Third Mediterranean Meeting on Morphology (MMM3)

Abstract

Iggy Roca (University of Essex) and Elena Felíu (Universidad Autónoma de Madrid)
Morphology and Prosody in Spanish Word Truncation

Prosodic Morphology (McCarthy & Prince 1986 et seq) has been the object of special attention in Spanish throughout the 1990s: Crowhurst (1992), Prieto (1992a, b), Harris (1994), Lipsky (1995), Colina (1996), Piñeros (1998, 2000a, b). In particular, (hypochoristic) truncates (TEREsa) have been argued to target a syllabic trochee, viewed as the Spanish minimal word by transitivity from syllabic foot binarity: [(té.re)]. The two types of truncates discussed have been analysed as the outcome of the two ranking alternatives below, both assuming among other constraint relations a dominant I-CONTIGUITY and a dominated MAX(SF-TF) (Piñeros 2000b):

1. ANCHOR(SF-TF)L >> HEAD(PrWd)MAX: TEREsa, MARGArita, DANIel, CONCEpción
2. HEAD(PrWd)MAX >> ANCHOR(SF-TF)L: crisTINA, magdaLENA, alFONSO, alFREDO

Benua (1995, 1997) highlights the theoretical repercussions of extending Correspondence Theory from reduplication (McCarthy & Prince 1993, 1995) to truncation:

1. further evidence for output-to-output correspondence (McCarthy & Prince 1995)
2. further evidence for emergence of the unmarked (McCarthy & Prince 1994)

Emergence of the unmarked in Spanish truncates would concern:

1. the targetting of a bisyllabic trochee
2. structural simplifications (no [p] in CONCEpción); no [r] in possible alFrEDO)
3. segmental changes (alPONCHO as an alternative to alFONSO)

Evidence for output-to-output correspondence would concern the extraction of the truncate from a surface base, rather than from an underlying representation. A number of questions with regard to emergence of the unmarked arise from the data given:

• why are there two truncation patterns, rather than just one (the least marked)?
• why are the two constraints involved in the reranking not homogeneous?
• why are segmental changes heavily skewed towards type B) truncation?

Further data give rise to further questions:

• hiPÓLitO: why does type B) truncation necessitate ANCHOR(SF-TF)R over I-CONTIGUITY?
• FederICO: why can ANCHOR(SF-TF)L intervene in type B) truncation, also over I-CONTIGUITY?
• ramONA: why can the first onset be omitted?
• FRANcisco: why can the target foot be just bimoraic (rather than fully disyllabic)?
• FLOrencia: why can the target foot be monomoraic?
• feDERIco: why is the truncate's foot neither left-anchored nor the head foot in the base?

The (non-hypochoristic) trisyllabic truncation pattern examined in Felíu (2000) also fails to match the optimal disyllabic prosodic word, by definition: ECOLOgista, PROLETArio, etc. Why should this pattern exist at all, therefore?

In addressing these questions, we will bring morphology into the picture, in an attempt to account for the whole range of truncation patterns. In particular, the task faced by the language (via the speaker) is that of creating new words from existing ones. Clearly, the existing base needs to be distorted phonologically in some way if the goal is to be achieved. The output must obviously be consistent with the phonological patterns of Spanish (magdalena cannot become mgdlnaaea, for instance, with Cs left-aligned and Vs right-aligned), and in preference with those which are less marked, by pressure from (a generalised) Occam's razor. This minimalist phonological target is formally achieved in OT by means of constraint ranking. Because the unmarked is thus favoured, it will often emerge, but not necessarily across the board, as the data confirm. Attainment of canonical morphological structure also plays an active role in Spanish truncation, as attested by the provision of a desinence and its shape (concha < concepción; pura, chona < purificación; suso < jesús; etc.).