Computer-Assisted Sociolinguistic Research Methods: 沖縄地域学リポジトリ

(1)

Author(s)

Ohyama, Nakakatsu

Citation

沖縄短大論叢 = OKINAWA TANDAI RONSO, 9(1): 19-45

Issue Date

1995-03-01

URL

http://hdl.handle.net/20.500.12001/10663

(2)

N akakatsu Ohyama

No matter what else we do, we must remember that if data are inadequate, there is always the danger that the theory and conclu-sions drawn from them could be unrelaible and misleading. (Wolfson, 1986, p. 689)

Multiple methods of data collection and analysis permit a more complete view of the research object and reveal at the same time the relative potential of different methods. (Grotjahn and Kasper, 1991, p. l11)

Contents I • Introduction

II. Sociolinguistic Qualitative Data Analysis in the Past Organizing Qualitative Data

Breaking Qualitative Data Synthesizing Qualitative Data

III. Recent Computer-Assisted Qualitative Analysis Inputting Qualitative Data

Coding and Subcoding Qualitative Data Retrieving Qualitative Data

IV. Modern Technologies in the Mechanical Phase of Field Work 1) Formatting the Text

(3)

3) Coding Segments of the Text 4) Sorting out Coded Segments Advantages and Disadvantages V. Summary

Bibliography

I . Introduction

This paper discusses the various strategies for organizing sociolinguis-tic qualitative data analysis. It will cover sociolinguistic data analysis of the past, recent computer-assisted qualitative data analysis, and modern technologies in the mechanical phase of field work. An effective computer software program will be introduced along with a discussion of the numer-ous advantages and obstacles that may result from advanced technologies.

The Need for This Study

In recent times, sociolinguists tend to collect qualitative information for their study. This inclination has dramatically increased during these years. Due to this phenomenon, some of them are having difficulties in examining their information. They need a research method facilitating high efficiency, reliability, and flexibility (Glaser and Strauss, 1967; Grotjahn & Kasper, 1991). There is a great demand to facilitate modern technologies for examining qualitative information (Lincoln & Guba, 1985).

But for a knowledge of computer-assisted qualitative data analysis, sociolinguists will face an obstacle in trying to manage a large amount of information. It goes without saying that the computer, through the mass production, is accessable to all individuals and is a useful machine for examining sociolinguistic qualitative information because of its mechanical storage. There is a need to know how to examine complex information quickly, inexpensively, and accurately.

(4)

II. Sociolinguistic Qualitative Data Analysis in the Past

Before the computer was available, sociolinguists had to process and examine information in the form of field notes, official information, news-paper articles, subjects' written memoranda, and interview transcripts_ Those who have not studied research methods might consider that qualita-tive data analysis seems to be an insurmountable task They occasionally fail in properly managing the information after collecting it_ Those who studied research methods might develop their research strategies so that they can examine their information by organizing, breaking data into manageable units, and then synthesizing it_

Organizing Qualitative Data

Organizing qualitative information entails reading all the information and numbering it sequentially and/or systematically_ This process helps the researcher understand the variety of incidents, events, and categories of the data_ The qualitative researcher, through observing and examining the information, can accumulate the same type of incidents_

Glaser and Strauss have stated that:

The analyst starts by coding each incident in his data into as many categories of analysis as possible, as categories emerge or as data emerge that fit an existing category_ While coding an incident for a category, compare it with the previous inci-dents in the same and different groups coded in the same category_ (Glaser and Strauss, 1967, P- 105-106)

As the sociolinguist keeps accumulating information, " ___ the previous incidents in the same and different groups coded in the same category" also starts to appear from the qualitative information organized on the specific

(5)

topic.

Breaking Qualitative Data into Manageable Units

Breaking qualitative information into manageable units entails creat-ing a small number of codcreat-ing classifications. Furthermore, it entails developing more coding classifications due to the fact that the small number of coding classifications is occasionally too inaccurate to indicate specific incidents. According to Glaser, researchers need to ask many questions of the qualitative information, such as "What is the purpose of the data study?" and examine each sentence.

The second rule is to analyze line by line, constantly coding each sentence. This may seem somewhat painstaking, but as codes emerge and saturate, it becomes easier and faster. It is necessary for achieving a full theoretical coverage which is thoroughly grounded. (Glaser, 1978, p. 57)

Synthesizing Qualitative Data

Synthesizing qualitative information usually involves splitting up a large amount of qualitative information to be coded in classifications, each with more than one labeled code. It facilitates sociolinguists to organize their qualitative information sequentially and/or systematically.

Needless to say, it is time-consuming to cut and paste manually without computer assistance. The sociolinguists spend their time cutting and pasting by using scissors, glue, scotch tape, and index cards. In order to sythesize qualitative information, they make many xeroxed copies of the field notes that include the raw information. Furthermore, due to the fact that qualitative data analysis involves endless tasks of coding and recoding, they are confronted with the difficulty of keeping track of all the informa·

(6)

tion. With computers, this complex problem can be solved.

III. Recent Computer-Assisted Qualitative Analysis

The computer has become more friendly and much easier to use. It has eased the qualitative data analysis and facilitated workable kinds of analy-sis which were not previously achievable. Sociolinguists, using word proces-sors, have developed research strategies to examine their qualitative infor-mation, including inputting, coding, subcoding, and data retrieval.

Inputting Qualitative Data

In order to examine qualitative information, sociolinguists use word processors to input the qualitative information, containing field notes, interview transcrips, and written memoranda. It is much easier to read information on a printed hard copy than on a screen. Reading the collected data is an essential part of the data analysis. Hammersley and Atkinson have stated that:

The first step in the process of analysis, then, is a careful reading of the data collected up to that point, in order to gain a thorough familiarity with it. At this stage the aim is to use the data to think with. (Hammersley and Atkinson, 1983, p. 178)

This data inputting process allows the sociolinguist to gain familiarity with the information.

Coding and Subcoding Qualitative Data

After the preparation process is complete, the sociolinguist must begin the step of coding and subcoding qualitative information. Glaser has stated

(7)

that:

The analyst codes for as many categories that might fit; he codes different incidents into as many categories as possible. New cate-gories emerge and new incidents fit existing catecate-gories. He may even code for what is not obviously stated. This maximizes allowing the best fit, the most workable ones and the core relevancies to emerge on their own. (Glaser, 1978, p.56)

Sociolinguists start with a process called open coding, coding the data every possible way. For determining classifications, they typically begin with a small amount of codes from the theoretical background. Then they develop coding categories that have emerged from their data and subse-quently group the classifications.

For example, a coding system was developed by Arizona State Univer-sity's (ASU) Job Language Training Program members, including myself. Grouped codes involve the collection of codes which were in the same type of category (See Table 1: an example of the coding system used by ASU Language Training Program). The ASU Job Language Training Program groups twenty five codes into six different categories. First, in the situation category, there are five codes; 1) situation-physical, 2) situation-social, 3) situation-temporal, 4) critical event, and 5) job change. Second, the follow-ing four codes are in the environment category; 1) quality improvement team, 2) temporary, supplement, and other job status, 3) job advancement, and 4) company policies. Third, there are three codes in the job category; 1) job tasks, 2) job coordination, and 3) job performance. Fourth, the next four codes are grouped as participants; 1) participants/limited English proficient workers, 2) participants/group leaders and trainers, 3) partici-pants/managers, and 4) participants/other workers. Fifth, there are four

(8)

codes in the attitudes and values category; 1) cultural values, 2) age and value gap, 3) attitudes, and 4) productivity. Sixth, the following five codes are in the training category; 1) sleeting participants, 2) scheduling classes, 3) other program ideas, 4) learning the job, and 5) training.

It is necessary to choose a specific code and analyze how to subcode data. For example, the ASU Job Language Training Program uses the major code "15" for limited English proficient (LEP) workers (See Table 1). In a page of interview transcripts, the first and fourth paragraphs are coded as "15" and other categories because they referred to LEP workers (See Table 2). The researchers do this coding first in pencil and then enter it into the computer.

Table 1

AN EXAMPLE OF THE CODING SYSTEM USED BY OUR ASU JOB LANGUAGE TRAINING PROGRAM

SITUATION: 17. Situation-physical 18. Situation-social 19. Situation-temporal 41. Critical event 48. Job change ENVIRONMENT:

26. Quality improvement team

TRAINING:

23. Sleeting participants 24. Scheduling classes 25. Other program ideas 32. Learning the job 35. Training

27. Temporary, supplemental, and other job status 28. Job advancement

(9)

THE JOB: 30. Job tasks 38. Job coordination 50. Company performance

PARTICIPANTS:

15. Participants/limited English proficient workers 34. Participants/group leaders and trainers

39. Participants/managers 43. Participants/other workers ATTITUDE, VALUES: 29. Cultural values 36. Age/value gap 37. Attitudes 40. Productivity

(Coded by ASU Job Language Training Program members in 1985)

Table 2 DOCUMENT A

GROUP LEADER INTERVIEWS

<

1

>

A : 15,9, 14, B : Y au have one person working for you from Africa, don't you? When you tell him some-thing, does he understand your oral instruc-tions?

(10)

<

1

>

A : 44, 45,

<

1

>

A : 20,

B What about reading the planning packs? C : He doesn't seem to have any trouble reading

English at all. He is very bright.

C : I just think it is hard for me to understand him sometimes. He has such a strong accent. B So is his pronounciation the problem?

C Yes, and he is very shy, too. He speaks very quietly. You practically have to put your ear on his mouth to hear what he is saying. That is a big problem.

<

1

>

A : 15, 18, 37 B How does he interact with the other workers? C He stays by himself. He eats lunch with us but

he will not say a word the whole time.

B : Do you think that is because he is worried about his English?

C I think he is just shy. His brother is very quiet, too.

(Interviewed by one of our project members in 1984)

Retrieving Qualitative Data

After the coding process, the researchers get a page from the query of everything coded "15" in their data. According to Glaser (1978), sorting data should involve the following procedures:

1 • Start to sort anywhere.

2. Begin sorting all categories and properties that relate to one core concept. This rule forces focus, selectivity, and delimiting

(11)

of the core concepts.

3 . Promote one core idea at a time to the center and demote the others to sub-core ideas.

4 . Memo any new ideas and then sort the memo into the outline. 5 • Carry forward to subsequent sorts any concepts that might

need to be used again.

6 . Integrate ideas; all ideas must fit somewhere in the outline or the outline must be changed.

7 • Sort in stages; it is necessary to resort and constantly correct and confirm the outline.

8 . Cut off rules including running out of memos, saturation of core concepts, and personal saturation and completeness. 9. Cut up memos as often as necessary to increase sortability.

Write "pass on" notes on memos to be carried forward and used again in other ideas.

10. Pace sorting. One good rule is to have a flexible but regular schedule for sorting. {p. 111-112)

The ASU Job Language Training Program shows us that computers can be used for the mechanical part of sorting. Their queries are done by the computer such as the Word II computer. (However, other software is available. They also use software called the Ethnograph, which will be described in a subsequent section.) Then, they subcode a major code of "15" into twenty-three categories (See Table 3).

Subcodes of the participants-LEP workers, consist of nine major sub-categories; 1) potential students, 2) attitudes of workers, 3) qualifications and job performance of LEP workers, 4) work history, 5) language ability, 6) personal interests, 7) years in the country, 8) family, and 9) age. In subcoding a major code of the participants-LEP workers, the researcher

(12)

also needs to see the other five codes; 1) isolation of LEP workers, 2) other workers who help LEP workers, 3) attitude toward the role of managers, 4) attitudes toward work associated with age, and 5) attitudes towards project-associated with age_ An example of subcoding is shown in Table 4_ The previous sentence "You have one person working for you from Africa, don't you?" is subcoded into 103 and 108 because this sentence is indicating nationality and specific individuality_

Table 3 SUBCODES OF (15)

102 _ potential students-number

103 _ potential students-langauge background and/or nationality 108 _ potential students-specific individuals

112 _ potential students-job position or status

115 _ potential students-other background characteristics (including educa-tion)

110_ attitudes & participation of LEP workers in learning English 104 _ attitudes of others towards LEP workers-other workers 105 _ attitudes of others towards LEP workers-managers 109_ attitudes of others towards LEP workers-group leaders 113 _ attitudes of others towards LEP workers-trainers 127 _ attitudes of others towards LEP workers-liason 202_ qualifications & job performance of LEP workers 206 _ work history

214 _ language ability 218 _ personal interests 219 _ years in the country

(13)

242. family 225. age

+

Interviewing a manager of C area EI: I went through the first and last

pages of the agenda with Tom and in general he agreed that most of the areas cited in the agenda were true at ABC company, however Tom had some specific concerns he wanted to raise with me and a good part of the interview dealt with these.

For example, he wanted to tell me about what he called the new culture of workers who want to run the shop but really are not ready for it.

(16)

around. They are young. They are not used to having much responsibility. Tom talked about one new worker who was from Germany. This worker had a hard time accepting the way the young workers behaved at ABC company.

Tom said this was especially true of the supplemental and contract workers. The supplemental workers said this program was particularly aimed at college students and housewives who had never worked or had worked a long time ago, and so they have a lot of trouble getting used to the work environment.

(Intereviewed by one of our project members in 1984)

2) Numbering the Lines of the Text

THE ETHNOGRAPH software autimatically formats the data file into a standard ASCII text file. But in order to select and manipulate text, it is important to number the lines of the text. To number the lines of text, use the "Number A File Procedure" of the "ETHNOGRAPH PROCEDURE MENU." When numbering lines, remember that the ETHNOGRAPH program disk must be in drive A, and the target disk containing the data file must be in drive B.

This system for numbering the lines of text is an interactive program. It requires the user to input information as the computer prompts. In order to begin the program, type "B>A: ethno A; xnumber." Then the researcher

(17)

needs to enter the already existing formatted file's name and the desired name of the numbered file at the computer's request. Then the computer will request the researcher to respond whether or not the file names are correct. If the user confirms the information, the computer will request what number to begin numbering by and then begin numbering the lines. If the user indicates that the file names are incorrect, the computer will ask him to enter the correct names. Through this step, any typographical errors can be corrected. After the name has been correctly entered, the computer will continue to the next step. For example, in the first file the researcher can begin from number one and 201 for the following file. Numbering example is on Table 6.

3) Coding Segments of the Text

After completing the numbered data file, the researcher must input codes on a screen or on a printed hard copy. Coding segments on a hard copy is much easier than on screen. First, it is necessary to separate the printed copy into sections by category and mark these sections with the desired code. These categories will form the basic outline of the final menu. When coding on a hard copy, he uses pencil. This process is called "code mapping." The coded information is put into the computer through the "CODE A FILE" program on the "ETHNOGRAPH PROCEDURE MENU." Like the "NUMBER A FILE," the "CODE A FILE" program is also an interactive program.

(18)

Table 6

NUMBERED VERSION OF FILE A: KUNI

ID: Interviewing a manager of ABC company 1

+Interviewing a manager of C area 2

EI: I went through the first and last 3 pages of the agenda with Tom and in 4 general he agreed that most of the 5 areas cited in the agenda were true at 6 ABC company, however Tom had some 7 specific concerns he wanted to raise 8 with me and a good part of the 9

interview dealt with these. 10

For example, he wanted to tell me 11 about what he called the new culture 12 of workers who want to run the shop 13 but really are not ready for it. 14 During work hours they waste time joking 15 around. They are young. They are not 16 used to having much responsibility. Tom 17 talked about one new worker who was from 18 Germany. This worker had a hard time 19 accepting the way the young workers 20

behaved at ABC company. 21

Tom said this was especially true of 22 the supplemental and contract workers. 23 The supplemental workers said this 24 program was particularly aimed at 25 college students and housewives who had 26

(19)

never worked or had worked a long time 27 ago, and so they have a lot of trouble 28 getting used to the work environment. 29

(Intereviewed by one of our project members in 1984)

The program will ask the researcher to specify each category's bound· ary in terms of number of lines and the number of codes to be contained within that boundary. The program allows the user any length of combina-tions extending up to three seperate codes per category. Any single code may be a combination of one to ten characters including spaces. The computer can call up any one of the categories according to that code.

Typographical errors can be corrected through a program called "CODE SUMMARY." Within the "CODE SUMMARY," the coding seg-ments of the text can be changed through reentering the code name and resetting the length of the coded segments. Coding example is on Table 7.

4) Sorting out Coded Segments

After all the text is coded, the researcher can search for coded seg-ments from the file by typing specific codes he wants to collect. When searching, the "SEARCH A FILE" procedure on the ETHNOGRAPH can be used. The "SEARCH A FILE" is also an interactive program.

First, the researcher must enter the file name to use it through the "CODE A FILE" program. Then the computer will call up and display that file.

The researcher will have a second opportunity to alter the code as mentioned in the previous "CODE SUMMARY" section. In order to identify segments, the new name must be used through the "SEARCH A FILE"

(20)

program.

The researcher may enter the code words for sorting out (e.g., "SEARCHING FOR SEGMENTS DEFINED BY CODE WORD: WORKER"). Sorting example is on Table 8.

ADVANTAGE AND DISADVANTAGES

1) Advantages

Some of the more major reasons why modern technology will simplify the task of qualitative data analysis and magnify the effect of qualitative analysis are: efficiency, reliability, and flexibility.

Table 7

CODED VERSION OF FILE A: KUNI

ID: Interviewing a manager of ABC company 1 +Interviewing a manager of C area 2 #-RAPPORT

EI: I went through the first and last 3-# pages of the agenda with Tom and in 4 general he agreed that most of the 5 areas cited in the agenda were true at 6 $-WORKER

ABC company, however Tom had some 7-#-$ specific concerns he wanted to raise 8 with me and a good part of the 9

interview dealt with these. 10 -$

#-CULTURE

(21)

about what he called the new culture 12 of workers who want to run the shop 13 but really are not ready for it. 14 !-SOCIAL !-JOKE

During work hours they waste time joking 15-! #-AGE-DIFF #-WORKER

around. They are young. They are not 16-# used to having much responsibility. Tom 17 talked about one new worker who was from 18 Germany. This worker had a hard time 19 accepting the way the young workers 20

behaved at ABC company. 21-#

#-SUPP-CONT

Tom said this was especially true of 22-# the supplemental and contract workers. 23 The supplemental workers said this 24 program was particularly aimed at 25 college students and housewives who had 26 never worked or had worked a long time 27 ago, and so they have a lot of trouble 28 getting used to the work environment. 29

(22)

Efficiency

Table 8

SORTED VERSION OF FILE A: KUNI

ID: Interviewing a manager of ABC company 1 +Interviewing a manager of C area 2 $-WORKER

ABC company, however Tom had some specific concerns he wanted to raise with me and a good part of the interview dealt with these.

#-AGE-DIFF #-WORKER

around. They are young. They are not used to having much responsibility. Tom talked about one new worker who was from Germany. This worker had a hard time accepting the way the young workers behaved at ABC company.

7-#-$ 8 9 10 -$ 16-# ' 17 : 18 : ' 19 : 20 : 21-#

Sociolinguists, using WordStar, can effortlessly type, delete, and edit research memos. The qualitative information, having been input, can be examined effectively through useful features such as editing text. This software program also provides a simple method to save research memos. Through this, sociolinguists can save many hours which might have been spent transcribing research memos by hand and spending their time typing, correcting, and editing them. This word processing system shortens the turn-around time between accumulating the information and preparing the information for examination.

(23)

format, number, and sort information files effortlessly. Without this type of software, the sociolinguists would have to reread research memos every time they need to retrieve the coded segments. In addition to this tedious task, they also waste their time on continuous copying, splitting, pasting, and filing for each code analysis. This software provides solutions to those monotonous tasks by saving the researcher's time. Furthermore, it can sort information queried by classifications including many pages of data since it does not require the whole text to be input. As a result, the Ethnograph can manage a large quantity of research memos, audio tapes, or video cassette tapes, and efficiently retrieve queried information in those files.

Reliability

Another major strength of computer-aided sociolinguistic data analysis is that the sociolinguist may effortlessly identify the coded segments and sort those that satisfy some conditions without losing information (Conrad and Reinharz, 1984). If sociolinguists examine qualitative information without utilizing any software, they might make a mistake or miss various potential linkages while splitting, pasting, copying, and filing in an effort to analize the series of research data.

In contrast, Seidel and Clark (1984) say the most dependable measure of the Ethnograph is its ability to do data retrievals based upon the co-occurence (or mixture) of classifications in each sentence. It enables one to check all the relevant information without losing any data coded on the same classification. Even if the researcher inputs additional classifications for data retrieval, it enables them to be accessible to the researcher by selectively sorting the classifications.

Flexibility

(24)

research memos without any restrictions. The Ethnograph additionally provides for the suitable coding and recoding of research memos as the perception of the information alters. This has major strengths for meeting the needs of sociolinguistic qualitative information analysis including the accessibility of the raw information and the capability to sort information selectively. It consistently allows the sociolinguist to do essentially any-thing involving research memos that can be done on paper, but much more flexibly.

Additional Addvantages

Additional strengths of the advanced technology are: portability, in-expensiveness, ease of mastering, and confidentiality (Becker, et al., 1984). The computer has been improved based on the modern technology and has been created into a more powerful machine. The computer has been formed in a smaller shape so that it can be moved around to anywhere the researcher wants to carry it.

With regard to cost, the advanced technology has also brought costs down due to mass production. In today's society, inexpensive computers are available to most researchers.

Regarding the ease of mastering, the modern technology has provided effortlessly mastered systems and an integrated data base system for the convenience of the researcher. Namely, difficult and sophisticated opera-tions have been improved and simplified for the researcher.

With regard to protecting confidentiality, much effort has been made to protect the researcher's right and obligation toward the qualitative informa-tion analysis. The computer protects confidentiality by prohibiting any unauthorized use. The authorized researcher typically uses an ID number for this purpose.

(25)

can be made. The mechanical aspects of qualitative analysis are more effortlessly done with the Ethnograph than contrasted to how they were achieved in the past without this computer software. The computer-assisted qualitative research strategy can simplify and quicken the task of the researcher and accelerate each step of the data analysis. This advanced technology will release researchers from monotonous tasks in typing, copying, pasting, and shuffling the data and allow them to concentrate on the more analytic task of qualitative information.

2). Disadvantages

In spite of various promised strengths, computer-assisted sociolinguistic data analysis must still conquer its weaknesses. Some of them are: the time consumed for typing research memos, limited storage capacity, strict directions for each margin space, and operation limitations. These follow-ing comments will focus on the weaknesses and the research for future studies.

The major weakness of the Ethnograph is that the data must still be input into the computer by hand. For the purpose of removing this weakness, voice input to a word-processing software might be developed (Becker, et al., 1984). For example, the sociolinguistic qualitative researcher could transcribe the qualitative data without scretarial tasks, utilizing the mechanical aspects of voice input. This mechanical task might increasingly simplify the preparation of raw research notes.

Another weakness of the Ethnograph is that converting, numbering, coding, modifying, and sorting program occupy a considerable percentage of its memory. Also, the Ethnograph has strict directions about the size of each margin space (Seidel and Clark, 1984). Since the text file stores the qualitative information in limited columns (through eight to forty four spaces from the left margin), the memory of the file quickly exceeds the

(26)

storage capability of the Ethnograph. This software can store about 320k per drive in its memory at each time. More qualitative information is needed by most sociolinguistic studies and hence more memory storage space is required.

The third weakness of this software is that it _also has rigid rules concerning the length of combinations extending up to only three separate codes per category_ Since any individual portion of information can only be marked by up to three codes, information can only be retrieved three ways. If the maximum number of possible codes per piece of information was increased, the manipulation of these codes would also increase allowing for a more complete retrieval of the input information.

The fourth weakness of this software is that sociolinguistic researchers are asked to do analytical tasks in the process of coding, modifying, and sorting the qualitative information. Its weakness stems from its incapabil-ity of analytic operation (Seidel and Clark, 1984). It only manages the mechanical operation of data analysis. For example, it can neither code the classification nor examine their continuing suitability. This software constantly requires researchers to input information based on their analyti-cal thinking_ It, then, provides its prompt to do the mechanical operation of qualitative data analysis. Its role for the qualitative researcher is to sort and display the raw information for a certain classification coded by the researcher. In no distant future, this is the area where artificial intelligence might be applied (Gerson, 1984). A computer software might identify incidents, concepts, or ideas, and it may be able to be utilized in making rational data deductions. If it is possible, computer-assisted qualitative data analysis can manage not only the mechanical application but also the intelligent tasks for the qualitative researcher. Namely, the advanced computer software could organize the qualitative information in place of a research assistant or as a secretary_

(27)

V. Summary

Computer-assisted sociolinguistic analysis has achieved considerable tasks for qualitative researchers. The computer software has unbound them from many tedious tasks and allowed them to concentrate their effort on more analytic objectives. The Ethnograph, in spite of its some weak· nesses, has made contributions to sociolinguistic qualitative information analysis. It helps researchers to locate segments of the information and code classifications. After classifying incidents, concepts, and ideas, researchers can effortlessly and efficiently make rational deductions so that they can generate interpretations of the qualitative data with this advanced computer software.

*This paper used the data which had been collected by Arizona University

] ob Language Training Program members, including the author. This

paper is the revision of the oral presentation in the TESOL'86 conference held in Los Angels.

Bibliography

Becker, Howard S., et a!. "Field Work with the Computer: Criteria for Assessing Systems," Qualitative Sociology. 7(1984), 16-33.

Bogdan, Robert C. and S.]. Taylor. Introduction to Qualitative Research Methods: a Phenomenological Approach. New York: John Wiley, 1975. Conrad, Peter and Shulamit Reinharz. "Computers and Qualitative Data:

Editor's Introductory Essay," Qualitative Sociology. 7(1984), 3-15. Crookes, Graham. "Second Language Speech Production Research," SSLA.

13(1991), 113-132.

Gerson, Elihu M. "Qualitative Research and the Computer," Qualitative Sociology. 7(1984), 61-74.

(28)

Chicago: Aldine, 1967.

Glaser, Barney. Theoretical Sensitivity. Mill Valley, CA: Sociology Press, 1978_

Grotjahn, R. and Gabriele Kasper_ "Methods m Second Language Research," SSLA 13(1991), 109-112.

Grotjahn, Rudiger_ "The Research Programme Subjective Theories," SSLA 13 (1991), 187-214.

Hammersley, Martyn and Paul Atkinson. Ethnograph: Principles in Prac-tice. New York: Tavistock Publications, 1983.

Kasper, Gabriele and Merete Dahl. "Research Methods in Interlanguage Pragmatics," SSLA. 13(1991), 215-248.

Lincoln, Yvonna S. and Egon G. Guba. Naturalistic Inquiry. Beverly Hills, CA: Sage, 1985.

Miles, Maatthew B. and A Michael Huberman. Qualitative Data Analysis: A Sourcebook of New Methods. Beverly Hills, CA: Sage, 1984. Nunan, David. "Methods in Second Language Classroom-Oriented

Research," SSLA 13(1991), 249-274.

Seidel, John V. and Jack A Clark. "The ETHNOGRAPH: A Computer Program for the Analysis of Qualitative Data," Qualitative Sociology. 7(1984), 110-125.

Wolfson, N. "Research Methodology and the Question of Validity," TESOL Quarterly, 20(1986), 689-699.