CHAT and CLAN for language sample analysis

What does getting started involve?

CLAN is available as a desktop application for Windows and Mac. You work in a transcript editor and run analyses through CLAN’s Commands window, using typed commands or the available buttons and dialogs.

Getting started involves preparing transcripts in CHAT format, setting up the language resources for grammatical analysis, and learning the commands and options for the measures you want. The clinical guide illustrates that workflow; the CLAN manual provides setup and command details.

What is different about a CHAT transcript?

CHAT places the spoken words on speaker lines and grammatical annotations on separate lines beneath them, called dependent tiers. This differs from SALT’s convention of marking bound morphemes within the transcribed words. In a CHAT grammatical-analysis workflow, you enter ordinary word forms and the annotation software supplies a separate representation of their morphology and grammatical relationships.

Two annotation lines you will encounter are:

• %mor: information about word classes and morphemes.
• %gra: information about grammatical relationships, such as which word functions as a subject.

You still mark repetitions, revisions, unintelligible speech, and relevant errors using CHAT notation. You also supply file headers identifying the speakers and language. These conventions tell the software what to analyze and how to treat particular productions. The CHAT manual defines the notation.

What analyses can I run?

You can run individual analyses, such as word-frequency counts or MLU, or use KIDEVAL, which runs a set of child-language analyses together and returns a spreadsheet of results. Its outputs include utterance and word measures, grammatical-morpheme information, and, where supported, DSS and IPSyn scores.

KIDEVAL requires at least 50 utterances. Some included measures need a particular number of qualifying sentences before a score can be calculated. The KIDEVAL section of the CLAN manual lists the measures and their requirements.

What do I need to check in the output?

For grammatical analyses, review both the annotations and the utterances used in the calculation:

Finestack et al. (2020) discuss reviewing automated grammatical analyses and illustrate how to interpret the resulting profiles.

If you use CLAN’s reference-comparison options, also examine the description of the comparison group: its ages, language, and elicitation context need to be relevant to your sample. Those options use language-sample collections and published reference values; CLAN does not supply one standardized normative assessment. The clinical guide explains the comparison workflow and its scope.

How can DSS and IPSyn help identify goal areas?

DSS and IPSyn both examine grammar. DSS scores grammatical forms within qualifying sentences; IPSyn records which grammatical structures appear across the sample. Their profiles organize that information into areas you can investigate for target selection:

The clinical cases in Finestack et al. (2020) illustrate how these analyses can inform further probing and target selection.

Useful CHAT/CLAN materials

• CLAN download and introduction: the software and links to its supporting documentation.
• A Clinician’s Complete Guide to CLAN and PRAAT: illustrated clinical workflows, transcript preparation, and analysis guidance.
• Assessment and Therapy Goal Planning Using Free Computerized Language Analysis Software, Bernstein Ratner & MacWhinney (2023): a hypothetical preschool example covering preparation, analysis, and clinical application.
• Using Computerized Language Analysis to Evaluate Grammatical Skills, Finestack et al. (2020): clinical cases showing how grammatical analyses add information beyond MLU.
• CHAT manual and CLAN manual: current references for transcription conventions, configuration, commands, and analysis requirements. Use these to check software details described in older tutorials.