• Tidak ada hasil yang ditemukan

Research Design

Dalam dokumen Case Study Research in Applied Linguistics (Halaman 119-125)

Excerpt 6 June 1988)

4.4 Research Design

Language loss occurs “… when [a] minority group member cannot do the things with the minority language that he [sic] used to be able to do.… Some of the proficiency he [sic] used to have is no longer accessi- ble” (Fase, Jaspaert, & Kroon, 1992, p. 8). It may also refer to incomplete or imperfect learning of a language spoken in childhood. (Kouritzen, 1999, p. 12)

She then provides definitions of other terms related to or used synony- mously with language loss in some contexts, such as language shift, lan- guage attrition or language regression, subtractive bilingualism, language death or obsolescence, and language change. The operationalization of the construct for her research purposes became clear in her recruitment and sampling/selection of subjects (see Section 4.5.2).

4.4.2 Establishing Chains of Evidence

Like most research, regardless of paradigm, case study research design entails establishing clear, credible, coherent, and strong “chains of evi- dence” (Gall, Gall & Borg, 2005; Krathwohl, 1993; Yin, 1994) or an “audit trail.” Yin (2003a) uses the phrase “the logic of design” to refer to this process of linking research questions and data as part of “a comprehensive research strategy” (p. 14). Decisions about research design—how many cases, which kinds of cases, contexts for cases, types of analyses foreseen or intended—are based on the kinds of evidence needed to speak to an issue with any authority (Bachman, 2004). Readers, in turn, can assess the strength of arguments and interpretations by seeing clear connections among the research questions, raw data, analyses of data, and conclusions (Gall et al., 2005).

Denzin (1994) discusses issues of representation and legitimation.

Representation refers to how we represent or position our participants, data, and interpretations, and also, perhaps indirectly, how we position ourselves as researchers in relation to those studied. Legitimation is the basis for the warrants or claims we make about our data and the authority of our reports (see Edge & Richards, 1998).

A concept related to warrants is the need to provide an audit trail (Gall et al., 2003, 2005; Merriam, 1998). This involves keeping records of rel- evant documentation for decisions made, data collection strategies, the development of instruments or protocols, and examples of analysis pro-

cedures. These records might be included in an abbreviated form in an appendix or, if not published, might constitute a paper trail of what was done and why, should questions arise later. Journal reviewers, dissertation committee members, or scholars wishing to replicate the study or follow similar steps may seek further information, but it can be difficult to recall the basis for decisions several years after having made them without keep- ing good notes or records.

4.4.3 Mixed-Method Designs

Case study is increasingly used in mixed-method studies such as pro- gram evaluations. Here I will provide some details about one such study in the context of a Japanese Foreign Language in the Elementary Schools (FLES) program in the United States. Antonek et al. (2000) reported on their evaluation of the program in its fourth year, which they had been investigating since its inception. Using a stratified random sample of 32 qualified students, they conducted oral proficiency interviews and a word recall assessment exercise with each one. They then described a typical day in the Japanese program and proceeded with a multiple-case study of students’ achievement, linguistic growth, and attitudes toward learning Japanese in the program:

We have conducted a case study of six upper-elementary learners to provide a more in-depth view of the older FLES learner. Their profiles enable us to document and examine more closely differential linguis- tic development and differential engagement with the Japanese FLES program of older learners who have had an uninterrupted sequence of instruction.… Having identified students who had relatively homoge- neous Japanese learning experiences, we selected three students who had received Pro-I [oral proficiency interview] ratings in the novice range and three students with intermediate range ratings. (p. 330)

They then reported on not only the proficiency measures for these stu- dents, but also the results of a learner questionnaire, an interview with a classroom teacher, and with the FLES (Japanese) teacher specifically, and examined the students’ Japanese homework. Next, they conducted a cross-case analysis of the six children, first describing each student in some detail and then building a composite picture of all six students in

terms of their achievement and attitudinal data. They concluded that mul- tiple measures in the assessment of language learning by children are very important because the learners demonstrated the “multifaceted, develop- mental nature of language learning” (p. 344). Citing one of the students, Lilly, as an example of a learner who had made “dramatic improvement” in oral proficiency despite not having a “completely positive attitude profile,”

they observed:

Labels alone do not tell us all that we need to know about a student’s proficiency level. Knowing the optimal conditions under which Lilly was learning Japanese provided an understanding of her dramatic profi- ciency gains that would not have been possible if only numbers had been collected. Lilly’s case demonstrates the importance of interviewing both FLS learners and their parents. (p. 347)

Besides mixed-method program evaluation, another common design is to conduct a survey (e.g., involving questionnaires) and then to follow up with a small number of respondents who indicate a willingness to take part in additional research and who represent important sectors or types of cases within the larger survey. The survey then also allows the researcher to establish the representativeness of the cases presented. Alternatively, the larger study might be (quasi-)experimental, with case studies included to provide a more concrete or more vivid description of the phenomenon explored in the larger study. Another approach is to conduct a larger quali- tative study, of multiple institutions, for example, and then to concretize the analysis by including case studies of individual institutions and their ecol- ogy of teachers/employees, administrators, or students/clients to illustrate general trends or differences among the sites. Other options for mixed- method study designs are described further by Caracelli and Greene (1993) and Tashakkori and Teddlie (1998). Qualitative and quantitative analyses may be conducted concurrently or sequentially (with either one preceding the other type), but with a synthesis of results at the end.

4.4.4 Single-Case versus Multiple-Case Designs

Yin (2003a) presents four types of case study research design in a 2 × 2 matrix (see Figure 4.3); along the horizontal axis (top) are single- versus multiple-case designs, and along the vertical axis (left side) are holistic

versus embedded units of analysis. In this way, he differentiates contexts, cases, and units of analysis but concedes that “the boundaries between the case and the context are not likely to be sharp” (p. 39). Holistic analyses would involve a more global and possibly more abstract consideration of one or more cases, such as organizations, programs, or participants.

Yin suggests that a single-case design is appropriate—and even

“eminently justifiable” (p. 45)—when studying “critical” or “revelatory”

cases (e.g., Schmidt’s (1983) study of Wes). In such instances, “the sin- gle case can represent a significant contribution to knowledge and theory building” (p. 40). It is also appropriate when studying extreme cases, either unique or representative/typical cases, longitudinal cases, or pilot cases in a planned multiple-case study. He cites Whyte’s (1943/1993) Street Cor- ner Society as an example of a revelatory case, “previously inaccessible to

Figure 4.3 Types of case study design. (From Yin, R.K., Case Study Research, 3rd ed., Sage, Thousand Oaks, CA, 2003, p. 40.) Copyright 2003 by Sage Publications. Reprinted with permission of Sage Publications.

scientific investigation” (p. 42).However, Yin cautions readers that single- case designs may be unpredictable, and they “therefore require careful investigation of the potential case to minimize the chances of misrepre- sentation and to maximize the access needed to collect the case study evi- dence” (p. 42). Then, returning to his other axis, within a single case there may be one main focus (unit of analysis) or multiple embedded units. For example, if the case was a school, one might examine teachers, students, or curriculum policy documents as separate units of analysis, or select several individuals from the same category for comparison; for an individual, one might study different skill areas (speaking vs. writing) as separate units.

Or one might narrow the unit of analysis further by examining particular types of linguistic structures in a study about language acquisition and/or use. In such studies, Yin stresses the importance of returning to the larger unit of analysis after examining the parts.

Multiple-case designs, if done well, can provide compelling evidence of a phenomenon and are preferable to single-case studies, according to Yin, although not all case study methodologists agree or feel so strongly about this matter (see also Chapter 2). However, this putative advantage must be weighed against the time and other resources needed to include additional cases and the resulting trade-offs in depth of analysis between a study of one and a study of two or more. Being a positivist, Yin makes the analogy of multiple cases and multiple experiments, and sees the advan- tage of multiple studies as a form of literal or theoretical replication. Hav- ing two or more cases can help assuage concerns that cases are unique in unforeseen ways.

4.4.5 Closed or Flexible Case Study Design

A final design feature is whether the study is closed (predetermined and inflexible) or flexible, evolving throughout the research process (Hatch, 2002; Yin, 2003a). Most qualitative researchers recognize the importance of being accountable to the unfolding data and situation and view flexibil- ity as a strength of interpretive research. That does not mean that the stud- ies should be unstructured or without focus, though. Research participants, supervisors, and institutional ethical review boards normally require that

The fact that Whyte’s book is currently in its fourth edition and has been translated into numerous languages is a testament to the enduring power of the case.

researchers plan their studies carefully and be as explicit as possible about the study’s goals and intended methods in advance. Besides allowing oth- ers to judge the soundness of the design and methods, for research ethics purposes it enables the researcher to be very clear about what the expecta- tions are for potential participants, in terms of both the time and the activi- ties they will engage in. They also need to know the focus (to the extent one can reveal all of the specifics in advance without influencing behav- iors unduly), the procedures, the benefits or risks to the participants, and the total duration of the study (see Section 4.14). A significant change in design, such as the inclusion of new types of subjects or new instruments, may require the submission of amendments to the original ethical review application or the submission of a new application.

Dalam dokumen Case Study Research in Applied Linguistics (Halaman 119-125)