Word stress in English can fall on any syllable but is typically circumscribed when the word is derived, as in the case of adjectives derived from substantives (nouns). As in all derived entities throughout grammar and culture, derived adjectives are subordinate in rank (hierarchical value) relative to the substantives from which they are derived. Moreover, where adjectives are formed by the addition of a suffix to the nominal base, their subordinate status can be expressed by a different place of stress in comparison with the deriving noun. Adjectival suffixes that displace the stress in this way are called “auto-” and “pre-stressed,” the first type pulling the stress onto itself (Japán > Japanése, grótto> grotésque), the second onto the penult or antepenult (ádjective > adjectíval, geriátric > geriatrícian, évidence > evidéntiary, Eúrope > Européan, Terpsíchore > Terpsichoréan, Hércules > Hercúléan [stress doublet = both stresses are extant]). Some suffixes occasion such stress shifts obligatorily; others do so facultatively, depending on the nature of the vowel (conventionally termed “strong” vs. “weak”), for instance súicide > suicídal, mícroscope > microscópic, but máyor > máyoral, eléctor > eléctoral, dóctor > dóctoral, pástor > pástoral, Chíle > Chílean, Ghána > Ghánaian.
The last set of examples is significant in that one often hears these words mispronounced in contemporary American (and, to a certain extent, British) speech, specifically with stress on the penult, thus non-standard mayóral, etc.; and Ghanáian, Chiléian (the spelling is irrelevant). For all three of these particular items, the Merriam-Webster Online Dictionary and the American Heritage Dictionary cite both pronunciations as equivalent while listing the traditional one first. It is clear historically that the newer stress is the one on the penult, which raises (NOT “begs” !) the question, why the shift?
The answer has nothing to do with the phonetic characteristics and everything to do with the semiotic characteristics of these words. In order to make sense of the shift, one must recognize the fact that prosody in languages like English––the (suprasegmental) placement of stress––has a SIGN FUNCTION, and that this function is coordinated with the function of suffixation. Specifically, where derivational morphology is concerned, as the derivative comes into more widespread use it tends to assert its independence formally and semantically from the deriving base. This process gathers strength if and when the derivative all but loses a synchronic connection with the deriving base, as is the case of pastor > pastoral, where the meaning of the latter refers to a literary genre and not an agricultural or ecclesiastical context (although the latter contexts may ultimately come under the sway of the more frequent reference, thereby shifting the stress to the penult for all meanings). Cf. áncestor > ancéstral, sépulchre > sepúlchral. Semiotically, what obtains in the interplay between prosody (stress) and segmental structure (derivation via suffixation) is a teleological tendency to align the prosodic markedness value of the deriving base with that of the derived form. Any derived form whose constituent structure is transparent, i. e. where the semantic link between base and derivative has been preserved, is evaluated semiotically as marked by definition. Thus in the case of the adjectives at issue, their derived status and hence their marked value tends to occasion the emergence of a form of the deriving base (here: a noun) that is likewise marked vis-à-vis the unmarked value of the noun. A shift of stress away from the syllable stressed in the noun onto another of its syllables closer to the end in the derived adjective produces a prosodic form of the deriving base that is marked vis-à-vis its unmarked underived counterpart. The markedness value of the adjective is, in other words, mirrored by the markedness value of its base, thereby promoting the structural coherence of the form.
Note that this analysis is confirmed by the fate of the adjective banal, where the traditional penultimate stress has been all but superseded by stress on the ultima in contemporary English, a reflection of the fact that the word has lost all semantic links to its etymologically recoverable deriving base, ban. The word illustrates what happens when the deriving base is monosyllabic: in the absence of an alternate syllable to which stress could shift in the derived adjective, it falls on the lone other available syllable, viz. that of the suffix.
A more straightforwardly simple way of expressing the process, of course, might be to say that derivation always tends to promote a contrast between deriving base and derivative. But this is a specious simplicity: contrast––not just between word classes––is so often violated as to be practically useless in explaining language change, whereas the coherence of markedness values is a telos that obtains universally in change. The contemporary shift of máyoral to mayóral and Chílean to Chiléian, etc., is thus to be understood not merely as a violation of the traditional norm but as a realization of an IMMANENT STRUCTURAL COHERENCE in the derivational morphology of English.