SALT for language sample analysis

What should I check before collecting a sample for a normative comparison?

SALT’s normative data come from samples collected using specified materials, prompts, and procedures. Check the database descriptions and collection protocols before recording. They identify the ages and language populations represented, how the samples were elicited, and the analyses available for comparison.

For example, SALT’s Narrative Story Retell database includes samples elicited with designated books. The protocol describes how the child hears the story before retelling it. Asking a child to generate a story from a different picture book changes both the material and the elicitation task; that sample does not reproduce the conditions underlying those norms.

The descriptions also identify which measures have comparison data. In that story-retell database, transcripts were scored for clause complexity and narrative structure, but not for the optional comprehension questions. Being able to score an additional feature does not necessarily mean that the same normative data include it.

What can I do with SALT without a normative comparison?

You can use SALT to analyze a conversation, narrative, or other sample collected for your own assessment question. Its reports can summarize utterance length, vocabulary, grammatical errors, mazes, and pauses, and list the utterances or words behind those summaries. Narrative and clause-level scoring can also be used without a normative comparison.

For example, a story elicited with a book outside SALT’s normative collection can still be examined for grammatical errors and narrative organization. Those analyses describe the child’s performance on that task; they do not provide a normative score for it. The SALT variables guide describes the available measures and follow-up reports.

How much language should I collect?

For open-ended conversation, SALT’s sample-length guidance suggests about five minutes or 40–50 utterances. For a story retell or expository task, collect the complete response rather than stopping when the child reaches a particular utterance count. Follow any additional requirements in the protocol you are using.

The recorded sample and the portion used for a particular comparison may differ. SALT allows samples to be equated by length for relevant comparisons; that is an analysis decision, rather than a reason to interrupt an unfinished story.

What SALT-specific notation do I need in the transcript?

You record the speaker turns and add SALT’s markers for morphemes, mazes, errors, and timing. These entries supply information used in the analyses. The Standard SALT Conventions defines the notation and provides examples.

C-unit segmentation. A communication unit, or C-unit, consists of a main clause with its dependent clauses. You separate independent clauses into different units while keeping a dependent clause with its main clause. For example, “We stayed inside because it rained” forms one C-unit; “It rained and we stayed inside” contains two. SALT also specifies how to handle fragments and other incomplete productions.

Morphemes, mazes, and errors. You mark bound morphemes, enclose mazes in parentheses, and enter error or omission codes. SALT uses those entries to count morphemes, separate maze words from the main utterance, and summarize marked errors.

Speaker turns and timing. Speaker labels identify who said each utterance. You use the recording to enter elapsed-time information and pause durations. SALT uses that information for speaking-rate, pause, and interaction measures; those reports require more than the child’s words alone.

Some collection protocols modify the segmentation rules. The bilingual Spanish/English story-retell description specifies modified C-units to accommodate Spanish subject omission and applies the modification to both languages.

What do I code, and what does SALT calculate?

SALT calculates results from the transcript and the linguistic information entered with it. The preparation differs by analysis:

For SI, you identify and code clauses. For NSS, you rate components such as referencing, cohesion, and conflict/resolution. The software summarizes those judgments; it does not assign them simply from the transcript’s wording. The SALT variables guide identifies the inputs required for the reported measures.

What will the normative comparison tell me?

SALT’s comparison reports place the child’s measures alongside the results for the relevant group, including group means and standard deviations. The comparison can be defined by factors such as age or grade and, where appropriate, sample length. Those choices determine the group behind the reported values.

The SALT tutorial by Heilmann and Miller includes annotated reports showing this relationship between the child’s results and the normative comparison. It also illustrates how individual measures contribute to a broader description of performance.

A comparison report can identify where performance differs from the group. The transcript and more detailed analyses show what contributes to that difference—for example, which grammatical forms were omitted or which narrative components received lower scores.

How can SALT help identify treatment or IEP goals?

SALT’s detailed analyses let you examine patterns within the broader measures:

• Grammatical forms: error and omission summaries can reveal recurring difficulties with particular forms, such as tense markers or pronouns. Reviewing the associated utterances helps specify the grammatical target rather than relying on a total error percentage.
• Clause structure: Subordination Index coding identifies clauses within C-units. Examining the coded utterances can help identify which complex sentence structures the child uses and which warrant further probing.
• Narrative organization: the Narrative Scoring Scheme examines components including character development, referencing, conflict/resolution, and cohesion. Difficulty making character references clear suggests a different target area from difficulty connecting events or explaining how a problem is resolved.

Useful SALT materials

• Database descriptions and protocols: the population, elicitation requirements, and coding associated with each set of normative data.
• Standard SALT Conventions: the transcription rules, with examples of segmentation, morpheme marking, mazes, and errors.
• Guide to SALT Variables: definitions of reported measures and the more detailed analyses available to investigate them.
• Coding aids: scoring guidance for SI, NSS, and other analyses that require additional coding.
• Systematic Analysis of Language Transcripts Solutions: A Tutorial, Heilmann & Miller (2023): clinical cases, annotated reports, and supporting materials. An author manuscript is available through the University of Wisconsin repository.
• Case-study practice: transcripts and clinical cases for practicing coding, analysis, and interpretation.
• Software comparison and purchasing information: current license options and included features.