The reality of phonological forms: a rejoinder

Abstract

I thank Dr. Fowler for her thoughtful comments on my paper (see Port, 2010; Fowler, 2010). I am especially pleased that she acknowledges several points of agreement between us in her first paragraph. I would like to begin by highlighting one of these points since I believe it is a radical view that she and I may share but which probably contrasts with the views of many linguists and psychologists. Certainly it contrasts with the traditional Chomskyan view. Fowler says, and I agree, a linguistic description ‘‘reflects what works for a language community. It is not what sits inside . . . someone’s mind or brain.’’ Instead, it is ‘‘on average what language users know.’’ This means a language is a set of phonological, lexical and syntactic categories shared as conventions across some community of speakers and some range of contexts. If this is where language lives, then looking at an individual speaker, one should not expect to always find the same categories specified in the same terms. Each speaker will develop their own speech motor patterns, suitable to their vocal tract, as well as perceptual patterns and categories that are idiosyncratic in detail across speakers but which work well enough to permit successful communication. In such a system, consonants and vowels as well as other linguistic units are only definable as social patterns of behavior. And we do not necessarily know how these patterns are instantiated within any speaker. My conclusion from this is that we no longer need to look for linguistic units like Cs or Vs, distinctive features, phones, words, etc. as identifiable psychological units in speakers’ minds. They are social objects that individuals, through imitation during language acquisition, are able to produce and perceive using idiosyncratically-defined perceptual chunks and motor routines of various types and sizes. Nevertheless, Fowler still wants to postulate consonants and vowels as real-time tokens ‘‘used by members of a community.’’ In other words, she apparently supports some uniform code of symbols that is the same across the brains of all speakers of a language and used for real-time production or perception of words. She gives several reasons to support the real-time role of phones or phonemes. The first source of evidence is the research conducted on speech errors since frequently speech errors can be succinctly described as cases of ‘‘switching’’ segments (e.g., the error sprit blain for intended split brain). But speech error data are not reliable. First, they depend on auditory phonetic transcription (i.e., reduction to written form) thereby forcing the use of a low bit-rate representation of phonetic symbols. Plus the transcription relies on an ill-understood subjective process. But worse, as acknowledged by Fowler, recent data suggest that speech errors may seem to be discrete mainly because they were collected using a discrete phonetic transcription. As soon as one looks at the real-time behavior of someone making a speech error (Goldstein et al., 2007), then continuous valued parameters, temporal overlap of gestures and other real-time phenomena are found that ruin the simple picture of discrete segment-to-segment speech production. The second difficulty