Introduction: Previous studies have found four different pitch configurations (i.e. continuation rises) in non-final clauses in Spanish (Estebas-Vilaplana et al., 2015). One of these patterns –which in the Spanish tradition is called patrón suspendido ‘suspended pattern’ (Navarro Tomás, 1918)– is the most frequent; it is realized as a sustained pitch and is also present in other Romance varieties (Frota et al., 2007; Feldhausen, 2010).

Aim: This work aims to provide an explanation by assessing the impact of several sociolinguistic variables that can serve to predict the choice of one pattern over the others. The sociolinguistic variables are dialect and gender.

Methodology: The study examines 535 continuation rises produced in elliptical clauses by 14 young educated speakers from 4 dialects of Peninsular Spanish (Castilian, Cantabrian, Andalusian and Spanish spoken in Catalonia). We used elliptical clauses because this type of sentences can only be produced with a continuation rise intonation. Speakers identified themselves either as male or female. Data were obtained by means of a Discourse Completion Test (Vanrell et al., 2018). Data were annotated in Sp_ToBI (Hualde & Prieto, 2015) using an automatic transcriber (Elvira-García et al. 2016). Two statistical analyses were carried out. The first analysis analyzes the distribution of each contour across the different dialects, sociolects and genders by estimating mixed effects logistic regressions. The second analysis uses the range of the last movement as the dependent variable and the sociolinguistic variables as fixed factors. In both analysis speaker and item are set as random variables. The statistical results reported here are the likelihood ratio tests for the models created.

Results: The data show that 5 intonational contours (instead of 4) are possible in elliptical clauses: H H%, L H%, L+H H%, L+H !HH%, H !H% (see Figure 1 for examples of two patterns). However, there are differences in the frequency distribution of the patterns by gender (χ2= 65.2, p<0.001), insofar as women tend to use more frequently suspended patterns (H H%, H H!%) than male speakers, and by dialect (χ2= 163.19, p<0.001). Moreover, dialect plays a role in the phonetic realization of the rises (χ2= 9.8904, p=0.01): while in some dialects H H% surfaces as a high plateau (Figure 2 left), in others it is realized as a rise from a high level in the last syllable (Figure 2 right).

Conclusions: The variation attested can be explained by the coexistence of “dialectal” or traditional patterns and more “standard” patterns. The same situation has been previously documented in y/n questions, which “standard” pattern is L* H%, but they are also produced with different vernacular nuclear configurations (Henriksen, 2012; López-Bobo & Cuevas, 2010). In addition, in the same contexts, women use sustained pitch more often than men. These results are in line with previous research showing that women tend to use less assertive language forms (Lakoff, 1973), given that sustained pitch has been described as useful for presenting apologies and doubt in Spanish (Hidalgo, 1998).