There are two main approaches to the syllable: a phonetic one and a phonological one. A phonetic approach tries to characterize syllables in terms of articulatory or acustic events. A phonological approach tries to identify syllables on the basis of the sound pattern attested in the languages of the world.
Why we have to look into syllables ?
Syllables as units of phonological analysis mainly because they influence the realization of allophones; put in another way, they affect the allophonic distribution of phonemes.
Let us analyze some English examples and/or phenomena in relation to this:
The generalization is: ˇ§English voiceless stops (plosives) are aspirated in syllable-initial positionˇ¨.
As can be seen, we need to refer to the syllable to explain this allophonic variation.
[p] in ˇ§happyˇ¨ is not aspirated, why? It belongs to two syllables in the sense that it functions as the coda for ˇ§ha-ˇ¨ and the onset for ˇ§pyˇ¨. The stress of the first syllable ˇ§attractsˇ¨ the [p] into it.
Listen to your ˇ§aˇ¨ in ˇ§badˇ¨ and ˇ§batˇ¨. Which sounds longer? YES! Before [d] it is longer than before [t]. This is a common process that takes us to the following generalization:
ˇ§English vowels are long when followed by a voiced obstruent in the same sˇ¨
Nouns stress placement
Generalization: ˇ§English nouns are stressed on the penultimate syllable if heavy, otherwise, they are stressed on the antepenultimate syllableˇ¨ (The above-mentioned explanation and examples are provided by Victor Prieto )
Each syllable has the simplified structure shown below:
A fairly typical sort of syllable has the structure CVC (where C = any consonant; V = any vowel). The vowel is called the nucleus (Nu.) which is the obligatory element of the syllable. The consonants which precede the nucleus are called the onset (On.), while the consonants which follow the nucleus are called the coda (Co.). The nucleus and coda are combined to form the rhyme (Rh.) of a syllable. A syllable has a rhyme, even if it doesn't have a coda. That is, the nuclear node will be immediately dominated by the rhyme node.
The world's languages have a variety of syllable types. English permits a wide range of structures, but in other languages the options may be limited to a single type. While the nucleus is obligatory, onsets may by obligatory or optional. In English onsets are optional (e.g. [feis] <face> ~ [eis] <ace>), but in some languages (e.g. Cairene Arabic) every syllable must begin with a consonant. Similarly, codas may be optional or absent. English selects the first of these choices (e.g. [sta:t] <start> ~ [sta:] <star>), but there are languages like the South American language Pirahã which does not permit codas.
The other parameter which languages set is whether the elements of each syllable branch or not. This determines whether a language permits consonant clusters or long vowels beside single consonants and short vowels. In English all three parts of the syllable can branch, with long vowels and diphthongs occurring in the nucleus (e.g. [ti:m] <team>, [taim] <time>), and consonant clusters occurring in the onset and coda (e.g. [klaim] <climb>, [b^lk] <bulk>). Other languages only permit short vowels and disallow consonant clusters. It is important to distinguish between tautosyllabic clusters (i.e., both consonants are in the same syllable as in [limp] <limp>) and heterosyllabic sequences of consonants (i.e. the consonants occur in separate syllables as in [.bæn.t@.] <banter>).
The types of consonant clusters which are permitted in the onset and coda are mostly governed by the Sonority Sequencing Principle (SSP) which states that the sonority of the segments in a syllable increases towards the nucleus and declines away from it. The segments of a language can be arranged into a sonority scale. The sonority scale for English given in (3), although it is also valid for other languages. Voiced segments are more sonorous than voiceless segments and sonorants are more sonorous than obstruents. Vowels are more sonorous than consonants. The scale ranges from the least to the most sonorous segments.
A Sonority Scale
Voiceless stops [p, t, k]
Voiced stops [b, d, g]
Voiceless affricates [t?/SUP>]
Voiced affricates [dzh]
Voiceless fricatives [f, þ, s, ? h]
Voiced fricatives [v, ð, z, zh]
Nasals [m, n, ng]
Liquids [l, r]
Glides [w, j]
As we shall se below, there are certain languages in which syllable quantity plays a role in determining the assignment of stress. Syllable quantity is partly determined by the number of segments in the rhyme of a syllable. The crucial distinction is between light syllables and heavy ones. A light syllable consists of a single short vowel in an open syllable. A heavy syllable consists of a long vowel or a diphthong, or a short vowel followed by one or more consonants in the same syllable. Any syllable which ends in a consonant, whether the vowel is long or short, is called closed and will be heavy. The following words from English illustrate these various syllable types. (L = light syllable; H = heavy syllable; O = open; Cl = closed)
(Source: The Encyclopedia of Language & Linguistics, ASHER)
(For related fields, please see Metrical Phonology)