Homophones are different words that sound alike, e.g., “bear” and “bare.” Since young children can often hear differences they cannot speak, they may produce homophones for words they know to be different. Thus, for example, a child may hear that “bike” and “bite” are different, but if s/he cannot yet produce word‐final consonants, s/he may utter both as /bai/. Two questions presently under investigation are: (1) Do young children actively seek to include or exclude such potential homophones in their early vocabulary? (2) Do children faced with such potential homophones produce subtle, subphonemic differences between them? This paper reports the results of a study aimed at answering the second of these two questions. Nine 20‐month‐old normally developing, English‐speaking children's utterances were tape recorded for acoustic analysis. These utterances were produced in situations in which homophones were likely to result from (a) reduction of initial consonant clusters, (b) stopping of initial fricatives, or (c) deletion of final consonants. Although it is hypothesized that the most likely sources of subphonemic differentiation would be VOT [for type (a) and (b) pairs] and vowel duration [for type (c)], F0 and formant frequencies were also measured and detailed auditory phonetic analyses were performed. Results from these analyses will be discussed in terms of individual and group patterns of differentiation between potentially homophonous word pairs. [Work supported, in part, by NIH.]

This content is only available via PDF.