The founder of modern phonology, the great Russian linguist Nikolai Trubetzkoy(Николай Сергеевич Трубецкой [1890-1938]), claimed that it is impossible to say whether English has distinctive voicing or distinctive protensity (tense vs. lax) in its obstruent system, but the concept of rule coherence with its reliance on markedness considerations now makes this agnosticism seem groundless. Positions of neutralization are diagnostic in this respect because neutralization rules provide contexts in which variation rules tend typicallyto produce diagrams of the markedness values of the terms of phonological oppositions. The relation between syllable peaks and contiguous obstruents in English is such that syllable peaks are [-long] before tense obstruents but are [+long] before lax obstruents, sonorants, and in final position. Hence beet is [bit], but bead, beam, and bee are [bi:d], [bi:m], and [bi:], respectively. Beyond the fact that the complementary distribution here is semiotically significant-–as a sign of the non-distinctiveness of quantity in English–-it is the stipulation of the tenuis and media obstruents as distinctively tense vs. lax that allows the variation rule to be coherent. Since tense obstruents are marked relative to the unmarked lax obstruents, and shorter realizations understood as abridgements of syllable peaks are marked relative to unabridged peaks (which are unmarked), the markedness values of the vowels replicate those of the contiguous obstruents. If the tenuis and media obstruents were assumed to implement the opposition voiced vs. voiceless (as they often erroneously are to this day), the variations would lack coherence because the markedness values of the obstruents would not match those of the vowels (voiced obstruents are marked, voiceless unmarked).
This little digression into the phonological problematics of English illustrates the methodological status of what is now commonly referred to as “independent motivation” in linguistic explanation. Trubetzkoy’s claim of irresolvability of the English tenuis/ media problem in its obstruent system can be seen as justified only as long as circularity is barred from explanations of language structure (as if language were not a hermeneutic object). The mutual dependency of the elements of the solution proposed above––the shorter realizations of the syllable peaks seen as abridgments rather than the longer ones as prolongations, the stipulation of protensity as the relevant phonological category rather than voicing, and the invocation of markedness considerations as the vehicle of grammatical coherence––testifies to the fact that these elements cohere as an ensemble of conditions informing the data. This is the structural coherence that emanates from an evaluation (intrinsically, in the grammar) of the units and the contexts in tandem, in a mutually dependent manner, so that randomness and arbitrariness are reduced to a minimum (if not always to nil ).
In an unavoidably circuitous way we now come back to the earlier examples of alternation between English tenuis and media obstruents in stem-final position. The alternation becomes coherent when we understand it to subsume two conditions:(1) the alternation is morphophonemic, hence the markedness values as between sounds and meanings will be chiastic (complementary);(2) the obstruents involved are distinctively tense vs. lax. The recognition of these two conditions enables us to assert a coherence based on markedness values. Otherwise the alternation would be strictly arbitrary, non-iconic, and non-coherent.
When units and contexts do not cohere, the typical outcome in the long run is a heightened tendency toward the reduction of such instances, to the limit of their wholesale elimination from the language. With reference to our English examples there are attested historical changes that confirm the correctness of the analysis, specifically by showing morphophonemic coherence in just the sense advanced to be the telos of the changes. Where coherence has already been reached, no further changes occur. Thus in the history of English there is evidence of generalization of either the tense or the lax obstruent in words which now regularly have orthographic s, e.g., enterprise, compromise, purpose, promise, practise ( = American practice). In Middle English texts one can observe the testing of the contemporary rule in the occasional writing of z instead of s, particularly in verbs (but not only). In a Milton manuscript, for example, one finds the spelling practiz’d; and as late as 1836, the pronunciation of the infinitive with a z is proscribed as vulgar by normative grammarians. The same difference in pronunciation as between the nominal and the verbal forms of the word evidently obtained for enterprise and compromise in Middle English, the difference here being in the particular obstruent that was generalized. In the case of practise and promise,it was the tenuis obstruent that was generalized;in enterprise and compromise it was the media obstruent. Exactly why it was s in the first pair and z in the second constitutes a separate problem that might be treated in the spirit of rule coherence and markedness, but I hesitate to offer an explanation. Perhaps, in the presence of a primarily or secondarily stressed vowel in the verb form there is a discernible tendency to generalize the unmarked media z (compromise, but also close, as in at the close of … ). Similarly, one might want to explain the gradual elimination of the pronunciation of greasy with [-z-] (but cf. its continued presence in Southern American dialects) as a case of an unmotivated alternation being dropped from the (standard) language:there is no chiastic distribution of markedness values within the category of nominals. The semantic split between louse and lousy would tend to confirm such an analysis, albeit obliquely.
All of the above goes to show just how valid explanatorily is the conception of phonology as a semiotic system.