Erratum to: Nat Lang Linguist Theory (2007) 25:273–313 DOI 10.1007/s11049-006-9017-2

This article has been published with the following errors: on page 278, three instances of ‘−singular’ should be ‘+singular’; on page 307, two instances of the symbol ∩ should be the symbol ∪. These issues are corrected below and should be regarded as the final version by the reader.

At the same time, I would not like to wholly discount the role of phonological similarity in providing a diachronic nudge for the morphologization of (15), and for phonological factors quite generally to provide a force in the grammaticalization of a formal morphological constraint such as (15). For example, consider the fact that amn’t is an impossible form in many dialects of British and North American English (Francis, 1985; Bresnan, 2001). This is arguably due to a synchronic filter banning the feature combination [+copula, +Pres, +neg, +Auth, +PSE, +singular] on a single syntactic node, which is resolved in various ways: in the dialect identified as “Nb 5” in Francis (1985) as isn’t (i.e., via deletion of [+Auth, +PSE]), while in North American English as aren’t (i.e. via deletion of [+singular]). On the other hand, the fact that amn’t is tolerated in the dialect identified as “Nb 1” in Francis (1985), and is even attested in children’s speech (as revealed by a search conducted on CHILDESFootnote 1) suggests that there is no active phonological constraint against this form in English; if there were, it should be rescuable in the phonology, e.g., by epenthesis of a vowel, or by deletion of one of the two coda nasals (cf. autumn∼autumnal). Rather, certain dialects have arguably morphologized a ban on the feature combination [+copula, +Pres, +neg, +Auth, +PSE, +singular] and respond to this morphosyntactic filter through various morphological repairs (e.g., feature deletion). While phonological pressures may play a role in shaping the diachronic development of purely morphological filters, the claim is that the synchronic representation of the ban on *amn’t and the *le lo constraint is morphological in nature: a ban on morphological feature co-occurence.

The example (95) on page 307 should read:

  1. (95)

    Interpretive possibilities for Impersonal pronouns:

    {[+Participant, +Author] ∪ [+Participant, −Author]

    ∪[−Participant, −Author]}