Guthrie Bilingual Corpus

Larry F. Guthrie
Far Western Research Laboratory

Participants: 14
Type of Study: classroom
Location: USA
Media type: audio
DOI: doi:10.21415/T5S307

Browsable transcripts

Download transcripts

Citation information

Publications using these data should cite:

In accordance with TalkBank rules, any use of data from this corpus must be accompanied by at least one of the above references.

Other references include:

Project Description

This subdirectory contains data from a detailed examination of the language use of a group of Chinese-American first-graders and their two teachers. The files were collected by Larry Guthrie of the Far Western Research Laboratory and donated to the CHILDES in 1985. They were reformatted into CHAT in 1986.

Although considerable information is available on language use in monolingual class-rooms, and to a lesser extent on that in Hispanic bilingual situations, very little is known about how Chinese children and their teachers construct interactions. The focus of the re-search was a bilingual class of students that alternated each half-day between a Chinese bilingual teacher and a teacher who did not speak Chinese. This provided the unique opportunity to examine the language of the same Limited-English-Speaking (LES) children with two different teachers. The first of these teachers not only spoke the students’ first language, Cantonese, but was also of the same cultural background. A woman in her early twenties, she had immigrated to the United States at the age of 9. Both her Cantonese and English were native-like. The other teacher was an Anglo male who had taught in Spanish-English bilingual programs, but had little prior experience with Chinese students.

Three basic questions directed the research. The first of these sought an in-depth de-scription of the classroom interaction between Chinese-American children and their teachers. How do teachers orchestrate lessons and how, in turn, do students respond? What variation, in both teacher and student language, is found across student English language proficiency groups? Second, we compared the interaction in the two classrooms. What differences occur between the ways in which the two teachers orchestrate lessons? What differences emerge in student language use? How do these differences compare across linguistic proficiency groups? Third, we asked what variations in teacher and student language might be found when this group of children moved on to second grade. Did these students experience difficulty in crossing the “border” between first and second grade, or in adjusting to the rule system of the new teacher?

Sociolinguistic methods were used to seek answers to these questions and to uncover the ways in which Cantonese-speaking children and their teachers constructed their inter-actions and used language. The study was conducted in three phases. In the first phase, target students and speech events (lessons) were identified. In the second phase, recordings of sample lessons were collected, transcribed, and analyzed. The third phase involved additional recording in reading lessons after target students had progressed to second grade. The procedures employed within each phase are described in more detail later. First, however, is a description of the setting in which the study was conducted.


The setting for the study was an elementary school with a predominantly Chinese population. The school was located near a large Chinatown community on the West Coast. There were approximately 644 students enrolled in Chinatown Elementary at the time of this study. The school population is relatively stable, but there are periodic influxes of new immigrant and refugee populations. Almost half (44.6%) of the school population was Chinese; the remainder of the students were largely Hispanic (19.9%), other Asian (20.5%), and Black (11.6%). Because of the ethnic quota system operative within the district, the school is now officially “closed” to new Chinese students, except those who live within the most immediate neighborhood. Most of the Chinese students at Chinatown Elementary are classified as either Limited-English-Speaking (LES, 28%) or nonEnglish-speaking (NES, 61%). These students, in turn, are placed in either a bilingual or regular class.

Within the Chinese community, the school has a good reputation. Most Chinese parents seem to feel more secure if their children are attending a school that is predominantly Chinese and has Chinese teachers. There have been reports of parents who submitted a falsified address, or used that of a relative, in order that their child might be allowed to attend the school.

The participants in Phase One were eleven first-grade Chinese-American students, se-lected on the basis of English-language proficiency. Prior to data collection, each teacher was asked to rank all students in the class on a four-point scale of oral English language proficiency. The bilingual teacher also provided similar information on students’ Chinese proficiency. These judgments were then verified through observations of potential target students. In this way, five students ranking at the low end of the scale (1-2), four ranking at the middle of the scale (3), and two fluent English speakers were selected.

Lessons: Reading & oral language

As mentioned earlier, the two participating teachers in the study taught in a half-day alternation bilingual program. Each teacher met with the students in the target class for half of each school day, and alternated between mornings and afternoons. One teacher was bilingual and biliterate in Chinese and English, and although the other spoke no Chinese, he did speak Spanish and had taught a self-contained Spanish bilingual class the year before. Both teachers had several years of experience.

Two types of lessons were selected for analysis in this report — reading with the bilingual teacher and oral language with the Anglo teacher’s class. All of the lessons in the CHILDES database are in English. Although the lesson content and focus differed somewhat across the teachers’ lessons, they were in many respects comparable. For two weeks prior to taping, classroom observers took descriptive field notes and coded for activity structures. These two lessons were found to be compatible in that they were both teacher-directed, student membership was approximately the same, and both teachers organized lessons around a basic question/answer format. Descriptions of the typical organization of each teacher’s lesson follow.

The bilingual teacher divided students into four instructional groups for reading: Flintstones, Roadrunners, Bugs Bunnies, and Snoopies. Each group met with the teacher for 15 to 20 minutes during each reading period, rotating according to the schedule set up by the teacher. Reading lessons were conducted in much the same way with each group. The teacher usually began by writing a list of vocabulary words on the board near the reading table. She then would introduce each word and ask students to read and say the words as a group. Individual students were then called on to read all the vocabulary words aloud. The next activity included story posters. Each poster contained a picture on the top and a story below. When she used the poster, the teacher would ask the students to look at the picture first, then ask them to describe the picture. Together, they would then read the story on the poster. When she used the book, she adopted the same approach as with the poster, beginning with a description of the picture, followed by reading. The final step in the typical reading lesson would be to ask the children to read the text silently, after which she asked them comprehension questions. To answer these, students were allowed to read an appropriate phrase or sentence from the text. Throughout the reading lesson, if students stumbled over a word, the teacher read it out and asked the student to repeat it.

The Anglo teacher divided his class for oral language into two instructional groups on the basis of oral English proficiency: Low and a combination of Middle and High. However, during the oral language period, only that group being taught by the teacher remained in the classroom; the other group met with another instructor in a different room. The overall procedures employed with each group were much the same. The Low group consisted of six students who sat in their assigned seats. For oral language, the teacher would join the group by pulling up an additional chair. Very often the lesson began with picture flashcards, which students were required to identify and describe. The Middle/High group was composed of nine students. They all sat at a table in the center of the room, where only the Middle group students normally sat. The teacher brought his own chair when he joined the group. Once again, the teacher usually began with picture flashcards, which the students were to identify. Chinese lessons taught by the bilingual teacher as well as seat work in the other teacher’s class were recorded as well.

Phase Two

In Phase Two, teachers and target students were recorded in different lessons: oral language and seat work in the Anglo teacher’s class and reading and Chinese in the bilingual teacher’s class. These were transcribed, coded, and analyzed. The following is an overall description of the activities within this phase of the study. Audiotape recordings were made through the use of a Marantz recorder, with two lavaliere microphones placed in the middle of each group’s table. All data collection for Phase Two was conducted over a 2-month period in the spring of 1982.

Two data collectors were present during each taping session, both fluent speakers of Cantonese, Mandarin, and English. One data collector took field notes on the activities of the focal group, recording information on the physical arrangement of the group, important nonverbal behaviors, the text and materials used, and other contextual information. The other data collector, meanwhile, monitored the audiotape through earphones. Because of incidental noise in the class and the voices of students in other groups, the earphones enabled the data collector to hear the speech of the teacher and target students much better. This data collector wrote down names and utterance fragments of speakers throughout the interaction to aid in subsequent transcription.

The audiotape recording of each lesson was transcribed by the data collector who monitored that taping session. The handwritten transcript was then entered into an IBM-PC used for the analysis. Those utterances in Chinese were transcribed in Chinese, and an English translation was provided in brackets. Descriptions of nonverbal behavior were included in parentheses.

Recordings took place between February 1982 and October 1982. Lessons 1 to 28 were recorded while students were in first grade. Lessons 30 to 38 were recorded while students were in second grade. The following tables list the target children, additional children in the classrooms, and the adults in the classrooms.

1. Guthrie children
CodeChild’s nameSexAge
*PHUPhung NgocF6;11
*AHTAh TayF?
*AHNAh NgatF6;10
*AHPAh PhangF6;6
*MEOMei OanhF8;0
*HIEHieu NghiF6;7

2. Guthrie – Other children
CodeChild’s NameSex
*WAIWai YeeF
*STUUnidentified Student 1
*STUUnidentified Student 2

3. Guthrie – Adults
CodeAdult’s NameDescription
*MARMary(bilingual teacher A)
*LARLarry(Teacher B)
*ELAElaine(Mary’s aide)
*ELLEllen(Larry’s aide)
*TRATracy(Mary’s student teacher)
*FOOMiss Foo


There are three subdirectories. The “Larry” subdirectory contains the files from the monolingual Anglo teacher’s class of first graders. The “Mary” subdirectory contains the files from the first grade class of the bilingual teacher. The “Maisie” subdirectory contains the files from the second grade teacher’s class in the follow-up study. The teacher was observed on three different days with three reading groups on each day. For the “Maisie” subdirectory, the following speakers were present:

4. Speakers in the Maisie data
SpeakerCode NameIdentity
MAIMaisie DeaTeacher
*There is no positive identification of speaker XYZ’s name, however the speaker may be “Sui-Wai”.


Utterances were coded using a system of conversational acts (C-acts) developed by Dore (1977). C-acts represent a taxonomy of speech act types that code utterances according to (1) the grammatical structure of the utterance, (2) its illocutionary properties, and (3) its general semantic or propositional content. Because of the different nature and focus of the present research, some modifications were made in the system as used in previous studies. These included both the addition and deletion of certain codes.

Forty-nine separate speech acts, each assigned a three-letter code, constitute the con-versational act system. These are grouped into six broad function types: (1) Assertions, which solicit information or actions; (2) Organizational Devices, which control personal contact and conversational flow; (3) Performatives, which accomplish acts by being said; (4) Requests, which solicit information or actions; and (6) Responses, which supply solic-ited information or acknowledge remarks (Dore et al., 1978, pp. 372-3). An additional category of special speech acts that codes activities such as microphone talk, laughing, and singing is also included. Conversational acts serving the Request function, for example, include Requests for Action (QAC), Product Requests (QPR), and Requests for Permission (QPM).

Coding proceeded as follows. First, the grammatical form and its literal semantic meaning were determined. Then a judgment was made as to the conventional force, or purpose, of the utterance. In this step, sequencing, reference, and other conversational cues, such as marked illocutionary devices and intonation, were taken into consideration. Utterances were thus placed first within the six broad function types, and then categorized as an individual conversational act. Throughout the coding, the contextual information contained in field notes provided an addition check for the validity.

Initial coding was conducted by the data collector who observed a particular lesson. To ensure inter-coder agreement, each taped session was then coded a second time by another member of the research team, all of whom had engaged in two weeks of training and practice. Discrepancies were resolved through discussion. Throughout the coding process, inter-coder agreement for individual lessons ranged from 0.90 to 0.96. It should be noted that conversational act coding has been shown to be highly reliable in other studies as well, with inter-coder reliability approaching 0.90.

Although utterances in Chinese were translated into English and entered as data, all coding was done on the original Chinese. In several instances, this procedure proved to be crucial, because the English translation would have received a different code.

Cicourel has compared three prevalent models of discourse: the speech act model, the expansion model, and the problem-solving model. His conclusion was that any one of these models in isolation is inadequate; some sort of integration is required. The method used in the present study represents an attempt at such an integration. By including both quantitative and qualitative analyses, the speech-act and expansion models were to some extent combined. This integration also helped to meet some of the criticisms leveled by Cicourel against the speech-act approach. Cicourel faulted the speech act model because it cannot easily account for 1) organizational features of interaction; 2) participant’s strategies, such as plans for elaboration; 3) the situated nature of discourse, such as situated meaning and context; and 4) the multiple functions of utterances. The present study overcomes these weaknesses by incorporating the following methodologies:

The use of conversational acts rather than other coding systems contributed to a miti-gation of some of the other weaknesses Cicourel identified in speech-act analyses. First, because conversational acts are sensitive to grammatical form, semantic content, and illocutionary force, and not just one of these, they provide a link between form and func-tion. As Cole, Dore, Hall, and Dowley (1978) pointed out, conversational acts mediate between the grammatical and the social, between the “grammatical forms and the interactional purpose for which they are used” (p.74). In other words, they integrate speakers’ interests and purposes.

Definitions and examples of conversational acts

Assertive Devices Organizational Devices Performatives Responsives Special Speech Acts