CHILDES English Korman Corpus

Myron Korman (1944-2008)

Browsable transcripts

Download transcripts

Link to media folder

Citation information

Project Description

This subdirectory contains the speech of British mothers to infants during the infant’s first year. These data are from the Myron Korman’s doctoral dissertation. The project fo-cused on maternal speech interactions with preverbal infants. The data were collected in Britain from middle-class mothers with their first children. The children ranged in age from 6 weeks at the outset to 16 weeks at the end of the project.

These data focus on the language of mothers of infant children. They are useful for understanding the input to the infant, but not for studying the child’s vocalizations, because these vocalizations are quite primitive and there is no attempt to capture the vocalizations in phonetic detail. Rather the focus is on the functions and pause characteristics of the maternal input.

The data were reformatted into CHAT in 1992. There are five files from each of the six mothers. The ages of the child for these five files are 6, 7, 11, 15, and 16 weeks. The main tier has utterances that are marked either as ATT for “attentional” or TUR for “turn-constructional.” The codes for the %spa tier are given in the 0funct.cdc file. In addition, the codes $t and $c indicate repetitions of content or temporal structure. The recordings at 7, 11, and 15 weeks have a %tim tier that contains two types of timing information. The first number gives the length of phrase in seconds; the second number gives the length of the pause after the utterance in seconds. When the mother is not talking and the tape recorder goes on and off, an @New Episode marker is inserted.

Participants

Participants were six primiparous mothers from the greater Nottingham community. They had been contacted through introductions and referrals made by health visitors serv-ing health centers local to their area. Mothers were told that the experimenter was interested in the development of their infant’s vocalization over time.

Every effort was made to acquire a homogeneous but balanced study sample. All mothers were in their 20s and all were middle class. All (save one who had been completing a first degree) had worked prior to marriage and pregnancy. Three were breast feeders and three used a bottle. The infants were three boys and three girls. The following table provides an account of various vital statistics of each family in the study.  

Table 1: Korman Mother Characteristics

	La.	Gl.	Cr.	Hi.	St.	Gi.
Mother’s Age	27	26	23	25	27	25
Occupation	Teacher	Clerical	Student	Nurse	Teacher	Chemist
Feeding	Breast	Breast	Bottle	Bottle	Bottle	Breast
Husband	Police	Chemist	Clerk	Plumber	Photog	Chemist
Infant’s Sex	Female	Female	Female	Male	Male	Male

Data Collection

Recordings were made of each infant’s auditory experience in the home for the whole of 24-hour periods. These recordings were made without an experimenter present and in-cluded a continuous record of the whole of all mothers throughout the day. Video record-ings were made of the mother in the home. Videotapings were limited to three monthly sessions nearest to the end, the midpoint, and the beginning of the period under investiga-tion (at 7, 11, and 15 weeks). Audio records were made at fortnightly intervals beginning at 6 and ending at 16 weeks. The recording apparatus consisted of a Revox 4000 reel-to-reel tape recorder, a voice key, a small lavaliere microphone, and 50 feet of thin wire.

Audio Sessions

Audio records were made at weeks 6, 8, 10, 12, 14, and 16. The sample was split into two groups of three infants each, and the groups were recorded on alternate weeks. It was explained that the study concerned the development of the infant’s vocalization over time. Mothers were shown the operation of the device and asked to keep the microphone pinned near to the infant at all times. The apparatus was usually dropped off in mid-morning and picked up the following day at approximately the same time. In each household, a hiding place was found and the equipment, except for the wire and microphone, was always out of view. All households were on two floors and in each instance the length of wire was checked to see if it allowed access to all floors and rooms before recording began.

Video Sessions

Videotaping took place in the participants’ homes in a room of their own choosing (usually a front or “best” room). Sessions were held at a time of day that each mother had indicated their infant would be most alert and active. Most sessions took place around midday. Mothers were asked to “play with the baby as you would normally do” and no other instructions were given or restrictions imposed. There was no attempt to “standardize” the location of taping sessions or the positioning of mother and infant. Mothers were never discouraged from stopping to chat with the experimenter and short breaks in the play activity were common. Each mother was simply asked to play with the baby in their normal manner wherever, whenever, and however it suited them to do so. This is not to say that the purpose of these sessions was not clear to each mother (i.e., to play with the baby), but to point up the fact that control of the activity in each session was left as much as possible in the hands of the mothers. The sessions themselves were intended to be as relaxed and nonrestrictive as possible so as not to constrict or inhibit the mothers’ natural responsiveness under the circumstances. Three toys were offered to provide an interactive alternative (a monkey hand puppet, a rattle, and a pop-up toy), but mothers were not encouraged to use them. In each session mothers could use their infants’ own toys, which were usually close at hand. The play sessions were, for the most part, friendly social visits during which the mother was asked to play or interact with her infant while a video record was made. A typical session might consist of the experimenter and mother first having coffee and a chat, the mother preparing the area and the baby for play, the recorded play sessions themselves, which stopped at the mothers’ convenience or the infant’s continuance, and then perhaps another cup of coffee and a further casual discussion of the activity just finished, the infant’s responsivity and growth generally, or the cost of coffee at Sainsbury’s.

Pause Analysis

A pause was defined as “any maternal silence of longer than 300 ms which ended at: (a) another maternal utterance, (b) an interruption from the experimenter, or (c) an intrud-ing and other than vocal interactive sound produced by the mother, such as rattle shaking or tapping noises.” As a consequence, this procedure resulted in the scoring of certain maternal pauses whose duration, although long, was nonetheless populated with some form of interactive behavior from one or the other of the partners in any exchange. Those especially long pauses that are scored will be made up of: a maternal vocal silence in anticipation of, or in response to, some infant behavior or some maternal activity that was done in complete silence.

Timing Procedures

The following timing procedures were used in analysis of the eighteen video observa-tions of the present study. Videotaped interactions were copied onto audiotape cassettes and timings were then undertaken with a hand held stop watch and a tape recorder. Maternal vocal behavior was timed from the perceived end of a phrase to the end of the next sequential phrase. Pauses were then timed from the end of the same first phrase to the onset of the second phrase and the durations of each sequential phrase and pause determined. Each set of timings was undertaken for a minimum of three trials until the experimenter was familiar with the sequence and satisfied with his result. The procedure was designed to familiarize the experimenter with the rhythmic and temporal patterning of the phrase/pause sequence about to be timed, and to help reduce the amount of time that might have been lost through guessing at the onset of a sequential vocalization. Average adult serial reaction time to auditory stimuli at irregular intervals is 335 ms to intervals of 500 ms. All transcription and timing was later checked against the videotapes themselves to assess contextual accuracy.

Reliability

Three raters (undergraduates) were asked to transcribe 100 sequential vocalizations of the same mother selected at random from available audio observations. That transcription resulted in a 97% agreement with the content and segmentation of the original transcript produced by the experimenter.

Six 30-second exchanges were selected at random and the mothers’ speech segmented and timed by the research assistant (an average of 21 phrases and pauses per mother; 126 in all). The resulting segmentation and the durations of phrases and pauses were then compared with measures of the experimenter. A comparison of segmentation resulted in 88.8% agreement. Those instances in which segmentation did not coincide (i.e., instances where two phrases were separated or joined by differing measurement of the same pause criteria) were later discarded from the final comparison of temporal durations (discarded were 14 phrases and 7 pauses of the original transcription). That final temporal analysis resulted in a comparison of 112 phrases and 119 pauses overall. In the measurement of phrase durations, agreement was 95.5% to within 300 ms and 84.8% agreement to within 200 ms.

Table 2: Korman Functional Utterance Types

Code	Function	Code	Function
STQ	tag questions	TC	turn-constructional
Bc	back-channel	Aux	auxiliary utterances
Ex	exhortations	ST	statements
Ph	phatics	Co	commands
Gr	greetings	Wh	wh-questions
Pe	performance	Y-N	yes-no questions
TQ	tag question	RI	rising intonation
PC	post completer

This taxonomy is an attempt to define and describe the uses and functions of mothers’ language as an interactive phenomenon. It is predicated on the assumption that maternal speech in any interaction is not only a mode of expression, but that it also constitutes a form of social organization whose overall structure is largely based in the coordination of engagement relative to the behavior of a prelinguistic infant. Its purpose is to provide a catalog of frequently occurring interactive uses or “functions” in maternal speech. Its goal is to provide insight into the ways a mother will integrate her language in the organization of an engagement — how she will cultivate and eventually achieve a framework of mutual orientation necessary for successful interaction; the vocal alternatives that are open to her in any exchange; and how her individual vocalizations are organized within an engagement to both exchange information and help keep it going.

It will be assumed a priori that maternal language in interaction is organized to both offer communication (whether it be real or imagined) and to assist in the interactive flow of behavior in any prelinguistic engagement. To define and distinguish those communicative vocalizations from those that perform an interactive (and/or regulative) function, a distinction had to be made between those vocal constituents intended as actual communication and those constructions having a “metacommunicative” or supportive role in engagements. In other words, we have to distinguish when a mother is “talking” to her infant using communicative constructions which indicate she is taking a “turn” at speaking and when her vocal behavior is simply something that is used to get or keep the infant’s attention. The terms that will be used to identify these two general types of interactive function are turn-constructional utterances (TC), and auxiliary utterances (Aux).

A turn-constructional utterance is one in which a mother takes a turn at verbally com-municating or “saying” something to her infant. The utterance she makes is intended to convey information, request it, or to direct an infant to a specific activity. In other words, the utterance is used to “construct” a mother’s turn at speaking whose purpose is an ex-change of verbal information in some form. For this group of mothers, this type of mean-ingful utterance was identified in three frequently occurring forms that are fundamentally standard English constructions: statements, commands, and interrogative questions.