Phonological underspecification and speech motor organisation.

Number 794
Year 1991
Drawer 14
Entry Date 11/05/1999
Authors Boyce, S. E., Krakow, R. A., & Bell-Berti, F.
Contact
Publication Phonology, 8, 216-236.
url http://www.haskins.yale.edu/Reprints/HL0794.pdf
Abstract [Introduction] Over the last few years, much work in phonology has been devoted to exploring the way features are specified for segments; in particular, to what extent feature specification may be underlyingly present and/or acquired by rule or default in the course of a derivation.While a numer of proposals have been made attributing various degrees of underspecification to abstract levels of the phonology (Kiparsky, 1985, Steriade 1987, Archangeli 1988), it has been generally assumed that where phonetic implementation comes into play, i.e. at the end of the derivation, segments are exhausitively specified. This view stands in contrast to that adopted in much of the literature on speech motor control, where the necessity of accounting for coarticulation across multisegmental spans has led researchers to assume that the input to the motor plan leaves a good deal of phonetic detail unspecified. Motor implementation in thse models isa sssumed to proceed by direct translation of specified features into articulatory/acoustic targets, leaving the position of articulators during an unspecified segment open to influences from the surrounding context. Thus, coarticulation is viewed as assimilation of specified features from surrounding segments onto an unspecified target. As a practical matter, researchers in the field have assigned feature specification for this purpose from: (i) surface phonological contrast and (ii) required articulatory/aerodynamic configurations. If neither source mandates specification, the segment is assumed to be unspecified for that feature and thus to have no independent target. For example, [+nasal] amd [-nasal] are feature values that characterise English stops at the surface level, and therefore these consonants are taken to require a relatively low or relatively high velum position, respectively. At the same time, because English vowels lack a surface contrast in nasality and because nasalisation of vowels is articulatorily possible (while a nasal /s/ os mpt), coarticulation researchers have assumed that English vowels are unspecified for nasality (e.g. Moll & Daniloff 1971; Kent et al. 1974). Note that, in this view, English vowels and consonants have the same [nasal] specification status underlyingly as at the surface. The case is different for a segment such as /s/. On the one hand, it is presumed to have the surface feature value [-nasal] for /s/ may be lacking at more abstract levels of the derivation. In English, as in other languages, some amount of nasalisation is generally present on vowels preceding nasal consonants (Clumeck 1976). In analogy to phonological analyses of assimilation processes such as vowel harmony, many studies of this phenomenon (Moll & Daniloff 1971; Kent et al. 1974; Hammarberg 1976) have treated the presence of even minimal acoustic or articulatory indicators of nasilisation as showing the spread of the feature specification [+nasal] into an unspecified doman. It was assumed that the intermediate level of implementation typical of the data came from an ability of the articulators to achieve opposite target configurations instantaneously; that is, it was assumed that the vowels acquired a full [+nasal] target but would not fully implement it, rather than that the vowels had independent but intermediate targest (see Kent eta al. 1974 for further discussion). In a seminal paper, Keating (1988b) examined theories of underspecification as they exist both in the motor control literature and in the phonological literature, in an attempt to reconcile the two. She accepted the motor control notion that there are phonetically unspecified segments, and that these segments have no inherent targets, but argued that, rather than indicating the presence of feature spread, the presence of intermediate levels of articulatory/acoustic implementation (e.g. slight lip protrusion for rounding in the context of a [+round] segment, or the presence of a weak nasal formant in the context of [+nasal] segment) could be taken to indicate presistence of underspecification into the motor planning level. Further, she suggested that if a segment normally analysed as unspecified for a particular feature showed an apparent target, i.e. showed apparent full implementation for that feature, a phonetic rule must have applied to supply that target. She proposed that segments are not exhaustively specified at the end of the derivation, and most radically, and interestingly, that phonetic data can be used to make inferences about the lack of specification at higher levels of the derivation. In this paper, we examine some of Keating’s arguments and introduce data of our own indicating that certain of her conclusions may be premature and/or insufficiently detailed. In particular, we attempt to show that, although Keating’s basic insight remains viable, much of her argument suffers from the nature of her assumptions about the phonetic implementation of targets, and that a greater attention to phonetic detail, and in particular to the variable of timing, is required in order to eliminate other interpretations of phonetic data. In opposition to Keating’s point of view, we present evidence to show that segments which lack specificaton by contrast criteria or by aerodynamic/articulatory criteria nevertheless exhibit characteristic articulatory positions associated with those features. Our data will focus on the features [round] and [nasal].
Notes

Search Publications