Systematic Variation in Spoken Word Production: Phonological, Lexical and Contextual Factors

During speech production, there is considerable variation in how a given segment is articulated, and past work suggests that this variability is not completely random; rather, how a sound is pronounced depends on a number of factors. I will discuss several factors that appear to influence how phonetic segments are acoustically realized, including characteristics of the word in which a segment is found, that word's phonological relationships to other words in the lexicon, and the sentential context in which that word is produced. I will argue that investigating the parameters that govern how different factors condition the acoustic-phonetic properties of spoken words can offer a window into the cognitive architecture underlying speech production. I will focus on some of my recent experimental work (with Sheila Blumstein and Megan Reilly) and discuss several possible future directions.