This repo contains the code for the paper Comparing Comparatives (Korat, Omer. Corpus Pragmatics (2018). https://doi.org/10.1007/s41701-018-0042-2). The code contains documentation and references to the appropriate parts of the paper. Data was collected from the Corpus of Contemporary American English (Davies, 2008).

To set up, unzip data.zip and set data.dir to the current working directory.

Each row is a numerical expression found in the raw text. Data were collected from the written (not spoken) portion of COCA, which contains around 400 million words.

Column descriptions are as follows:

num Raw numerical value of expression

is_p 1 iff the expression is a percentage (e.g. "50 percent", "50%"), 0 otherwise

is_frac 1 iff the expression is a fraction (e.g. "1/2", "one half"), 0 otherwise

nat_hyp 1 iff the expression satisfies the natural roundness hypothesis described in the paper (not used in the code)

neg 1 iff the expression is negated (e.g. "no more than 3"), 0 otherwise

mt 1 iff the expression contains 'more than', 0 otherwise

al 1 iff the expression contains 'at least', 0 otherwise

lt 1 iff the expression contains 'less than', 0 otherwise

am 1 iff the expression contains 'at most', 0 otherwise

@book{davies2008corpus, title={The corpus of contemporary American English}, author={Davies, Mark}, year={2008}, publisher={BYE, Brigham Young University} }