T-units / C-units

T-units and C-units

T-units and C-units are the two competing ways of segmenting a transcript into analyzable chunks for syntactic analysis.

What T-units / C-units measures

A T-unit ("minimal terminable unit") was defined by Hunt in 1965 as one independent clause plus any subordinate clauses attached to it — the smallest stretch of language that can be terminated as a grammatical sentence. A C-unit ("communication unit") was defined by Loban in 1976 as an independent clause with its modifiers that can also stand as a complete conversational turn, even if it is not a full grammatical sentence. The choice between them is methodological: T-units are stricter and better suited to written or academic-register samples, C-units are more permissive and capture the realities of conversational speech.

Formula

Count T-units or C-units by applying Hunt (1965) or Loban (1976) rules to the transcript

Normative ranges and benchmarks

  • For conversational samples in typically developing school-age children, C-units and T-units usually agree within 5%
  • For highly elliptical conversation (yes/no answers, single-phrase replies), C-units include utterances T-units exclude
  • For narrative retells and expository monologues, T-units and C-units converge nearly perfectly
  • The choice of unit must be reported in the methods section — it changes every downstream metric
  • Subordination index and clause density should always use the same unit they were originally normed on

Normative bands are central estimates drawn from the cited literature. Individual variation is wide — always cross-reference against the source paper and your assessment's own manual before quoting a cut-score in a report.

Clinical use

Most school-based clinicians report C-units because they work on conversational samples, and C-units do not punish a child for producing a grammatically incomplete but pragmatically adequate turn. Research reports more often use T-units because they are stricter and more reproducible. The critical clinical rule is never to mix them in a single report — if you compute MLU-M on T-units but look up norms that are based on C-units, the difference is large enough to move a child across a clinical cut-score. Document your segmentation choice and stay consistent across the child's longitudinal record.

The fastest way to invalidate a language sample is to mix T-units on Monday and C-units on Friday. Pick one at the start of the year and write it into every report template you touch.
Pick one and never mix

References

  1. Hunt, K. W. (1965). Grammatical Structures Written at Three Grade Levels. NCTE.
  2. Loban, W. (1976). Language development: Kindergarten through grade twelve. NCTE.
  3. Scott, C. M., & Stokes, S. L. (1995). Measures of syntax in school-age children and adolescents. LSHSS, 26(4), 309–319.