“ Me and her ” meets “ he and I ” : Case , person , and linear ordering in English coordinated pronouns

Some accounts of English pronoun case in coordination (Emonds 1986, Sobin 1997) have argued that the observed data on the phenomenon are the result of a naturally acquired grammar in which coordinated pronouns are universally accusative (Emonds’ N[ormal] U[sage]), pitted against an explicitly learned system in which coordinated pronouns are prescribed to be nominative in syntactic environments where a non-coordinated pronoun would be nominative (Emonds’ P[restige] U[sage]). The conflict between the natural NU grammar and the allegedly unnatural PU grammar results in a peculiar case distribution in which case interacts not only with syntactic function (and imperfectly so), but also with conjunct ordering (1st conjunct vs. 2nd conjunct) and pronoun agreement features (1sg vs. 3sg), often in ways that are difficult to explain. In this paper, the results of a sentence judgment experiment show that EVEN WHEN SPEAKERS CONFORM TO PU TO A HIGH DEGREE, they impose two additional constraints when coordinating a pronoun with a full-NP, one that limits 1sg nominative to second-conjunct position (the “X and I” constraint), and one that limits 3sg nominative to first-conjunct position (the “s/he and X” constraint). A survey of advice literature on English usage suggests that the “X and I” constraint is likely the result of the way in which PU is taught. As for the “s/he and X” constraint, a largescale corpus study suggests that (a) this constraint is systematically violated when X = you, and (b) both the constraint and the violation fall out of a more general system of tendencies that governs the ordering of pronouns in coordination. According to the proposed system, when a pronoun is coordinated with a full-NP, the pronoun will tend to be ordered first, and when a pronoun is coordinated with another pronoun (other than 1sg nominative), they will tend to be ordered according to person, in the order 2 < 1 < 3. Finally, the corpus data show that in both subject and object position, 1sg coordinated pronouns are more likely to be nominative than 3sg pronouns. I correlate this with the finding that, regardless of case, 1sg coordinated pronouns are more frequent than 3sg coordinated pronouns, and argue that in subject position, greater frequency entails greater susceptibility to prescriptive pressure, thus leading to more nominative 1sg. Likewise, as suggested by Boyland (2001), increased frequency in subject position leads to increased overextension into object position. This analysis makes empirical predictions about the range of possible case-marking patterns employed by individual speakers, and these predictions are supported by preliminary evidence. In addition to improving on our understanding of NU-PU interaction, this analysis (a) provides a first step toward a typology of pronoun ordering tendencies cross-linguistically, and (b) advances additional evidence for the role of frequency in driving pattern extension.