Search: add option to ignore marginal/loan phonemes

[defseg]

Wutung has only one seemingly native word with /k/, kacengceng 'cicada', and no other words with velars. So /k/ would be marked marginal in an IPHON inventory for Wutung: (k).

Northwest Mekeo has /t l/ only in loans, e.g. /tauli/ < Eng. towel, /loβiaŋa/ < Iafia or Lapeka loviana. No coronal consonants occur in native vocabulary. So /t l/ would be marked as loan phonemes in an IPHON inventory for Northwest Mekeo: {t} {l}.

In SQL, the doculect_segments table has boolean marginal and loan columns to represent whether a phoneme is marginal or only present in loans. (marginal and loan both being true means the phoneme is marginal in native vocabulary but has been reinforced by loans.)

Since the properties of core phoneme inventories (i.e. phoneme inventories excluding marginal and loan segments) are relevant to the study of universals - see Blevins's paper on Mekeo for an example - it should be possible to exclude marginal and/or loan phonemes from search.

[defseg]

What format should this have in the query language?

• Should there be a binary switch applying to the whole search, or should it be possible to specify for each inventory search term whether it should include or exclude marginal/loan phonemes?

There may be use cases where it would be useful to have granular control over inclusion or exclusion of marginal/loan phonemes. I can't foresee them, but it's better to be more featureful than less - we shouldn't disallow granular control in the syntax. So control should be granular.

• What should the default behavior be?

• What should the syntax look like?