An analyzer is an FST intended primarily to convert
word forms into a [morphological analysis][analysis]. For example,
if you apply the Cree analyzer to the word form ê-wâpamit, it will
return the analysis PV/e+wâpamêw+V+TA+Cnj+Prs+3Sg+1SgO, meaning
the lemma is "wâpamêw", it's a Transitive Animate
Verb, in the Conjunct form, in the Present tense with
a 3rd person singular [actor] and a first-person singular
[object].
A descriptive FST operates on orthographically dubious text. This means that the text may be written in a non-standard orthography, make some common spelling mistakes, and may be lacking diacritics or aspiration ('h') where they would be required in "correct" text. Think descriptive linguistics. In practice, we use descriptive analyzers to accept input and word forms in whatever way people may type it---we allow for some wiggle-room and forgive spelling mistakes, in the hope that the FST will find the most appropriate analysis.
A finite-state transducer is a computational formalism that is useful for defining functions that convert an input string into an output string. In practice, linguists can use this to convert a word form into its [lexeme]+[tags] and back again. See main article: Finite-state transducer.
When referring to "the FST", we (ALTLabbers) tend to mean the source
code (*.lexc, *.xfscript, *.regex, *.twolc files) and resulting
"compiled" data structures (*.hfst, *.hfstol files) we used to
break-down and describe Plains Cree words.
A generator is an FST that converts a
[morphological analysis][analysis] into one or more word forms. For
example, if you apply the Cree generator to the analysis
miskinâhk+N+A+Distr, it will output the word form miskinâhkonâhk.
A lemma is the "dictionary form" of a word. The lemma is usually a possible word form of the word. In the case of Cree non-dependent nouns, the lemma is the singular form of the noun. In the case of dependent nouns, the lemma is the singular first-person possessor form. In the case of VII/VAI/VTI verbs, it's the independent, present tense, third person form of the verb. In the case of VTA verbs, it's the independent, present tense, third person actor/obviative goal form of the verb.
A lemma can be considered the "canonical" form of the word. This is the form of the word produced by the FST, and written as the key in the dictionary XML file.
A normative FST operates on orthographically-correct text. This means, no spelling errors, accents in all the right places, and in the exactly correct letter case. In practice, we use normative generators that, given an [analysis], output a word forms that is perfectly written and spelled-out.
A word form (also spelled wordform) is a specific grammatical realization of a word. For example, the lemma eat has several different word forms including eat, ate, eaten, and eats. Likewise, the Cree word wâpamêw has many possible word forms, including kiwâpamin, kiwâpamatin, and wâpamêw, among many others.