There are still problems with the description of the syllable: an unanswered question is how we decide on the division between syllables when we find a connected sequence of them as we usually do in normal speech. It often happens that one or more consonants from the end of one word combine with one or more at the beginning of the following word, resulting in a consonant sequence that could not occur in a single syllable. For example, 'walked through' wɔ:kt θru: gives us the consonant sequence ktθr.

We will begin by looking at two words that are simple examples of the problem of dividing adjoining syllables. Most English speakers feel that the word 'morning' mɔ:nɪŋ consists of two syllables, but we need a way of deciding whether the division into syllables should be mo: and nɪŋ, or mom and ɪŋ. A more difficult case is the word 'extra' ekstrə. One problem is that by some definitions the s in the middle, between k and t, could be counted as a syllable, which most English speakers would reject. They feel that the word has two syllables. However, the more controversial issue relates to where the two syllables are to be divided; the possibilities are (using the symbol . to signify a syllable boundary):

i) e.kstrə

ii) ek.strə

iii) eks.trə

iv) ekst.rə

v) ekstr.ə

How can we decide on the division? No single rule will tell us what to do without bringing up problems.

One of the most widely accepted guidelines is what is known as the maximal onsets principle. This principle states that where two syllables are to be divided, any consonants between them should be attached to the right-hand syllable, not the left, as far as possible. In our first example above, 'morning' would thus be divided as mo:.niŋ. If we just followed this rule, we would have to divide 'extra' as (i) e.kstrə , but we know that an English syllable cannot begin with kstr. Our rule must therefore state that consonants are assigned to the right-hand syllable as far as possible within the restrictions governing syllable onsets and codas. This means that we must reject (i) e.kstrə because of its impossible onset, and (v) e.kstrə because of its impossible coda. We then have to choose between (ii), (iii) and (iv). The maximal onsets rule makes us choose (ii). There are, though, many problems still remaining. How should we divide words like 'better' betə? The maximal onsets principle tells us to put the t on the right-hand syllable, giving be.to, but that means that the first syllable is analyzed as be. However, we never find isolated syllables ending with one of the vowels I, e, as, A, Q, U, SO this division is not possible. The maximal onsets principle must therefore also be modified to allow a consonant to be assigned to the left syllable if that prevents one of the vowels ɪ, e, æ, Λ, ɒ, ʊ from occurring at the end of a syllable. We can then analyze the word as .ə, which seems more satisfactory. There are words like 'carry' kasri which still give us problems: if we divide the word as kas.ri, we get a syllable-final as, but if we divide it as kasr.i we have a syllable-final r, and both of these are non-occurring in BBC pronunciation. We have to decide on the lesser of two evils here, and the preferable solution is to divide the word as kær.i on the grounds that in the many rhotic accents of English, this division would be the natural one to make.

One further possible solution should be mentioned: when one consonant stands between vowels and it is difficult to assign the consonant to one syllable or the other - as in 'better' and 'carry' - we could say that the consonant belongs to both syllables. The term used by phonologists for a consonant in this situation is ambisyllabic.