Why do we need parsed corpora?
- The best way to investigate the syntax of a language is to ask
native speakers for acceptability judgments.
- But sometimes such judgments are not available.
- In such situations, we use corpora (= collections of texts).
- Corpora become more useful when they are annotated with different
types of linguistic information.
- The earliest annotated corpora were annotated with part-of-speech
information (= tagged corpora).
- The corpora we will be focusing on are parsed corpora, where each
sentence is annotated with part-of-speech information as well as with
further (morpho)syntactic information.
- It is possible to study syntactic variation and change using tagged
corpora - up to a point.
- But it is much easier to study these phenomena using parsed corpora.
- Moreover, in the general case, syntactic variation and change aren't
tied to particular lexical items or parts of speech.
- Loss of V2
- Loss of topicalization
- OV --> VO phrase structure
- Rise of do support in questions
- Zero relative clauses versus other types (wh-, that)