Robert Albert Felty

Previous research has shown that listeners can adapt to particular samples of noise, a phenomenon known as “frozen noise” [Langhans and Kohlrausch, J. Acoust. Soc. Am.91, 3456–3470 (1992)]. However, no studies have reported a similar effect for multi-talker babble. The results of this study comparing open-set word recognition in multi-talker babble showed that listeners are significantly more accurate when the babble is fixed than when the babble is random. This documents the effect the authors refer to as “frozen babble.”

In speech perception and spoken word recognition, a lexical "neighbor" of a target word is a phonetically similar word that competes with the target during lexical access. A commonly‐used definition is that neighbors differ from one another by a single segment. While this basic definition accounts for many results of lexical decision and word recognition tasks among monosyllabic words, it also suggests that approximately half of the words in the English lexicon are hermits (words without neighbors). To generate a more inclusive definition of neighbor, we created a representative sample of the lexicon (N=1428) to examine spoken word recognition errors, and we used the incorrectly selected competitors as a means of indexing "neighbor" status. Analysis of over 15,000 errors reveals several patterns: (1) Errors tend to be of higher frequency than the target word; (2) Less than 1/3 of the errors were neighbors according to the traditional definition of neighborhood density; and (3) Errors differed in phonetic similarity more as the S/N ratio decreased. We use these data to test several alternative definitions of "neighbor" and their ability to predict actual word recognition errors. [Work supported by NIH‐NIDCD R01 00111 and T32 00012]

Theories of speech perception and spoken word recognition widely assume that phonetically similar words compete for selection during lexical access. These competitors ‐ called lexical neighbors ‐ are classically defined as differing by a single segment, with little attention paid to other levels of phonological structure such as featural content or syllabic structure. This paper reports on analyses of over 15,000 word recognition errors made on a representative sample of the lexicon (N=1428) that includes the same variety of syllable structures, lengths, and lexical frequencies that exists in the English lexicon. The analyses reveal that competitors share segmental information as has been previously suggested, but that the classic definition of neighbor cannot capture the level at which featural and syllabic properties of the target constrain the error responses. We will discuss several analyses on these issues and discuss the implications for our understanding of the organization of the lexicon. [Work supported by NIH‐NIDCD R01 00111 and T32 00012]