Embedded appositives (release 2)

Christopher Potts and Jesse Aron Harris

In earlier work, Potts claims that certain kinds of appositives are always interpreted as speaker-oriented. This claim has since been challenged persuasively. This data set addresses the question of how frequent such non-speaker-oriented readings are in discourse, and it should provide some clues as to what conditions favor speaker-orientation and what conditions disfavor it.

We began with 177 million words of novels, newspaper articles, and TV transcripts. With a simple regular expression search, we found 278 examples of appositives syntactically embedded inside the complements to attitude verbs. We went through these examples by hand, developing, where possible, textual arguments for what the intended appositive interpretation was: text-level or embedded. These arguments were then assessed by two independent annotators

Optionally restrict to an embedding level*
Text Embedded 		Optionally specify the embedding predicate**
add allege announce argue
ascertain ask believe call on
claim doubt find out hope
make sure mention quote remark
report say suggest sure
tell think understand

*The judgments are supported by textual arguments that have been independently assessed by two assessors. Multiple choices return the union of the categories.

**Multiple choices return the union of the categories. No choice means complete freedom on the embedding predicate.

Results

Creative Commons License

This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 License. Last update: 2009-03-29