“Pour distinguer la syllabe réelle ou physique, de la syllabe d’usage, il faut observer que toutes les fois que plusieurs consonnes de suite se font sentir dans un mot, il y a autant de syllabes réelles qu’il y a de ces consonnes qui se font entendre, quoiqu’il n’y ait point de voyelle écrit à la suite de chaque consonne: la prononciation suppléant alors un e muet, la syllabe devient réelle pour l’oreille, au lieu que les syllabes d’usage ne se comptent que par le nombre des voyelles qui se font entendre et qui s’écrivent. Voilà ce qui distingue la syllabe physique ou réelle, de la syllabe d’usage. Par exemple, le mot armateur seroit en vers de trois syllabes d’usage, quoiqu’il soit de cinq syllabes réelles, parce qu’il faut suppléer un e muet après chaque r; on entend nécessairement aremateure. Bal est monosyllabe d’usage, et dissyllabe physique. […] A l’égard de la diphtongue, c’est une syllabe d’usage formée de deux voyelles, dont chacune fait une syllabe réelle, Dieu, cieux, foi, oui, lui.” (Charles Pinot Duclos, Commentary to the Grammaire général et raisonnée de Port-Royal, 1754)
‘To distinguish the real or physical syllable [that is, the “underlying” syllable] from the everyday syllable [that is, the “surface” syllable] one should note that whenever several consonants in a row are felt in a word, there are as many real syllables as many consonants are heard, even if there is no vowel written after each consonant: the pronunciation then supplying a silent e, the syllable becomes real for the ears, instead of this the everydays syllables can be counted by the number of vowels that can be heard and that are written. So this is what distinguishes the physical or real syllable from the everyday syllable. For example, the word armateur [‘shipowner’] would be three everyday syllables, although it would be five real syllables, because a silent e has to be supplied after each r; one inevitably hears aremateure. Bal [‘dance’] is an everyday monosyllable and a physically bisyllabic. […] As for the diphthong, it is an everyday syllable formed of two vowels, of which each makes a real syllable, Dieu [[djø] ‘god’], cieux [[sjø] ‘heavens’], foi [[fwa] ‘faith’], oui [[wi] ‘yes’], lui [[lɥi] ‘him’].’
If underlying syllables are surface syllables, the number of vowels gives the number of syllables. If underlyingly there are only CV (or VC) “syllables”, the number of consonants give the number of syllables. In any other case indeterminacy arises: in the case of a word like Geoffrey [ʤɛfɹi] standard government phonology cannot tell whether it’s [ʤɛ.fɹi] or [ʤɛ.f∅.ɹi]
Syncope is a widely attested phenomenon, usually occurring in the _CV environment. According to Kaye et al. (1990, reference on the schedule page) the (proper) government between a vowel and the preceding one is responsible for the muteness of the syncopated vowel: family [faməli] or [famli], but camel [kaməl], but *[kaml].
Two predictions:
Moroccan Arabic nouns of the same binyan surface differently, according to the consonants they contain: [mɨrd] ‘sickness’ vs [sdɨr] ‘chest’. [rd] is a C-to-C governing domain, [dr] is not.
C is muteness, V is loudness. Government spoils its target, licensing supports its target.
Consonant lenition is observed in “coda” position (i.e., _# and _C), as well as intervocalic position (V_V). When not before a vowel, a consonant is not licensed (causing lenition). When intervocalic a consonant is governed (since the following vowel does not govern the preceding vowel, so it governs the consonant). Governed consonant lose their muteness, become more sonorous.
Lenition is not observed (or observed to a lesser extent) in the “coda mirror” position (i.e., #_ and C_). In both of these cases a consonant is followed by a vowel, which licenses it, and preceded by an empty vowel (yes, every word begins with an empty vowel!), which absorbs the government of the following vowel.
| target | trigger | government | licensing |
|---|---|---|---|
| V | V | syncope | ? |
| C | V | lenition | support |
| C | C | CC cluster | ? |
| V | C | — | — |