In 1972 Aravind Joshi created an unpublished manuscript titled “Factorization of Verbs: An Analysis of Verbs of Seeing” that examined decomposing the semantic meaning of around 60 verbs of seeing into distinct factors that represented basic units of meaning. These factors could then be composed together to build up a frame that represented the meaning of a verb. In 2015 James Pustejovsky and Joshi published a paper “Lexical Factorization and Syntactic Behavior” that examined Joshi’s factorization framework in light of more recent distributional semantics, using Pointwise Mutual Information (PMI) to establish and explore the notion of Factor Expression Likelihood (FEL) – how likely a specific factor in a verb’s frame is to be expressed syntactically in a sentence. This work uses a corpus tool, Sketch Engine, and a large web corpus of English, enTenTen 2012, to examine Joshi’s factorizations from a distributional standpoint. I seek to find if and how individual factors are expressed syntactically through the use of modifiers, and which patterns emerge that can be linked to the insertion or removal of certain factors. I then modify Joshi’s original factorizations in light of the data. This is done to motivate the creation of a metric for the FEL beyond just PMI – and since I have confirmed that specific factors do link to certain modifiers, it stands to reason that if one knows a verb’s factorization, one should be able to calculate how likely a modifier is to occur with that verb. This should also be extendable to calculate the likelihood of any syntactic realization of any factors of a verb’s factorization.
- What You Don’t See is What You Get: Lexical Factorization of Verbs of Seeing
- Todd Curcuru
- James Pustejovsky (Advisor)
- Brandeis University, Graduate School of Arts and Sciences; Master of Arts (MA)
- Master of Arts (MA), Brandeis University, Graduate School of Arts and Sciences
- Brandeis University
- Brandeis University, Graduate School of Arts and Sciences
- 10192/31073; 9923879925501921
- Copyright by Todd Curcuru 2015
- Interdepartmental Program in Linguistics and Computational Linguistics
- English
- Thesis