Transcending the lexical vs. grammatical divide regarding the mass/count distinction. Evidence from corpus studies and acceptability surveys in French and Dutch

- The lexicalist and grammatical approach fail to account for all properties of the mass/count distinction.
- The mass/count distinction is probabilistic rather than categorical.
- Acceptability ratings of mass/count preferences are related to corpus frequencies.

The lexico-grammatical status of mass and count nouns has been extensively debated over the last decades. We discuss how the two main approaches (lexicalist and grammatical approach) both fail to account for all the properties of the distinction. An alternative lexicalist view is proposed, approaching mass and count features as probabilistic features rather than as categorical ones. Building on Allan (1980), we deployed corpus data and acceptability ratings to examine the mass-count preferences of French and Dutch nouns that allow for both a mass and count interpretation. Our findings indicate that acceptability ratings are related to corpus frequencies. More specifically, we found that attested mass usages (even with very low frequencies) of nouns are more acceptable than nouns with unattested mass usage, that more frequent mass usages are more acceptable than less frequent ones and that the degree of acceptability of a mass usage of a lexical item depends on its relative frequency in discourse. In addition, we found that acceptability ratings of mass usage are sensitive to semantic-pragmatic modulation of the sentence context.