External Allomorphy as Emergence of the Unmarked

Lexical Classes in Phonology

“…Our claim is that there is: In English, the s-plural This avoids both construction-specific rules and unenlightening listing of allomorph environments and follows the OT analysis of allomorphy and lexical selection in many earlier studies, beginning with Mester 1994 for Latin (see Anttila 1997, Burzio 1994, Kager 1996, Mascaró 1996a, Perlmutter 1998, Russell 1995, Tranel 1996a. This leads to a TETU-effect for such listed pairs of allomorphs (Mascaró 1996b)--here, postnasal voicing in tandem with default voicelessness--in thoroughly foreign territory, where IDENT-F otherwise prevents any changes in voicing. Similar TETU-effects with affixal elements in Japanese arise with Sino-Japanese counters (Ito and Mester 2003:138-141) and verbal suffixes (Ito and Mester 2004).…”

Optimizing by accident: A/an allomorphy and glottal stop

“…(2) Mascaró (1996) {a,an} book ONSET NO-CODA {a,an} egg ONSET NO-CODA  a.book * * a.egg **! * an.book * **!…”

“…In some dialects (including my own), Syllabic Nasal Formation applies only to morpheme-internal /əN/ sequences, producing the contrasts in (20) Mascaró's (1996) prediction in (10)-that an should be selected iff its /n/ surfaces as an onset-fails to be borne out. But this finding does not automatically doom the POA-explaining approach endorsed by Mascaró (1996) and others. Is there a way to reanalyze a/an that would preserve the spirit of (1a) without making the problematic stipulation that an must provide an onset?…”

“…' Mascaró (2007:716) Some other well-known cases of POA from the literature are listed in (7); note that all of these alternations seem to conspire to yield CV syllables, similar to English a/an. 3 (7) More examples of POA (Mascaró 1996(Mascaró , 2007Lee 2009 On the other hand, non-optimizing and anti-optimizing allomorphy are also attested; two well-known examples are given in (8): (8) a. Tzeltal perfective: /-oh/ after monosyllabic stems, /-ɛh/ after polysyllabic stems (Mascaró 2007:715-716) b. Haitian Creole definite: /-a/ after V, /-la/ after C (Embick 2010:125-129;Bonet, Lloret & Mascaró 2006) ) To account for (8)a and other non-optimizing alternations, Mascaró (2007) argues that there are two kinds of allomorphy, internal and external, with only the latter driven by phonological markedness constraints. Paster (2015:237-240), however, points out various complications that result from this addition to the grammar.…”

“…For the sake of concreteness, let's consider two concrete analyses of a/an, one based on assumption (1a) (POA explained in the grammar) and one based on assumption (1b) (POA by accident). Mascaró (1996) offers an analysis in the spirit of (1a), in which a and an are equally faithful candidates and the choice between them comes down to syllable markedness constraints-a classic emergence-of-the-unmarked (TETU) effect:…”