Do these two pronunciations constitute renderings of a minimal pair in the phonological sense of the term? If they do, then the syllabic and non-syllabic n-sounds seem to be allophones of different phonemes.
Let's clear this up!
For two sounds (= phones) to be called allophones of the same phoneme two conditions must be fulfilled:
- the distributions of the phones must be predictable and
- no meaning difference is established if one phone is substituted for the other in the same context.
Recall that the syllabic [n̩] in the casual pronunciation of lightening is the result of a structure simplification process like the change of, say, bread 'n' butter or fish 'n' chips. What's left of the weak syllable is a nasal consonant. The vowel - here schwa - is deleted and the following consonant2 occupies the peak position; the consonant becomes syllabic. By this syllabic variant the syllable structure of the word is preserved. In the citation form the nucleus position is held by the schwa, the coda is occupied by the nasal. In the casual form we only have a nucleus, no coda, and this nucleus position is necessarily occupied by the /n/. As all phonemes in nucleus position are by definition syllabic, it's redundant to assign a syllabic status to the /n/. Moreover, a syllable is a syllable and a phoneme is a phoneme, i.e. they belong to different levels of analysis.
So the two n-sounds are incommensurable from a structural point of view. I wonder if there're any auditory or acoustic differences between [ˈlaɪtn̩ɪŋ] and [ˈlaɪtnɪŋ].
1 My thanks go to Martin J Ball for providing this beautiful example.
2 The interesting question is: Which consonants can be syllabic and which cannot?