As discussed in Chapter 7, at first blush, it might seem perfectly reasonable to assume thatlinguistic structure is influenced at least in part by speakers looking to (presumablyunconsciously) ease their own articulatory burden. Just as quickly, however, such a hypothesisencountered insurmountable difficulties, and we concluded that “ease-of-articulation”influences on linguistic sound structure are at the very least several times removed from the worldof individual speakers.

There are, however, variations on the fatally flawed theme of speaker-controlled approaches tosound patterning in general – and neutralization in particular – that we turn to now.These approaches center on the hypothesis that speakers control their speech in order to ease theperceptual burden of their interlocutors. We consider, in turn, Lindblom's “H&Hhypothesis”, Jun's “production hypothesis”, and Steriade's “P-maphypothesis”. We conclude by discussing some outstanding issues that arise from theseaccounts.

LINDBLOM’S H&H HYPOTHESIS (1990)

Lindblom (1990) does not specifically address the issue of neutralization. Nonetheless, as weshall see, his “H&H” (hyper- and hypo-articulation) proposal readily lends itself to such anapplication. The underlying assumption of this speaker-oriented approach to speech organization isthat language users have an awareness of listeners’ ability to extract information fromthe speechsignal. Consequently, speakers adjust their speech such that it varies along ahypo-articulation–hyper-articulation continuum. Hypo-articulation involves a “default to somelow-cost form of behavior”. If speech sounds are hypo-articulated up to or surpassing the point oftheir unique discriminability, neutralization may result. Hyper-articulation involves a higher-costform of behavior in which speech gestures are greater in magnitude and/or duration, thus enhancingthe ease with which listeners recover information, perhaps especially in contexts in whichnon-hyper-articulated speech sounds would run the risk of non-recoverability. In short: (1) speakersare constrained by certain production limitations that are both physiological and cognitive inorigin; and (2) speakers are aware of the receptive constraints of listeners, which are social andcommunicative in origin. Thus, overall, his model is certainly speaker-oriented, but speakersnonetheless take listening constraints into consideration as they plan and implement theirspeech.