Acoustic and Temporal Analysis for Assessing Speaking

Abstract

Objectively measurable features of a speaker's speech are certainly relevant to true score variance in assessing speaking skills. Ideally, speech proficiency scores should be more a function of the comprehensibility of speakers' pronunciation and the degree to which speakers' discourse is responsive to the communicative demands of the tasks. Speech science has made progress toward identifying acoustic features of pronunciation that affect comprehensibility. It has become common for elements of speech accent to be detected by instrument and computer‐assisted acoustical analysis (e.g., Computer Speech Laboratory [CSL] or Praat), which characterizes different accents by examining patterns of fundamental frequencies or F 0 formants. Until somewhat recently, research examining characteristics of second language (L2) speech production had been concerned with phonemic segmental phenomena, that is, the “accuracy” of non‐native speaker (NNS) consonant and vowel formation . Accordingly, acoustic characteristics of those productions had mainly focused on single parameters of English: vowel duration differences in voicing contrast, acoustic vowel spaces, and voice onset time of stop consonants. Currently, however, the consensus is moving toward an appreciation of the role that differences in speaking rates, intonation patterns, and other prosodic features (suprasegmentals) may play in intelligibility and listeners' assessments. Non‐native temporal (e.g., pause structures) and acoustic patterns (e.g., tone choices) account for native listeners' perception of L2 English learners' speech as “accented.” Acoustic parameters combining rate, pauses, and intonation of speech have been also conjointly investigated for assessing NNS' oral proficiency. This chapter starts with a brief background to acoustic analysis in applied linguistics, introduces various speech recognition and acoustic parameters used in assessing speaking, and illustrates application processes and examples of those analyses. It explores new conceptualizations of acoustic measures in speaking assessment and discusses challenges of this approach and future directions.