ASDBank AAC Minimally Speaking Autism Corpus


Aparna Nadig
School of Communication Sciences and Disorders
McGill University

website
Angela MacDonald-Prégent
School of Communication Sciences and Disorders
McGill University
Participants: 18
Type of Study: naturalistic
Location: Canada
Media type: video
DOI: doi:10.21415/JAJE-QP45

Browsable transcripts

Download English transcripts

Media folder

Citation information

Publications using this data should cite:

MacDonald-Prégent, A., & Nadig, A. (2024). Augmentative and Alternative Communication Intervention Corpus: Minimally Speaking Autistic Children.

Warnings

Warnings concerning the use of this data include:
  1. Limited attention was paid to the correct transcription of speech errors.
  2. Limited attention was paid to overlapping utterances and hesitations.
  3. Limited attention was paid to the consistent transcription of word approximations.
  4. All child vocal emissions are transcribed. A vocal emission is defined as the target child using their voice to emit ANY sound, vocalization, word approximation, word or full communicative utterance. This may result in incorrect parsing on the %mor and %gra tiers.
  5. This study includes the coding of non-spoken modalities of communication (i.e., gesture and AAC use), which are included in our transcripts on their own individual tiers.

Restrictions and Pseudonyms

We set corpus specific access on the use of this data. Access is to be restricted to faculty, SLPs, or postdocs. Should students intend to use this corpus, they must rely on their faculty advisors for access.

Families provided informed consent to participate in the research study (n = 18). This included being advised that de-identified data (e.g., containing no personal or identifying information) may be shared in a data repository. Therefore, first names in transcripts were replaced by placeholders (e.g. CHILDNAME, SIBNAME) to preserve anonymity. Consent for video recordings to be shared on online data sharing repositories was optional and was obtained from a subset of the families (n = 14), whose video is provided along with transcripts.

Project Description

Data collection

The purpose of this project was to teach minimally speaking autistic children the use of augmentative and alternative communication (AAC) methods through a caregiver-mediated AAC intervention study. The study was approximately 16 weeks in duration and included pre-intervention (two 5 minute-samples), mid-intervention (one 5-minute sample), and post-intervention (two 5-minute samples) data collection. All participation was remote; videos were obtained by research staff via a video-conferencing platform, or by upload of videos collected by parents on their mobile device or laptop. The language samples that comprise this corpus were collected at all three time points. Language samples were transcribed from videos of five-minute interactions during which the caregiver and child participated in an activity of the caregiver’s choosing. AAC devices were introduced through the intervention and were incorporated during the mid and post-intervention language samples.

Sampling procedure

This study was conducted remotely from the city of Montreal, Canada, with a culturally and linguistically diverse sample of families residing in the Montreal area or in other Canadian provinces. With the exception of one child who was on the waitlist for an autism evaluation, all had a confirmed diagnosis of Autism Spectrum Disorder (American Psychiatric Association, 2013). Caregivers reported English to be the dominant language for 15 of the 16 children with one child reported to be equally proficient in English and Romanian. The intervention was conducted in English; caregivers were able to communicate in English. To be eligible to participate in the intervention study, children needed to meet our definition of minimally speaking, being over 30 months and having inconsistent phrase speech, the definition used in the Autism Diagnostic Observation Schedule, 2nd edition manual to assign children to the lowest language level/module 1 (ADOS-2 manual; Lord et al., 2012, pg. 9-13). This was confirmed by Angela MacDonald-Prégent via review of caregiver-child interaction videos obtained pre-intervention.

Transcription Protocol

Two English-speaking research assistants who were undergraduates in Linguistics transcribed the interactions using conventions outlined by the Codes for the Human Analysis of Transcripts (CHAT) transcription system (MacWhinney, 2023). Transcribers were naïve to study hypotheses and time point of data collection. Each video was viewed independently by the transcribers. The main transcriber completed the transcription, which was then reviewed by the second transcriber; both transcribers served as either the main or second transcriber for different files. The second transcriber reviewed the video for utterance breaks, adherence to CHAT conventions, and the possibility of deciphering unintelligible utterances. Any major discrepancies with the original transcription were flagged for discussion. Transcribers worked in a quiet room and wore headphones while transcribing the recorded interactions using ELAN, a software for annotating and transcribing video or audio recordings (https://archive.mpi.nl/tla/elan). The full duration of the interactions were transcribed. When applicable, utterances were determined by either a clear pause followed by a breath, or clear suprasegmental features such as the rising intonation of a question. All other utterances were determined by a clear pause followed by a breath. If an utterance was difficult to comprehend, transcribers were told to treat the utterance as unintelligible, and the portion or full utterance was marked with an “xxx”. If there was an overlap between speakers, the transcriber was to first pay attention to the parent utterances, then the child utterances.

Project-specific conventions

The child’s age reflects his/her age at the time of the interaction. Where available, parent education levels for caregivers are provided in the speaker tier. These reflect the highest degree received and caregivers fell into one of three categories: below university, university, above university.

Codes

The codes used are described in this file.

Biographical Data

Biographical data for the child include: age, gender, 1st language, number of languages the child is exposed to, first nations membership, visible minority status. Biographical data for the parent include: gender, education, immigration status, province of residence, first nations membership, visible minority status, 1st language and number of languages spoken. Eight of the 18 children were from visible minority groups. Please see this file for details. This file also includes spoken language production data for child and parent at the beginning of the intervention study (T1).

Acknowledgements

We would like to sincerely thank all the families who participated in this study. We would also like to thank 1) the Centre for Research on Brain, Language, and Music for funding and 2) the Miriam Foundation for providing a portion of the materials used to create the AAC communication boards used in this intervention study.