StringNet as a Computational Resource for Discovering and Investigating Linguistic Constructions

We describe and motivate the design of a lexico-grammatical knowledgebase called StringNet and illustrate its significance for research into constructional phenomena in English. StringNet consists of a massive archive of what we call hybrid n-grams. Unlike traditional n-grams, hybrid n-grams can consist of any co-occurring combination of POS tags, lexemes, and specific word forms. Further, we detect and represent superordinate and subordinate relations among hybrid n-grams by cross-indexing, allowing the navigation of StringNet through these hierarchies, from specific fixed expressions (“It’s the thought that counts”) up to their hosting protoconstructions (e.g. the It Cleft construction: “it’s the [noun] that [verb]”). StringNet supports discovery of grammatical dependencies (e.g., subject-verb agreement) in noncanonical configurations as well as lexical dependencies (e.g., adjective/noun collocations specific to families of constructions).