speech recognition

subsentence = sentence | prenex subsentence

prenex = terms ZOhU #

sentence = (FA KOhA) ... bridi-tail

bridi-tail = selbri tail-terms | gek-sentence

selbri = tag [[jbocre: NA [jbocre: tag|tag]] ... selbri-6

For each sentence, if the bridi-tail is selbri tail-terms:

(FA KOhA) ... tag [[jbocre: NA [jbocre: tag|tag]] ... selbri-6 tail-terms: reduces to tag zo'u [[jbocre: NA ku zo'u [jbocre: tag zo'u|tag zo'u]] ... (FA KOhA) ... selbri-6 tail-terms

If the bridi-tail is gek-sentence:

(FA KOhA) ... gek prenex ... sentence gik prenex ... sentence tail-terms: reduces to gek prenex ... (FA KOhA) sentence gik prenex ... (FA KOhA) sentence tail-terms

Finally, the tail-terms of each sentence are reduced in the same way as the pre-selbri terms have been before: starting from the first one, tagged terms and negations go directly to the prenex, KOhA terms go to the prenex and leave a KOhA behind, and connected terms and termsets expand into a gek-sentence.

To tidy up, we now split every multiple term prenex into single term ones:

term ... ZOhU subsentence: (term zohu) ... subsentence

and we move all pre-selbri terms behind the selbri:

([[jbocre: [jbocre: FA|FA]] KOhA) ... CU selbri FA KOhA] ... /VAU#/: reduces to selbri FA KOhA] ... FA KOhA] ... /VAU#/

text has been reduced to:

subsentence = selbri-6 FA KOhA] ... /VAU#/

| gek subsentence gik subsentence

| NA KU # ZOhU subsentence

| tag /KU#/ ZOhU subsentence

| tag quantifier KOhA xi number relative-clauses ZOhU subsentence

(Where selbri-6 is an untagged/unnegated selbri, KOhA is one of da, de, di when quantifier is present, else it is one of ko'a, ko'e, ... fo'u.)