You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Knowing the form of the verb is important for detecting clause-level constructions like passives
Knowing the difference between numbers and other N/D words can be useful, and the distinction is made in PTB and UD (including the nummod relation distinct from det)
Rather than implement a full set of morphosyntactic features, an incremental step is to import fine-grained POS tags. UD has XPOS tags which for English follow the PTB tagset.
Can go directly in .cgel file with :xpos feature, alongside the existing lexical categories
For now, limit to CD (number), MD (modal aux), and the following for all other verbs: VB (plain form), VBD (preterite), VBG (gerund-participial form), VBN (past participle), VBP (present tense except 3sg), VBZ (present 3sg)
For verbs, CGEL makes some finer-grained distinctions, like modal present vs. preterite, and incorporates negation clitics. And the tags collapse some of the paradigm cells for be. But these further distinctions are groupings of closed-class lexemes, so it isn't strictly necessary to have finer-grained tags.
Importing all XPOS tags might cause confusion as there are significant boundary differences for categories like P and D in CGEL versus other schemes (as we explored here)
The text was updated successfully, but these errors were encountered:
nummod
relation distinct fromdet
):xpos
feature, alongside the existing lexical categoriesCD
(number),MD
(modal aux), and the following for all other verbs:VB
(plain form),VBD
(preterite),VBG
(gerund-participial form),VBN
(past participle),VBP
(present tense except 3sg),VBZ
(present 3sg)The text was updated successfully, but these errors were encountered: