Skip to content
James Tauber edited this page Jan 30, 2018 · 12 revisions

A catalog of types of ambiguity we face and how we'll deal with them.

This is incomplete and a work in progress. It's a proposal designed to start conversation. Not everything (or even much) is set in stone.

Basic Syncretism

I include here things that can be modelled as one property being underspecified either partially or completely (we might want to further divide that into two categories). Note that partial underspecification could be modelled as complete underspecification by splitting the property into multiple properties (e.g. splitting case into N,A vs G,D or gender into animate vs inanimate)

Examples are:

  • nominative vs accusative neuter [99+50+33+32]
  • the gender of genitive plurals
  • masculine vs feminine in two-termination adjectives
  • feminine vs non-feminine in non-nominative and non-accusative-plural three-termination adjectives

For the first one, I'm leaning towards just using C for the case.

For the others, now see Proposal for Gender Tagging.

Extended Syncretism

I include here things that involve ambiguity between two properties (but still fixed pairs of possibilities)

  • 1st person singular vs 3rd person plural
  • 1st declension genitive singular vs accusative plural [30+3]

Related to this is the "double" syncretism between accusative singular masculine and neuter on the one hand and nominative and accusative singular neuter on the other hand. If we model the latter as CSN then we've lost the former (which, if by itself could be modelled as ASY). So, in a sense CSN and ASY are syncretic (but also share an overlapping cell).

Non-Examples

  • middle vs passive (this shouldn't be modelled as ambiguity; we should just model the form not the semantics)

Cross-Overs

  • neuter adjectives functioning as nouns
  • neuter adjectives functioning as adverbs
  • verbs fossilised as particles

The main issue with the second and third is, even if we eliminate part-of-speech, do we tag the inflectional properties or not?

Indeclinables

These go away if we just mark these indeclinable.

  • numbers

Indeclinable Part-of-Speech Classification

These go away with a morphological part-of-speech system.

  • adverb vs preposition [23]
  • conjunction vs adverb [15]
  • adverb vs particle [4]
  • conjunction vs particle [4]
  • conjunction vs preposition [4]
  • interjection vs particle [3]
  • conjunction vs adverb vs particle [3]
  • conjunction vs adverb vs preposition [1]