The roots of the Interaction Hypothesis lie in the Input Hypothesis put forth by Krashen (e.g., 1980, 1985) which is briefly summarized below. As a preface to the following discussion, it must be pointed out that the author considers the Interaction Hypothesis to be an extension, or revision, of the Input Hypothesis, not a replacement.

Much attention in the field of SLA turned toward the input available to learners when Krashen put forth his opinion that "the Input Hypothesis may be the single most important concept in second language acquisition today" (1980, p. 168). Krashen presented the Input Hypothesis as part of a network of interrelating hypotheses which attempts to answer the crucial theoretical question of how we acquire language (see Krashen, 1985, pp. 1-4 for an overview). The "acquisition-learning hypothesis" distinguishes acquisition, a subconscious process resulting in implicit knowledge of the language, from learning, a conscious process resulting in "formal" knowledge of the language. The "natural order hypothesis" claims that there are definite and predictable regularities in the order of acquisition (not learning) of grammatical structures. The "monitor hypothesis" also stresses the learning-acquiring distinction in claiming that the monitor is used to edit and make corrections on the output of the acquired system. The "affective filter hypothesis" claims that although input may be understood, it may not reach the internal language processor (Chomsky's [1975] Language Acquisition Device [LAD]) if the acquirer is unmotivated, lacking in self-confidence, or anxious. Bringing together the notions put forth by the hypotheses mentioned above is the Input Hypothesis which relates to acquisition, not learning, and claims that new words and structures are acquired when they are contained in meaningful messages which are understood by using context, extralinguistic information, and the linguistic competence of an acquirer who is open to the input.

Several authors in the field of applied linguistics and SLA were not convinced that the Input Hypothesis was the panacea Krashen purported it to be. White (1987), for example, points out several problems with Krashen's Input Hypothesis as it is currently formulated. White illustrates Krashen's duplicity of meaning for his notion of i+1, (both meanings can be seen in the quotation in Chapter 1). Krashen (1985) uses i+1 to indicate both the learner's competence at the stage after i (the definition White employs) and the next structure to be acquired. White contends that "while there is something essentially correct about the Input Hypothesis, in its present form it suffers from a number of defects" (1987, p. 96). Of relevance to the present study, White sees limitations in Krashen's view of both comprehensible input and simplified input. As for comprehensible input, Krashen's emphasis on the role of context and meaning, contends White, disregards the extent to which change in the learner's grammar can emerge as a result of the learner's current lexical and syntactic knowledge alone. White contends that "the driving factor for grammar change is that the input is incomprehensible, rather than comprehensible" (1987, p. 98). White explains that it is when a learner does not understand that a re-analysis of the grammar will be forced, in order to make sense of the input. White contends that certain aspects of grammar are internally driven, and that the Input Hypothesis fails to consider cases of acquisition in which the input does not help at all. According to White, Krashen overemphasizes the role and benefits of simplified input. White explains how simplified input could have unsatisfactory consequences for acquisition, since it often deprives the learner of essential information about language. This point is made lucid with the following: "many forms of simplified input would result in i-1, rather than i+1!" (1987, p. 96). White believes that a theory of SLA should include an input hypothesis and that Krashen's formulation should be adjusted rather than abandoned.

One adjustment to the Input Hypothesis which received considerable attention and motivated extensive subsequent research emerged from the work of Long who focused on interactional modifications in NS-NNS discourse (Long, 1980, 1981). The following discussion will begin by reviewing Long's laying of the theoretical groundwork for the distinction between input and interaction as they occur in NS-NNS conversations and chart the evolution of the Interaction Hypothesis.

Like Krashen, Long saw great promise in the notion of making input comprehensible to language learners and recognized the contribution to that end made not only by accessing learners' current knowledge, but also by using context and extralinguistic information. In an effort to refine the hypothesis, Long (1983c) summarizes the four pieces of indirect evidence Krashen offers to support his claims: (1) adults tend to adjust their speech to young children acquiring their first language (i.e., caretaker speech), (2) speech by NSs is modified for use with NNSs in much the same way as caretaker speech (i.e., foreigner talk), (3) the silent period some young L2 learners go through suggests they are building up competence via listening and understanding language, (4) research suggests that teaching methods which provide sufficient comprehensible input are more effective than teaching methods which do not. Long (1983c) points out the weaknesses in these arguments and provides three additional pieces of evidence for the Input Hypothesis: (5) immersion programs appear to be more successful than foreign or L2 programs, (6) for students in immersion programs, additional exposure to the target language outside the school does not seem to facilitate acquisition, (7) acquisition is either severely delayed or does not occur at all if comprehensible input is unavailable.

In an effort to define and describe comprehensible input, Long (1981) describes the patterns which emerged from research in this area which demonstrate the NSs' tendency to simplify their speech in various ways resulting in the phenomenon known as foreigner talk (FT). Long (1981, 1983a, 1983b) reviews research which reveals that differences between standard English and FT appear in three areas: phonology, lexis, and syntax. Long (1983a) explains that these adjustments appear to occur across many groups, regardless of socioeconomic status, age, and prior FT experience, although there appears to be some individual variation. To date, there had been no research on the question of what causes NSs to use FT, but Long (1983a) suggests that a variety of factors are at work, and that the initial type of FT may be rapidly replaced by another type of FT based on such triggering factors as the NS's opinion of the NNS's comprehension of the NS's speech based on the NNS's feedback. Concerning the effect of FT on learners' comprehension, Long (1983b) points to many studies of NS-NNS conversations which described the NSs' use of a simplified variety of their language, characterized by shorter utterances, lower syntactic complexity, and avoidance of low frequency lexical items and idiomatic expressions, presumably used to enhance NNSs' comprehension. Long (1983b) contends that input probably has to be comprehensible to learners for acquisition to occur. However, Long (1983a) points out that to date there had been no research investigating the effects of the adjustments in FT on SLA.

An important aspect of Long's work is the distinction he draws between input modifications and interactional modifications. Long (1981) puts forth the distinction that input refers to the linguistic forms used, while interaction refers to the functions served by those forms. Long (1983b) points out that this distinction is apparent in research to the extent that analyses of input consider only the forms that the learner hears in isolation, while analyses of interaction also consider the functions of those forms in discourse which includes the NNSs' participation. Long (1981) reports the findings of studies that have investigated NS-NNS conversations and claims there is evidence of interactional modification as follows: (1) a NS may use a variety of devices presumably intended to facilitate comprehension and participation by the NNS, (2) a NS may use a variety of techniques to help sustain conversation and to lighten other aspects of the NNS's interactional burden, and (3) a NS may do a lot of work to avoid conversational trouble and fix up the interaction when it does occur. Long (1983b) classifies these interactional modifications into the following categories: confirmation checks, comprehension checks, clarification requests, self- and other repetitions, and expansions. To provide empirical support for the distinction between input modifications and interactional modifications, Long (1981, 1983b, 1983c) reports a study (Long, 1980) in which conversations during tasks requiring a two-way exchange of information in NS-NNS dyads contained statistically significantly greater amounts of interactional modifications than similar conversations in NS-NS dyads while there was no significant difference in the amount of input modifications. Furthermore, no significant difference in either method of modification appeared between these two groups during conversations which required only one-way exchange of information. These results, compared and contrasted with the findings of previous research, led Long to propose that interactional modifications "are more extensive and more consistently observed than input modifications, and often occur when the latter do not" (1983b, p. 126). However, it seems that this has not necessarily been the case in language classrooms. Long and Sato (1983) point out the vast differences in the interactional structure found between adult ESL classroom discourse and informal NS-NNS conversations outside the classroom. The classroom discourse they examined was virtually devoid of two-way information exchange and was replete with display questions (questions to which the answer is known to the person asking) while display questions occurred only extremely rarely outside the classroom. The input available to learners in the classroom was very limited in structure and lexis and had little or no communicative value.

Accepting the importance of comprehensible input in SLA, Long (1983c) examines the ways in which input can be made comprehensible and asserts that input modifications may facilitate understanding but speech modifications alone are often insufficient since they serve only the immediate needs of communication, not the learner's future interlanguage development, and that the use of context is more likely to aid acquisition because it can supply unfamiliar linguistic input that is comprehensible. Long (1983c) claims, however, that the most important and most widely used method of making input comprehensible is to modify the interactional structure of conversation (through the devices described by Long [1981] and listed above). Long (1983b) claims that it is the NSs' adjustments to the interactional structure of conversation that are probably more important for providing comprehensible input.

However, the question of how comprehensible input affects the learner's future interlanguage development remains unanswered. Long (1983a) proposes an "indirect approach" (1985, p. 378) to demonstrating a causal relationship between environmental features and interlanguage development. Long (1983a) explains that if it could be shown that these linguistic and conversational adjustments promote increased comprehension of input (A-B), and also that comprehensible input promotes acquisition (B-C), then it could safely be deduced that the adjustments promote acquisition (A-C) . Long (1983a) claims that there is sufficient evidence in support of the two relations (A-B and B-C) which warrants the deduction (A-C) but does acknowledge, however, that direct tests of this deduction (A-C) are clearly needed. In fact, what Long (1983a) offers are research findings which are limited but encouraging in their support of the first relation (A-B), very weak for the second relation (B-C), while as he admits, direct evidence of adjustments promoting acquisition is nonexistent. Long (1985) repeats his claim that the literature provides considerable evidence for the second relation (i.e., comprehensible input promotes acquisition). However, contrary to his earlier view (Long, 1983a), Long (1985) states that there is little evidence to support the first relation (i.e., linguistic/conversational adjustments promote comprehension of input) and presents the results of a controlled study involving lecture comprehension which investigates the extent to which modified target-language input affects the comprehension attained by college students who are NNSs of the target language, English. Long (1985) demonstrates that the learners displayed significantly better comprehension and reported significantly higher perceived comprehension of a lecturette recorded in English by a NS using FT features (including reduced syntactical complexity, rephrasings and restatements, slow speech rate, and clearer articulation) than for a lecturette containing the same content but intended for a NS audience and recorded by the same NS without these FT features. These results provide clear evidence consistent with the hypothesis that speech modifications result in greater NNS comprehension, as represented by the first relation (A-B). Having shown this convincing evidence in support of the first relation, Long contends:

In light of his extensive examination of this issue, Long (1983c) proposes that classroom instruction may be enhanced by providing learners with input made comprehensible through interactional modifications made among students in small groups performing tasks requiring two-way exchange of information. Long (1983b) calls for further research which investigates the effects of varying aspects of input and interaction on the course, rate, and ultimate attainment of the SLA process. The present study is an attempt to answer this call in that it scrutinizes the pivotal word presumably in Long's statement that "one of the main goals of research on NS-NNS conversation ... is to determine how input is made comprehensible to the acquirer, and thereby (presumably) usable for SLA" (1983b, p. 131).

Aston (1986) reviews studies in this area which indicate that certain conversational procedures occur more frequently in native-learner interaction or learner-learner interaction than in native-native interaction. Aston (1986) refers to these procedures for dealing with conversational trouble as "trouble-shooting procedures" (1986, p. 129). Aston (1986) points out that other researchers contend that these procedures enhance acquisition since they result in the negotiation of meaning which makes input comprehensible to the learner, and that various pedagogical proposals have been made on the basis of this argument. Aston (1986) argues that the mere presence of these procedures does not necessarily ensure that negotiation of meaning takes place, nor does their presence automatically result in understanding. Aston suggests, conversely, that interactions in which these procedures are very frequent may be frustrating for learners and hence pedagogically undesirable and states:

Allwright, guided by the contention that language lessons are best seen as instances of "collective interaction" (1984b, p. 8), proposes two forms of what he calls the "Interaction Hypothesis" (1984b, p. 5). The first form hypothesizes that the processes of classroom interaction determine what learning opportunities become available to be learned from. The second form hypothesizes that the process of classroom interaction is the learning process (Allwright describes this as the "strong" version of the Interaction Hypothesis, 1984b, p. 9). These hypotheses are based on the findings of an investigation which compared self-report data to classroom transcriptions (Allwright, 1984b). This analysis was undertaken to shed light on three issues: (1) why learners do not learn everything they are taught, (2) how learners manage to learn things they are not taught, and (3) how learners manage to learn things they are taught. Allwright (1984b) speculates that if the "strong" version of this hypothesis proves to be true, learners would learn best those items that they have personally worked for. Yet he points out that it would be extremely interesting to find that learning can be a "'spectator sport'-- that personal involvement in the interaction is less important than the opportunity to witness the making comprehensible of input, whoever does the actual interactive work" (Allwright, 1984b, p. 11).

To conclude this discussion and present the current status of the Interaction Hypothesis, the author turns to Ellis (1990) who reviews the theoretical claims advanced by the Interaction Hypothesis, examines supporting empirical evidence, evaluates the hypothesis, and puts forth revisions.

Summarizing the findings and opinions of many researchers (notably Krashen's Input Hypothesis, discussed above; Long's focus on interaction, discussed above; and Pica's emphasis on social relationships between interlocutors, discussed below), Ellis (1990) demonstrates the Interaction Hypothesis to be a hierarchical three-part statement. First, comprehensible input is necessary for SLA. Second, modifications to the interactional structure of conversations which take place in the process of negotiating a communication problem help to make input comprehensible to an L2 learner. Third, two conditions create increased opportunities for interactional restructuring: (1) tasks in which there is a need for the participants to exchange information, and (2) situations in which the conversational partners share a symmetrical role relationship.

Ellis claims that "what is missing from the hypothesis is a theoretical account of how input made comprehensible through interactional modification results in acquisition" (1990, p. 27). According to Ellis, SLA involves three basic procedures, all of which must take place for acquisition to occur; they are: noticing, comparison, and integration. Noticing entails the learner attending consciously to linguistic features in the input. Comparison involves the learner in identifying the difference between features noticed in the input and features currently in the learner's interlanguage. Integration takes place when the learner stores in long term memory linguistic information based on new hypotheses constructed as a result of the comparison of input and current interlanguage. Based on the evidence available to date, Ellis contends that interactionally modified input may facilitate noticing and comparison insofar as it "may be effective in drawing a learner's attention to features that would otherwise be ignored" (1990, p. 29). However, there are likely to be additional factors that govern these processes; Ellis offers a provisional list: task demands, unusual features, markedness, the learner's first language (L1), and individual learner differences. With respect to integration, Ellis (1990) contends that there is a strong likelihood that the integration of new structures depends to a large extent on factors other than input.

Based on this account of the process of SLA, coupled with other factors including the lack of direct evidence to support the Interaction Hypothesis as it is currently stated, Pica's emphasis on negotiation of meaning (e.g., 1991a, 1991b, 1992, discussed below), and Swain's notion of comprehensible output (1985, discussed below), Ellis (1990) suggests a number of revisions to the Interaction Hypothesis, as follows: (1) comprehensible input is facilitative but neither necessary nor sufficient for SLA; (2) input made comprehensible through modifications, especially those brought about through negotiation, makes acquisition possible provided learners notice new features and compare them with their current interlanguage; (3) interaction that requires learners to modify their initial output facilitates the process of integration.
 

Extending the discussion of the previous section which briefly introduced research in this area and subsequent theoretical claims which led to the formulation of the Interaction Hypothesis, this section begins with a closer look at research which describes input modifications made by NSs when addressing NNSs. Studies which investigate the effects of these modifications on learners' comprehension are then reviewed. Finally, based on the evidence available to date, a summary statement is made as to the relationship between input modifications and learners' comprehension.

Though input modifications and interactional modifications are often impossible to separate, the studies included in this section focus mainly on modifications made to the input made available to L2 learners.

Studies Describing Input Modifications

The first step made toward an explanation of the relationship between input and its subsequent comprehension by L2 learners was to describe the input available to them. The body of descriptive research amassed includes different types of learners in different learning environments. This investigation has shown these features to resemble, in some respects, features of what has been referred to in L1 research as motherese or caretaker speech and commonly called baby talk outside of the field of applied linguistics. Speech containing these features when used by NSs of a particular language in addressing NNSs of that language is referred to as foreigner talk (FT). The existence of this phenomenon has been documented by several descriptive studies.

One of the influential studies in this area was reported by Henzl (1973) who examines the syntax, lexicon, phonology, and speech rate of the stories her subjects (8 NS nonteachers) told to two separate audiences (adult NSs, and adult foreign language [FL] students). Many subsequent studies investigating foreigner talk refer to Henzl's (1973) article as having identified NSs' ability to adjust the complexity of their speech to match a perceived ability level of a NNS audience. Henzl (1973) claims that the lexicon used in addressing the FL students contains a greater use of basic vocabulary while the lexicon used in addressing NSs exhibits extremely rich diversity. Henzl also describes the speakers' tendency to use "concrete terms" (1973, p. 210) and to omit or replace colloquial expressions and idiomatic phrases when addressing the FL students. Notable examples of studies motivated, at least in part, by Henzl's early work are Gaies (1979), Wesche and Ready (1985), and, of course, subsequent work by Henzl (1979). These studies and other studies discussed below have provided evidence to support the claims made initially by Henzl.

One question that has been addressed since it was raised by Long (1983a) is the question of what causes NSs to use FT. In a comparative analysis of discourse between NS-NNS of low proficiency, NS-NNS of high proficiency, and NS-NS dyads in telephone interviews, Gass and Varonis (1985b) attempt to identify which features of a NNS's speech elicit the use of FT from a NS interlocutor. Gass and Varonis (1985b) offer evidence which shows that the comprehensibility of NNS speech is an important determiner of the use of FT by NS interlocutors. Also, the NS's perception of the NNS's comprehension appears to be an important factor in eliciting NS speech modification. Gass and Varonis (1985b) contend that this perception is shaped by the NS's ongoing assessment and reassessment of the NNS's ability to understand.

The term teacher talk evolved to refer to the adjustments language teachers make to their speech when addressing their students. Studies of teacher talk have sought to describe and quantify certain modifications teachers make in their classroom speech including adjustments to the rate of speech, lexicon, and syntax. These adjustments are apparently made by teachers in an attempt to facilitate learners' comprehension. This discussion now turns to studies of teacher talk which have described teachers' speech in addressing different types of learners in different learning environments.

In an effort to describe the nature of linguistic input available to learners in formal L2 learning situations and to provide evidence of teachers' use of linguistic and communicative strategies which might facilitate SLA, Gaies (1977) recorded four different levels of adult ESL classes at three different points in time. In addition to the transcriptions of the classroom recordings being analyzed, the teachers' oral classroom language was compared to recordings of their speech made outside the classroom. Gaies (1977) reports an overall process of syntactic simplification in the teachers' oral classroom language which has characteristics similar to the speech of adults in their interactions with children in the early stages of first language acquisition. Furthermore, the teachers used more complex syntax with the higher level classes, resulting in a significant relationship between the syntactic complexity of the teachers' oral classroom language and the level of proficiency of their students. Gaies (1977) contends that the nature of the teachers' oral classroom language is governed by the perception they have of the NNSs' proficiency.

In a descriptive study attempting to determine which characteristics of teachers' vocabulary elaborations would be helpful and which harmful to students' comprehension, Chaudron (1982) recorded, transcribed, and analyzed the interaction in subject-matter lessons taught to ESL learners at several school levels. Chaudron (1982) isolated and described several features of teachers' speech which are assumed to be used to call the learners' attention to new items and to enhance the learners' comprehension. Chaudron (1982) offers not only examples from the data which appear to document the effectiveness of these elaborations, but also examples which illustrate how ineffective teachers' elaborations can be. Chaudron (1982) cautions that despite the probable clarity offered by these techniques, it should not be taken for granted that the learners have understood. Chaudron (1982) explains that teachers' elaborations often involve quite complex syntactic and semantic rearrangements with little explicit indication to link them to the original lexical item. Therefore, the elaborations will not always be meaningful for ESL learners who are still unfamiliar with the language. Due to the ambiguity of many of the elaborations contained in the data, Chaudron concludes:

As part of an ethnographic investigation of a multilingual, multicultural kindergarten classroom, Kleifgen (1985) focused on 1 teacher's speech, in the regular-content classroom, when addressing small groups of students comprised of both NSs of English and NNSs of differing English proficiency levels. The teacher in Kleifgen's study devoted more explanation and demonstration time to the students who required more assistance from her than to other students. These practices contrast with teaching patterns found in previous studies which dealt with older learners; this age difference could be one explanation for the differing findings. For example, Schinke-Llano (1983) found that teachers interacted more with NSs than with NNSs in middle grade content classrooms. Similarly, Seliger (1977, 1983) found that adult ESL students who initiated interactions more frequently received more feedback, whereas those who displayed little linguistic initiative received less attention from teachers. The teacher in Kleifgen's study (1985) diverged from the apparent norm by attending to the more reticent students. Kleifgen's study (1985) corroborates earlier research which found that teachers varied their speech adjustments according to the level of the students' competence in the target language (Gaies 1977, 1979; Henzl 1973, 1979). Furthermore, the speech of the teacher in Kleifgen's study (1985) when addressing an individual child varied as a function of that child's increased proficiency over time. Finally, according to Kleifgen (1985), the investigation of the teacher's speech in this study demonstrates that foreigner talk not only can be used but is used in the regular-content classroom as a way to achieve successful communication with NNSs.

Studies Investigating the Effects of Input Modifications on Learners' Comprehension

In a controlled study involving lecture comprehension, Long (1985) investigates the extent to which modified target language input affects the comprehension attained by college students who are NNSs of the target language, English. Long (1985) reports that the learners displayed better comprehension of a lecturette recorded in English by a NS using FT features (including reduced syntactical complexity, rephrasings and restatements, slow speech rate, and clearer articulation) than for a lecturette containing the same content but intended for a NS audience and recorded by the same NS without these FT features. Other researchers have also offered evidence to support this claim. For example, Kelch (1985) found that university ESL students completed dictation with significantly greater accuracy and reported greater perceived comprehension of the passage when the rate of the dictation was slower. However, Kelch (1985) found no significant difference in accuracy or perceived comprehension for grammatically modified dictation texts.

Parker and Chaudron (1987) investigate the effects of input modifications on reading comprehension. Although their investigation does not specifically address teacher talk, it is included in this discussion since the findings do cast light on the issues. Parker and Chaudron (1987) distinguish between three types of modifications which may facilitate the comprehension of L2 input: two being modifications of input (1) simplification and (2) elaboration, and (3) modifications of interaction. Parker and Chaudron (1987) propose that the modifications most critical for comprehension are in fact repetitions or redundancy, and clearer signaling of the thematic structure of the communication; they include these modifications under the rubric of modifications of input but distinguish them from simplifications by calling them elaborative modifications. Parker and Chaudron (1987) contend that although elaborative modifications may be promoted through verbal interaction, they do not depend on an interactive setting. Thus, Parker and Chaudron (1987) report a small-scale study designed to explore the effects of elaborative modifications on reading comprehension in which subjects read two passages (one with increased elaboration and one with redundancies eliminated and thematic structure reduced) then completed a cloze test to measure comprehension. They failed to find a significant effect for the elaborated passage over the non-elaborated passage.

Summary

In summation, there is support for the contention that the input modifications described above attain their goal of making input comprehensible to learners. The mixed findings described above notwithstanding, there is evidence that teacher talk does exist and that certain features facilitate comprehension. It has been shown that teachers can select vocabulary items known to the learners (e.g., Henzl, 1973); clearly this adjustment increases learners' comprehension. Phonological adjustments such as slower rate of speech, increased stress, and increased duration of pauses can also facilitate comprehension (e.g., Long, 1985). Findings are not as convincing concerning adjustments made to syntax. However, as a whole, a skilled teacher's "intuitive feel for what makes input simple or complex for a given group of learners" (Ellis, 1992, p. 174) can lead to increased comprehension for that group of learners.
 

It is the contention of the Interaction Hypothesis that interactional modifications are more prevalent and more effective in facilitating comprehension than are the input modifications discussed above. The distinction between input modification and interactional modification drawn by Long (1980, 1981, 1983b, 1983c) is presented earlier in this chapter. The discussion now turns to a review of studies which describe interactional modifications and investigate their effect on learners' comprehension. Finally, based on the evidence available to date, a summary statement is made as to the relationship between interactional modifications and learners' comprehension.

Though input modifications and interactional modifications are often impossible to separate, the studies included in this section focus mainly on modifications made to the interactional structure of discourse engaged in by L2 learners.

Studies Describing Interactional Modifications

One of the contributions made by Long was to develop a system by which to classify interactional modifications (summarized in Long, 1983b). His work has been very influential and has motivated, at least in part, others to examine interaction between and among various combinations of interlocutors in various environments, both inside and outside of the classroom, through a variety of research approaches.

Scarcella and Higa (1981) examine input and interaction in child and adolescent SLA by testing two hypotheses in a carefully controlled study. The main hypothesis is that older L2 learners do more "negotiation work" than younger learners (of the same L2 proficiency level) when participating in conversations with native-English-speaking adults. That is, they use more conversational management devices which facilitate understanding and sustain the conversation. A major sub-hypothesis examined is that younger learners receive more simplified input than older learners when participating in conversations with native-English-speaking adults. Adult native speakers of English were paired with 8- to 9-year-old children to form adult-NS/child-NNS dyads, then with 15- to 16-year-old adolescents to form adult-NS/adolescent-NNS dyads. The subjects then participated in a block-building task which was recorded, transcribed, and analyzed. The analysis provides evidence to support both of the stated hypotheses. That is, older learners used more conversational negotiation devices and techniques than younger learners, and adult NSs provided more simplified input to the younger learners than to the older learners. One finding that is of special interest to the present study is that while the NNS adolescents used wh-questions, yes/no questions, and confirmation checks, these techniques were virtually absent from the child L2 data. One interpretation of this finding offered by Scarcella and Higa (1981) is that since older learners have been shown to have an advantage over younger learners in the beginning stages of learning, perhaps the work they do in negotiation serves them better than the simplified input the younger learners are supplied with in their conversations with adult NSs.

Schinke-Llano (1983, 1986) designed a study to characterize the linguistic environment experienced by fifth- and sixth- grade LEP students. One portion of the study was devoted to classroom observation (Schinke-Llano, 1983) and focused on 12 monolingual English-speaking classroom teachers conducting content lessons in several academic areas in which English was the medium of instruction, not the target. Lessons were audio-taped and transcribed. Analysis of the transcriptions focused on the interactions between teachers and students. The results of the analysis offer evidence that the teachers observed interact differently with the LEP students in their classes than they do with non-LEP students. First, the teachers interact significantly less frequently with LEP students than with their native-English-speaking counterparts. Second, the majority of interactions that did occur between the teachers and LEP students were managerial in nature while the majority of teachers' interactions with non-LEP students were instructional in nature. There was no significant difference between the duration of LEP-student/teacher interactions and the NS-student/teacher interactions. Schinke-Llano (1983) presents the argument that the teachers "exhibit differential treatment of LEP students by virtue of their perceived inability to function in the content classroom" (1983, p. 158). Schinke-Llano (1983) proposes that such differential treatment may have adverse consequences for students' self-esteem, second language acquisition, and mastery of content subjects; however, the results of the study do not test this claim. Another phase of the study focused on teacher-student interactions specific to joint cognitive activities (Schinke-Llano, 1986). In this experimental portion of the study, the fifth- and sixth-grade teachers instructed LEP students of differing proficiency levels, and NS students in completing a catalog order blank; the activity was conducted in teacher-student dyads. The activities were recorded, transcribed, and analyzed; analysis included both quantitative and qualitative features. The purpose of the experiment was to compare the instructional interactions between teachers and LEP students with those between teachers and NS students, focusing on differential language use in structuring the task situation for the two groups of students. Results of the study show that the teachers did in fact utilize language differently to structure task situations for LEP students than for NS students. The quantity of speech used by the teachers was significantly higher with the LEP students than it was with the NS students. There was no significant difference between the LEP-student/teacher dyads and the NS-student/teacher dyads in the amount of time spent on the task. The nature of the instructional interactions was different between groups in that the interactions with LEP students were more teacher-regulated, and the steps of the task were made more explicit for the LEP students. Schinke-Llano (1986) speculates that such differential treatment may have negative consequences for SLA and cognitive skills development but cautions that the data presented by this study do not constitute evidence for this claim.

In a descriptive study of teacher-student interaction, Gaies (1983) focused on learner feedback which is defined as "information provided by a learner to a teacher about the comprehensibility and usefulness of some prior teacher utterance(s)" (p. 192). Gaies postulates that when participating in verbal communication, "learners regulate the nature of the content to which they are exposed in the classroom and the rate at which it is presented" (p. 196). By recording, transcribing, and analyzing the classroom interaction of dyads (teacher-student) and triads (teacher-students) participating in a communication task, Gaies (1983) provides evidence that the learners do, to a certain extent, have control over their own intake. Gaies (1983) classified learner feedback based on a model of four types of pedagogical moves (responding, soliciting, reacting, and structuring) presented by Bellack, Kliebard, Hyman, and Smith (1966). Utterances were assigned to these major categories on the basis of functional properties rather than syntactic features. Gaies also developed several subcategories for the particular data collected. The analysis reveals a teacher-centered approach, probably due in large part to the nature of the task observed. However, the learners were able to control the discourse through the use of questions to solicit from the teacher the information they felt necessary. Gaies summarizes his findings as follows: (1) Collectively, the learners made use of feedback in all four of the major categories; however, not all of the learners used all of the moves; (2) There was considerable variation among learners in the amount of feedback provided; (3) Reacting moves were by far the most frequent form of feedback and structuring moves occurred the least frequently; (4) In each of the triads, one learner provided considerably more feedback than the other. In order to investigate the effect learner feedback produces on classroom discourse, Gaies (1983) uses a model presented by Glucksberg and Krauss (1967) to classify the post-feedback utterances made by the teachers into one of five categories (verbatim repetition, reduced repetition, expanded repetition, restructuring, and question). This analysis reveals little in the way of a relationship between learner feedback and teacher post-feedback. The only immediately apparent pattern detected is the tendency of teachers to expand utterances rather than to reduce them in length.

In an effort to describe and compare the input and interactional features of communication activities as carried out in small groups vs. teacher-fronted formats, Pica and Doughty (1985) found the teacher-fronted activities provided more grammatical input and a greater number of certain features of negotiation, while the small-group activities provided individual students with more opportunities to use the target language. These findings led Pica and Doughty (1985) to conclude: "We feel that small-group use of communicative activities can be effective in the ESL classroom but that its benefits may be more limited than had previously been assumed" (p. 132). In reflecting on the results, Pica and Doughty (1985) hypothesize that two-way communication tasks carried out by pairs of learners may ultimately be shown to be most conducive to negotiated modification of interaction, and hence to second language acquisition.

In a continuing attempt to describe the effects of task type and participation pattern on language classroom interaction Doughty and Pica (1986) compare teacher-fronted and group work on a problem solving task. Using "required information exchange," they compared the results of a similar study (Pica & Doughty, 1985, described above) which used "optional information exchange." Doughty and Pica (1986) explain that information gap refers to "the existence of a lack of information among participants working on a common problem" (1986, p. 307); however, there are different types of gaps: those in which information exchange is required among all participants, each of whom possesses some piece of information not known to, but needed by, all other participants (i.e., the task used in Doughty & Pica, 1986), and those in which information exchange is optional as all participants possess the same information and decide whether or not to contribute to the solution of the problem (i.e., the task used in Pica & Doughty, 1985). Doughty and Pica (1986) found interactional modifications (including clarification requests, confirmation checks, and comprehension checks) to be significantly more frequent in the group activities with required information exchange than in the teacher-fronted classroom. This was unlike the findings of Pica and Doughty (1985) in which the teacher-fronted lesson contains slightly more conversational adjustments than the optional information exchange task in groups.

Varonis and Gass focused on NNS discourse to present a model for the negotiation of meaning which they call "non-understanding routines" (1985b, 73). As the present study employs this model to analyze its data, a detailed description appears below (in Chapter 3); however, a brief explanation is included here as their work further documents interactional adjustments. The model presented by Varonis and Gass (1985b) consists of two main parts. The first part of the model is a trigger which indicates non-understanding, the second part is the resolution of the non-understanding. This model is also utilized and refined in subsequent studies. Gass and Varonis (1985a) compare the interactional structure elicited by different communication tasks by examining the negotiation of meaning accomplished by adult NNSs in dyads and triads participating in one-way tasks and a two-way task. Gass and Varonis (1986) examine differences between male and female speech by examining the negotiation of meaning accomplished by adult NNSs participating in communication tasks. Varonis and Gass (1985a) apply the model to adult NS-NNS interaction in describing miscommunication in a service encounter telephone conversation. In a comparative analysis of adult discourse between NS-NNS of low proficiency, NS-NNS of high proficiency, and NS-NS dyads in telephone interviews, Gass and Varonis (1985b) demonstrate that the model is also applicable to the investigation of interaction between a wide range of interlocutors.

Extending the line of reasoning which led to the investigation of the amount of interactional adjustments elicited by different types of communication tasks, Kennedy and Newton (1991) investigate the grammatical consequences of different tasks undertaken by adult NNSs. Their results suggest that communication tasks for language learning can be designed to influence not only the occurrence of more or less negotiation but also the use of particular linguistic structures. Their investigation also led them to suggest the following contention:

Studies Investigating the Effects of Interactional Modifications on Learners' Comprehension

The review above offers evidence for Long's claim (1981) that a NS may use a variety of devices presumably intended to facilitate comprehension and participation by the NNS and documents the existence of interactional modification. The discussion now turns to those studies that have investigated the relationship between interactional modifications and subsequent comprehension.

Evidence of interactional modifications resulting in increased comprehension is offered by Pica, Young, and Doughty (1987) who conducted an experimental study with college level ESL students. The study compared the comprehension of 16 NNSs of English on directions to a task presented by a NS of English under two conditions which consisted of two types of modifications of baseline task instructions: premodified input, with decreased complexity and increased quantity and redundancy; and interactionally modified input, which allowed for NS-NNS interaction. Two experimental groups listened to directions to place fifteen items on a board. One group heard instructions with modified input (e.g., by shortening sentences and removing embedded clauses), the modifications being derived from baseline data, but did not have the opportunity for interaction. Another group heard the baseline directions but was given the chance to interact with the person giving the directions. The study proposed two major hypotheses. First, certain linguistic adjustments would arise spontaneously during interaction. Specifically, there would be more input (i.e., a greater number of words), there would be more redundancy, and the linguistic adjustments would be less complex than the original baseline data. Second, the subjects that had the opportunity for interaction would show greater comprehension in selecting the correct item, placing the item in the correct position, and in total comprehension score. They concluded that the premodified input was not a significant factor in the NNSs' comprehension; however, interactionally modified input played a critical role in comprehension. The first hypothesis was supported in that there was more input and more redundancy; however, the linguistic adjustments were not less complex. The second hypothesis was strongly supported which lead to the conclusion that "interactional modifications of input did, in fact, lead to significantly greater comprehension than conventional ways of simplifying input ..." (Pica et al., 1987, p. 745). Consequently Pica et al. (1987) claim these results provide guidelines for restructuring interaction in the classroom and conclude that there is "enough evidence to infer that interactional modifications do in fact serve as an important stimulus or vehicle for the repetition of input ... crucial to comprehension" (1987, p. 753).

A similar contention is made by Pica, Doughty, & Young (1986) whose investigation indicated that the most significant aids to comprehension brought about by interaction were increased quantity and redundancy of input while a reduction in the syntactic complexity of the input was observed to play no significant role in its comprehension.

In two experimental studies, Tanaka (1991) and Yamazaki (1991) also examined the comparative effects of two types of modifications made to baseline task instructions (i.e., premodified input, and interactionally modified input) as outlined by Pica et al. (1987). Both studies investigated the comprehension and acquisition of new lexical items embedded into the directions of a communication task conducted with rather large groups (n=79 in Tanaka, 1991; n=127 in Yamazaki, 1991) of high school students studying English as a foreign language (EFL). The components of the studies which address acquisition will be discussed below. However, concerning comprehension, the results of both studies support the findings reported by Pica et al. (1987) in that significantly greater comprehension scores were achieved by the subjects given the opportunity for interaction. Tanaka (1991) and Yamazaki (1991) note that the modified interaction treatment also led to more input, as was the case in Pica et al. (1986) and Pica et al. (1987).

Loschky (1989) reports research in which 41 adult beginning learners of Japanese participated in an experimental study to examine the effects of the same variables described by Pica et al. (1987) (i.e., input modifications and interactional modifications of baseline instructions to a communication task) on comprehension and the acquisition of locative patterns (prepositions) and new lexical items (concrete nouns) embedded into the task instructions. Again, the component of the study which addresses acquisition will be discussed below. However, concerning comprehension, Loschky concludes that his results "add confirming evidence to the claim (Long, 1981; Pica et al., 1987) that negotiation of meaning is a powerful tool in facilitating comprehension" (1989, p. 122) as the subjects who achieved the highest comprehension scores were again those who were involved in interaction.

Pica (1991b) analyzes NSs' utterances of negotiation during communication tasks carried out in NS-NNS dyads. The results not only offer additional evidence for negotiation of meaning facilitating NNSs' comprehension of L2 input, but the detailed analysis also reveals specific ways the NSs' utterances helped the NNSs understand. The NSs helped by clarifying L2 forms, the meanings they encoded, and some of the structural relationships into which they could enter.

In an experimental study with college-level ESL students, Ehrlich, Avery, and Yorio (1989) investigate the relationship between negotiation of meaning and comprehensible input. Using NS-NS and NS-NNS pairs to perform a picture drawing task, negotiation is examined within a discourse framework. Ehrlich et al. (1989) question the assumption that more meaning negotiations mean more comprehensible input and contend that it is the point in the discourse at which the negotiations occur that determines their success or failure. The discourse strategies used by the subjects display characteristics of one of two categories: skeletonizing and embroidering (Brown's terms, 1972). Skeletonizing provides only the bare events of a narrative, whereas embroidering describes the same events with a greater amount of expansion and embellishment. Based on an analysis of transcriptions of the discourse, Ehrlich et al. (1989) report that when non-understanding (Varonis & Gass's term, 1985b) occurred, the subjects either attempted to repair the utterance or they abandoned negotiation of meaning. Repairs appear as either repetitions or expansions, while abandonments are either explicit or implicit (i.e., acknowledged or unacknowledged). As for the relationship between discourse strategy and task success, Ehrlich et al. report that "the embroidering strategy ... can lead to confused, problematic discourse in the NS-NNS interactions and may impede the production of comprehensible input" (1989, p. 411) (cf. Chaudron's critique of elaboration [1982], discussed above). The results led Ehrlich et al. (1989) to "agree with Aston (1986), who questions the prevailing assumptions ... that the mere presence of meaning negotiations within an interaction guarantees comprehensible input" (p. 411).

Summary

In summation, there is support for the contention that the interactional modifications described above attain their goal of making input comprehensible to learners. Pica offers the following explanation:

This discussion now turns to what is ultimately the most important question for the present study: What has all of this got to do with learning? Notwithstanding her continuing, extensive investigation of negotiated interaction and endorsement of its benefits for L2 learners, Pica points out that:

Studies Investigating the Effects of Interaction and Comprehension on Vocabulary Acquisition

Using both qualitative and quantitative analyses of longitudinal data collected by tracing the linguistic development of 2 young ESL learners, Ellis (1985a) established an important link between the interactions in which the subjects engaged and their syntactic development. Ellis (1985a) points out that obtaining comprehensible input is not really the result of separate contributions of the NS and the learner but from their joint endeavors in ongoing interaction to establish and maintain a topic (i.e., negotiation). By analyzing the longitudinal data, Ellis (1985a) documents learners' utterances of increasingly complex syntax brought about through interactional adjustments. This showed "how important cooperation between the interlocutors was in the process of building utterances that lay outside or on the edges of the learner's competence" (Ellis, 1985a, p. 79). It is important to note that the new forms produced by the learners were not the result of their being supplied with simplified input, rather they came about as a result of problems which were negotiated by the learner maximizing existing resources assisted by the interactional adjustments of the NS.

Turning specifically to vocabulary retention, Craik and Tulving (1975) propose that "depth" of processing is a significant factor. Craik and Tulving (1975) report that words which were more deeply encoded or sustained greater degrees of semantic involvement during the treatment phase of experimental studies were associated with higher retention levels. Their L1 research, however, did not deal with newly acquired words, but with words already known to the subjects. Hauge (1987), in discussing the implications of such findings on L2 vocabulary instruction, contends that "vocabulary that is taught in context, that encourages active processing, that is intense, and that is focused on the interests of the learner seems to be more effective than other techniques" (p. 220).

Two experimental studies, Tanaka (1991) and Yamazaki (1991), investigate the comprehension and acquisition of new lexical items embedded into the directions of a communication task. The results of both studies support the findings reported by Pica et al. (1987) in that significantly greater comprehension scores were achieved by the subjects given the opportunity for interaction than for subjects who heard premodified task directions but had no opportunity for interaction. Both studies (Tanaka, 1991; Yamazaki, 1991) also investigated acquisition of these lexical items by utilizing both self-report data and a pretest/posttest design. Self-report data was collected through the use of "Uptake Recall Charts" (Slimani, 1989) passed out immediately after completion of the task. These charts asked the subjects to list any words or phrases they felt they had learned during the task. Acquisition was also measured by results of posttests administered at three different times after completion of the task. The first posttest, which required the subjects to translate a list of new lexical items introduced in the task into their L1, was administered 2 days after completion of the task. An identical test (except for the order of the words on the list) was administered 4 weeks after completion of the task. The third posttest, administered 10 weeks after completion of the task, required the subjects to match the new lexical items with corresponding pictures. Analysis of the Uptake Charts showed that subjects given the opportunity for interaction during the communication task in both studies reported learning a greater number of words than subjects who heard premodified task directions. As for the posttest scores, the results in the study reported by Tanaka (1991) show significantly higher scores for the subjects given interaction group in all three posttests while Yamazaki (1991) reports significantly higher scores for the interaction group in the first posttest but no significant difference between groups in either the second or the third posttest.

In another study motivated by the findings of Pica et al. (1987) which also incorporates posttest scores as a measure of acquisition of both locative patterns (prepositions) and new lexical items (concrete nouns) embedded into the task instructions, Loschky (1989) also provides support for the claim that negotiated interaction facilitates comprehension. However, in contrast to the two studies just described, Loschky (1989) reports that the findings of his study do not support the claim that interaction promotes acquisition. The treatment in Loschky's experiment (1989) involved three groups of subjects participating in a communication task (i.e., those completing the communication task by listening to: (1) baseline instructions, (2) premodified instructions, and (3) those engaging in negotiated interaction to complete the task). The difference between pre- and posttest scores for all three groups of subjects shows significant improvement; however, there was no difference in gains between the three treatment groups in lexis or morphosyntax. Loschky's study (1989) incorporated two measures of vocabulary knowledge: "item recognition," in which subjects were asked if they had any recognition of words on a list, and "associative recognition," a forced-choice test similar to that used in the present study to measure receptive vocabulary knowledge. As for item recognition, Loschky concludes:

In summation, considering the mixed findings of the limited research that investigates the relationship between interaction and comprehension, and lexical acquisition, taking any definitive stance would be premature. It could be that this lack of conclusive evidence owes more to insufficient variety and volume of investigation than to a major flaw in the hypothesis that learners' acquisition can be enhanced as a direct result of their interaction and comprehension.
 

In deference to the subjects in the present study, this discussion begins with a review of the research which describes participation by young L2 learners. The studies are generally ethnographic in nature and approach participation as it is brought about through the learners' social contact. The discussion then moves to studies which investigate participation in language learning activities for learners of differing ages and learning environments. Finally, based on the evidence available to date, a summary statement is made concerning the role of learner participation in SLA.
 

Wong-Fillmore (1985) reports a descriptive study of LEP students attending American elementary schools which focuses on the subjects' learning the usual subject matter of school rather than focusing on ESL instruction and reports certain classes to be decidedly more beneficial for LEP students' English development than others. One of the characteristics of the successful classes deals with learner participation as determined by the teachers' allocation of turns. Wong-Fillmore (1985) reports that teachers in successful classes were consistent in following a well-established set of procedures for turn allocation and generally managed to call on everyone at least several times during each lesson. In less successful classes, however, procedures for participation were often unclear, and the most aggressive students were called on frequently while those who were less so got fewer turns to participate. Hence, according to Wong-Fillmore, the students who were slighted by the teacher got "less practice in using the new language and less of the feedback that is available through this kind of participation in lessons" (1985, p. 32) (cf. Schinke-Llano, 1983, 1986; Kleifgen, 1985 discussed above). This speculation drawn from the observations is an extension of Wong-Fillmore's earlier work (1976) in which she concludes that social contact is critical and necessary for L2 development. Her beliefs based on this earlier research may have influenced the contentions she put forth in the subsequent study.

Strong (1983) explores the relationship between "social styles" and the acquisition of English by 13 Spanish-speaking kindergartners over 1 school-year in a bilingual classroom. The seven social styles Strong (1983) attempts to operationalize and investigate are: talkativeness, responsiveness, gregariousness, assertiveness, extroversion, social competence, and popularity. Strong (1983) found that the faster learners were more talkative and responsive than slower learners. In other words, the faster learners were better at managing the interaction they were involved in. Strong concludes that "contact with English speakers alone does not enhance language learning, but that the active use that is made of the extra input is what counts (1983, p. 256).

In an ethnographic study, Saville-Troike, McClure, and Fritz (1984) examined the communicative tactics of 20 ESL students in a public elementary school in the United States and found that conversational tactics in social interaction had no apparent relationship with either English development or academic success. Saville-Troike et al. explain:

Several of the findings of a retrospective analysis which examined factors influencing elementary ESL students' academic achievement reported by Saville-Troike (1984) are of relevance to the present study. In investigating productive competence in English, it was found that the only factor significantly correlated with reading achievement was "the number of different vocabulary items each child used" (Saville-Troike, 1984, p. 207). This finding prompted Saville-Troike to conclude that "vocabulary knowledge in English is the most important aspect of oral English proficiency for academic achievement" (1984, p. 216). This same investigation of English language proficiency revealed that "accuracy in English morphology and syntax in spoken language appears to make little difference in academic achievement" (Saville-Troike, 1984, p. 206). Also, verbosity had a positive but non-significant correlation both with language test scores and with reading achievement. Furthermore, an analysis of interaction data, both in and outside the classroom, showed that, with the exception of vocabulary, very little carry-over from forms presented in the structured ESL lessons to those language forms actually used by the students for genuine communicative purposes in less structured events. These findings led Saville-Troike to the conclusion that "spoken practice in English may not be necessary for the development of English proficiency and may retard it in some instances" (1984, p. 216). As a result of this investigation, Saville-Troike contends that "we must begin to place more emphasis on vocabulary learning and less on grammar and pronunciation" (1984, p. 217).

Saville-Troike (1985) focuses on 3 non-English-speaking students in the English-speaking social milieu of kindergarten. One of the most interesting findings Saville-Troike (1985) reports is that at first the learners used their native languages indiscriminately, but as they gradually became aware that they were not being understood by others, they tried alternative communicative tactics before resorting to more drastic measures. Saville-Troike explains:

In an investigation of participation by adult ESL students, Seliger (1977) postulates the existence of two types of language learners characterized by the extent to which they interact in the L2. As per Seliger (1977), learners who interact intensively, who seek out opportunities to use the L2 and who cause others to direct language at them are termed high input generators (HIGs) while learners who either avoid interacting or play relatively passive roles in language interaction situations are termed low input generators (LIGs). Seliger (1977) grouped the students into one of the two categories based on a quantitative measure of verbal interaction in L2 classes then tested directional hypotheses to determine whether or not the HIGs would perform differently than the LIGs on four specific measures. One of these hypotheses was designed to test achievement, defined as performance on examinations given at the end of the 15-week semester as compared to performance on placement examinations given at the beginning of the same semester. Seliger (1977) reports strong positive correlations for interaction and final tests. Seliger claims that the end result of the HIGs' behavior is "a competence which develops at a faster and perhaps qualitatively better rate" (1977, p. 274). Seliger warns that the results of this research apply only to the group studied and that the study focuses on "subjects who were at two extremes on the interaction continuum" (1977, p. 275). However, Seliger (1977) does contend that HIGs will not only benefit from formal instruction, but also from their exploitation of other practice opportunities beyond what is presented formally, while LIGs will also benefit from formal instruction, but they are also dependent on it.

Seliger (1983) takes a closer look at his assumption that the active learning style of HIGs will lead them to "acquisition at a faster rate" (p. 252) than the more passive learning style of LIGs. Seliger (1983) presents the argument that when learners initiate interaction they obligate themselves to attend to the response of their interlocutors; this is when, according to Seliger (1983), learners change the input into intake. Furthermore, Seliger's position (1983) is that a learner who initiates interaction must attend carefully to his or her own output in order to make the intent of the utterance clear. Thus, Seliger claims, interaction which is learner initiated "may be a better source of material with which the learner can form or modify internal hypotheses about the target language" (1983, p. 253). Seliger claims that HIGs "progressed at a faster rate than learners who interacted little in the classroom" (1983, p. 262) and that "HIGs showed a lower percentage of errors that could be traced to transfer from the learner's first language" (1983, p. 262). Seliger (1983) contends that a major contributing factor leading to the advantages enjoyed by HIGs over LIGs as language learners is the notion that their learning behavior supplies them with additional feedback which gives them increased opportunities for hypotheses testing. However, a close examination of the results he reports offers little evidence to support these bold claims.

_ Day. 84. LL._ Day (1984) conducted a study motivated by the findings of Seliger (1977) described above. Although Day (1984) addresses basically the same issues as Seliger (1977), Day's study (1984) is better designed and more carefully conducted. Day (1984) examines the level of participation by adult students in the ESL classroom and its relationships with proficiency and with the use of English outside the classroom. Day (1984) found evidence for the existence of an "interaction continuum" (Seliger, 1977, p. 275) since there were students who could be labeled HIGs and LIGs. However, Day's (1984) study offers no evidence for classroom participation being a factor in the English proficiency of the subjects. Day (1984) uses two measures of English proficiency: the Bachman-Palmer oral interview (1982) which measures communicative competence, including grammatical competence; and a cloze test, constructed and validated for the level of the subjects, to measure formal knowledge of English. The students' language use outside the classroom (L1 and L2) was measured by a self-report questionnaire which had been piloted and deemed appropriate for the subjects of the study (Day, 1984). In conducting a statistical analysis of the results, Day (1984) finds no support for the hypothesis that the HIGs perform better than the LIGs on either of the measures of English proficiency, nor is there a significant relationship between classroom participation and use of English away from the classroom. While recognizing that more research into the variety of ways learners develop competence in a second language is needed, Day proposes that "the small amount of research to date indicates that using the target language in the ESL classroom may account for an insignificant amount of the variation in second language proficiency" (1984, p. 95).

Day (1985) closely examines the data collected from the self-report questionnaire mentioned above (Day, 1984). A detailed analysis of the responses correlated with the two measures of proficiency (Day, 1984) revealed no significant correlation which led Day to conclude: "For our subjects, their reported use of English in context, away from the ESL classroom, did not predict English proficiency" (1985, p. 262). As a result of this analysis, Day continues: "Further, we learned that evidence purporting to support the claim that the level attained by ESL students is related to their use of English outside the classroom is mixed and questionable" (1985, p. 265).

Allwright (1980) reports on the contributions made by individual learners in a university level ESL class and discovers one learner participating to a much greater extent than the other learners in the class. Although the findings are not conclusive, Allwright (1980) contends that these "real attempts at real communication" (p. 185) can benefit those who are not actually participating in the classroom discourse.

Empirical evidence of this phenomenon is offered by Slimani (1989) who investigates the relationship between classroom interaction and learning outcomes. Slimani uses Allwright's (1984b) notion of "uptake," which she defines as the learners' "perceptions of what they have learned from the interactive events they have just been through" (Slimani, 1989, p. 224), to examine learners' beliefs about learning. Slimani (1989) recorded language lessons and distributed "Uptake Recall Charts" at the end of every observed lesson. The chart asks learners to report, in terms of grammar, words and expressions, pronunciation, and spelling, in as much detail as possible, what points they recalled from the lessons they had just attended. About 3 hours later the learners also completed an "Uptake Identification Probe" which asked for specific information to isolate newly uptaken items. At the end of the 6-week observation period, an "Operating Principle Questionnaire" was distributed to determine whether the learners' perceptions of what they believed they had allocated their attention to during instruction corresponded to their actual behavior in the classroom. The recordings of the lessons were transcribed, and the items claimed to have been learned were then traced in this classroom data in order to examine what in particular happened in the interactive episodes in which they occurred which might have caused them to have been learned. As a result of this detailed examination of the data, Slimani (1989) reports that topicalization (i.e., who initiates the topics of interaction) by the learners is influential in accounting for learners' claims about uptake. Slimani (1989) also reports that the learners' responses to the Operating Principle Questionnaire show that input provided by the teacher for the whole class was perceived to be potentially most influential on learning. Slimani's analysis of the data (1989) shows that the discourse initiation appears to be predominantly in the hands of the teacher. However, Slimani's analysis (1989) of the responses on the Uptake Recall Charts and the Uptake Identification Probes reveals that learners are more likely to profit from episodes topicalized by fellow classmates. Furthermore, Slimani explains that:

However, an interesting argument for learner output is presented by Swain (1985) who contends that what is essential for L2 acquisition resulting in nativelike proficiency is the learners' production of "comprehensible output" (p. 236) which she defines as "output that extends the linguistic repertoire of the learner as he or she attempts to create precisely and appropriately the meaning desired" (p. 252). Swain (1985) tested a group of sixth-grade French immersion students and a comparison group of sixth-grade native-French-speakers on their grammatical, sociolinguistic, and discourse competence. The results of the comparison show that in three of the four grammatical categories tested, native speakers scored significantly higher than the immersion students, "indicating clearly that, although the immersion students are doing quite well, they have not acquired nativelike abilities in the grammatical domain" (Swain, 1985, p. 238). Similarly, Swain (1985) reports that native speakers performed significantly better on the sociolinguistic tasks than the immersion students; however, the differences between groups on the measures of discourse competence were small and mixed, depending on the particular trait being measured. Swain (1985) claims that these results show that in spite of 7 years of comprehensible input, the target system has not been fully acquired. Swain (1985) suggests that the missing ingredient could be that the immersion students did not have ample opportunities to produce comprehensible output which could facilitate acquisition in three ways. Swain (1985) explains that first, opportunities for meaningful use of one's linguistic resources could arise when one is "pushed" (p. 249) in output; this could come as a result of feedback to the learner which indicates that his or her output has been unsuccessful in some way (what Schachter refers to as "negative input" [1984, p. 168]). Second, according to Swain (1985), producing comprehensible output gives the learner opportunities to test out hypotheses about the target language. Third, Swain (1985) contends, producing comprehensible output may force the learner to move from semantic processing to syntactic processing of the target language. It must be pointed out, however, that these contentions are based on Swain's investigation of grammatical, sociolinguistic, and discourse competence and may not be relevant for lexical acquisition.

To describe Swain's notion (1985) of comprehensible output, Pica, Holliday, Lewis, and Morgenthaler (1989) investigate how L2 learners respond linguistically when NSs signal difficulty in understanding them. Results of their study provide empirical validation for the theoretical construct of comprehensible output. Through analysis of the data collected for this study, Pica et al. (1989) also reveal the extent to which NNSs' production of comprehensible output is influenced by the linguistic demands of NS signals of comprehension difficulty as well as by the type of communication task in which the subjects participated. In the study reported by Pica et al. (1989) NS/NNS dyads participated in three types of communication tasks which were recorded, transcribed, and analyzed in order to test four different hypotheses concerning the kinds and amounts of interlanguage modifications the NNSs would make during the three tasks. Relevant here, however, is the finding reported by Pica et al. (1989) that NNSs tended to modify their output most often when NSs signaled an explicit need for clarification which led the authors to conclude that during their study "comprehensible output was alive and well and was very much an outcome of linguistic demands placed on the NNS by the NS in the course of their negotiated interaction" (p. 83).
 

In summation, research in the area of the relationship between learner participation and SLA is still in its infancy and the range of variables and interrelationships among them is overwhelming. Ellis offers a good assessment of where the field stands at present:

This section relates the issues concerning the theories of vocabulary learning to the theories of teaching the lexicon of a second or foreign language. As a preface to this discussion, the contention put forth by Gass (1988b) should be considered. In reviewing the important research concerning L2 vocabulary acquisition, Gass (1988b) proposes that much of the research in this area has dealt with descriptive aspects of the lexicon but not with the establishment of a theory of the lexicon. Gass calls this lack of concern with lexical theory "striking" (1988b, p. 92) considering the centrality of the lexicon to both acquisition and use of the L2, and points out that lexical errors obstruct communication more so than grammatical and phonological errors.
 

L2 Vocabulary Knowledge Described

In an effort to describe what it means to learn vocabulary, Richards (1976) presents eight broad assumptions to characterize lexical competence; they are as follows:

Developing an L2 lexicon is different from learning other aspects of the target language in at least two ways. First, L2 learners come to the task with a broad background of knowledge and experience which includes the learners' L1 proficiency. Thus, it is conceivable that in some cases of L2 lexical acquisition all that is necessary is to learn new labels for known concepts and meanings. However, this concept should be viewed with caution since learning the vocabulary of another language is only one aspect of L2 acquisition; furthermore, as van Els, Bongaerts, Extra, van Os, and Janssen-van Dieten (1984) state "conceptually, words from L1 and L2 will often correspond to a considerable extent, but hardly on a truly one-to-one basis" (p. 251). Paradoxically, this crossover could actually confound learners' efforts. Second, it has been recognized that vocabulary is one area of language that needs continued growth and development for both native and non-native speakers (Celce-Murcia & Rosensweig, 1979; Rivers, 1983). Since it is believed that vocabulary development should be a lifelong undertaking, this concept has made an impact on L2 pedagogy.

Vocabulary seems to be very important and usable for learners, especially in the early stages of learning. Shapiro (1978) explains how "functional communication can take place through an ordered series of lexical elements" (p. 253). Blum and Levenston (1978) explain how L2 users often make themselves understood with few words, which suggests that vocabulary can be of extreme importance to learners from the very first stages of L2 acquisition. Evidence that some learners will concentrate on words that carry meaning, presumably in order to allow them to communicate, is seen in Yoshida's study of a 3-year-old's L2 lexical development over a 7-month period for which she gives the following account:

Krashen and Terrell (1983) assert that vocabulary is acquired via comprehensible input and that its development is very important for the acquisition process. According to Krashen and Terrell (1983), while the learner's attention is devoted to vocabulary recognition in the prespeaking stage, learning vocabulary continues to play an important role through the early speech stage and after speech emerges. Krashen and Terrell (1983) contend that since vocabulary is basic to communication, acquirers must be able to produce lexical items to convey their meaning.

Similarly, Rivers (1983) states that an extensive vocabulary is essential for comprehensive communication. She also explains that in a sense, vocabulary cannot be taught since "selection for storage is an individual act of the learner" (p. 130). Recall also the earlier contention made by Corder (1967) that it is the learner who controls which elements of the input will become intake.

While recognizing the monumental task of developing an L2 lexicon, van Els et al. (1984, p. 250) contend that there are two sides to the problem of vocabulary learning: "semantization" (when the meaning of a word becomes clear) and "internalization" (when the word is stored in memory in such a way that the learner can use it effectively). Concerning semantization, Beheydt (1987) explains that "the semantization process is a permanent process that must be stimulated by repetitive, elaborative mental practice" (p. 65). Van Els et al. (1984) also point out some of the obvious factors influencing the learning of vocabulary which include characteristics of the learners (motivation, proficiency, and learning strategies) and characteristics of the actual teaching situation (the teacher, and the material).

L2 Vocabulary Learning Strategy

One common thread that runs through the current accounts of learning an L2 lexicon is that different learners employ different techniques and strategies. Accounts of this phenomenon attribute these differences to the learning environment or to individual learner characteristics such as age and L2 proficiency. In describing some of the techniques for vocabulary development used by L2 learners, Rivers cites Naiman, Frölich, and Stern (1978) and states:

The differences in approach adopted by learners are also broken into two categories by Hatch who distinguishes between "data gatherers" and "rule formers" (1974, p. 7). Ellis (1985b) contends that vocabulary is very important for the first type of learner since they are more interested in fluency than accuracy. However, it is interesting to consider that Hatch's second group of learners (the rule formers) could very well be memorizing vocabulary lists much like the "studiers" described by Ellis (1985b).

Summary

Learning an L2 vocabulary differs from learning an L1 vocabulary, and also differs from learning other features of the L2. However, in vocabulary acquisition, as in all aspects of language learning, "understanding and learning are not all-or-nothing-processes" (Beheydt, 1987, p. 60); a learner's ultimate attainment of a single word or structure is rarely achieved on its initial encounter. Vocabulary learning is a very complex process, which is essential for successful L2 acquisition, and will inevitably lead different learners to tackle the task in different ways. The lexicon an L2 learner can use, however limited, can be a very useful tool even in the earliest stages of learning. The strategies learners use in developing an L2 lexicon vary tremendously. Different types of learners use different strategies which allow them, consciously or unconsciously, to expand their vocabularies.
 

L2 Vocabulary Teaching Theory

Vocabulary received considerable attention during the reign of the Grammar-Translation Method, but it fell from favor in the Audio-Lingual days when students were learning patterns of syntax into which they could later plug lexical items. The recent trend toward communicative language teaching has brought three major changes to the theories underlying the teaching of the L2 lexicon.

First, vocabulary teaching no longer takes a secondary role to the teaching of syntax. In an effort to resurrect vocabulary teaching, Judd (1978) went so far as to say that "since many of the errors in syntax will only disappear with time, classroom exercises might be better devoted to vocabulary enrichment" (p. 75). Earlier cries were also heard; Carter and McCarthy refer to Wilkins (1972) who stated that "without grammar very little can be conveyed, without vocabulary nothing can be conveyed" (Carter & McCarthy, 1988, p. 42). While current pedagogy has not gone to the extent of ignoring syntax in favor of vocabulary, it does recognize the importance of an equal balance between the two. As Celce-Murcia and Rosensweig state "neither impoverished structure nor vocabulary is desirable ... both should be properly taught since concentration on one of these areas--to the exclusion ... of the other--has negative consequences" (1979, p. 242).

Communicative teaching methods by definition provide learners with opportunities to convey actual meaning through task-oriented activities mediated through the L2. This idea of actual meaning constitutes the second major change in theory behind the teaching of L2 vocabulary. Krashen and Terrell contend that "acquisition will not take place without comprehension of vocabulary" (1983, p. 155) since recognizing the meaning of key elements in the utterance is essential for comprehension and, they contend, comprehension leads to acquisition. Krashen and Terrell claim that "it appears to be the case that 'memorized' or 'drilled' vocabulary does not stick; words learned by rote or drill do not enter permanent memory storage ... long-term retention occurs only with meaningful exposure in situations in which real communication takes place" (1983, p. 156). Although they offer no direct evidence for this contention, Krashen and Terrell (1983) encourage "true vocabulary acquisition" through activities which are not necessarily vocabulary builders, but focus attention on communication. Rivers (1983) describes several vocabulary-oriented activities designed to teach single words but stresses that "the types of activities developed should, whenever possible, involve using the words in the context of a meaningful utterance or series of utterances in a communicative exchange" (p. 128). However, Nation (1982) in reviewing the empirical evidence available cautions against jumping on the context bandwagon prematurely with the following reminder: "In fact, we know that, 'words taught in isolation' are retained very well indeed, both in large quantities and over long periods of time" (p. 22) (cf. Judd, 1978; Krashen and Terrell, 1983). Nation does not necessarily recommend teaching vocabulary in lists but points to the reality that "however attracted we are to the idea that 'vocabulary should be taught in context' this idea remains a statement of belief rather than a conclusion based on experimental evidence" (1982, p. 24).

The third major change in the outlook toward teaching vocabulary brought about by recent pedagogy is the recognition of the individual learner and the interest in attuning the lexicon to each learner's needs. According to Carter and McCarthy, "the recent trend ... has been to assist the learner to learn, ... and to personalize vocabulary expansion according to needs, purposes and goals" (1988, p. 49). In explaining just how important a personalized vocabulary is for an L2 learner, Rivers states that "learners must eventually develop their own vocabularies if they are to be able to use the language freely in the types of situations into which their personal or career interests will propel them" (1983, p. 118).

L2 Vocabulary Teaching Strategy

Approaches to teach L2 vocabulary can be generally classified into two categories: those that employ a focused approach and those that employ an unfocused approach to presenting new lexical items. Van Els et al. (1984) contend that there are essentially three ways of presenting new L2 words; however, all three imply a focused approach. They are as follows:

The most prominent of the unfocused approaches is put forth by Krashen and Terrell whose Natural Approach is based on the following premise:

Summary

In summary, current theory has placed renewed importance on teaching the L2 lexicon and claims that meaningful use of the L2 is most effective. It is also believed that the needs, interests, and goals of the learners should be considered when selecting the vocabulary to be presented. Approaches to teaching L2 vocabulary reflect either a focused or an unfocused presentation of words. L2 pedagogy seems to be steering away from the rote-based memorization of vocabulary items and aiming toward using the language in meaningful communication while attempting to meet the needs of individual learners. This shift seems to be in response to the inability of rote-memorization to account for the vast amount of vocabulary knowledge acquired by L2 learners which is beyond the realm of what could possibly be memorized. However, there seems to be an equally insufficient amount of evidence to support the shift in teaching pedagogy which calls for new lexical items to be introduced in such a manner that their meanings will become clear through the context or other information supplied.
 

The above review begins with a description of the Input and Interaction Hypotheses. The Input Hypothesis contends that new words and structures are acquired when they are contained in meaningful messages made comprehensible through the use of context, extralinguistic information, and the linguistic competence of an L2 learner who is open to the input. The Interaction Hypothesis is an extension of the Input Hypothesis which contends that it is the NSs' adjustments to the interactional structure of conversation that are probably most important for providing comprehensible input. This review then describes numerous studies which examine, and for the most part support, the contention that input and interactional modifications facilitate comprehension. However, there are few studies that investigate whether input made comprehensible through interactional adjustments facilitates acquisition, and this review shows that the existing studies offer mixed findings. No studies to date address these variables with young L2 learners.

Studies which investigate the role of learner participation are also reviewed. Again results are mixed; however, it appears that learners should have the opportunity to participate actively in classroom interaction if they are so inclined. Findings from several studies suggest, however, that learners who do not participate may benefit from the interaction their classmates engage in.

This review then describes vocabulary learning as a very complex process which is essential for successful L2 acquisition and will inevitably lead different learners to tackle the task in different ways. Approaches to teaching L2 vocabulary are described in terms of either employing a focused or an unfocused presentation of words. L2 pedagogy seems to be steering toward an increased use of an unfocused approach; however, there seems to be an insufficient amount of evidence to support this shift.

The present study attempts to address these remaining problems to the extent that it is designed to investigate the relationships among interaction, comprehension, and acquisition which occur through an unfocused presentation of new lexical items. By employing both individual and group communication tasks, the present study also investigates the role of learner participation. Additionally, this is an initial attempt to investigate these variables with young L2 learners.