Indefinite Determiners

Problem:

However, the class indefinite attributive pronouns (PIAT) or indefinite determiners was very heterogeneous in distribution now, especially in respect to articles in the surrounding, as the following examples illustrate:

• manch ein Mensch
• die beiden Verwandten
• ein jeder Mensch
• jeder Mensch
• ich habe wenig gelacht
• ich habe nur ein wenig gegessen
• ich habe etwas gesehen
• ein bißchen Schokolade
• er sah etwas gequält aus

Tagging practice IMS/TUE:

As described in B.2

Tagging practice STTS:

We decided to distinguish the group into indefinite determiners that can occur together with another (non-indefinite) determiner (an article, a demonstrative or possessive pronoun), and into those, that must occur on their own. This property should be lemma-based, i.e. only depending on the ability of the given indefinite determiner to combine with a determiner (independent of the question if there is a determiner in the actual context). Adverbial readings should be tagged as adverbs, not as indefinite pronouns.

• solch/PIDAT ein Mensch
• in keiner/PIAT Form
• er sah etwas/ADV gequält aus
• etwas/PIAT Schokolade

Tests:

I, II

1. Granularity changes: put PIDAT and PIAT into one class (as is in ELM-DE).
2. Change in definition: context depencency of PIAT vs. PIDAT as compared to lemma based disambiguation (as it is now).