As different as their linguistic backgrounds are, Presidents Barack Obama and George W. Bush share one phonetic trait (at least) that is a non-standard (dialectal) feature of American English, namely the “de-voicing” of obstruents after sonorants in final position. This happens most frequently in the case of the desinence {-s} of the possessive and the plural; hence in their speech what in standard English is [–z] comes out as [-s]. The final sound in doors, bills, moons, rooms, where a liquid or a nasal consonant is followed by what would be pronounced [-z], is rendered by them both as [-s]. If the more expansive definition of sonorant is used to include vowels as well, then for Messrs. Obama and Bush one can observe the same “devoicing” after stem-final vowels, except that it is not as sustained as in position after consonants.
How to explain this phonetic departure from the norm in speakers who in (most) other respects, albeit in varying degree, speak Standard American English? Is it a willful distortion? an instance of imperfect learning? a spelling pronunciation? No, it is none of these.
In order to understand this trait, we must first disabuse ourselves of the conventional definition of the sound at issue as “devoiced,” i. e., the “voiceless” counterpart of “voiced” /z/. From the strictly phonetic point of view, English [s] is indeed voiceless, and [z] voiced, hence the seeming aptness of the designation of the process observed in the two presidents’ speech as a “devoicing.” Traditionally, these obstruents are classified as, respectively, tenuis and media (pl. tenues/mediae). In the languages of Europe, this phonetic distinction translates into the phonological opposition between either the protensity feature (tense vs. lax) or the sonority feature (voiceless vs. voiced). In a language like French, for instance, the paired series of obstruents s/z. p/b, t/d, k/g, etc. are all opposed by the feature tense vs. lax, whereas in Russian they are opposed by the feature voiceless vs. voiced. These features are phonologically distinctive in these two languages. However, from the phonetic point of view, they typically go together in the physical realization of the sounds at issue, such that while the French /s/, for instance, is phonologically tense, it is also phonetically (non-distinctively, redundantly) voiceless (the vocal bands do not vibrate in its production); conversely, its counterpart /z/ differs from it phonologically by being lax but is phonetically (non-distinctively, redundantly) voiced (the vocal bands do vibrate in its production). In Russian, the same phonetic parallelism holds despite the difference in phonological values: Russian distinctively voiceless obstruents are redundantly tense, their voiced counterparts redundantly lax.
Unfortunately, descriptions of English have persisted in misidentifying which of these conjugate designations are distinctive, which redundant. All those phonetic traits that typically obtain in the realization of the media (“voiced”) members of the obstruent series in a language like French (or German or Serbo-Croatian, for that matter)––delayed onset of voicing, incomplete voicing, etc.––also pertain to English. But hierarchically these phonetic data are ancillary to the understanding of which feature is distinctive, which non-distinctive. The surest diagnostic in all such cases is not the phonetic realization per se but the manifestation of the two values of the phonological opposition in mutually exclusive contexts––otherwise known as positions of neutralization. Typically, in positions of neutralization, it is the so-called unmarked member that appears to the exclusion of its marked counterpart. Thus, in Russian, in final position before a pause the distinction obtaining elsewhere between voiced and voiceless obstruents is suspended, and it is the unmarked voiceless member that appears; hence rod [rot], xleb [xl’ep], voz [vos], etc.
The phonetic implementation of the opposition is not just a physical event: it is a sign of the relational values that define the opposition. That is the function of neutralization: it implements in sign-theoretic terms the values that define the phonological structure as a system. Without this sign function of the phonetic implementation, the phonology of a language would be neither a structure nor systematic––and would be neither learnable nor perptetuable. The sign function is the raison d’être of the phonetic implementation. Pace all the standard handbooks, it has nothing to do with such purely physical (phonetic) considerations as “economy of effort” or “assimilation.”
The interesting thing about neutralization is that while it is typically manifested without exception, this is the case only when the conditions of its manifestation are either positional or make reference to distinctive rather than redundant features. In the case at hand, Messrs. Obama and Bush’s trait of “devoicing” final desinences after liquids and nasals, what we have is a kind of neutralization that also has a sign function, but a slightly different one from straightforward neutralizations in which the unmarked term appears to the exclusion of the marked term of the opposition. The fact that voicing is non-distinctive in English liquids and nasals––while being normally voiced but positionally also voiceless, these sonorants cannot be opposed by the feature voiced vs. voiceless (unlike, say, Burmese)––is something that the phonology of English manifests sequentially as a sign-theoretic fact by allowing, as a matter of idiolectal variation, for either the “voiced” (i. e., lax/media) value [z] of the norm to appear after them; or, as in the speech of Messrs. Obama and Bush, the non-normative [s]. Here the typical result of neutralization––the appearance of the unmarked member, which would be the phonetic realization of /z/, alias [z] in English––does not obtain in non-normative speech precisely because the neutralization makes reference to a phonetic position (initial, medial, final), not to a sequential phonetic context.
Messrs. Obama and Bush’s departure from the norm follows as a matter of course, once we understand that the very indifference to a tense or lax realization of {s] as the possessive/plural desinence in English is wholly coherent with the semiotic nature of the phonetic realization, which is to signify the non-distinctiveness of protensity in sonorants.
MICHAEL SHAPIRO