Recognition of first words of isolated spoken word pairs with ambiguous word boundaries

Abstract

A series of experiments was conducted to test whether acoustic-phonetic information elucidates word boundaries. Word pairs were presented to subjects in gated format, where discrimination of the first word was marked by the first or last phoneme of the second word (e.g., RAN–PUMP or RAN–DUMP, respectively). Subjects were asked to provide either a written perceptual identification or use a numeric keypad to indicate the number of words at each gate (lexical enumeration). Results were consistent across response measures; perceptual identification of first words and lexical enumeration required less acoustic-phonetic information when first words were marked early in second words. In addition, the interval of silence between the word pairs was manipulated in two experiments to ascertain whether interword pause (30, 90, 150, or 210 ms) would result in differential perceptual identification and lexical enumeration. The interval of silence was not a reliable cue for first-word recognition or word counting and, thus is at odds with theories of auditory word recognition that incorporate pause as a factor influencing the activation levels of lexical candidates.