What exactly is the definition of a syllable?

Question

I do not consider myself a linguist. I just teach English to Japanese audience. So please excuse my ignorance if this is too basic a question.

What exactly is the definition of a syllable? What I often see is essentially this: a syllable is a series of sounds with exactly one nucleus, which is a vowel, with optional consonants surrounding it. This works for me because I think the distinction between vowels and consonants are relatively clear (except for glides, IMHO).

It gets confusing, however, when you factor in syllabic consonants, which they say can also be nuclei of syllables. I think what we have here now are two circular definitions:

• A syllable is a series of sounds with exactly one nucleus, which is either a vowel or a syllabic consonant, with optional consonants surrounding it.
• A syllabic consonant is a consonant that can be a nucleus of a syllable.

The only way to break this cycle is to define either one differently, without referencing the other. One such definition of a syllable I found is this: a syllable is marked by a peak in sonority. Admittedly, this is not a precise definition, but I do think it captures the underlying idea. But some would argue that it just paraphrases the criteria of a phone being able to be a nucleus of a syllable or not, as its sonority level being high or low. More importantly, it does not give a concrete threshold of sonority level for a phone to be able to be a syllable nucleus, so it still does not answer when a consonant is considered syllabic and when it is not.

I wonder if the notion of syllable is innately understood by native speakers of a particular language, but cannot be objectively defined, particularly not by a quantifiable measure. Perhaps what is often presented as the definition of a syllable is actually a mere description?

To wit, I would like to know if there is a definition of a syllable and that of a syllabic consonant that collectively are not circular in nature.

Thank you.

P.S. If I am not responding to your comment or answer immediately, that is probably because it is taking me a lot of time to take in your comment or answer. I am a complete layman, so just reading through your comment or answer can be a big task for me. Please pardon my tardiness.

Answer 1

I think it will be different depending on the theory that you choose to work with, and unfortunately, I've never studied formal linguistic theories, so I can't give you an actual definition. Sorry.

Although this is not a complete answer, I wanted to talk about an area that I think you did not mention in your question: the importance of the "syllable" in the prosodic systems of many languages. "Syllables" are often relevant to rules about accent, stress or pitch. Another area where we might find rule related to syllabification is minimum length requirements. Swahili is supposed to have a bias against monosyllabic words in various contexts, which leads to forms like [ḿ.bwa] "dog" (mentioned in user6726's answers here and here).

In the context of Japanese, the "syllable" has been argued to be a relevant unit for rules about the placement of pitch accent in certain varieties: see Mechanical snail's answer here.

Answer 2

There is some unclarity in your question, but that is largely the fault of linguistics as a field, which I regret to say doesn't have a solid grip on what a "definition" is. The problem is that we use definitions to encapsulate all of our knowledge about a thing, and we also use them to formally relate one thing to another thing in a mathematical theory of phonological things. So I could interpret the question as asking "what are the necessary and sufficient criteria for analyzing some substring of speech as a syllable", and "what are the required and optional properties of a syllable in a formal theory of phonology, and how do they relate to other phonological objects"? To complicate the matter a bit, you could also ask about "phonetic syllables".

There is a quasi-standard formal theory of prosodic objects stemming from work in the 70's and 80's, which posits a set of prosodic objects – mora, syllable, foot, "phonological word". Wording from the bottom up, a mora may immediately dominate a continuous sequence of segments. A syllable may immediately dominate a continuous sequence of segments and moras. A foot may immediately dominate a continuous sequence of syllables. A phonological word may immediately dominate a continuous sequence of feet. (Of course people have made claims for other possibilities, this is just the bare bones that is well-supported). There is a further issue regarding how many things of a higher level can immediately dominate a thing of a lower level, for example can a segment be in two syllables at the same time?

Unfortunately, there is a traditional term "syllabic", as in "syllabic consonant", which refers to something slightly different. Some languages have "special" versions of consonants known as "syllabic", for example in Serbo-Croatian, Swahili, Ewe and Sanskrit – also English (pervert, button), there are versions of m,n,l,r which are phonetically distinct from regular m,n,l,r, and which phonological behave as though they are the equivalent of vowels within a syllable.

We can interpret the expression "syllabic consonant" as meaning "consonant which is the nucleus of a syllable", since that corresponds to how people generally use(d) the expression. Notice however that our theory does not yet have a construct "nucleus" – we also have to settle on a theory of what the primes of representation are. I've given you the more sparse theory of syllables above, in which a "syllabic consonant" would be a consonant that is the only thing immediately dominated by a mora (which is then immediately dominated by a syllable). Some people have intermediate objects "onset", "nucleus" and "coda", so a "syllabic consonant" is a consonant that is immediately dominated by a nucleus.

The circularity that you find is because of a mixing of formal definition and diagnostics. Vocoids are generally parsed into a moraic / nuclear position, but that is not part of the definition of syllable, unless you have a metatheory of "definition" where the empirical properties of an object are implicit in (derivable from) the definition of the object. In my opinion, focusing on "definitions" has been an epistemological dead-end in linguistics. The relevant question usually is not "how do you define X", it is either "what are the properties of X?", "what is the evidence that X exists?" or related "how do you show that this is an X?".

Unfortunately, invoking "sonority" in theorizing about syllables is a problem since there's no independent means of computing sonority. A major defect in sonority-based theories of the syllable is that they tend to assign primacy to sonority, thus you make your decisions about syllabification based on assumptions about sonority. But sonority is determined based in part on syllabicity (a syllabic consonant is apparently high sonority).

Answer 3

I have a personal definition of syllable, which assumes that consonants as well as vowels can be stressed or unstressed. The definition is: a syllable is a syllabic sound together with any preceding stressed consonants and any following unstressed consonants. This is intended to describe the fact that consonants preceding the syllabic of a syllable are more strongly articulated than consonants in the same syllable that come after the syllabic sound.

My account is similar to the account of syllables in Saussure's Course in General Linguistics (except instead of stress, Saussure appeals to a feature explosive/implosive).