Towards a linguistically motivated computational grammar for Hebrew

What it is about

Wintner writes a small HPSG grammar of the Modern Hebrew noun phrase, the first HPSG analysis of a Semitic language, implemented in the AMALIA framework. It is a deliberately narrow, work-in-progress grammar, tested on hand-built examples and not on a corpus. The central claim is that ha- is an affix on words, not phrases, so definiteness is a boolean head feature (DEF) set by a lexical rule that applies only to words with DEF negative. Construct-state (cs) nouns and adjectives are left unspecified for DEF and inherit definiteness from their complement, which gives two definiteness combinations for constructs against four for free genitives. The paper argues against the DP hypothesis for Hebrew: possessives are complements, not specifiers, and a DEP feature forces prosodically weak words (construct forms, prepositions, some quantifiers) to combine with their obligatory complement before any modifier.

Ideas

• I1. Make DEF a lexical toggle with the paper's rule: the definite rule applies only to bare nominal words, so construct forms cannot become definite directly. A check on the tokens: ha- should never attach to a construct noun. Use it in Nouns, adjectives, pronouns and numerals and the construct-state notes in The Noun, the Adjective and the Numeral.
• I2. A DEP-like flag on words that must attach to their complement first, applied as a deterministic attachment rule before modifiers are attached. This is a rule for the construct noun in Syntax: the clause and the sentence and for 7. Syntax.
• I3. Possessive suffixes as complements of the noun rather than determiners. The paper's examples (kol sefer sheli, "all my book" with the possessive after the noun) become a small test set against the labels in treebanks/he.

For Mova

I1 goes into Nouns, adjectives, pronouns and numerals and The Noun, the Adjective and the Numeral. I2 belongs in Syntax: the clause and the sentence, with a parallel note in 7. Syntax. I3 is a test on treebanks/he for the labels in 11. Modern Hebrew in Universal Dependencies. The grammar is HPSG and not used by Mova; only the analyses are candidates.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1621753.1621768

@inproceedings{wintner1998linguistically,
  title = {{Towards a linguistically motivated computational grammar for Hebrew}},
  author = {Wintner, Shuly},
  booktitle = {Computational Approaches to Semitic Languages},
  year = {1998},
  doi = {10.3115/1621753.1621768},
  url = {https://doi.org/10.3115/1621753.1621768}
}

References

2 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Steven Abney. 1987. The English Noun Phrase in Its Sentential Aspect. Ph.D. thesis, Massachusettes Institute of Technology, Cambridge. find
2. Esther Bentur, Aviella Angel, and Danit Segev. 1992. Computerized analysis of Hebrew words. Hebrew Linguistics, 36:33-38, December. (in Hebrew). find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-1998-wintner-linguistically-motivated-computational-grammar-hebrew/