Recurrence quantification analysis of sentence-level speech kinematics

Eric S. Jackson, Mark Tiede, Michael A. Riley, D. H. Whalen

Research output: Contribution to journalComment/debatepeer-review


Purpose: Current approaches to assessing sentence-level indices were calculated: percent recurrence (%REC), speech variability rely on measures that quantify variability percent determinism (%DET), stability (MAXLINE), and across utterances and use normalization procedures stationarity (TREND). that alter raw trajectory data. The current work tests Results: Percent determinism (%DET) decreased only the feasibility of a less restrictive nonlinear approach— for the most linguistically complex sentence; MAXLINE recurrence quantification analysis (RQA)—via a procedural decreased as a function of linguistic complexity but example and subsequent analysis of kinematic data. increased for the longer-only sentence; TREND decreased Method: To test the feasibility of RQA, lip aperture (i.e., as a function of both length and linguistic complexity. the Euclidean distance between lip-tracking sensors) was Conclusions: This research note demonstrates the feasibility recorded for 21 typically developing adult speakers during of using RQA as a tool to compare speech variability across production of a simple utterance. The utterance was speakers and groups. RQA offers promise as a technique produced in isolation and in carrier structures differing to assess effects of potential stressors (e.g., linguistic or just in length or in length and complexity. Four RQA cognitive factors) on the speech production system.

Original languageEnglish (US)
Pages (from-to)1315-1326
Number of pages12
JournalJournal of Speech, Language, and Hearing Research
Issue number6
StatePublished - 2016

ASJC Scopus subject areas

  • Language and Linguistics
  • Linguistics and Language
  • Speech and Hearing


Dive into the research topics of 'Recurrence quantification analysis of sentence-level speech kinematics'. Together they form a unique fingerprint.

Cite this