Support for context effects on segmentation and segments depends on the context

Abstract

Listeners must adapt to differences in speech rate across talkers and situations. Speech rate adaptation effects are strong for adjacent syllables (i.e., proximal syllables). For studies that have assessed adaptation effects on speech rate information more than one syllable removed from a point of ambiguity in speech (i.e., distal syllables), the difference in strength between different types of ambiguity is stark. Studies of word segmentation have shown large shifts in perception as a result of distal rate manipulations, while studies of segmental perception have shown only weak, or even nonexistent, effects. However, no study has standardized methods and materials to study context effects for both types of ambiguity simultaneously. Here, a set of sentences was created that differed as minimally as possible except for whether the sentences were ambiguous to the voicing of a consonant or ambiguous to the location of a word boundary. The sentences were then rate-modified to slow down the distal context speech rate to various extents, dependent on three different definitions of distal context that were adapted from previous experiments, along with a manipulation of proximal context to assess whether proximal effects were comparable across ambiguity types. The results indicate that the definition of distal influenced the extent of distal rate effects strongly for both segments and segmentation. They also establish the presence of distal rate effects on word-final segments for the first time. These results were replicated, with some caveats regarding the perception of individual segments, in an Internet-based sample recruited from Mechanical Turk.

Not just a function of function words: Distal speech rate influences perception of prosodically weak syllables

“…Historically, distal rate effects on phoneme-level perception, e.g., for consonants like /g/ vs. /k/ differing in voice-onset time, have been recognized to be fairly fragile, with generally small effects on phonetic boundary shifts (Kidd, 1989;Summerfield, 1981;Wade & Holt, 2005). Recent studies have generalized knowledge of distal rate effects on speech perception by considering a wider array of phoneme-level perceptual ambiguities, both ambiguities in type of phoneme, e.g., / / versus / :/ in Dutch (Bosker, 2017), and number of phonemes, e.g., Canadian oats versus Canadian notes (Heffner et al, 2017;Reinisch, Jesse, & McQueen, 2011;Reinisch & Sjerps, 2013), and singleton/geminate contrasts (Mitterer, 2018) and have demonstrated effect sizes more comparable to those seen the Bdisappearing word^cases. While variability in study designs obviously raises caveats to any generalizations across these studiesincluding in the type of dependent measure used (e.g., temporal boundary shifts vs. proportion of responses indicating an extra unit)it tentatively appears to be the case that distal rate effects on phoneme-level perceptual ambiguities (vowel-length contrasts, phoneme count, singleton/geminate) may be less robust than distal rate effects on syllable-level perceptual ambiguities (leisure or time, I align vs.…”

“…Recent studies have begun to more systematically investigate how distal context speech rate can affect the interpretation of upcoming speech material, systematically focusing on the role of distal speech rate in segmentation (e.g., Canadian notes vs. Canadian oats; Heffner, Newman & Idsardi, 2017). However, the study by Heffner et al (2017) did not focus on how distal rate may affect numbers of syllabic units.…”

“…Recent studies have begun to more systematically investigate how distal context speech rate can affect the interpretation of upcoming speech material, systematically focusing on the role of distal speech rate in segmentation (e.g., Canadian notes vs. Canadian oats; Heffner, Newman & Idsardi, 2017). However, the study by Heffner et al (2017) did not focus on how distal rate may affect numbers of syllabic units. Our hypothesis is that distal context speech rate affects the dynamics of neural entrainment for distinct lexical parses (Brown, Dilley, & Tanenhaus, 2014;Tavano & Scharinger, 2015), which would be expected to result in different numbers of syllabic units.…”

The singleton-geminate distinction can be rate dependent: Evidence from Maltese

“…At first sight, this may seem like a small difference, but there are two points to be considered here. First of all, a context is usually considered distal in the speech-rate normalization literature if it is at least a syllable away from the target (see, e.g., Heffner et al, 2017). This definition is supported given the importance of the syllable for rhythmically structuring language(s) (Ramus, Nespor, & Mehler, 1999).…”

“…However, the question whether there is rate-normalization or not may be ill-posed at a general level. Recent evidence suggests that rate normalization may be important for some contrast but not for others (Heffner et al, 2017). It is therefore worthwhile to test whether a quantity distinction might be rate-dependent.…”