English is a language with primary and secondary stress, which means that words typically have one and only one syllable with a strongly individuated stress (primary stress) but may also have syllables with weaker stress (secondary stress). The longer the word, the more likely is the incidence of secondary stress. Thus a word like disestablishmentarianism (which The Oxford English Dictionary Online defines as ‘advocacy of disestablishment’ and qualifies by noting “usu. only as a factitious long word”) is pronounced with one primary stress (on the sixth syllable) and one secondary stress (on the third syllable) sensu stricto, the remaining eight syllables being unstressed.
Whether a word has constituent structure (= has more than one identifiable semantic or morphological element) may play a role in assigning a secondary stress to one of the syllables, but need not. Thus agentives in {-er/-or} like writer, prestidigitator, etc.––regardless of length––all treat this derivational suffix as being unstressed, i. e., bearing no secondary stress. On the other hand, in the contemporary speech of American adolescents––and of younger speakers of American English generally––a common word like student is increasingly to be heard with a clear secondary stress on the element [-ent], which is completely at variance with the traditional norm. This change may be due to the conceivable reconstrual of this element as a suffix, since the base [stud-] also occurs in study and studious, thereby lending plausibility to the analysis of student as having a constituent structure. Note, moreover, that no other interpretation can explain the emergence of secondary stress in this word as a change in contemporary American English pronunciation.
Why such a secondary stress does not also emerge in agentives in {-er/-or} may be due to the fact of the difference in length between post-tonic elements. If we regard {-ent} as an emergent morphological constituent (suffix) as a result of, or concomitant with, its being assigned secondary stress by younger speakers, then the fact of its having three sounds rather than two (by comparison with {-er/-or}) may be the threshold for such a change. This analysis would be wholly consistent with the general situation in English (and language in general), whereby suprasegmental (prosodic) features like stress are invariably dependent on the segmental structure, including the derivational morphology of words.
MICHAEL SHAPIRO