BarkMiner wrote: ↑15 Apr 2023 05:03
Quick question, if my speakers interpret a consonant cluster as a unique phoneme, does it make sense to Romanize it to the same characters as the consonants in the cluster, or use a modified character to match with how my speakers write and think about it. Specific example: the speakers of my language have separate characters for the consonants and consonant cluster /x/, /w/, /xw/. Would it make sense to write /xw/ simply as <xw>, or to mark it with a diacritic or something like <x́>?
We come to the philosophical question what a phoneme is.
Some grammars analyze consonant clusters
as if they were phonemes because it makes phonotactic analysis easier. But I think the realizations of a phoneme in the strict sense must be single sounds.
What is a single sound is the second question. In practice, you can well analyze the same thing either as the cluster /xw/ or a coarticulated sound /x
w/. The latter one is a single sound. Yes, coarticulations should in principle be simultaneous but in practice the coarticulated feature can well partially precede or follow the main feature. Similarly two adjacent sounds can well realize partially simultaneously.
Back to the original question if you should Romanize them as one-to-one match with the native script (transliteration) or more phonemically (transcription). It depends on if you want to show us something about the script or just about the language's phonemic structure. Usually imitating the native script is found more artistic by conlangers.
I personally doubt if people really perceive any sounds as single entities before formal training, i.e. learning to write. IMO, phonome analysis is always just analysis, for describing the language, not to find any mental reality.