CHILDES English Howe Corpus

Christine Howe Education Strathclyde University cjh82@cam.ac.uk website

Browsable transcripts

Download transcripts

Citation information

Howe, C. (1981). Acquiring language in a conversational context. New York: Academic Press.

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

Project Description

This directory contains transcripts from 16 of the 24 Scottish mother–child pairs ob-served by Christine Howe while playing with toys in their homes in Glasgow. Each pair was recorded twice and the recordings are 40 minutes in duration. The data was coded extensively for actions and situations. The children are aged 1;6 to 1;8 at the first recording and 1;11 to 2;1 at the second recording. There are two files per participant, thus the corpus contains 32 data files.

The procedure for participant recruitment was to place a notice in the local newspaper. The most straightforward method of obtaining a reasonable sample of mothers and children would have been random or quota sampling from a local authority list. Unfortunately, the local authority in question refused to cooperate and more indirect methods had to be used. As a start, an article was written for the local newspaper explaining the aims of the study in deliberately vague terms and asking mothers to volunteer children in the age range of 15 to 18 months. Notices making similar requests were posted in likely public places, including doctors’ waiting rooms, baby clinics, university common rooms and centers for further education. A social worker persuaded one of her clients to take part. Finally, a month after recruitment had started, health visitors from two of the baby clinics made contact with offers of help. One suggested sitting in on an afternoon session and asking attending mothers to participate. This was done. The other offered names and addresses of every mother with a child of the right age in her area. The first eight in the alphabetical list were contacted and six said they were interested in taking part. By this time, some of the first volunteers had marshaled their friends into participating, and at the end of the period available for sampling, 33 mothers had volunteered their children. Two mothers were considered unsuitable because they had delegated childcare to a grandmother and an employed nanny. The remaining mothers had volunteered children in the age range of 13 to 21 months. It seemed sensible to choose the mothers with the 24 children nearest in age to the mean of 17 months and use others for pilot work.

Most of the final group lived in a small university town or nearby villages. Despite the fact that sampling was anything but random, the group was represented well by gender, birth order, and social class of family. Twelve children were boys and 12 were girls. Seven were only children, seven were the youngest of two, four were the oldest of two, four were the youngest of three, and two were twins without other siblings. The fathers of 13 children had professional or managerial occupations, whereas the fathers of the remaining 11 children had skilled or semiskilled manual occupations. The first group was designated “middle class” and the second group “working class,” The following table shows the sex, birth order, social class, and recruitment method for the children (all identified by pseudonyms).

Table 1: Howe children

Name	Sex	Birth Order	Social Class	Method
Barry	Male	3rd of 3	Working	Clinic
Eileen	Female	Only	Working	Clinic
Faye	Female	1st of 2	Middle	Other Mother
Graham	Male	1st of 2	Middle	Notice
Ian	Male	2nd of 2	Working	Article
Jason	Male	1st of 2	Middle	Notice
Kevin	Male	2nd of 2	Middle	Article
Lucy	Female	2nd of 2	Middle	Other Mother
Melanie	Female	1st of 2	Working	Clinic
Nicola	Female	3rd of 3	Working	Clinic
Oliver	Male	Only	Middle	Other Mother
Philip	Male	2nd of 2	Working	Article
Richard	Male	2nd of 2	Middle	Article
Sally	Female	Only	Middle	Article
Wayne	Male	Only	Working	Article
Yvonne	Female	Only	Working	Social Worker

Each session was videotaped in the home. The first 20 minutes consisted of a play session with the children’s own toys and the second 20 minutes consisted of a play session with a special set of toys presented in the following order:

1. jigsaw puzzle,
2. plastic postbox with holes in the top for geometric shapes,
3. plastic doll with clothes, teaset, cot, and brush,
4. lorry,
5. jeep and horsebox,
6. model zoo animals with fences, cages, and keeper,
7. interchangeable heads, arms, and legs which could be assembled into postmen, firemen, and policemen,
8. cardboard building blocks with pictures on every face,
9. fluffy puppet, and
10. picture story.

Once recording sessions were completed, the tapes were immediately transcribed. The videotapes were played back in the order of recording on a video taperecorder connected to a television screen and an audio taperecorder. The transcriber sat 6 feet from the screen holding an electrically powered pad that moved paper across a frame at the rate of 6 inches per minute. The rolls of paper used with this device had lines at 1/4-inch intervals. Thus, it was possible to know within 2 1/2 seconds when any mark on the paper was made, and 2 1/2 seconds became the basic time interval for analysis.

The moving paper device, the audio taperecorder set to record and the video taperecorder set to playback were started in that order. Watching only the child, the transcriber noted changes of action and object using a short-hand code. The code essentially used hieroglyphics to represent actions, including gaze, and the first two letters of names to represent the objects of actions. Every time the child vocalized, a dash was drawn on the left of the paper to be filled in later. Every vocalization was being re-recorded on audiotape. The second tape was replayed and the child’s behavior transcribed in the same way. Then the audio taperecorder was switched off and the whole procedure repeated for the mother. The transcription of all other participants required a third run.

The next stage was transcribing the vocalizations and inserting them in the behavioral record. English words were transcribed as English words and other sounds were transcribed with some attempt to represent them using English syllables. Speakers varied in their intelligibility and the tapes varied in the amount of background noise. The mean percent of intelligible utterances was 94% for the children and 98% for the mothers. Reliability of transcription and coding completed a year later, resulted in 91% agreement for transcription and 83% agreement for the nonvocal behavior and speech in every 2.5 second period where speech occurred. Mean Length of Utterances and Type–Token Ratios for each recording session are presented below:

Table 2: Howe MLU and TTR at 1st and 2nd Recordings

Name	1st MLU	1st TTR	2nd MLU	2nd TTR
Barry	1.30	0.27	2.56	0.20
Eileen	1.47	0.18	1.57	0.38
Faye	1.27	0.17	1.65	0.23
Graham	1.09	0.26	1.32	0.51
Ian	1.53	0.38	2.04	0.41
Jason	1.00	0.17	1.17	0.40
Kevin	1.16	0.45	1.87	0.50
Lucy	1.09	0.17	1.15	0.23
Melanie	1.19	0.43	1.55	0.41
Nicola	1.21	0.21	1.67	0.55
Oliver	1.53	0.23	2.04	0.23
Philip	1.22	0.17	1.54	0.26
Richard	1.30	0.16	2.13	0.19
Sally	1.33	0.39	1.98	0.44
Wayne	1.33	0.18	1.39	0.26
Yvonne	1.72	0.36	1.69	0.37