98 Rosalind A. M. Temple
contribution of the field, if it may be called such,^2 is the accurate description of
patterns of phonetic variation in naturalistic data, on which theoretical constructs
may be built. However, notwithstanding the work of Stuart-Smith, Scobbie and
their collaborators, reported here and elsewhere, and phonetically informed varia-
tionist analyses of, for example, t-glottalisation (e.g. Docherty & Foulkes 2005),
insufficient attention has been paid to the phonetic substance of some major con-
sonantal variables. The present paper focuses on one such variable, perhaps the
most widely studied consonantal variable in English sociolinguistics, whose social
indexicality has been shown to be restricted to relatively few dialects, but which
has garnered so much attention within and beyond sociolinguistics because of its
claimed implications for phonological theory.
The variable deletion of coronal stops in word-final clusters (e.g. stopped pro-
nounced as variably [stɒp ̚t] or [stɒp ̚]) seems to occur in all varieties of English
and has been one of the most studied variables in the variationist sociolinguistics
of the language. It has been used as a diagnostic in debates about the origins of
African American Vernacular English (AAVE) since the late 1960s (e.g. Wolfram
1969) and more recently it has figured prominently in the exploration of cross-
dialectal differences (e.g. Santa Ana 1992; Smith et al. 2009), the acquisition of
variable constraints (e.g. Guy & Boyd 1990; Roberts 1997; Smith et al. 2009) and
particularly the relationship between variation and phonological theory (e.g. Guy
1991; Guy & Boberg 1997; Bermúdez-Otero 2010a and 2010b; Coetzee & Pater
2011). The phonological model most widely applied to the variable has been one
rooted in Lexical Phonology (LP), which characterises (t,d)^3 as an iterative deri-
vational rule that applies variably in the lexical and postlexical phonology. The
analysis is motivated crucially by there being a consistent (statistical) morphologi-
cal constraint on (t,d) whereby monomorphemic forms undergo deletion of the
final consonant considerably more frequently than bimorphemic forms. However,
findings from several recent studies (e.g. Tagliamonte & Temple 2005; Smith et al.
2009; Guy et al. 2008; Hazen 2011) have introduced an element of doubt as to the
role of this particular constraint, thus undermining the LP account of the variable.
Temple (ms) goes a step further in an exploration of some of the theoretical and
methodological issues which arose during the research reported in Tagliamonte
& Temple (2005), arguing that once the morphological constraint is called into- See Celata & Calamai and Stuart-Smith et al., this volume, for brief discussions of the scope
of the term ‘sociophonetics’. - The variable notation will be used here as a shorthand means of referring to both the vari-
able rule which deletes word-final coronal stops in clusters and the set of consonants affected
by that rule.