The role of juncture cues and phonological knowledge in English syllabification judgments

Listener syllabification judgments vary on words with VCCV sequences. This variability seems to present a challenge to phonological theory, which predicts invariant behavior following from rules or principles. Judgment variability might be better accounted for by positing that syllable production and perception are linked by the signal. According to such a hypothesis, variable listener judgments would result from variable speaker productions. To determine whether phonological or phonetic factors better account for listener syllabification judgments, different speakers produced nonsense words, listeners syllabified them, and then stimuli characteristics were used to predict syllabification. Results showed that, in spite of production variability, listener judgments were nearly invariant on words with medial sequences that formed illegal onset clusters to the second syllable and on words with first syllable stress, suggesting that these tokens were syllabified according to categorical phonological knowledge. Other judgments could not be similarly explained, but instead were best predicted by gradient phonetic patterns cuing juncture. The results disconfirm the hypothesis that syllable production and perception are directly linked via the signal. Instead, they suggest a two-step model of English syllabification in which listeners rely on juncture cues to determine syllable boundaries only after phonological knowledge fails to indicate a unique boundary location.