Open LinguList opened 3 years ago
The name for the sound inventory is now Language.sound_inventory
, so adding a Language.grapheme_inventory
is trivial, and would take the segments
and not the tokens
of a language.
The wordlist class computes occurrences of sounds for both segmented BIPA sounds and for segmented non-bipa sounds, so this is really parallel, the only difference is that a list of segmented graphemes does not allow access to feature data in CLTS.
Graphemes require valid Segments, but Segments are not always valid BIPA, so we should distinguish a grapheme_inventory (which shows only occurrences) and a Sound inventory (which shows the occurrences for bipa-normalized segments).
Methods differ, as Grapheme-inventories can only be compared by Jaccard.