Is it ever normal for word boundary information to be available for shaping in a writing systems where word boundaries usually lack visual indication?

It has been argued that to render a Northern Thai word in isolation correctly in the Tai Tham script, one does not need to know the relative order of vowels and subscript consonants. For example, there is only one one word composed of base consonant HIGH HA, SIGN U and subscript LOW YA; SIGN U goes bottom left. If one adds just the letter MA at the end to form a different word, there is again only one arrangement that forms an actual word; SIGN U goes bottom right relative to the HIGH HA. However, if the MA were part of a following word, (there are such words) the first arrangement, and not the second, would be the only arrangement that yields 'correct' Northern Thai.

Justification seems to be made in the Thai manner - inter-word gaps as such are not adjusted, though punctuation spaces are the preferred target for adjustment, and the gaps between characters seem to be adjusted independently of grammatical structure.

So, is word boundary information, e.g. from ZWSP in the backing store, often available for the positioning of dependent glyphs?