Can someone explain why would we syllabize constant as /kɔ̃-stɑ̃/, so V$CCV but actif as /ak-tif/ (VC$CV)? French tends to have open syllables so why wouldn't we syllabize actif as /a-ktif/?

Merci d'avance.

  • I'm not familiar with your notation, what's V$CCV and VC$CV ? Could you be a little more clear on that part? It would make answering your question easier. Commented Nov 28, 2018 at 13:31
  • @TeleportingGoat V = vowel, $ = syllable break, C = consonant. lmc has omitted the /k/ of constant, though.
    – Luke Sawczak
    Commented Nov 28, 2018 at 14:45
  • 1
    "Constant" is very much / /kɔ̃s.tɑ̃/ for me. The /ɔ̃/ is very clearly long in my pronunciation, which only happens in closed syllables. Commented Nov 28, 2018 at 16:49
  • Actually, I'd extend that to any word internal /sC/ cluster: preceding vowels all show the allophonic variant you'd expect from a closed syllable: espoir [ɛs.pwɑːʀ̻] rather than *[e.spwɑːʀ̻], austère [o̝ːs.t̪ɛːʀ̥] rather than *[o̝.st̪ɛːʀ̥], histoire [ɪs.t̪wɑːʀ̻] rather than [i.st̪wɑːʀ̻]. Trying to analyse them as /V.sC/ seems like trying too hard to make word internal syllabic onsets match word initial ones that's at odds with the fact of the language (as I speak it at least) Commented Nov 28, 2018 at 17:00
  • @Eauquidort Je dois réfléchir sur ça...
    – Luke Sawczak
    Commented Nov 28, 2018 at 19:04

1 Answer 1


This is due to what's known as the Maximal Onset Principle: when dividing syllables, we place as long a cluster in the onset as possible, provided that it doesn't violate the phonotactics of the language.

As you might imagine, the condition is the hard part to define. What's allowed and what isn't?

Here's a case where there's no violation: if we take the word bacon /bakɔ̃/, the principle demands that we put the /k/ in the onset of the second syllable since French syllables can begin with /k/.

Here's a case where there is a violation: if we take the word strengthen /ˈstrɛŋθən/, the principle demands that we put only the /θ/ in the second syllable since nasal + fricative is not attested in English.

So then looking at constant and actif, we would ask for each of them, how much of that cluster can we fit in the onset? And it turns out that we can put /st/ together into the onset for constant, given words like stage. However, we can't get both /kt/ into the onset for actif, since we don't find any word that begins with "kt". In fact, two stops in a row is generally rare. Hence, we are forced to divide the syllable between /k/ and /t/.

A couple of notes apply. First, you might ask: surely we can come up with a rule that determines allowable onsets? Do we really have to look at the inventory of words to look for attestations?

  • This is something that linguists do indeed wonder, and there are certain tentative rules, such as "the cluster must always increase in sonority in an onset and decrease in an onset" — one problem being that the onset cluster /str/ should violate that rule since /s/ is more sonorous than /t/ but less than /r/.

  • Moreover, these rules also vary between languages: in English, we eliminate the /p/ in pneumatic, psychology, and pterodactyl, but in French, words like pneu, psychologue, and ptérodactyl all have a pronounced /p/ — the last being particularly interesting since it's two stops in a row, which as we just saw above is not too common.

  • And they also vary over time; the onset /s/ + stop has a troubled history in Romance languages, leading to phenomena like the Greek name Stephanos turning into Esteban in Spanish and both Étienne and Stéphane in French depending on the phonotactics at the different times the word was re-imported.

Second, even the Maximal Onset Principle has its detractors. One way that we can test whether the syllable is a real cognitive entity or a mere abstraction of our analysis is through priming. The priming effect on a word has been observed to be stronger if you use a complete syllable of the word rather than stopping short at a fragment, and similarly if you use a complete syllable but do not add part of the next syllable.

  • For example, balcon is primed most effectively by /bal/, and less by /ba/ or /balk/.

  • However, there's evidence for ambisyllabicity as well, in which a segment belongs to both a coda and an onset. The English word balloon turns out to be primed equally well by /bə/ or /bəl/. This is not the case with the French word balloune, which is primed better by /ba/ than by /bal/.

  • Hence, even the concept of the syllable may not be consistent across languages.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.