TRANSCRIPTION OF INTONATION OF JEREZANO ANDALUSIAN SPANISH UNA TRANSCRIPCIÓN DE LA ENTONACIÓN DEL DIALECTO JEREZANO DE ANDALUCÍA, ESPAÑA NICHOLAS C. HENRIKSEN University of Michigan [email protected] LORENZO J. GARCÍA-AMAYA University of Michigan [email protected] Artículo recibido el día: 13/02/2012 Artículo aceptado definitivamente el día: 25/06/2012 Estudios de Fonética Experimental, ISSN 1575-5533, XXI, 2012, pp. 109-162 Transcription of intonation of Jerezano Andalusian Spanish 111 ABSTRACT Dialect variation in intonation is a widely attested phenomenon in the cross- linguistic literature (see Warren 2005a) and also in the literature on Spanish (Prieto & Roseano 2010, Sosa 1999). Prieto and Roseano (2010) provide an edited volume on the transcription of Spanish intonation for ten different dialects of the language. Two of these chapters characterize the intonation of Spanish spoken in northern and central Spain, but no information is offered on varieties spoken in the southern Andalusia area. This paper is designed to fill this gap in the literature as it investigates the acoustic properties of a series of F0 contours produced in an intonation survey by nine speakers from Jerez de la Frontera, a coastal city located in the southwest province of Cádiz, Spain. Speech data were analyzed in Praat (Boersma & Weenink 2011) following the original Sp_ToBI transcription system (Beckman, Díaz-Campos, McGory, & Morgan 2002) and the revised system put forth in Estebas-Vilaplana and Prieto (2008). Results show that neutral and biased statements were communicated by the L* L% vs. L+H* L% contrast, respectively. Confirmation questions show a variety of patterns based on the speaker’s belief of the proposition in question. Two configurations are possible for information- seeking wh-questions, H+L* L% and L+¡H* L%. Commands are the only prag- matic intent for which the M% boundary tone is used, and vocatives are characterized by L+H* HL% in neutral and marked intents. Although Jerezano Andalusian speakers share intonational characteristics with other varieties of European and American Spanish, there are important differences that motivate a thorough investigation of this language variety. Finally, dialect comparisons are provided to achieve a more comprehensive account of intonational variation in Spanish. Keywords: intonation, Sp_ToBI, Jerezano Andalusian Spanish, dialect variation. RESUMEN La variación dialectal de la entonación es un fenómeno ampliamente corroborado en la literatura especializada en el análisis lingüístico de las lenguas (ver Warren 2005a) así como en múltiples publicaciones sobre el español (Prieto & Roseano 2010, Sosa 1999). Prieto y Roseano (2010) ofrecen un volumen editado sobre la transcripción de la entonación española de diez dialectos diferentes. Aunque dos de estos capítulos especifican la entonación del español peninsular en las zonas norte y central, no se aporta ninguna información sobre las variedades habladas en el sur de Andalucía. Este proyecto está diseñado para suplir esta carencia y de este modo, EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 112 Nicholas C. Henriksen & Lorenzo J. García-Amaya ofrecer una investigación sobre las propiedades acústicas de una serie de contornos F0 producidos en una encuesta entonativa realizada por 9 hablantes de Jerez de la Frontera, ciudad localizada al suroeste de la provincia de Cádiz, España. Los datos de producción oral fueron analizados con Praat (Boersma & Weenink 2011) siguiendo tanto el sistema de transcripción para Sp_ToBI (Beckman, Díaz- Campos, McGory, & Morgan 2002), como el sistema revisado de Estebas- Vilaplana y Prieto (2008). Los resultados muestran que las afirmaciones neutrales y no neutrales se produjeron mediante el contraste L* L% vs. L+H* L% respec- tivamente. En cuanto a las preguntas de confirmación, éstas muestran una variedad de patrones basados en la creencia que el hablante tiene de la pregunta en cuestión. Sobre las preguntas QU (o pronominales), dos configuraciones resultaron posibles: H+L* L% y L+¡H* L%. Las órdenes resultaron ser la única intención pragmática para las que se usa la frontera tonal M%, y por último, los vocativos se produjeron mediante L+H* HL% en intenciones de habla neutrales y marcadas. Aunque los hablantes del acento jerezano del andaluz comparten características entonativas con otras variedades del español europeo y americano, hay diferencias importantes que motivan una investigación profunda de esta variedad. Por último, se proporcionan comparaciones dialectales con la intención de conseguir una explicación más completa de la variación entonativa del español. Palabras clave: entonación, Sp_ToBI, jerezano, español andaluz, variación dialectal 1. INTRODUCTION The goal of this paper is to present findings on the basic intonational tunes used in Jerezano Andalusian Spanish within the Sp_ToBI system of intonational labeling (Beckman et al. 2002, Estebas-Vilaplana & Prieto 2008). The Sp_ToBI system is based on the Autosegmental-metrical (AM) approach to intonational phonology (Goldsmith 1979, Gussenhoven 2004, Ladd 2008, Pierrehumbert 1980, Pierrehumbert & Beckham 1988). The AM framework posits intonational structure by means of two tones, L(ow) and H(igh), which may associate with metrically strong (i.e., stressed) syllables or utterances edges. Tones that associate with metrically strong syllables are called pitch accents, and their main function is to enhance the prominence of these syllables. Pitch accents may be further separated into two subtypes: nuclear pitch accents, which associate with the last stressed syllable of the utterance; and prenuclear pitch accents, which associate with all prior stressed syllables. Tones that associate with phrase edges can be of two types, boundary tones or phrase accents, and their main function is to delimit or mark EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 Transcription of intonation of Jerezano Andalusian Spanish 113 utterance boundaries (initial or final). As for labeling procedures, pitch accents are marked with a star diacritic (*) to indicate their association with metrically strong syllables, and boundary tones are marked with a percentage diacritic (%) to indicate their alignment with phrase edges. AM analyses of Spanish intonation are common in the literature (Beckman et al. 2002, Face 2002, Face & Prieto 2006/7, Henriksen 2012, in press, Hualde 2002, Sosa 1999). In terms of the characterization of Andalusian Spanish at the segmental level, previous works have focused on syllable-final /s/-aspiration, merger of the /s/-/θ/ distinction, velarization of /n/, the fricative pronunciation of /č/, and weakening of trill /r/, to name a few relevant phenomena (see Alvar 1996:233-258 for general overview; see also Henriksen & Willis 2010). The fact that Andalusian Spanish would not exhibit variation at the intonational level is almost unexpected, given what has been reported for other Spanish dialects known to undergo processes of phonological innovation. We also know that that there is much inter-dialectal variation in Spanish intonation, so a reasonable hypothesis is that not all intonational patterns documented for speakers of Madrid Castilian Spanish (e.g., Estebas-Vilaplana & Prieto 2010) extend to the southern part of the peninsula. This paper is designed to fill this gap in the literature as it investigates the acoustic properties of F0 contours produced by nine speakers from Jerez de la Frontera, a coastal city located in the province of Cádiz, Spain1. There is much debate on the historic motivations leading to the modern differences between the Andalusian and Castilian varieties of Peninsular Spanish. Researchers point to the heavy Arab/Mozarab influence, repopulation from northwestern Spain, and linguistic drift as possible motivating factors for the phonological innovations of Andalusian Spanish (Penny 2001:118). The first Moorish capital in Spain was the Caliphate of Córdoba, lasting from the middle of the eighth century until the beginning of the eleventh century, and this marked the peak of Moorish domination and presence in the Iberian peninsula. This strong Arabic presence in the south may be argued as the basis for innovative Andalusian speech patterns, but the principal phonetic features of Andalusian Spanish appear to derive from rustic Castilian and Leonese dialects brought in by the reconquest (Narbona, Cano, & Morillo 1998:39). In fact, Narbona et al. (41) suggest that many of the linguistic 1 The city of Jerez de la Frontera is located in southwest Spain, with a population of roughly 210,000 inhabitants. It is located in the coastal province of Cádiz, bordered by the Spanish provinces of Huelva, Sevilla, and Málaga, as well as the Atlantic Ocean, the Mediterranean Sea, and the Strait of Gibraltar. Cádiz is considered a ceceo area within the Andalusian Spanish continuum (Alvar 1996:250). EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 114 Nicholas C. Henriksen & Lorenzo J. García-Amaya features of Andalusian Spanish arose in the late Middle Ages, after much of the Moslem population had decreased. The phonetic features that define contemporary Andalusian Spanish developed at different times and in different regions of the southern part of the peninsula. The two most notable categories of phonetic change include merger of sibilants as seseo/ceceo and widespread weakening or loss of syllable- and word-final consonants (e.g., /s/-aspiration or deletion). These processes are commented as early as the writings of Antonio de Nebrija, Juan de Valdés, and Damasio de Frías (Mondéjar 1979). Studies of contemporary Andalusian Spanish phonetics and phonology include Gerfen (2002), Villena Ponsoda (2008), Ruiz-Sanchez (2008), Torreira (2006, 2007), and Parrell (2012). As for work on features of Jerezano Andalusian Spanish, we have data on seseo/ceceo (Carbonero, Álvarez, Casas, & Gutiérrez 1992, García-Amaya 2008) and trill weakening (Henriksen & Willis 2010). Unfortunately, information on the intonation of speakers from the Cádiz province is currently unavailable. In fact, to the best of our knowledge, there are limited findings on the intonational patterns for speakers of Andalusian Spanish generally (cf. Congosto Martín 2011). For this paper, we aim to determine the usefulness of the Sp_ToBI labeling systems, and the revisions of Estebas-Vilaplana and Prieto (2008) in particular, for this undocumented language variety. In the present study, a new contribution to the description of Andalusian Spanish intonation is provided. To this end, we base our analysis on a corpus that includes varied syntactic structures with different pragmatic meanings. This allows for a broad initial approach to Andalusian Spanish intonation and opens questions for future research on more specific issues. A second contribution is to enrich our current knowledge of intonational variation in Spanish. With this objective, we use the findings of the chapters in Prieto and Roseano (2010) as a reference point for many dialectal comparisons. Specifically, the dialects investigated in Prieto and Roseano (2010) are Castilian, Cantabrian, Canarian, Dominican, Puerto Rican, Venezuelan Andean, Ecuadorian Andean, Chilean, Argentinean, and Mexican Spanish. An advantage of the Prieto and Roseano (2010) methodology is that all chapters are based on a common data collection protocol in which informants respond verbally to a series of controlled situations designed to elicit a wide range of intonational contours in a naturalistic setting. In this paper the methodological approach is very similar, and our goal is to put forth principled dialectal comparisons so that a more comprehensive understanding of pan-Hispanic intonation may be achieved. Given these motivations, this paper is structured as follows. Section 2 is a summary of the data elicitation protocol and provides important characteristics EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 Transcription of intonation of Jerezano Andalusian Spanish 115 about the speaker sample. Section 3 introduces the inventory of pitch accents and boundary tones of Andalusian Spanish based on the proposals of Beckman et al. (2002) and Estebas-Vilaplana and Prieto (2008). In Section 4 we provide a description of the intonation patterns in different sentence types (e.g., statements, yes/no questions, wh-questions, etc.) containing various pragmatic meanings such as insistence, disbelief, obviousness, etc. The most noteworthy findings are summarized in Section 5, and we provide a brief cross-dialectal comparison with data found for other varieties of Spanish. Section 6 concludes and offers topics for future research. 2. MATERIAL AND METHODS Six female and three male speakers of Jerezano Andalusian Spanish were recruited for participation in this study. Their ages were between 18 and 64, and the average age was 41. Age-based variation in intonation has been documented elsewhere (Warren 2005b, Warren & Britain 2000), but it was decided to recruit speakers from a broad age sample so that a wide variety of intonational contours could be gathered for this initial attempt to document Jerezano Andalusian Spanish intonation. All speakers were born and raised in Jerez de la Frontera and had lived the majority (or all) of their adult lives in the city when speech data were analyzed. Two speaking tasks were implemented in the methodology. In the first task, nine speakers read aloud test sentences designed to elicit pragmatic contexts for different types of declaratives, interrogatives, exclamations, and vocatives. The intonation survey (i.e., discourse completion task) contained 69 test sentences, for a total 621 productions (69 sentences x 9 speakers) submitted to acoustic analysis. This was a questionnaire based on Prieto and Roseano (2010), developed first in Prieto (2001). Speakers read each pragmatic context in silence prior to reading aloud each test sentence2. The discourse contexts and test sentences are provided in the Appendix. It is important to mention that the questionnaire is centered on common everyday contexts and that the test sentences reflect realistic communicative answers. This is not always possible with standard laboratory practices, which may not take context into account and often use test words or 2 Note that in the Prieto and Roseano (2010) methodology, informants were asked to respond aloud to the pragmatic contexts that were read to them. In the current study, subjects read the pragmatic contexts in silence and then read aloud the responses provided in the survey. EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 116 Nicholas C. Henriksen & Lorenzo J. García-Amaya phrases that are uncommon in everyday speech (see Lickley, Schepman, & Ladd 2005 for discussion). The second task was a follow-up questionnaire administered to examine two types of yes/no questions in further detail: inner negation confirmation yes/no questions, and outer negation confirmation yes/no questions. Following Armstrong (2010), this task was created to gain insight into the intonation of a wider variety of confirmation questions than what is offered in Prieto and Roseano (2010). Speaking procedures were identical in both tasks, although the second task was conducted one year after the first task. For the second task, speech data were elicited from three speakers only: two female speakers and one male speaker from Jerez de la Frontera. Speech data were analyzed using the Praat acoustic analysis software (Boersma & Weenink 2011). The intonation labeling system was the Sp_ToBI system (Beckman et al, 2002) and its revised version (Estebas-Vilaplana & Prieto, 2008) based on the AM approach to intonational phonology (Ladd 2008, among others). 3. JEREZANO ANDALUSIAN SPANISH INTONATIONAL PHONOLOGY In this section we present an inventory of the pitch accents and boundary tones observed in the Jerezano Andalusian Spanish corpus. For each accent, a schematic configuration is provided along with the Sp_ToBI label (Beckman et al. 2002, Estebas-Vilaplana & Prieto 2008). 3.1. The pitch accents Beckman et al.’s (2002) Sp_ToBI system included three bitonal pitch accents for Spanish. Two accents were rising (L*+H with posttonic F0 peak and L+H* with a peak aligned at the end of the tonic syllable), and one was falling (H+L* with an F0 valley within the accented syllable). This proposal is in line with early analyses of Spanish rising accents that make a distinction between early and late rising peaks (Face 2002, Hualde 2003, Sosa 1999). Monotonal H* was also included for cases when no F0 valley was observed prior to the tonic syllable. Estebas-Vilaplana and Prieto (2008) have since put forth a revised Sp_ToBI system. Two labeling innovations are of note. First, the revised Sp_ToBI adheres EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 Transcription of intonation of Jerezano Andalusian Spanish 117 to the three-way distinction in rising accents described in Face and Prieto (2006/7). The analysis of Face and Prieto (2006/7), based on findings of Face (2006), assumes the following three-way contrast: L*+H (low F0 throughout the tonic syllable with a posttonic rise); L+>H* (F0 rise on the tonic syllable with a posttonic peak); and L+H* (F0 rise and peak within the tonic syllable). In the present study we adhere to the three-way distinction proposed in Face and Prieto (2006/7), given the empirical findings on the three accent types in Castilian Spanish in particular (Face 2002, 2004, 2006). A second important innovation in the revised Sp_ToBI system is the introduction of L*, which phonetically contains a low valley throughout the tonic syllable. It is observed most commonly in broad focus statements and information-seeking yes-no questions. Table 1 presents the pitch accents observed in the Jerezano Andalusian Spanish corpus. Based on the present analysis, it remains unclear whether the ¡L+H* configuration exhibited in narrow focus contradiction statements (see section 4.1.2.1) is an allotonic variant of L+H* or H*, thus it is not listed here. The L+¡H* accent is listed in Table 1 due to its productivity in the Jerezano corpus. For discussion of the upstep diacritic, see the Discussion section of this paper. Note also that H* has two phonetic possibilities. For wh-questions, prenuclear accents are high throughout the stressed syllable. For echo yes/no questions, F0 is rising after the prenuclear peak and continues to do so throughout the remainder of the utterance, as documented in Gussenhoven and Rietveld (2000) and Haan (2001) for Dutch3. 3.2. The boundary tones It is standard practice in the AM framework to assume a binary contrast based on monotonal boundary specifications: L% vs. H%. However, Estebas-Vilaplana and Prieto (2008) argue that bitonal boundary tones are also necessary within the labeling system. Their revised system allows for the possibility of bitonal nuclear accents followed by bitonal boundary movements posttonically. Complex boundary tones are documented widely in all chapters of Prieto and Roseano (2010), and these include configurations such as LH%, HL%, HH%, and LM%. 3 Specifically, Haan (2001:112) describes the H*H% melody as: following the high F0 target on the [preceding] accent, pitch remained level for a brief stretch; from this high target, the final rise took off. This is what was observed for yes/no echo questions in the Jerezano corpus. See also Henriksen (2012) for further discussion on the H* accent in Spanish. EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 118 Nicholas C. Henriksen & Lorenzo J. García-Amaya MONOTONAL PITCH ACCENTS L* This accent is phonetically manifested as a low steady tone throughout the tonic syllable, towards the bottom of the speaker’s range. It is found in the nuclear position of broad focus statements and information-seeking yes/no questions. H* This accent is phonetically manifested as a high with no preceding F0 valley at the baseline of the speaker’s tonal range. It is also observed as continuous rising F0 after a L+H accent in echo yes/no questions. BITONAL PITCH ACCENTS L+H* This accent is perceived as a rising pitch mo- vement over the accented syllable, with the peak located at its end. It is commonly found in nuclear position of biased statements. L+¡H* This accent is realized phonetically as a very steep rise to a peak located in the accented syllable. It is used in information-seeking wh-questions. L+>H* Restricted to prenuclear position, this accent is interpreted as a rising pitch movement over the accented syllable; the peak is aligned with the posttonic syllable. It is observed commonly in broad focus statements. L*+H This accent is realized as an F0 valley on the accented syllable with a subsequent rise on the posttonic syllable. It is found in prenuclear po- sition of information-seeking yes/no questions. H+L* This accent is manifested as a descending pitch movement over the accented syllable, low pitch being located right at its end. It is observed in neutral and imperative wh-questions. Table 1. Schematic representations of monotonal and bitonal pitch accents in Jerezano Andalusian Spanish. EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162 Transcription of intonation of Jerezano Andalusian Spanish 119 To exemplify, LH% is realized phonetically as an F0 valley throughout posttonic material followed by a rise. Estebas-Vilaplana and Prieto (2010) present additional evidence on the H% vs. HH% contrast in statements and disjunctive questions in Castilian Spanish. The mid level M% boundary tone was also posited to account for a half-rise or mid level plateau after the L+H* or H* pitch accents. There is cross-linguistic motivation to posit mid tones in sentence-final position, as in Greek (Arvaniti & Baltazani 2005) and German (Grice, D’Imperio, Savino, & Avesani 2005). In Greek, however, the label that is used is !H%, or a downstepped version of the H boundary tone. Beckman et al. (2002) and Estebas-Vilaplana and Prieto (2008) opt for the more transparent M% label in these cases. The use of the M% tone at the right periphery is productive in many of the varieties described in Prieto and Roseano (2010). MONOTONAL BOUNDARY TONES L% L% is manifested as either low pitch near the speaker’s baseline, or as a falling movement. It is observed at the end of statements and neutral and imperative wh-questions. M% M% is perceived as relatively mid-pitch. In the Jerezano corpus this target is reached from a high point and is restricted to commands. BITONAL BOUNDARY TONES HL% HL% is attested in vocatives. The descending movement is typically implemented on an extra- long syllable. HH% HH% is phonetically realized as a sharp rise usually reaching the highest level of a speaker’s range. It is observed in most yes/no question types. Table 2. Schematic representations of monotonal and bitonal boundary tones in Jerezano Andalusian Spanish. Table 2 presents the inventory of boundary tones used in the Jerezano Andalusian Spanish corpus. Jerezano Spanish presents two monotonal tones, L% and M%, and EFE, ISSN 1575-5533, XXI, 2012, pp. 109-162