Speech sound identification influenced by adjacent “restored” phonemes

Abstract

When listeners' identifications of speech sounds are influenced by adjacent sounds, do they operate only on the physical phonetic characteristics of there sounds or could they also use just their linguistic identity? We tested this by leading subjects to restore or induce the noise-obliterated medial consonant in VCV utterances by first presenting them with several prior utterances where this medial consonant was consistently the same, either a /b/ or /d/. Included as V1 were synthetic vowels from the /i-u/ continuum. As expected, more /u/s were identified out of this continuum in the environment of physically present /d/s than /b/s [Ohala et al., J. Acoust. Soc. Am. Suppl. 1 64, S18 (1978)]. But restored /d/s and /b/s had the same effect (though with slightly less intensity) thus indicating that the influence of context need not operate only via physical phonetic features. We discuss the implications of this finding for “direct realist” theories of speech perception and for claims of the existence of invariant phonetic elements. [Supported by the Cognitive Science Program, UCB.]

Context Effects in a Double-Weak Theory of Speech Perception

“…A bias represents a constant shift of evaluation functions that does not vary over levels of the stimulus properties that have been actively manipulated in a n experiment. See Ohala and Feder (1987) for what appears to be an illustration of a transsegmental bias effect induced by phonemic restoration. See also note 7 below.…”

On the perception of “segmental intonation”: F0 context effects on sibilant identification in German

“…2 Following a reviewer's request, we point out in this context that the assumption of a "subtraction mechanism" which underlies speech perception and allows listeners to compensate for variation in sound segments is still compatible with variation-driven sound change. As is stressed by Ohala and Feder [37] and summarized by, for example, Kleber et al [40], sound changes can occur despite a "subtraction mechanism" due to a misalignment between the production and perception of variation [85]. For example, L2 speakers of a language who are not entirely familiar with the coarticulatory patterns of that language cannot (fully) subtract the corresponding variation and hence still perceive the coarticulatory variation in speech sounds.…”

“…See, e.g. Ohala and Feder [37], Harrington et al [39], and Kleber et al [40] for related work and findings on the perceptual compensation of coarticulation are sensitive to the aperiodic pitch impressions of fricatives and relate them to the F0 context in normally voiced utterances.…”

“…Similarly, but in the opposite direction, Kleber et al [40] found lowered F2 values to still trigger /ʊ/ (rather than /ɪ/) identifications, if this lowering was attributable by listeners to the low F2 level of a subsequent back vowel. To sum up in the words of Ohala and Feder [37]: "There is, in fact, abundant evidence, both from perceptual experiments as well as phonology […], that listeners identify speech sounds in part by normalizing them with respect to their phonetic context".…”

“…The experimental approach used in the present study is inspired by the feature-parsing model of Gow [36] and the work of, for example, Ohala and Feder [37], Fowler et al [38], Harrington et al [39], and Kleber et al [40] on the perceptual compensation of coarticulation. The basic idea behind this approach is the following: If listeners interpret a certain aspect of a sound segment as being due to an external factor, then they "subtract" this aspect from the sound segment when identifying its phoneme category.…”