CHILDES English Tommerdahl Corpus

Jody Tommerdahl College of Education University of Texas at Arlington joditom@uta.edu website

Browsable transcripts

Download transcripts

Link to media folder

Citation information

Tommerdahl, J. and Kilpatrick, C. (2014). The Reliability of Morphological Analyses in Language Samples. Journal of Language Testing, 31,1, 3-18.

Tommerdahl, J. and Kilpatrick, C. (2013). Analyzing Reliability of Grammatical Production in Spontaneous Samples of Varying Length. Journal of Child Language Teaching and Therapy, 29,2, 171-183.

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

The Participants were recruited from the local community of a large city in central England through a variety of playgroups in a spread of socio-economic areas. Participants were 13 females and 10 males, ranging in age from 2;6–3;6 (mean = 35.6 months). Inclusion criteria were similar to those of Johnson and Tomblin (1975), which required the child to be monolingual and of normal hearing ability. The following additional criteria were gained through parental report:

1. The child had not been referred to speech and language therapy.
2. The parents did not feel that the child began using language later than his or her peers.
3. No one in the immediate family had been suspected of having language or communication difficulties.
4. The parents did not suspect that the child had language or communication difficulties; and
5. The child had no known neurological disorders.

Two spontaneous language samples were recorded for each participant, with recordings taking place no more than one week apart. The sessions were controlled as much as possible (Muskett et al., 2012) to have nearly identical contexts. Recording took place in a Flexible Learning Room fitted with four hidden cameras, five microphones, and a host of media systems controlled from an adjacent gallery. A large selection of toys was available for play, including a model farm, several toy animals, building blocks, vehicles, a tea set, and a large rug with pictures of roads and a local village. In addition, a circus scene was projected onto a whiteboard on the wall. The same toys were available for each session and were placed in a box before each child’s arrival.

For each recording, the child played in the playroom with the same caregiver (a parent or grandparent) and was recorded for approximately 35 minutes. No scripted elicitation took place, but before beginning the recording, caregivers were asked to attempt at some point to initiate a conversation about something that had happened in the recent past in order to encourage the children to use a wider variety of verb tenses than what they might use during the normal course of playing with toys (Crystal, 1982). Otherwise, the caregivers were asked to carry on a normal conversation with the child as they would while playing together at home.

A trained speech and language therapist transcribed each language sample orthographically in its entirety and divided each sample into utterances according to the CHAT Transcription Format (MacWhinney and Snow, 1990). A trained Linguistic graduate student counted each morpheme for appropriate usage. In order to assess the accuracy of the transcriptions and counts, approximately 10% of the samples were transcribed by the first author and compared with the original transcriber’s work to achieve an agreement value of 0.88. Accuracy checks resulted in a score of 94%.

The videos linked to the transcripts play well through QuickTime with the exception of these eight that only play smoothly through VLC: fmw2, jessica1, jessica2, mcc1, mcc2, megan1, megan2, and wjaw2.