There are two types of syllabification in English, the first is orthographic and the second is phonetic. The following references are have been extracted from JC Wells' Longman Pronunciation Dictionary, 3rd edition 2008 (LPD).
(LPD, p. 801) Phonetic (spoken) syllables must not be confused with orthographic (written) syllables. An orthographic syllable is a group of letters in spelling.When a word is plit across two lines of witing, it should be broken at an orthographic syllable boundary. (Word processors do this autoatically with a hyphenation program.) In some cases an orthographic boundary may not correspond exactly to a phonetic syllable boundary. For example, in the word happen the spelling includes two ps, and the orthographic syllabification is hap.pen. But the pronunciation has only a single p, and the syllables are /ˈhæp.ən/.
(LPD, p. XXIII 3 The English phonemic system and its notation
It is generally agreed that phonetic syllable divisions must as far as possible avoid creating consonant clusters which are not found at the edges of words. This is the phonotactic constraint. Thus windy might be ˈwɪn.di or ˈwɪnd.i, but it could not be ˈwɪ.ndi (because English words could not begin with "nd"). LPD takes the view that the syllabification of this word actually parallels its morphology: wind+y, /ˈwɪnd.i/. For the same reason language must be /ˈlæŋ.gwɪdʒ/, not /ˈlæŋg.wɪdʒ/ or
The principle that LPD adopts is that consonants are syllabified with whichever of the two adjacent vowels is more strongly stressed. If they are both unstressed, it goes with the leftward one. A weak vowel counts as
less stressed than an unstressed strong one.
In general this principle is subject to the phonotactic constraint. However, there are some cases where accurate prediction of allophones requires us to override it.
(i) Certain unstressed syllables end in a strong short vowel, even though words cannot. In nostalgia the t is unaspirated (as in stack (stæk), not as in tack), so the syllabification is (BrE) nɒsˈtældʒ.ə.
The differences are resolved in relation to a particular variety of the language at a particular time in its evolution.
Note that the pronunciation in the Cambridge dictionary is /nɒs ˈtæl.dʒə/ for the UK and /nɑːˈstæl.dʒə/ for the US. So, the stressed consonant t is (rather abnormally) not aspirated in this pronunciation in the UK case, but in the US case it is normally unaspirated since s precedes. In both cases the syllabification of "dʒ" with shwa diverges from more traditional English as the syllabification with the strong syllable has been disregarded.
/tekˈnɒl.ə.dʒi/ is correct according to LPD as both the phonotactic constraint and the principle of more strongly stressed vowel agree with making l part of the syllable with the ɒ nucleus, but is not correct as pertains to the principle of syllabification with the leftward syllable in the case of unstressed adjacent syllables (In LPD, the pronunciation is /tekˈnɒl.ədʒ.i/; both ə and i are weak vowels, and so are equally unstressed; it is admitted that a certain degree of stress has to be associated with strong vowels.).