising effect on the comprehension
of written input if the learners themselves are actively involved in signaling their problems and in trying to solve
them.
However, this may not be straightforward in the traditional language classroom. Students may be inhibited
and lack the motivation to spontaneously signal their
comprehension problems. As this study has shown, what
may be helpful in this respect is for learners to confront a
text that interests and intrigues them and a task (linked to
this text) that not only motivates them but almost obliges
them to comprehend the input, and thus to remove the
main obstacles they face on the way to comprehension.
In addition, for learners to appeal to others for help, the
teacher has to take care that the learners perceive the
class climate as nonthreatening. Noncomprehension
should not be associated with ignorance; comprehension
problems should be seen as starting points for learning
rather than as infringements on the smooth educational
process that the teacher has in mind.
The study also provides a strong indication that,
once learners make the move of bringing their comprehension problems to the surface, they can help one another out in solving them. For teachers of (multilingual)
classes that contain a range of levels of language proficiency, the results of the pair negotiation condition must
sound quite encouraging. They reveal that, both for high-
438
ly proficient pupils and less proficient pupils, peer negotiation, especially in heterogeneous combinations, has
great potential for promoting comprehension.
Nevertheless, even if learners can learn quite a lot
from each other (more than they often think), they may
not be able to learn everything that way. Teachers, being
the cognitively and conversationally most capable interlocutors, can stretch the rich potential of negotiations to
its very limits, especially if they do not take all initiative
and responsibility for solving comprehension problems
out of the pupils’ hands. By supporting the pupils’ negotiations positively and unobtrusively, by building verbal
scaffolds on the pupils’ contributions and hypotheses, by
putting pupils back on the right track when they fail to
solve comprehension problems, and by endorsing the
right explanation of unfamiliar input when pupils have
reached it, teachers can make sure that all processes of
negotiation of meaning eventually lead to success (i.e.,
enhanced comprehension).
This study also calls for an exploration of how
teachers can be trained best to play a less dominant role
in reading education. After all, the majority of studies of
classroom interaction still reveal that teachers tend to
dominate interaction too much and, in this way, relieve
students of chances to solve problems themselves and
learn from doing so.
REFERENCES
ALMASI, J. (1995). The nature of fourth graders’ sociocognitive conflicts in
peer-led and teacher-led discussions of literature. Reading Research Quarterly,
30, 314–351.
ASTON, G. (1986). Trouble shooting in interaction with learners: The more
the merrier? Applied Linguistics, 7, 128–143.
BLOOME, D., & GREEN, J.L. (1992). Educational contexts of literacy. In
W.A. Grabe (Ed.), Annual review of applied linguistics (Vol. 12, pp. 49–70). New
York: Cambridge University Press.
BROWN, A., & PALINCSAR, A. (1989). Guided, cooperative learning and
individual knowledge acquisition. In L. Resnick (Ed.), Knowing, learning, and instruction: Essays in honor of Robert Glaser (pp. 393–451). Hillsdale, NJ:
Erlbaum.
BRUNER, J. (1978). The role of dialogue in language acquisition. In A.
Sinclair, R.J. Jarvella, & J.M. Levelt (Eds.), The child’s conception of language
(pp. 241–256). Berlin: Springer-Verlag.
DAY, R. (Ed.). (1986). Talking to learn: Conversation in second language
acquisition. Rowley, MA: Newbury House.
DERWING, T. (1996). Elaborative detail: Help or hindrance to the NNS listener? Studies in Second Language Acquisition, 18, 283–297.
EEDS, M., & WELLS, D. (1989). Grand conversations: An exploration of
meaning construction in literature study groups. Research in the Teaching of
English, 23, 4–29.
EHRLICH, S., AVERY, P., & YORIO, C. (1989). Discourse structure and the
negotiation of comprehensible input. Studies in Second Language Acquisition,
11, 397–414.
ELLIS, R. (1994). The study of second language acquisition. Oxford,
England: Oxford University Press.
ELLIS, R., TANAKA, Y., & YAMAZAKI, A. (1994). Classroom interaction,
comprehension, and the acquisition of L2 word meanings. Language Learning,
44, 449–491.
FAERCH, C., & KASPER, G. (1986). The role of comprehension in secondlanguage learning. Applied Linguistics, 7, 257–274.
GALLAWAY, C., & RICHARDS, B. (Eds.). (1994). Input and interaction in
READING RESEARCH QUARTERLY July/August/September 2000 35/3
language acquisition. Cambridge, England: Cambridge University Press.
GASS, S., & VARONIS, E. (1994). Input, interaction, and second language
production. Studies in Second Language Acquisition, 16, 283–302.
GOLINKOFF, R. (1983). The preverbal negotiation of failed messages:
Insights into the transition period. In R. Golinkoff (Ed.), The transition from
prelinguistic to linguistic communication (pp. 57–78). Hillsdale, NJ: Erlbaum.
HAWKINS, B. (1985). Is an “appropriate response” always so appropriate?
In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp.
162–178). Rowley, MA: Newbury House.
KRASHEN, S. (1985). The input hypothesis: Issues and implications.
London: Longman.
KUHN, D. (1989). Children and adults as intuitive scientists. Psychological
Review, 96, 674–689.
LONG, M. (1985). Input and second language acquisition theory. In S. Gass
& C. Madden (Eds.), Input in second language acquisition (pp. 377–393).
Rowley, MA: Newbury House.
LONG, M. (1987). Native speaker/nonnative speaker conversation in the
second language classroom. In M. Long & J. Richards (Eds.), Methodology in
TESOL: A book of readings (pp. 339–352). Rowley, MA: Newbury House.
LOSCHKY, L. (1994). Comprehensible input and second language acquisition. What is the relationship? Studies in Second Language Acquisition, 16,
303–323.
PARKER, K., & CHAUDRON, C. (1987). The effects of linguistic simplifications and elaborative modifications on L2 comprehension. University of Hawaii
Working Papers in ESL, 6, 107–133.
PICA, T. (1994). Research on negotiation: What does it reveal about second-language learning conditions, processes and outcomes? Language
Learning, 44, 493–527.
PICA, T., DOUGHTY, C., & YOUNG, R. (1986). Making input comprehensible: Do interactional modifications help? ITL Review of Applied Linguistics, 72,
1–25.
PICA, T., HOLLIDAY, L., LEWIS, N., BERDUCCI, D., & NEWMAN, J.
(1991). Language learning through interaction. What role does gender play?
Studies in Second Language Acquisition, 13, 343–376.
PICA, T., YOUNG, R., & DOUGHTY, C. (1987). The impact of interaction on
comprehension. TESOL Quarterly, 21, 737–758.
PINE, J. (1994). The language of primary caregivers. In C. Gallaway & B.
Richards (Eds.), Input and interaction in language acquisition (pp. 15–37).
Negotiation
Cambridge, England: Cambridge University Press.
REUTZEL, D.R., & HOLLINGSWORTH, P. (1988). Highlighting key vocabulary: A generative-reciprocal procedure for teaching selected inference types.
Reading Research Quarterly, 23, 358–378.
SCARDAMALIA, M., & BEREITER, C. (1987). Knowledge telling and knowledge transforming in written composition. In S. Rosenberg (Ed.), Advances in
applied psycholinguistics, volume 2: Reading, writing, and language learning
(pp. 142–175). Cambridge, England: Cambridge University Press.
SWAIN, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C.
Madden (Eds.), Input in second language acquisition (pp. 235–253). Rowley,
MA: Newbury House.
VAN DEN BRANDEN, K. (1997). Effects of negotiation on language learners’ output. Language Learning, 47, 589–636.
VARONIS, E., & GASS, S. (1985). Nonnative/nonnative conversations: A
model for negotiation of meaning. Applied Linguistics, 6, 71–90.
VYGOTSKY, L. (1978). Mind in society. Cambridge, MA: Harvard University
Press.
WELLS, G. (1985). Language development in the pre-school years.
Cambridge, England: Cambridge University Press.
WELLS, G. (1986). Variation in child language. In P. Fletcher & M. Garman
(Eds.), Language acquisition: Studies in first language development (pp.
377–395). Cambridge, England: Cambridge University Press.
WHITE, L. (1987). Against comprehensible input: The input hypothesis and
the development of second-language competence. Applied Linguistics, 8,
95–110.
YANO, Y., LONG, M., & ROSS, S. (1994). The effects of simplified and
elaborated texts on foreign language reading comprehension. Language
Learning, 44, 189–219.
Received 9/10/98
Revision received 3/10/99
Accepted 8/9/99
AUTHOR NOTE
This research project was funded by the Belgian National Fund for
Scientific Research (NFWO).
439
Editing test
Note: This test was used to measure the participants’ level
of Dutch language proficiency. For the sake of exposition,
the original Dutch editing test was translated to English. A
number of words (taken from an existing story for 10year-olds) were randomly inserted in an original text written by the researcher. The participants’ task was to scrap
the superfluous words. The superfluous words are italicized below.
Wortegem Castle is very old. Around and the castle lies
were a moat, which you can it cross all by using a worse
boat. Getting into the castle is not there that simple very.
Since no one has lived there window for behind three
centuries, it over is no use at pushing the doorbell. You
will only hour get in when you can and force the front
gate. The castle void has many surprises for its visitors.
There garage are harnesses in the hall warm. The most
beautiful had room of the be castle, the library, in is situated in the cellar and evening contains, besides a sat precious crate full then of toys, thousands of books. In the
doggie middle of this room there knocked stands a sofa
less, and against while the wall there’s a but fireplace.
Some people say that in his the attic of the castle were a it
treasure lies hidden. To railroad reach the attic you have
it to find the secret staircase in the him laboratory. The
last he man glass who entered the castle has to never returned. He was any suffocated in the bedroom by to a
strange gas. Ever since and no one has dared to service
enter the castle me.
440
READING RESEARCH QUARTERLY July/August/September 2000 35/3
Part of Chapter 1 of the story and its accompanying comprehension test
(In the introductory paragraph of this first chapter, Harry finds himself in front of the castle, wondering how to get in. The only aid he has available is a booklet written by Grandpa Pete, which
contains some information on the castle. The rest of the chapter is quoted below (translated from
Dutch), first in the unmodified input version, and then in the premodified input version.
1. Unmodified input version (translated from Dutch)
“Getting into the castle doesn’t seem that simple,” Harry mumbled to himself. Between him and
the castle gate there was a moat which was about twenty meters wide. “What is the world record
for the long jump?” Harry wondered. “Eight meters?” The water of the moat, which once must
have been blue, had silted up to a black, muddy porridge that could compare with the most dangerous swamp. “Let’s measure how deep that moat is.” Harry walked toward a birch tree, broke
off a branch that was as tall as himself, and pushed it in the water. The branch went deeper and
deeper, and then it suddenly vanished. “Huhm,” Harry sighed. To his right he noticed a small
boat, lying against the bank of the moat. “Perfect,” the detective said with delight. His good luck
didn’t last long. Right at the moment that Harry stepped up to the boat, its bottom was pierced by
the head of a strange beast. “Good heavens!” Harry yelled. “A rhinoceros! How on earth does a
beast like that end up in here! Don’t you belong in the zoo, my friend?” Starting back, he took
Grandpa Pete’s booklet out of the inside pocket of his coat. Would it say anything on the animal?
Hastily he opened the booklet, looked for the right page and read: “The rhinoceros wading in the
water of the castle moat finds itself in a continuous state of drowsiness. As a guard of the castle it
can be ruled out, as a trip to the other side of the moat with a traveler on its back is one of its favorite occupations.” Was there no other possibility? Harry walked further along the moat and suddenly saw a drawbridge. It lay across the moat and led toward the castle gate. Fate would have it,
however, that it looked so dilapidated that the weight of a colony of ants would probably be too
much.
2. Premodified input version (modified passages have been underlined)
“Getting into the castle doesn’t seem that simple,” Harry said to himself. Between him and the
castle gate there was a moat. The moat was about twenty meters wide. “What is the world record
for the long jump?” Harry wondered. “Eight meters?” Once the water of the moat must have been
blue. Now it had become black mud. As a result, the moat was a dangerous swamp. “Let’s see how
deep that moat is.” Harry walked toward a tree. He broke off a branch that was as tall as himself,
and pushed it in the water. The branch went deeper and deeper. Suddenly it vanished. “Huhm,”
Harry sighed. To his right he saw a small boat, lying against the bank of the moat. “Perfect,” the
detective happily said. His good luck didn’t last long. Just as Harry stepped up to the boat, a
strange beast pushed his head right through the bottom. “Good heavens!” Harry yelled. “A rhinoceros! How does a beast like that end up in here! Don’t you belong in the zoo, my friend?” He
started back. Out of the inside pocket of his coat he took Grandpa Pete’s booklet. Would it say anything on the animal? Hastily he opened the booklet and looked for the right page. There he read:
“The rhinoceros that lives in the water of the castle moat is always sleepy. The beast doesn’t guard
the castle well at all. Do you know why she’s not such a good guard? Well, she loves making a trip
to the other side of the moat with a traveler on her back.” Was there no other possibility? Harry
walked further along the moat and suddenly saw a drawbridge. It lay across the moat and led towards the castle gate. Fate would have it, however, that the bridge looked terrible. The weight of a
colony of ants would probably be too much for the bridge.
(Comprehension test to both versions) (translated from Dutch)
What should Harry do to get to the castle gate?
• Jump across the moat
• Swim to the other side of the moat
• Take the boat and row to the other side of the moat
• Jump on the back of the rhinoceros
• Step over the drawbridge
Negotiation
441
Flowchart of interventions made by the researcher during collective negotiation
All pupils have finished reading the chapter?
No?
Wait
Yes?
Elicit indicators of noncomprehension by way of a comprehension check
formulated in motivating terms1
Spontaneous indicator?
No?
Yes?
Specific comprehension check on one crucial word2
addressed to one particular pupil
Rebound indicator to
other pupils
Can pupil explain the word?
Yes?
No?
Pupils negotiate?
Yes?
No?
Elicit possible explanations of unfamiliar input by
addressing particular pupils
Monitor turn-taking3. Make sure not only brightest pupils take a turn4. Give
every pupil ample time to finish his/her turn. Do not switch to another pupil
when a pupil has trouble saying what he or she wants to say but try to push
output5. Give hints if pupils get stuck collectively.
Unfamiliar input explained correctly?
Yes?
No?
Endorse explanation by way of
repetition or paraphrase
Extract all positive elements from reply6. Elicit further negotiation by
pupils7. If necessary, provide hints about meaning of unfamiliar input.
Elicit further indicators along same scenario
No further indicators?
Add one concluding general check addressed to whole class and follow above scenario. Otherwise round off.
Notes on the flowchart
1 The first comprehension check was of a general kind, addressed to the whole class. However, instead of dryly asking “Who wants to ask a question about the meaning
of a word?” or “Any questions?” the researcher usually used a more motivating formula like “Okay guys, this is your chance. If there are any words of which you think,
‘Gosh if I knew what that word meant I would have the solution to Harry’s problem, tell me.’ Such general checks addressed to the whole class were not repeated when
no (further) indicators were produced. Instead, a comprehension check of a particular word was addressed in the first place to the whole class and, if necessary, to a
few particular pupils (usually two or three of the more assertive pupils whom the researcher suspected of not knowing the meaning of the particular word at issue).
More than one pupil was subjected to such a check, both to create a feeling of shared incompetence and to avoid one particular pupil being singled out as ignorant.
2 Words were considered crucial if comprehending their meaning was essential to the comprehension of the whole chapter.
3 Monitoring turn-taking has to be distinguished from rigidly controlling turn-taking. The main thing the researcher saw to was that, at any moment, only one pupil held
the floor; he or she immediately intervened the moment two pupils were talking simultaneously. Second, pupils were not allowed to barge in on another pupil’s turn if
the latter had not finished. As long as they abided by this principle, the pupils were allowed to take a turn even if the researcher had not explicitly given them permission to do so. Not that they did so as a rule; in practice, the pupils almost always raised their hands if they wanted to make a contribution, unless they got overexcited.
The researcher’s turn, then, was very often confined to granting permission by naming a pupil.
4 This step usually entailed appointing a pupil who had not yet taken a turn after the negotiation had been going on for quite a while. This time, however, the researcher
asked a question that the pupil could answer correctly.
5 Pushing output (cf. Swain, 1985) was usually done by asking clarification questions, making confirmation requests, giving subtle hints, or making absurd or funny remarks that were likely to elicit further comment (in fact, the constructive power of humor in maintaining or establishing spontaneous conversation with 11-year-olds
should not be underestimated). Evidently, pushing was not prolonged ad infinitum; if it was clear, after sufficient waiting time and at least two attempts to push, that the
pupil would not come up with any better response, the researcher switched to another pupil.
6 Incorrect explanations of unfamiliar input were not bluntly rejected. As far as possible, all useful elements were extracted from the pupil’s reply and highlighted as a contributive element in the negotiation.
7 When a word was incorrectly explained, and all pupils agreed with this incorrect explanation, it was explicitly rejected by the researcher. When a word was only partially explained, further negotiation was elicited, but in this case pupils were reassured that they were on the right track.
442
READING RESEARCH QUARTERLY July/August/September 2000 35/3
Excerpt from a collective negotiation during the pilot study (T = teacher/P = pupil)
(A number of other words and expressions have already been discussed during this negotiation round)
T
Okay, any other words or sentences that you don’t understand? (...) Yes.
P1
Uhm, pierced!
T
Pierced, pierced, okay, could you read the sentence? Where does the word occur?
(several pupils bidding for a turn: Sir! Sir!)
P1
T
Uhm, as...no, no, wait. (more bidding going on) (reading aloud) Right at the moment that Harry stepped up to
the boat, its bottom was pierced by the head of a strange beast.
Right. You’ve all found that? (...) So, who’s got an idea? What does it mean? (...) Pierce.
P2
It was lying on the bottom.
T
Lying on the bottom, you say. On the bottom of what?
P2
Of the boat.
T
Everybody agrees?
P3
No, no, it’s not in the boat...it’s too heavy.
T
Yes, that might be true. Does everybody know what a rhinoceros is?
P4
Yeah, it’s a big, big animal, like very ugly, very very ugly, and it’s got a horn on its face, like that (probably
showing).
P3
And it weighs more than 500 kilos. (laughter)
T
P5
Well, that’s true, I don’t know myself how much it weighs, but it really weighs a lot. And, as you say, probably
too much for that little boat.... So, so, if it’s not lying, what does pierce mean? Remember, the text says, right
when Harry stepped up to the boat, the bottom of the boat was pierced by the head of the rhinoceros.
Drill, drill! (part of the Dutch equivalent for pierce can be translated as “drill”)
T
Uhu uhu...what’s drilling?
P5
(probably making the noise of a drill) Dududududu....
T
Right, with a drilling machine you can drill.
P6
Through a wall.
T
Right.
P7
Yeah, yeah, sir, it makes a hole in the bottom.
T
Did everybody hear what N. said? N., say it again.
P7
Well, it...uhm, the rhinoceros goes with its head right through the bottom.
P8
Yeah, yeah, with its horn. (noise)
T
Yeah, just picture it, so where was the rhinoceros?
P8
Under the boat.
T
Yes, and then it pierces the bottom, it drives, it drills its head through the bottom...or it...
P8
Pushes its head.
T
Right, it pushes its head through the bottom.
Negotiation
443