03-02-2026, 09:14 AM
Positive autocorrelation
This relative scarcity of short function-word tokens could also help explain parts of the positive autocorrelation of word lengths in the VMS (long-long-short-short instead of long-short-long-short). In normal texts, articles/prepositions/“and” as separate tokens naturally create a long-short alternation. If such function material is instead attached to the following word (consistent with qo as a function marker, and to a lesser extent o), that rhythm disappears: several words in a row become “bloated” or remain “bare” – which yields length clustering.
Especially since in the standard EVA transcription, higher qo-density is associated with measurably stronger word-length clustering (a positive correlation).
This relative scarcity of short function-word tokens could also help explain parts of the positive autocorrelation of word lengths in the VMS (long-long-short-short instead of long-short-long-short). In normal texts, articles/prepositions/“and” as separate tokens naturally create a long-short alternation. If such function material is instead attached to the following word (consistent with qo as a function marker, and to a lesser extent o), that rhythm disappears: several words in a row become “bloated” or remain “bare” – which yields length clustering.
Especially since in the standard EVA transcription, higher qo-density is associated with measurably stronger word-length clustering (a positive correlation).