The Role of Interaction in Selected Studies on Second Language Acquisition

Nov. 17, 1997

The distance between theory and practice of research is sensed in this assignment. The different forms of the major research approaches were not accurately embodied in the studies reviewed for this paper. Although the authors of the reviewed studies claimed their studies were empirical, the characteristics of empirical research were not entirely exhibited in the studies. This paper reviews 22 studies on SLA and interaction to find patterns in their research design. A further purpose of this paper is to investigate the general drawbacks or problems that are found common in those studies.

The studies in this paper have tested empirically the hypothesis of interaction and its role in SLA. Interaction has attracted more interest in the second language acquisition (SLA) research in the last decade. Attention to pedagogical processes is responsible in part for that growing interest in studying the influence of interaction on second language acquisition.

The type of interaction that influences SLA occurs among non-native speakers of L2 (NNS) and, more importantly, between NNS and native speakers (NS). Long (1990) proposed that interaction is necessary for SLA (but insufficient). He points out that it is essential for such a valid description of SLA to have a "synthesis of well-attested empirical findings about process and product in interlanguage development related to universals and variance in learners and learning environments" (Long, 1990, p.650). In his sample of well-established findings, Long discussed several aspects of linguistic environment: The amount and kind of verbal interaction that are important factors in SLA. The interactional features are supposed to promote SLA. Three aspects of verbal interaction can be distinguished: input, production, and feedback. Input is the language offered to the language learner by native speakers (or by other learners); the L2 learners use the input to form hypotheses about the language. Production (or output) is the language spoken by the language learners themselves. Feedback is the reaction offered by the conversational partners to the production of the language learner. Optimizing the interaction implies improving the quantity and quality of input, production, and feedback.

This paper will examine a selected number of empirical studies that participated in researching the interaction hypothesis in empirical methods. It will look for global methodological patterns in the design of those studies. This review will include a brief discussion of the research processes that this study has gone through.

Research has universal characteristics, which consumers expect to find. However, every field has its character which researchers in that field attempt to maintain. Given that, it is important to have a glance at the way SLA specialists view/conduct research in their field. Numerous SLA researchers have written books that characterize research in SLA. Among these researchers are J. D. Brown, D. T. Campbell, S. M. Gass, Donna M. Johnson, Dian Larsen-Freeman, M. Long, L. Selinker, P. Selinker, and B. W. Tuckman. These researchers have formulated most of the research designs that characterize SLA research nowadays.

Gass, S. M. (1994) suggests an anatomy or format that is typical in SLA research papers which she claims is expected to be exhibited in them:

1. Abstract
2. Introduction

1. Statement of topic area
2. Statement of general issues
3. Statement of purpose
4. Literature review

1. Method

1. Subjects

1. How many?
2. Characteristics

1. Materials and instruments
2. Procedures
3. Analysis

1. Results
2. Discussion/Conclusion
3. Notes
4. References
5. Appendices if any.

In general, this format is reflected in the studies reviewed in this paper. Perhaps the unity between these studies in audience, scope, and theme has caused them to have almost the same format.

Brown (1988) provides a straightforward model of research in SLA. This model best serves the purpose of this study, which is to compare the design and research methodology of a number of selected SLA studies.

J. D. Brown (1988) described thoroughly the statistical research which "fits into the framework of different types of research and describes the characteristics of sound statistical research." He put research in two categories: (1) secondary research and (2) primary research. He explains that the primary research is derived from primary sources of information like a group of students who are learning a language, whereas the secondary research is derived from secondary sources like books about students who are learning a language. The secondary research covers straightforward insights into SLA and analyses of theoretical movements. Primary research covers case studies (longitudinal) and statistical studies (cross-sectional). The statistical studies have two categories: surveys and experimental studies. Brown (1988) characterizes the statistical studies as being systematic, logical, tangible, replicable, and reductive.

As this paper spans the selected empirical SLA studies, it will be noticed that the study of interaction and SLA involves interrelated aspects like language pedagogy, discourse analysis, communication, sequencing, repair, feedback, attention to errors, and error correction. The involvement of these aspects is due to the nature of interaction, which is a complicated process that SLA researchers attempt to break into smaller processes.

The studies reviewed in this paper cover most of the areas which SLA researchers consider legitimate and typical areas which SLA research is usually expected to address (Brown, 1988; Larsen-Freeman & Long, 1992; Johnson, 1992). The major components that typical SLA research design has are exhibited in the studies reviewed in this paper. However, these studies differ in the details of research procedures like grouping and subject selection. Some studies did not report the process of selecting subjects for the study or the way they were assigned to their groups. Moreover, some studies reported that statistical analysis was used to analyze data but no statistical tests are named or reported except for some frequencies.

Abstracts are varied in so many ways. Some abstracts are wordy and provide unnecessary theoretical arguments that are not needed by research consumers. Therefore, I had to use the original articles which I found for all abstracts except for two dissertations. Some abstracts had no mention of subjects or research methodology. Abstract writing is very critically important.

Most of the studies are experimental which means that a theoretical concept is tested by comparing the scores of a control group to those of an experimental group. Most of the studies used native speakers for control.

A considerable number of the reviewed studies did not show adequate control over a critical element of SLA research which is testing tools used to collect data. Pilot testing of instruments is an aspect I noticed neglected in both the research process and the research report of a number of studies I reviewed (table 1.). Pilot testing of tools gives more reliability to the tool to be used to collect data. Some studies used some forms of standardized tests like TOEFL to measure the L2 proficiency of L2 speaking participants. The ongoing controversy in SLA research now is whether standardized tests are reliable enough as a measurement of L2 proficiency or not. Some researchers prefer to find their way out of that mess by constructing their pilot-testing own tools which is more reliable if compared to standardized tests.

The dominant analysis approach toward SLA in the reviewed studies is the discourse analysis approach (DA). Larsen-Freeman and Long (1991) provided a chronological review of the history of data analysis in SLA research. They indicated that attention is directed to discourse analysis approach (DA) in SLA research especially in the late 1980s (pp. 52-80). Only four studies used theperformananalysis approach (PA) and no studies were found using the contrastive analysis (CA) or error analysis (EA) approaches to analyze the data collected.

Audiotaping and videotaping are used in most of the studies in this paper. The studies have used transcription using different conventions that are accredited in the field of discourse analysis. Interviews are used in some studies as a supplement for the audiotaped conversations. They are used to provide background information about the participants.

The studies used for this review paper have used combinations of research methods and tools. Some studies have more than one experiment carried out at the same time. Therefore, they are, in other words, combinations of a number of mini-studies that collaborate to make one major project. The combining approach is justified and practiced increasingly in SLA research, which supports combining qualitative with quantitative approaches and using more than one approach in one study. Team research has contributed to this trend practiced in SLA research.

In general, examining SLA in a controlled experimental environment has a number of merits, the first of which is the ability to draw causal inferences by looking at effect on the dependant caused by the manipulation of the independent variable. Second, replication of results is open for future researchers. The third merit is the ability to select a limited number of variables for the study. However, one of the problems noticed in SLA studies is the number of interacting variables. The studies reviewed in this paper have reflected that problem. In addition, the question of validity is a problem haunting SLA researchers since it comes to mind when working either with an artificial language or in a controlled lab setting.

As required for this paper, this part will address the research process I went through to complete it.

1. This paper has more influence on me than I have on it. As I finished the review table and note cards, I looked back at the research process I went through to come up with this limited number of pages. This assignment has guided me to research avenues I have not been to before.
2. Looking at 22 studies that discussed one topic is a very informative experience. Finding this number of studies on the same topic was the first hardship I came through. The required description that the studies had to share was the secret behind that hardship especially the empirical approach. I found that the number of theoretical research papers on a certain topic is far bigger than the number of empirical studies on the same topic. This finding applies to all the databases I searched (DAI, PsyLit, MLA, ERIC, and LLB databases).
3. ERIC database seems to be wide open in regulations relating to abstract writing. There is a wide range of variance between the abstracts I found on ERIC. I decided to remove those studies from my list. Some studies used a standard style to write the abstract, whereas some other studies did not differentiate between an abstract and a summary.
4. Common sense search* was a very helpful hint that helped me in my search for empirical studies. As I have experienced in the Stapelton Library, this kind of search is highly effective.
5. Some journals publish indexes at certain times of their history. Those indexes are very helpful for researchers. However, only a limited number of journals provide that service.
6. SLA research is moving toward the empirical type of research. An earlier paper helped me realize that fact which led me to find the appropriate topic to investigate within the field of SLA research.
7. The Internet is full of useful resources on empirical research. However, search engines and contributors are part of the problem that researchers encounter finding those sites. Many contributors have great web sites but need to publicize their web sites on search engines and frequently visited academic web sites. It seems that it is assumed by many web authors that they are going to be recognized once they post their web sites which is not true. To find these web sites, I used the common sense search method by visiting the departments of English and Linguistics in some of universities. That technique resulted in a number of ongoing research projects as well as abstracts of theses and dissertations defended in those departments.
8. I realized that some research projects might have been well carried out but were poorly reported. The opposite could be true as well. On the other hand, the studies I reviewed have had great ideas that are not well carried out. Some researchers might consider replicating those studies with the elimination of the factors that caused the study to have drawbacks.