Temporal speech parameters

What Speech Gap actually measures: the timing structure of spontaneous speech, rather than its content.

The measurement family

Temporal speech parameters describe how someone speaks over time — speaking rate, articulation rate, how often pauses occur, how long they last, and how speech and silence alternate. They are extracted from a short spontaneous-speech sample and require no linguistic interpretation of what was said.

This is what makes them attractive for repeated clinical measurement: they are low-burden, language-portable in principle, and do not require a clinician to be present or a transcript to be produced.

Cross-language validation

Kálmán et al., Temporal Speech Parameters Detect Mild Cognitive Impairment in Different Languages: Validation and Comparison of the Speech-GAP Test in English and Hungarian (DOI) is the key reference for sensitivity across two languages with different structure — one resource-rich, one less-resourced. This connects directly to Gábor Gosztolya’s broader research programme on the linguistic resource gap.

Complementary linguistic features

The BPD narrative-speech study (Felletár et al., DOI) linked linguistic formulation to silent and filled pauses within the BPD group. That result is what motivates combining temporal parameters with linguistic features:

  • self-reference (first-person singular use)
  • content-word density
  • pronoun use
  • syntactic complexity
  • task context

This combination is specific to the BPD case and is a research direction, not an implemented feature set. → Borderline personality disorder

The pause taxonomy problem

Current Speech Gap software collapses several distinct events into “silence.” The planned extension separates them:

EventCurrentlyPlanned
Silencesilencesilence
Inhalationtreated as silencedistinguished
Exhalationtreated as silencedistinguished
Filled pausepartly handleddistinguished
Speechspeechspeech
Uncertain interval—explicitly labelled

→ Respiratory signal extension

Speech Gap · Respiratory signal extension · Longitudinal within-person measurement · Gábor Gosztolya · Borderline personality disorder