Abstract
Aims and objectives/purpose/research questions:
The main goal of this paper is to analyse how social factors determine the degree of occurrence of intonational features of Basque in Spanish in the Basque Country (i.e. Basque Spanish).
Design/methodology/approach:
We concentrate on information-seeking yes/no questions. In Castilian Spanish, these end in a rising contour, whereas in Basque they end in a rising–falling contour. The data were gathered through sociolinguistic interviews with 12 speakers of Basque Spanish with different linguistic profiles: monolingual Spanish; first language Spanish–second language Basque; and L1 Basque–L2 Spanish.
Data and analysis:
172 information-seeking yes/no interrogatives were obtained from conversational speech. Their final intonational contours were annotated in the Spanish Tone and Break Index model of intonational analysis.
Findings/conclusions:
79% of all information-seeking yes/no questions had final configurations with a rising–falling circumflex contour. Only 21% had the final rising contour of Castilian Spanish. Speakers differed in their frequency of occurrence of falling contours, but the differences did not correlate with the speakers’ linguistic profile (monolingual vs bilingual). Rather, higher percentages of yes/no questions ending in a falling contour were found among speakers who had (a) a higher degree of contact with the Basque ethnolinguistic group, and (b) more positive attitudes towards the Basque language and the Basque ethnolinguistic group.
Originality:
Methodologically, this study is original because the intonational analysis is carried out on natural speech rather than on read or elicited speech. This study is also original from a theoretical point of view because it is the first one to underline the role that subjective factors such as linguistic attitudes play in the adoption of features of a language variety from another contact variety.
Significance/implications:
Our research opens up a path to continue investigating the weight of subjective social factors such as linguistic attitudes in explaining the variation in the influence of one language variety over another.
Introduction
The present study is part of a wider research project that seeks to determine whether there are intonational features of Basque in the Spanish varieties of areas where the two languages are in contact, and to observe to what extent social factors and the degree of contact with Basque can determine the degree of linguistic convergence between these two languages. Unlike other studies in which prosodic convergence is shown in language contact situations, our study seeks not only to prove prosodic convergence in Spanish in contact with Basque, but also to explore what factors may explain to a greater extent the degree of convergence. In this article, we defend the idea that there are subjective factors, such as attitudinal factors, that may explain why speakers who are in contact with several languages end up converging on common features to a higher or lower degree.
Prosodic aspects of language contact have received increasing attention recently, showing that intonational features can be transferred between two languages in contact. A representative list could include Queen (2001), Elordieta (2003a), Colantoni and Gurlekian (2004), Mennen (2004), Elordieta and Calleja (2005), O’Rourke (2005, 2012), Alvord (2006), Romera et al. (2008, 2009), Simonet (2008, 2010, 2011), Benet et al. (2011), Colantoni (2011), Gabriel et al. (2011), Meisenburg (2011), Rakow and Lleó (2011), Pešková et al. (2012), Queens (2001), Robles-Puente (2012, 2019), Sichel-Bazin et al. (2012), Romera and Elordieta (2013, 2019), van Rijswijk and Muntendam (2014), Delais-Roussarie et al. (2015), Elordieta and Irurtzun (2016), Kireva and Gabriel (2016), Muntendam and Torreira (2016), Masa and Elordieta (2017, in press), Troncoso-Ruiz and Elordieta (2017), Lai and Gooden (2018), Baltazani et al. (2019a, 2019b), Congosto Martín (2019), Fernández Rei (2019), Kozminska (2019), Muñiz Cachón (2019) and Uth (2019). It has been claimed that prosodic features are sufficiently salient to identify language varieties (cf., among others, Barkat et al., 1999; Leemann & Siebenhaar, 2008), so it is not surprising that they can be borrowed in language contact situations.
For Iberian Spanish, there is growing work on Spanish intonation in geographical areas in which Spanish is in intensive contact with another language. In all these studies, it is claimed that the varieties of Spanish spoken in those areas adopt intonational features (pitch accent type and tonal alignment) from the language they are in contact with. Simonet (2008, 2010, 2011) offers a detailed study of convergence in segmental and intonational phonology and phonetics between Spanish and Catalan in Majorca. Romera and Elordieta (2013) studied the linguistic behaviour of monolingual Spanish speakers from the mainland that are relocated in Majorca, and observed that the rapid adoption of linguistic features of Majorcan Catalan was achieved through daily communication in Spanish with native speakers of Majorcan Catalan. The variety of Spanish of these bilingual speakers shows intonational features of Majorcan Catalan (by a process of first language (L1) to second language (L2) transfer). Other researchers have also found similarities between Barcelona Spanish and Central Catalan (Romera et al., 2008, 2009). As for northwestern Spain, Troncoso-Ruiz and Elordieta (2017) and Muñiz Cachón (2019) describe the presence of intonational features of Asturian in the variety of Spanish spoken in Asturias, and Fernández Rei (2019) shows that the variety of Spanish spoken in Galicia is also heavily influenced by the intonational system of Galician.
Similar situations of linguistic convergence have been described for other varieties of Spanish showing features of other languages. For instance, Colantoni and Gurlekian (2004) claim the presence of Italian intonational traits in Argentinian Spanish (especially in Buenos Aires), O’Rourke (2005, 2012) and Muntendam and Torreira (2016) find prosodic features of Quechua in Cuzco Spanish, and Alvord (2006) shows that prosodic transfer occurs from English onto the dialect of Spanish spoken by Cuban immigrants in Florida.
With regard to Spanish in areas of the Basque Country, Elordieta (2003a, 2006) and Elordieta and Calleja (2005) describe and analyse the prosodic patterns of Spanish spoken by native speakers of Northern Bizkaian Basque. They observe features in the intonation of neutral declarative clauses with subject–verb–object order that are different from the typical features of Castilian or standard Spanish (Estebas-Vilaplana & Prieto, 2008, 2010; Face, 2008; Hualde & Prieto, 2015, among others). Elordieta (2003a, 2006) and Elordieta and Calleja (2005) attribute these features of Basque Spanish to convergence with Northern Bizkaian Basque. Robles-Puente (2012) also found evidence for convergence between Basque and Spanish in the intonation of absolute interrogatives in Bilbao Spanish. Elordieta et al. (2016) drew similar conclusions analysing the intonation of absolute interrogatives in natural speech by monolingual Spanish speakers, L1 Spanish/L2 Basque speakers and L1 Basque/L2 Spanish speakers in different locations of the Basque Country and Navarre.
Despite these findings, the discussion persists as to whether convergence may be explained by internal or external factors (Winford, 2005, 2013). Some factors that have been pointed out as potential causes of change have to do with internal language tendencies, or changes already initiated in the language (Poplack & Levey, 2010). Other authors point to the intensity of contact or the relative prestige of languages (Thomason, 2003; Thomason & Kauffman, 1988). To these we should also add other psycho-social factors such as the linguistic identification of the speaker (Bucholtz & Hall, 2003, among others) and the linguistic attitudes shown by the speakers towards each of the varieties.
We argue that in the explanation of prosodic convergence between Spanish and Basque we should account not only for objective or external factors, such as the speaker’s linguistic profile, his or her command of each language, or his/her degree of contact with Basque speakers, but also for his/her attitude towards Basque and the Basque-speaking ethnolinguistic group. Along with other factors, attitudes are also expected to be responsible for the degree of prosodic convergence shown by the speakers. That is, speakers who have a more positive attitude towards Basque and the Basque-speaking group will also produce higher rates of prosodic traits present in Basque.
Basque and Spanish in contact: social and linguistic situation in the Basque Country and Navarre
Basque is spoken in the southwestern corner of France and in northern Spain (the Autonomous Community of the Basque Country and Navarre). In the Spanish part, contact between Basque and Spanish has been long and extensive, starting with the ancestor of the Spanish language, Latin (Gorrochategui, 1995; Zuazo, 1995). As Perales and Cenoz (2001) point out, the sociolinguistic relationship between Basque and Spanish has been of a diglossic nature until recently. Basque was traditionally used in informal situations, whereas Castilian Spanish was considered the language of culture and was reserved for formal situations. Since the early 1980s, however, both languages were recognized as official languages in the Basque Country and in Navarre, and a process of linguistic normalization was initiated by the institutions. Especially in the Basque Country, education models and language policies in favour of the revitalization of Basque have led to an increase in the number of competent speakers and to a more favourable attitude towards their promotion (Etxebarria, 2015; Gobierno Vasco et al., 2016).
At present, the contact situation between these languages can be characterized as a “social bilingualism of diglossic character” (Etxebarria, 2000), meaning that the two languages are used in different degrees according to the situation, and not necessarily corresponding to formal or informal levels. All speakers in the Basque Country and Navarre know and use Spanish on a daily basis. This is not the case for Basque. The most recent study on the situation of both languages comes from the sociolinguistic survey elaborated by the Basque Government in 2016 (Gobierno Vasco et al., 2016). The data in this survey indicate that Spanish is the L1 (i.e. the language learned at home or spoken up to the age of 3 according to the survey questions) for the majority of the inhabitants of in the Basque Country (cf. Figure 1) (total population over 16: 1,864,000 in 2016).
First language of the Basque Country’s population (cf. Gobierno Vasco et al., 2016).
With regard to the knowledge about and use of Basque, although Basque may not be their L1, there is also a great number of speakers that know and use Basque regularly on a daily basis (Cf. Figure 2): 33.9% speak Basque regularly (631,000 people); 19.10% (356,000) are considered “passive” speakers, that is, they understand Basque but do not speak it; and 47% (877,000) are monolingual Spanish speakers.
Percentage of speakers of Basque in the Basque Country (cf. Gobierno Vasco et al., 2016).
In Navarre, the degree of knowledge about and use of Basque is much smaller. Only 11.9% of the population are Basque native speakers or bilinguals in Basque and Spanish (cf. Gobierno de Navarra, 2015). The opportunity for schooling in Basque is also smaller. Hence, the transmission of this language occurs almost exclusively in the family.
The contact situation between Basque and Spanish in the Basque Country and Navarre arising from the coexistence of the two languages has given rise to the mutual influence of one language on the other. In this paper, we focus on one specific aspect of the influence of Basque on Spanish. Similar to the situation in Majorca described in the introductory section, the adoption of linguistic features of Basque by the Spanish variety of the Basque Country and Navarre (at least in regions of the Basque Country and Navarre where this contact has been more extensive in time) may have arisen through the variety of Spanish spoken by native speakers of Basque, who transfer some of their Basque features to their Spanish. Then, when communicating with them in Spanish, native speakers of Spanish (monolinguals or L1 Spanish/L2 Basque) adopt these features to a greater or lesser extent, depending on the degree of competence in both languages and on social factors.
The influence that one language can exert on another in situations of contact does not always have identical results. In addition to internal linguistic factors of each language, social factors play a fundamental role in the convergence of linguistic features (Thomason, 2003; Thomason & Kauffman, 1988; Winford, 2005, 2013, among others). In his pioneering work, Weinreich (1953) already considered as influential factors aspects such as the density of the speakers of each language, the sociopolitical relations of groups, the relative prestige of languages, or individual factors such as the way in which each language was learned or the attitudes that the speakers show. In general, the studies on language contact considered the relative prestige of languages, the population density or the duration and intensity of contact as the most relevant social factors that may condition convergence (Winford, 2013). However, individual factors have not been taken into consideration as often. A feature that in our opinion plays a relevant role in speakers’ convergent or divergent behaviours is the attitudes that s/he shows towards the languages in contact. In this regard Weinreich (1953, pp. 3–4) stressed as important social factors the attitudes of the speakers towards their own language/s, towards the culture of each community and towards bilingualism. There have been numerous studies on attitudes in different contact situations (Auer & Hinskens, 2005; Ihemere, 2006; O’Rourke, 2005; Ries, 2014, among others), but these studies have almost always dealt with phenomena related to the survival, use or preservation of the languages in contact, not so much with the influence of speakers’ attitudes in the convergence of linguistic features.
Oppenheim defines linguistic attitudes as follows:
[. . .] a construct, an abstraction which cannot be directly apprehended. It is an inner component of mental life which expresses itself, directly or indirectly, through such more obvious processes as stereotypes, beliefs, verbal statements or reactions, ideas and opinions, selective recall, anger or satisfaction or some other emotion and in various other aspects of behavior. (Oppenheim, 1982, p. 39)
It is generally accepted that attitudes entail three components: cognitive; affective; and behavioural (Ryan et al., 1982). Attitudes are cognitive because they imply beliefs and the attribution of traits towards what surrounds us. They are affective because they imply feelings. And they are behavioural because they predispose the individual to behave in a certain way (Garrett et al., 2003, p. 3). This latter aspect has aroused controversy. For some researchers, individual behaviour does not always correspond with attitudes. Other factors may interfere in this correspondence; group pressure, social correction, or the speaker’s own interests in certain contexts may prevail over attitudes and beliefs (Garrett et al., 2003, p. 9). Despite this argument, most authors agree that the attitudes expressed by speakers can be a valid indicator of the individual’s linguistic behaviour. However, as we said, there are no studies that relate language attitudes and convergence of linguistic features. 1 Our own studies on Spanish and Catalan from Majorca (Romera & Elordieta, 2013) show that speakers of Spanish who showed more favourable attitudes towards Majorcan Catalan present a higher percentage of Catalan prosodic features in their Spanish discourse. We found that for Spanish speakers who had arrived recently in Majorca (they had been living on the island 2 to 8 years before the date of the recordings) objective factors such as the intensity or duration of the contact had a lower weight than the attitude that the individual showed towards integration in the target society.
For the case of Spanish in contact with Basque, in this article we defend the hypothesis that individual subjective factors have a substantial impact on the degree of prosodic convergence, especially those related to the more or less favourable attitude of speakers of Spanish to Basque. This factor is evidently not isolated from other external factors that are usually recognized as relevant in linguistic contact, such as the L1 of the speaker, the intensity of the contact, and the knowledge about and use of the other language. However, we believe that attitudes should be considered as an element that may explain to a greater extent the adoption of prosodic features of Basque, along with the aforementioned factors.
Study
Selection of speakers
The study presented in this article forms part of a larger project which investigates the possible correlation between the intonational patterns and language attitudes in six places where Basque and Spanish are in contact in the Autonomous Communities of the Basque Country and Navarre. Three urban areas and three rural areas were selected, thus accounting for the different degree of contact with Basque as a variable. The urban places are cities where a local variety of Basque is present or has been present until recent times. Thus, in the province of Bizkaia speakers from the city of Bilbao and the town of Lekeitio were interviewed; in the province of Gipuzkoa, speakers from the city of San Sebastian and the town of Ibarra were interviewed; and in Navarre, speakers from the city of Pamplona and the town of Lakuntza were interviewed. 2
Six speakers were recorded in each place, fulfilling several sociolinguistic profiles: two monolingual Spanish speakers; two L1 Spanish/L2 Basque speakers; and two L1 Basque/L2 Spanish speakers (one male and one female in each of the three groups). They were between 35 and 55 years of age, with medium or higher level of education. We selected the age band of 35–55 because they represent the population that is more dominant and influential in the socioeconomic life of the survey points. At the same time, most of the people in this age band have a medium to higher level of education.
Thirty-six speakers were recorded for a total of approximately 30 hours of quasi-spontaneous speech, obtained by a semi-directed interview (described in the “Our comparable data on the intonation of information-seeking absolute interrogatives in Madrid Spanish” subsection). The goal was to obtain a closer representation of the speech patterns of real speakers and to avoid a more standard or formal style of speech that can arise in reading or controlled tasks (cf. footnote 6). Although we are aware that a larger pool of speakers would be needed for a more representative sample, the 36 speakers recorded can provide first-step evidence for the relationship between linguistic production and attitudes towards the Basque language and the Basque ethnolinguistic group in areas of language contact between Basque and Spanish. It is also important to bear in mind that acoustic analyses of non-scripted, natural data are more time-consuming in natural speech than in scripted, carefully controlled laboratory speech. There is an initial demand in utterance segmentation that does not exist or is much more straightforward in scripted laboratory speech, as the individual utterances to be analysed have to be extracted from a continuous signal where boundaries are sometimes not clear and where overlapping from one or more interlocutors may happen, as well as distortions or interruptions. On the other hand, the annotation of prosodic features in spontaneous speech is especially complicated by uncontrolled microprosodic effects caused by obstruents that require interpretations and recalculations from the researcher. Also, phonological processes of resyllabification and phonetic lenition of consonants and vowels are frequently observed in informal speech registers and faster speech rates, commonly associated with conversational speech. Resyllabification and lenition processes cause difficulties in establishing clear syllabic boundaries, which are necessary for a correct interpretation of the tone-to-syllable associations that are central in any analysis of intonational phonology and phonetics. In sum: recording, segmenting and analysing a large amount of data from natural speech would require the participation of a comparatively larger set of human annotators than were available for the present project.
In this article we present the results obtained from of a subset of the recordings made, that is those from the cities of Bilbao and San Sebastian. 12 speakers from these cities were recorded, 6 from Bilbao and 6 from Donostia-San Sebastian. Bilbao and San Sebastian were chosen because they are good representatives of the variety of Spanish spoken in the Basque Country from demographic, geographical and linguistic standpoints. These two cities have 343,430 and 181,652 inhabitants, respectively, and the urban areas they form together with smaller towns that surround them have 858,236 and 327,428 inhabitants, respectively (Eustat, 2019). The total number of 1,185,664 inhabitants amounts to 54% of the population of the region officially called the Autonomous Community of the Basque Country (population 2,188,017). Bilbao and San Sebastian are also the capital cities of Bizkaia and Gipuzkoa, two of the three provinces that constitute the autonomous region. And importantly, although Spanish is the main language in Bilbao and San Sebastian, they are cities where vernacular varieties of Basque are either still spoken (a variety of Gipuzkoan in San Sebastian) or have been spoken until the beginning of the last century (northern Bizkaian, in Bilbao). Next to Bilbao and San Sebastian there are towns where Bizkaian and Gipuzkoan Basque are still spoken, and historically there has been a close contact with inhabitants of those towns, in the form of local immigration to the two capital cities or of commercial relationships. In recent years there has been an increase in speakers of standard Basque in Bilbao and San Sebastian, as in the whole region. Thus, Bilbao and San Sebastian constitute good sources of the types of speakers that are necessary in order to carry out the main goal of the present work.
As we said, the speakers belonged to three sociolinguistic profiles: monolingual speakers of Spanish, with little or no knowledge about Basque; native speakers of Spanish who had learned standard Basque at school or later in life; 3 and native speakers of Basque who had learned Spanish in their childhood simultaneously or sequentially, but with a dominant presence of Basque in the family. In all cases, the variety of Spanish was the one spoken in Bilbao or San Sebastian, and the varieties of Basque were Bizkaian and Gipuzkoan. In each of the two cities there were two speakers of each of the three sociolinguistic profiles (one man and one woman). Table 1 summarizes the profile of the speakers that provided us with the data to be analysed in this article.
Number, gender and sociolinguistic profile of speakers selected for this study. All between 35 and 55 years of age and with middle-higher education.
Recordings
The recordings consisted of semi-directed sociolinguistic interviews (Silva-Corvalán, 2001) between a speaker and an interviewer. In 10 of the 12 interviews, the interviewers were not the authors, but two young females trained by the authors to conduct the interviews. These interviewers had been born and raised in the same city as the speakers, and lived there at the time of the interview. The interviewers also served the role of selecting speakers who fulfilled the social and linguistic profiles described in the previous subsection. In two of the recordings it was the first author who conducted the interviews. The recordings took place in quiet rooms at the speakers’ homes or workplaces, in order to facilitate an optimal level of confidence and comfort for the speaker. The motivation for fostering a closer degree of acquaintance between subject and interviewer was to allow for a more relaxed environment for the conversation, which could favour a more natural and relaxed production from the subjects.
The interviews consisted of three modules, in which the interviewer asked questions to the subject. The first module contained questions on the subjects’ degree of knowledge about the two languages in contact, Basque and Spanish. The second module had questions related to the degree of use of each of these languages. In the third module, the subjects were questioned on their attitudes towards Basque and the Basque ethnolinguistic group: for instance, whether they thought that speaking Basque improved the social image of a person, whether in their opinion knowing and speaking Basque was useful in their personal and professional lives, or whether they thought that Basque should be taught obligatorily. 4 The questions were written on sheets of paper in the form of bulleted topics rather than in the form of full questions, the idea being that the interviewer posed the questions in as natural a style as possible, not as read speech.
There were two interviews in each recording session. The first interview was one in which the subjects of our experiment were asked questions by the interviewers. With such an interview, declarative utterances were obtained from the subjects. In order to obtain interrogative utterances from them, we asked the subjects to take the role of interviewers and ask their interlocutors (the interviewers of the first part) the same questions that they had been asked. This way, a number of absolute and partial interrogative utterances (i.e. yes/no and who, what, when where, and why questions, respectively) were recorded from the subjects.
The interviews were recorded with a Tascam DR-100 digital recorder, through a built-in omni-directional microphone pointing towards the subjects but able to capture the speech of the two participants. The audio was recorded with a sampling rate of 44,000 Hz, in .wav format. The conversations were also recorded on video, with a Sony video camera held on a tripod, with the objective of analysing the speakers’ level of (dis)comfort, relaxation or nervousness while answering the questions about their attitudes towards Basque and the Basque ethnolinguistic group.
In all, 9 hours and 20 minutes of conversations were recorded for the 12 subjects, with an estimated average of 46 minutes per subject.
Types of sentences analysed
From the recorded speech material, 366 declarative utterances, 172 absolute interrogatives and 207 partial interrogatives were segmented. These three types of utterances are usually considered as the basic types of utterances in prosodic studies across languages, with the assumption that they offer sufficient evidence for the main aspects of the intonational and prosodic structure of a language (sometimes with the addition of exclamative or imperative sentences and various subtypes of declarative and interrogative sentences, such as topicalized and focalized constructions, echo-, confirmation, rhetorical or declarative questions, etc.). The reason for the higher number of declarative utterances is natural: the subjects produced more utterances in their responses than in their questions. The number of utterances of the three types were rather balanced across the 12 subjects: per speaker, we collected an average of 30 declarative utterances (standard deviation (SD) 1.93), 14 absolute interrogatives (SD 2.53) and 17 partial interrogatives (SD 3.72). All the utterances fulfilled the criteria of being full utterances, that is, not being suspended, cut, rephrased, interrupted by coughs, laughter, or overlapping speech from the interlocutor.
For the present article, we are focusing only on absolute or yes/no interrogatives. Besides space limitations, this choice is motivated by the fact that these are the sentence types that have the most significantly different intonational contours in Castilian Spanish and Basque. Hence, any intonational features in yes/no interrogatives in the Spanish variety of the Basque Country (in Bilbao and San Sebastian) that differ from the typical Castilian Spanish features and which resemble those of Basque are more easily discernible.
The prosody of information-seeking absolute interrogatives in Spanish
In our corpus, the absolute interrogatives, polar questions or genuine questions (Escandell-Vidal, 1998, 1999, 2017), are information-seeking questions, which encode the genuine need the speaker has for an answer from the hearer that may fill the speaker’s information void on the matter of the utterance. In the traditional descriptive literature on absolute questions, these interrogatives have also been referred to as unmarked, pragmatically neutral, prototypical, canonical or default yes/no questions. Researchers working in different frameworks of intonational analysis agree on a description for Castilian Spanish information-seeking absolute questions, at least for elicited or read speech: these interrogatives are characterized by an initial pitch accent with a peak on the posttonic syllable, followed by a steady descent in fundamental frequency up to the final stressed syllable of the utterance, where a final steep rise is observed (Estebas-Vilaplana & Prieto, 2008, 2010; Face, 2004, 2008; Henriksen, 2010; Hualde & Prieto, 2015; Navarro Tomás, 1944; Quilis, 1993; Sosa, 1999, among others). Many of the descriptions are couched within the framework of the autosegmental–metrical model of analysis of intonation, especially the Spanish Tone and Break Index model (Spanish ToBI, or Sp_ToBI, for short). The intonational labelling of the most common contour of yes/no questions in Central Castilian Spanish in Sp_ToBI is L+<H* L* H%, that is, an initial rising accent with a peak realized on the posttonic syllable (L+<H*) and a nuclear accent in a low tone (L*) followed by a high rising boundary tone (H%). An example is illustrated in Figure 3, from Estebas-Vilaplana and Prieto (2010, p. 30).
Fundamental frequency contour of an absolute interrogative in Castilian Spanish. After an initial peak on the first stressed word, the nuclear accent is low (L*) and rises subsequently to an upstepped high pitch level at the end of the utterance (HH% in Estebas-Vilaplana and Prieto’s (2010) notation, transcribed in more recent work in the Spanish Tone and Break Index as ¡H%).
In non-neutral absolute interrogatives in Castilian Spanish, a final falling contour is observed, rather than a rising one. There is a rising accent with a peak on the tonic syllable, very often with an upstepped pitch level (L+(¡)H*, in Sp_ToBI), followed by a falling boundary tone (L% or HL%). Final rising–falling contours of this type (i.e. circumflex contours) are found in non-neutral absolute interrogatives in Castilian Spanish, such as yes/no echo-questions, used to express surprise at what the interlocutor has just said or to beg a clarification from the interlocutor of what (s)he has just uttered (Estebas-Vilaplana & Prieto, 2010; Hualde & Prieto, 2015, p. 378). Escandell-Vidal (1998, 2017) generalizes this type of contour to all questions that are “attributed to other”:
[. . .] it is possible to give a unified account of all interrogatives with a rising–falling final contour in terms of a single and systematic property, namely the attribution of the utterance’s content to another individual. The intonation thus encodes a procedural restriction on the construction of higher-level explicatures, which guides the hearer to an interpretation that must be fully compatible with the assumption that the speaker does not present herself as responsible for the representation contained in her utterance [. . .]. (Escandell-Vidal, 1998, p. 185)
These are marked interrogatives, whereas the interrogatives that are gathered from the recordings of our corpus are pragmatically neutral, information-seeking questions.
Other types of non-information seeking questions end in a fall, such as confirmation-seeking and imperative yes/no questions, but in these types of interrogatives the nuclear accent is of a falling type (H+L* L%; Estebas-Vilaplana & Prieto, 2010, p. 29), not a rising one as in Basque Spanish information-seeking questions.
It is important to point out that the generalizations established above are based on elicited speech. In conversational speech, the facts may be different. Torreira and Floyd (2012) found that 79% of 1070 yes/no questions in a corpus of conversational speech in Madrid Spanish were of the circumflex type (L+¡H* L%). The authors point out that the “attributed” meaning claimed by Escandell-Vidal (1998, 1999, 2017) for this type of contours is found in only a minority of cases (16%). Although the pragmatic and discourse functions in which the circumflex contour occurs is rather disperse, the generalization that Torreira and Floyd (2012) posit is that this type of contour is mainly used as a “topic follow-up”, and secondarily as a signal that the speaker is “maintaining the course of action” (responding to a previous question, providing news receipt, initiating a repair, checking the listener’s attention during a telling, or providing a pre-announcement during a telling). In our corpus, however, the interrogatives were all purely information-seeking, in which the interviewer wants to gather information from the interlocutor, so the pragmatic uses are very different. Interestingly, the interrogatives that showed a final rising contour in Torreira and Floyd’s (2012) work had the role of “topic proffers”, that is, topic initiators. They can also appear in other contexts that “disrupt the ongoing course of action”, in the authors’ words (such as changing the interlocutor, or deviating from the main topic being discussed). In general, then, low rises are used to ask about a new topic or a topic that is introduced for the first time. Conversely, only 2% of yes/no interrogatives with falling circumflex contours were observed in this type of context.
For our purposes, the relevance of this difference in the association between final contour and discourse function is that the yes/no questions in our corpus had a use that resembled more the topic-proffer function, as the interviewer had to ask about different topics, without engaging in a subsequent exchange of ideas and thoughts about the topics being asked. Thus, they could be considered on a par with the interrogatives displaying final rising configurations in Torreira and Floyd’s (2012) corpus; namely, the so-called neutral, unmarked information-seeking absolute interrogatives.
Henriksen et al. (2016) analyse the intonational contours of polar questions in spontaneous speech in the variety of Iberian or Peninsular Spanish spoken in the region of Castile-La Mancha (Manchego Spanish). The most common nuclear contours are L+H* L% and L+H* H%. They claim that the low-rising nuclear contour L* H% that has been reported for yes/no interrogatives in Peninsular Spanish is used in formal contexts mainly. Following Escandell-Vidal’s (1998, 1999) theory, the authors conclude that a falling circumflex contour L+H* L% is used more often in interrogatives in which the speaker expresses thoughts attributed to others (the interlocutor or somebody else). For instance, yes/no echo-questions are pronounced with this contour. The rising contour L+H* H% is used when the speaker does not know the answer to the question or is not committed to the propositional content of the question, that is, a speaker-attributed strategy.
Our comparable data on the intonation of information-seeking absolute interrogatives in Madrid Spanish
In order to have a first-hand assessment of the frequency of occurrence of final rising and falling configurations in absolute interrogatives in Madrid Spanish, we recorded seven speakers from Madrid within the same age band and the same level of studies as the speakers in the Basque Country (i.e. 35–55 years of age and medium–higher education). The recordings were made with the same methodology, that is, semi-directed interviews. As in the interviews in the Basque Country, the conversations had two parts. First, the subjects were asked questions about their personal profile, tastes and habits by a research assistant, whom they did not know (that way we obtained declarative utterances from the subjects). 5 There were no questions about the Basque Country or about the Basque language, as the goal was not to gather information on the speakers’ attitudes towards Basque or the Basque ethnolinguistic group. After that, the subjects had to ask the same set of questions to the research assistant, acting as interviewers (thus obtaining interrogative utterances from the subjects). Out of a total of 176 absolute interrogatives, the majority of the utterances had a final rising configuration (119, that is, 68% of the total). 57 utterances had a falling final contour (32% of the total), with the configurations H* L% or L(¡)H* L%. 6 Thus, the quantitative results from our corpus of Madrid Spanish are that final rising contours are more frequent than falling ones, contrary to the results obtained by Torreira and Floyd (2012). At this point, we cannot know whether the differences between the two studies are due to the nature and the size of the corpora, but what is relevant for our purposes is that the nature of the corpora (i.e. the nature of the conversations) were the same in Madrid Spanish and Basque Spanish. That is, the results are directly comparable quantitatively.
Information-seeking absolute interrogatives in Basque
Information-seeking absolute interrogatives in Basque are characterized by a high-falling or a low-falling nuclear contour. Figure 4 contains an example of a yes/no question in the Northern Bizkaian Basque variety of Gernika, with a rising–falling circumflex contour at the end of the utterance (example from Robles-Puente, 2012, p. 258). Figure 5 contains a sample pitch contour of a yes/no question in standard Basque, uttered by a native speaker of the Gipuzkoan dialect (example from Elordieta & Hualde, 2014, p. 457). Further examples can be found in Elordieta (1998, 2003b), Aurrekoetxea et al. (2011), Gaminde et al. (2011), Robles-Puente (2012), and Elordieta and Hualde (2014).
Fundamental frequency contour of an absolute interrogative in Gernika Basque (from Robles-Puente (2012, p. 258)). After an initial rise on the word badalonara ‘to Badalona’, the pitch rises and falls on the final syllable of the verb doa ‘goes’. The pitch movement in the verb could be analysed as a L+¡H* accent, followed by a L% or HL% boundary tone.
Fundamental frequency contour of an absolute interrogative in standard Basque, pronounced by a speaker of Gipuzkoan Basque. After an initial rising (L+<H*) accent, the pitch level is maintained high until the end of the utterance, where another pitch accent is produced on the verb, followed by a fall (L% boundary tone).
Younger generations are introducing the use of final rising intonations for interrogatives, possibly as an influence from Spanish (cf. Elordieta, 1998, 2003b; Gaminde et al., 2011), but its use is still reduced among native speakers of Basque. It is more frequent among L1 Spanish/L2 Basque speakers (Gaminde et al., 2011).
As we will see in the next section, in the variety of Spanish spoken in the Basque Country information-seeking yes/no questions most commonly present a final contour in which there is a rising accent with a peak on the tonic syllable, very often with an upstepped pitch level (L+(¡)H*, in Sp_ToBI), followed by a falling boundary tone (L% or HL%). Thus, instead of the typical final rising configuration of Castilian Spanish with a low tone on the nuclear accented syllable followed by a steep rise (L* H%), the majority of information-seeking absolute interrogatives in Basque Spanish are characterized by a final circumflex contour, in a rising–falling configuration. 7 It is reasonable to presume that such a final contour may be attributed to influence from the Basque final falls in absolute interrogatives.
Absolute interrogatives in other northern varieties of Spanish
It must be said that final falling contours in absolute interrogatives are very common in northwestern Spanish, that is, in the Spanish varieties spoken in Galicia, Asturias and Cantabria (cf. Cuevas-Alonso & Bleorţu, 2012). 8 In these varieties, the falling patterns coexist with the central Castilian final rising interrogatives, which are more common in urban areas. However, the final descending patterns are not of the same circumflex type as that of Basque Spanish. In Galician, H+L* L% is the common pattern, with a fall that starts already by the end of the pretonic syllable, although there are subdialectal differences as well (cf. Fernández Rei, 2007, 2016; Pérez Castillejo, 2012, 2014, among others). In Asturian, the descriptions of the yes/no intonation contours agree that the most common pattern can be annotated as H* (H)L%, that is, without a rising nuclear tone such as the one we have found in Basque Spanish (cf. Alvarellos Pedrero et al., 2011; López Bobo et al., 2008; Muñiz Cachón, 2013; Muñiz Cachón et al., 2008; Troncoso-Ruiz & Elordieta, 2017, among others). In Cantabrian Spanish, the final nuclear contour L+(¡)H* L% is found in rural areas (Cuevas Alonso & López Bobo, 2011). The pitch level rises on the tonic syllable, and the pitch level continues ascending on the posttonic syllable, where a fall to a mid-level is observed (at least not to the base level, according to the descriptions). 9 This contour is different in shape from the one commonly found in Basque Spanish, where the peak is usually realized on the tonic syllable. It is noteworthy that Spanish intonation in Galicia is heavily influenced by Galician intonation (see Fernández Rei, 2016, 2019), and Asturian Spanish also shows traits of Asturian (Muñiz Cachón, 2013, 2019; Troncoso-Ruiz & Elordieta, 2017). That is, there is an effect of the adstrate language due to language contact. The same situation would seem to have taken place in the Basque Country. 10
Results
Falling and rising final contours
In our corpus with 12 speakers from Bilbao and San Sebastian, 136 of the 172 information-seeking absolute interrogatives end in a falling contour. That is, 79% of polar questions have a final falling configuration (compared to 32% in Madrid, as reported in the previous section; cf. footnote 6 as well). The most common nuclear contour has a circumflex rising–falling shape (69.5% of the yes/no questions). Of these, the majority have a rising nuclear accent with a peak on the stressed syllable. In terms of the autosegmental–metrical model of analysis of intonational phonology, this nuclear accent would be labelled L+H*. Half of the yes/no questions with this nuclear accent have a high tone substantially upstepped with respect to previous H tones, L+¡H*. 11
After the peak of the nuclear accent, the pitch level starts descending almost immediately in most instances. A L% boundary tone is proposed for these cases (almost half of the information seeking yes/no questions in our corpus, 47% of the total). Figure 6 shows an intonational contour with a L+H* nuclear accent followed by a L% boundary tone (example from a male monolingual Spanish speaker from San Sebastian). In 27.1% of falling yes/no questions the high pitch level reached in the nuclear stressed syllable can be maintained onto the final syllable, before falling. For these cases, a bitonal falling boundary tone must be proposed, HL%. Figure 7 illustrates an example with a L+¡H* HL% nuclear contour (example from a female L1 Spanish speaker from Bilbao).
Fundamental frequency contour of an absolute interrogative in San Sebastian Basque, pronounced by a male Spanish monolingual speaker. After an initial peak, the pitch level descends gradually until the last word of the utterance, where a rising pitch accent is produced, followed by a fall (L% boundary tone).
Fundamental frequency contour of an absolute interrogative in Bilbao Basque, pronounced by a female first language Spanish speaker. After an initial peak on the first stressed word, the pitch level is maintained until the last word of the utterance, where an upstepped rising pitch accent is produced, followed by a HL% boundary tone.
Finally, in a few instances (3%), the fall in pitch can be rather moderate, only slightly lower than the H* tone in the accent or the H of the complex boundary tone. A downstepped !H% boundary tone has been posited there.
In sum, in our corpus with 12 speakers from Bilbao and San Sebastian, a total of 79% of the 172 information-seeking absolute interrogatives end in a falling contour. The most common nuclear configuration is a circumflex rising–falling contour, with the shape L+(¡)H* (H)L%. The parentheses around the diacritic for ‘upstep’ indicate that the accentual H* tone can be upstepped, and the parentheses around the high tone in the complex boundary tone indicate that the boundary tone can be either L% or HL%.
On the other hand, 18% of the information-seeking yes/no questions end in a rising configuration, that is, they have a final H% or upstepped ¡H% boundary tone. Of these, the majority have a rising nuclear accent, L+(¡)H*. The nuclear contour would thus be transcribed as L+(¡)H* (¡)H%. An example is found in Figure 8 (example from a male L1 Basque speaker from Bilbao).
Fundamental frequency contour of an absolute interrogative in Bilbao Basque, pronounced by a male first language Basque speaker. After an initial peak on the first stressed word the pitch level ascends gradually until the last word of the utterance, where a rising pitch accent is produced (with an upstepped level), followed by a H% boundary tone.
Only 6% of the absolute interrogatives present the typical nuclear contour in Castilian Spanish absolute interrogatives of a steep final rise, from a low tone in the nuclear accent to a high boundary tone, L* H%. Finally, 3% of the interrogatives end in a sustained pitch.
To summarize our results, 79% of a total of 172 information-seeking absolute interrogatives in Bilbao and San Sebastian Spanish end in a falling configuration (i.e. 136 interrogatives), and 21% (36 interrogatives) end in a rising or sustained final configuration. And of these, only 6% (10 interrogatives) present the L* H% configuration of Castilian Spanish, the rest having a rising nuclear accent, L+(¡)H* (¡)H%).
In comparison, in our corpus of Madrid Spanish with roughly the same size and the same type of polar interrogatives (176 information-seeking polar questions), only 32% of the polar interrogatives ended in a falling configuration (with five of the seven speakers showing an average percentage of only 16%). Of these, H* L% was the most common nuclear contour, followed by L+(¡)H* L%. Thus, more than two-thirds of the information-seeking absolute interrogatives ended in a final rise (68%).
In this subsection, we have provided a qualitative and quantitative report of the main final contours found in information-seeking yes/no questions in spontaneous speech in the varieties of Spanish spoken in Bilbao and San Sebastian. We have seen that there are two nuclear configurations: most of the interrogatives (79%) end in circumflex falls with the shape L+(¡)H* (H)L%; and 21% of the interrogatives end in final rises, with the contour L+(¡)H* (¡)H%. Given the difference in the frequency of occurrence of final circumflex falling contours in Basque Spanish compared to a central Spanish variety such as the one in Madrid, we might hypothesize that the higher frequency of occurrence of final circumflex contours in Basque Spanish is due to influence from Basque. That is, that the variety of Spanish spoken in the Basque Country has “picked up” the falling contours of Basque. However, as we will see in the next subsection, there is variation among speakers in the frequency of appearance of final circumflex contours in Bilbao and San Sebastian.
Association of final contours with sociolinguistic variables
We proceed to explore possible correlations or associations between such contours and the sociolinguistic variables corresponding to the subjects of our investigation. That is, we examine whether and to what extent the percentages of occurrence of falling and rising contours are related to sociolinguistic factors of the speakers. These variables are: (a) city; (b) linguistic profile; (c) attitude towards Basque and the Basque ethnolinguistic group; and (d) degree of contact with the Basque language and with the Basque ethnolinguistic group. The values for each of these variables are presented in (1):
(1) a. City: Bilbao/San Sebastian
b. Linguistic Profile: Spanish monolingual/L1 Spanish/L1 Basque
c. Attitude towards Basque and the Basque ethnolinguistic group: a quantitative value in a scale from 1 (more negative attitude) to 4 (more positive attitude).
d. Degree of contact with the Basque language and with the Basque ethnolinguistic group: a quantitative value in a scale from 1 (no contact) to 4 (intense contact).
No differences were found between the two cities our subjects came from, Bilbao and San Sebastian. In Bilbao 80% (n = 65) of the interrogatives ended in falls, and 20% (n = 16) ended in a rising intonation. In San Sebastian, the percentages were 77.5% (n = 69) and 22.5% (n = 20), respectively. As expected, the differences are not statistically significant (Chi-square (χ2) = 0.188, p = 0.665).
The linguistic profile of the speakers did not produce any differences in the frequency of occurrence of falling and rising nuclear contours, either. Table 2 shows the absolute numbers and percentages (in parentheses) of interrogatives ending in falling or rising intonation in information-seeking yes/no interrogatives in each of the three linguistic profiles: monolingual speakers of Spanish; L1 Spanish/L2 Basque speakers; and L1Basque/L2 Spanish speakers. A cross-tabulation test revealed no correlation between the linguistic profile of the speakers and the percentages of final falls and rises (χ2 = 2.443; p = 0.295). Although monolingual speakers appear to have a lower percentage of final falls than L1 Spanish and L1 Basque speakers, and hence a possible difference could be claimed between monolinguals and bilinguals, this difference does not amount to a correlation; once the frequency is analysed for the two cities independently, such a possibility disappears. In Bilbao (cf. Table 3) and San Sebastian, (cf. Table 4) the bilingual speakers do not present distinct or significantly different frequencies of falls and rises in comparison to monolinguals (χ2 = 12.000; p = 0.285; and χ2 = 6.000; p = 0.306, respectively). No correlation was found between the linguistic variable and the speaker’s gender, either (χ2 = 14.5; p = 0.413; cf. Table 5).
Numbers and percentages of information-seeking yes/no interrogatives with rising–falling and rising final contours (falls and rises, respectively) for each linguistic profile in the two cities combined, Bilbao and San Sebastian.
Numbers and percentages of information-seeking yes/no interrogatives with rising–falling and rising final contours (falls and rises, respectively) for each linguistic profile in Bilbao.
Numbers and percentages of information-seeking yes/no interrogatives with rising–falling and rising final contours (falls and rises, respectively) for each linguistic profile in San Sebastian.
Numbers and percentages of information-seeking yes/no interrogatives with rising–falling and rising final contours (falls and rises, respectively) according to the speaker’s gender, in Bilbao and San Sebastian.
Table 6 shows the percentages of final falling and rising contours for each of the speakers. As we can see, final circumflex contours (labelled as “Falls” in Table 6) are the most common fundamental frequency configurations found at the end of absolute interrogatives for all speakers, the smallest percentage of final falls being 64%. The 64–100% range amounts to stating that final rising–falling (i.e. circumflex) configurations occur in at least two-thirds of all information-seeking absolute interrogatives in Bilbao and San Sebastian.
Numbers and percentages of information-seeking yes/no interrogatives with rising–falling and rising final contours (falls and rises, respectively) for each of the 12 subjects in our investigation.
Notes: FBiL1B, female Bilbao L1 Basque; FBiL1Sp, female Bilbao L1 Spanish; FBiMon, female Bilbao monolingual; FSSL1B, female San Sebastian L1 Basque; FSSL1Sp, female San Sebastian L1 Spanish; FSSMon, female San Sebastian monolingual; MBiL1B, male Bilbao L1 Basque; MBiL1Sp, male Bilbao L1 Spanish; MBiMon, male Bilbao monolingual; MSSL1B, male San Sebastian L1 Basque; MSSL1Sp, male San Sebastian L1 Spanish; and MSSMon, male San Sebastian monolingual.
The list of topics for the direct and reverse interviews included questions that prompted the interviewee to give his/her opinion on issues such as whether or not: Basque improves the social image of an individual; people living in the Basque Country and Navarre should know Basque; education should be in Basque; Basque is useful professionally and for everyday life; Basque is in a risk to be lost; Basque is a language spoken by a minority; and there exist differences between people who speak Basque and people who do not: they like listening to Basque.
As a general tendency, the speakers expressed positive attitudes towards the Basque language and the Basque ethnolinguistic group. Our results support the results of previous studies regarding perceptions and opinions by Basque citizens on the situation and future of Basque (Amorrortu et al., 2009). Most of our speakers have a positive consideration of Basque as a professionally useful language and also as a useful language for living in the Basque Country (66.7% and 83.3%, respectively). Very few consider Basque to be in danger of being lost (33.3%), and most think it is spoken more often now than before (66.7%). Almost all our subjects consider that education in the Basque Country should be in Basque (91.7%), and the majority do not establish differences in the population according to whether a person speaks Basque or not (66.7%). Besides this consensus, when asked about whether Basque should be spoken in the Basque Country, 11 of the 12 subjects thought that speaking Basque should not be mandatory. The majority expressed initial agreement with the need for Basques to speak Basque, and that there is also a need to preserve Basque culture, heritage and identity through the language, but everyone agreed that forcing speakers to use a language is impossible. Everyone said that they have friends and family who had had no chance to learn Basque, or they themselves were in this situation and find it difficult to learn it now. As one subject said: “[learning Basque in adulthood] requires a huge effort that not everyone can make.” The idea that many of the subjects expressed was that children could be a “solution”, as children “can learn languages easily” (Amorrortu et al., 2009) and they will make Basque a language effectively spoken by all. The high consensus observed for almost all of our subjects that schooling should be in Basque is correlated with this idea.
In addition to a qualitative analysis, and in order to quantify differences among speakers, a score was also assigned to each subject reflecting his/her global attitude towards Basque and the Basque ethnolinguistic group. All responses were coded on a scale of 1 to 3, from least to most positive. Following the results of previous studies on attitudes towards Basque (Amorrortu et al., 2009; Ortega et al., 2015; Puigdevall et al., 2018), responses agreeing with a useful consideration of Basque in everyday life, with the need to preserve this language, with the fact that education should be in Basque, or with the opinion that Basque should be spoken by everyone in the Basque Country were taken to be the most positive (i.e. 3). Responses representing a partial agreement with these ideas were coded as moderately positive (2). Finally, responses that did not agree with these ideas were interpreted as the least positive (1). Then the mean value of these ratings was calculated for each speaker and the resulting number was assigned to each of them and taken as an attitudinal index.
Table 7 shows the values obtained according to the speakers’ language profile. As can be seen, most of the speakers (75.1%) independently of their L1 are located between the values 2.11 and 2.56, which confirms the general positive attitude described above. Only three speakers (24.9%) show fewer positive attitudes and have lower values, between 1.56 and 1.78.
Attitudinal values according to the subjects’ linguistic profile (for subjects’ codes see ‘Notes’ in Table 6).
Similar results are obtained with respect to the variable degree of contact. To calculate a global contact value for each speaker, the same procedure was performed as in the case of the attitudes. Speakers’ responses regarding their use of Basque and their contact with Basque speakers were scored between 1 and 3 (1 ‘minor contact’, 2 ‘contact’, and 3 ‘extensive contact’). 3 was assigned when Basque was used at work, with friends, when listening to the radio, watching television, reading books, newspapers or consulting the Internet. Additionally, speakers answered two questions about their overall degree of contact with Basque speakers at work and in their personal lives; these were also coded accordingly. Again, the mean value of their responses was obtained, and then assigned to each speaker as a contact index. The results are shown in Table 8, grouped according to the speaker’s linguistic profile.
Contact values according to the subjects’ linguistic profile (for subjects’ codes see ‘Notes’ in Table 6).
As Table 8 shows, the majority of speakers are situated in the lower part of the scale, with contact values between 1.00 and 1.44 (66.7% of the subjects, in blue). The participants’ answers reveal that monolingual speakers (33.3%) do not try to speak Basque, and in their environment (family, friends and work) the dominant language is Spanish. In the group of speakers who know Basque and/or who speak it to different degrees, 16.7% of L1 Spanish speakers and 16.7% of L1 Basque speakers also confess not to have many opportunities to speak or listen to Basque daily in their environment. Only the remaining 33.4% say they use it frequently at work, with friends or in the neighbourhood (with values between 1.57 and 2.14, in yellow).
The question is now to what extent the attitudes towards Basque and the Basque ethnolinguistic group and the degree of contact with the Basque language and with the Basque ethnolinguistic group correlate with the percentage of rising–falling final contours in our speakers. As we have seen above, the differences in the percentages of final falls in absolute interrogatives are not extremely large; the baseline (i.e. the speaker with fewer percentages of falling contours) is 64%.
There is a clear relation between the percentages of final rising–falling contours and the linguistic attitudes and the degree of contact: the speakers who have a higher degree of contact and better attitudes have higher percentages of final falling contours. As shown in Table 9, the attitudinal value given to each speaker corresponds to a higher or lower degree of final falling contours. Those who showed more positive attitudes (between 2.22 and 2.56; in dark grey) produced between 77% and 100% of falls, while those with only positive or less positive attitudes (1.56–2.11; in light grey) produced between 64% and 75%. Table 9 reveals a clear tendency: as attitudinal values increase, the percentage of falls increases. 41.2% of speakers with attitudinal values from 1.56 to 2.11 produce between 64% and 75% of final circumflex contours (i.e. rising–falling), and 58.3% of the speakers who have more positive attitudes (2.22–2.56) produce more final circumflex contours, between 77% and 100%. Nonetheless, the correlation is not a perfect one. For instance, MBiL1Sp and FBiMon speakers should have fewer percentages of final falls, given their lower positive attitudes, and FSSMon and MBiL1B should show higher percentages of falls, given their more positive attitudes (see ‘Notes’ to Table 6 for subjects’ codes – e.g. MBiL1Sp, FSSMon, MBiL1B, etc. – referred to in the text). A linear regression analysis, performed using SPSS software, shows that the correlation is significant, with medium strength (R2 = 0.466, F (1, 10) = 8.721, p = 0.01). The scatter plot in Figure 9 illustrates this.
Attitudinal values and percentages of falls for each speaker (for subjects’ codes see ‘Notes’ in Table 6).
Linear regression between % of falls and attitudinal value.
We checked whether the degree of contact could better explain the percentages of falls. The cross-tabulation between the contact values of each speaker and their percentages of final falling contours is shown in Table 10.
Degree of contact values and percentages of falls for each speaker (for subjects’ codes see ‘Notes’ in Table 6).
As can be observed in Table 10, there is also a relationship between the value representing the degree of contact with Basque and the Basque ethnolinguistic group and the relative production of interrogatives with final falls. Again, speakers with lower contact levels produce lower percentages of final falling contours (between 64% and 80%; in light grey), while speakers with a higher contact value produce between 83% and 100% of final falls (in dark grey). In order for the correlation to be perfect, some speakers should show smaller percentages of falls according to their low values for degree of contact (e.g. FBiMon, MBiMon, MBiL1Sp or MSSMon). Despite this, a linear regression analysis reveals that the strength of the correlation between these two variables is high (R2 = 0.697, F (1, 10) = 22.959, p = 0.001). Figure 10 illustrates this.
Linear regression between % of falls and contact values.
Our results have revealed that the production of rising–falling contours has a medium correlation with the subjects’ attitudes towards Basque and the Basque ethnolinguistic group on the one hand, and a stronger correlation with the degree of contact with Basque and the Basque ethnolinguistic group on the other hand. These results lead us to think that, although each of these variables has a clear influence on the production of falls, it may be the case that both variables together are able to explain a greater number of cases. A multiple regression analysis with two predictors (attitudinal value and contact value) shows us that this idea is in fact correct; the two variables together can account for 80.7% of the results (i.e. R2 = 0.807), and the correlation is stronger than with the two variables analysed separately (R2 = 0.807, F (2, 9) = 18.844, p = 0.001). A comparison of the two factors confirms that ‘Contact’ is a stronger predictor than ‘Attitude’ (t = 3.992, p = 0.003 for the former, and t = 2.273, p = 0.049 for the latter).
Discussion and conclusion
The results of our investigation show that in the variety of Spanish spoken in the Basque Country (Bilbao and San Sebastian) 79% of all information-seeking yes/no questions (136 of a total of 172) have final configurations with a rising–falling circumflex contour, with the form L+(¡)H* (H)L%. Final rising or sustained pitch configurations are found in 21% of the information-seeking yes/no questions (36 interrogatives). Of these, only 10 interrogatives present the L* H% configuration of Castilian Spanish (6% of the total number of absolute questions), the rest having a rising nuclear accent, L+(¡)H* (¡)H%). There are no subjects with fewer than 64% of rising–falling nuclear contours, and two of the subjects have all interrogatives ending in this type of contour.
The circumflex nuclear contour of information-seeking yes/no interrogatives in Basque Spanish is found in Castilian Spanish in absolute interrogatives that are not information-seeking in nature but rather have other pragmatic nuances, at least in elicited speech (from read sentences or from Discourse Completion Tasks). Escandell-Vidal (1998, 1999, 2017) claims that a circumflex contour can appear in yes/no echo-questions, used to express surprise at what the interlocutor has just said or to beg a clarification from the interlocutor of what (s)he has just uttered (cf. also Estebas-Vilaplana & Prieto, 2010; Hualde & Prieto, 2015, p. 378). Torreira and Floyd (2012) find this contour in yes/no questions that serve the discourse function of signalling that the topic of the discourse is being followed up, or that the “course of action” is maintained. In our corpus, the absolute questions were of the genuine information-seeking type, which in Castilian Spanish have been reported to have a final rising configuration. In any case, in order to be able to establish a more direct comparison with central varieties of Peninsular Spanish, we recorded seven speakers from Madrid in conversations of the same type as those in the Basque Country. The results showed that two-thirds of the information-seeking absolute interrogatives ended in a rising configuration (a percentage that rose to 84% for five of the seven speakers). Thus, the relative frequency of appearance of rising and falling configurations was roughly the opposite in Basque Spanish and Madrid Spanish.
In principle, one could consider the hypothesis that the high frequency of final falling circumflex configurations in Basque Spanish is due to influence from Basque. However, we found no significant differences in the frequency of occurrence of falling or rising contours in absolute interrogatives depending on the city of origin (Bilbao or San Sebastian) or the linguistic profile of the subjects (i.e. whether they were monolingual speakers of Spanish, L1 Spanish/L2 Basque speakers or L1 Basque/L2 Spanish speakers). Rather, the hypothesis that we defend in this article is that more or less favourable attitudes have an impact on the production of features of Basque. That is, those speakers whose perceptions or attitudes towards Basque are more positive will be more likely to incorporate features of this language. In particular, we should perceive more production of Basque prosodic features in their variety of Spanish. The results of the attitudinal values obtained show that the majority of the subjects of our recordings (75%) have a positive or very positive attitude towards the Basque language and towards the Basque ethnolinguistic group, while 25% exhibited a less positive attitude. All the subjects in our recordings have a general positive attitude. Our results indicate that the correlation between attitudes and prosodic features of the Basque language is significant. Clearly, speakers with positive or very positive attitudes are grouped at the top of the scale with high percentages of falling questions, and speakers with less positive attitudes are placed at the bottom of this scale. However, this correlation is only of a medium strength. At least two speakers do not fit the expected behaviour.
The male and female monolingual subjects from Bilbao (MBiMon and FBiMon) should show a lower percentage of falling questions according to their attitudinal values. Neither of these two speakers shows strong negative attitudes towards Basque or towards the Basque group, but their perceptions are less favourable than the rest. FBiMon comes from a Basque speaking family on her mother’s side, but Basque was lost in the family two generations ago. She would like to speak this language and regrets that she could not learn it as a child, but she confesses having abandoned the idea of doing so. She never liked to study languages, and had a hard time learning Basque in school. This is why, when she finished compulsory education, she did not feel like continuing to learn Basque. She has made several attempts in her life, but has always been frustrated with the results because of the amount of effort she needed to put into it. She has not needed to speak Basque to find a job at any time, and this language is not essential for her to communicate with her friends or her family, although she is willing to have her son (2 years old at the time of the interview) learn Basque from an early age.
The MBiMon speaker does not exhibit openly negative attitudes, either. He recognizes the importance of speaking Basque in the Basque Country and the right of speakers to use this language. He has Basque-speaking colleagues in his job and feels comfortable listening to this language. However, Basque is not a priority for him at work or with friends, and he believes that other languages such as English are more important. If he decided to study a language at this moment, he would not study Basque. It seems important to him that children study Basque, but trilingual education models (Spanish/Basque/English) seem the best option for him.
Thus, these two subjects do not exhibit strong negative attitudes towards Basque, but neither one feels favourable to learn it or speak it. However, in both cases their results seem to be above expectations, FBiMon stands 8% above the baseline of 64%, and MBiMon is 11% above this line.
Another speaker that shows a possible divergent behaviour is the male L1 Basque subject from Bilbao (MBiL1B), who produces an almost identical number of falling questions as FBiMon (71.5%) even though he shows a rather positive attitude to Basque and the Basque-speaking ethnolinguistic group (2.11 out of a maximum of 3). MBiL1B is a native Basque speaker from a Basque-speaking family who although living in Bilbao, has been educated in Spanish and has not had much opportunity to speak Basque outside the family. He uses only Spanish in his profession, and Basque with friends and family. He is in favour of all measures that have been implemented for the revitalization of Basque, although in his opinion there are “more seductive” actions that can better convince speakers to speak Basque. He feels great affection for Basque, knows well its grammar and structure and enjoys listening to its different varieties, but he also considers it necessary to know other languages that have more prominence in the world, such as English. He sees many positive aspects in Basque society, but he also feels critical about some other chauvinistic attitudes among the Basque population.
A possible explanation for this behaviour may be related to other factors that are intervening in the linguistic behaviour of the speakers. As we have seen, the degree of contact with Basque and the Basque-speaking ethnolinguistic group is a relevant factor that has a great impact in linguistic production. There is an even stronger correlation between this variable and the percentages of rising–falling final contours among our subjects than the one with the variable of sociolinguistic attitudes. However, although the correlation was significant, linear regression analyses did not show a perfect fit. Degree of contact alone would not explain the behaviour of MSSMon, who presents 80% of falls with the lowest degree of contact (1), or why MSSL1Sp has the lowest percentage of falls (64%) with a higher degree of contact (1.29). A clearer correlation though was obtained when both social factors (the attitudinal values and the contact values of each subject) were combined. Then 80% of the differences in the percentages of circumflex contours could be accounted for.
A social variable to consider potentially in future work would be the degree of contact with varieties of Spanish outside the Basque Country. It seems that this variable could explain the behaviour of MBiL1B, the subject whose attitudinal value towards Basque and the Basque-speaking ethnolinguistic group is positive but has a percentage of falling circumflex contours at the end of interrogatives that is relatively lower than expected for this attitudinal value (71.5%). This subject has a substantial exposure to other Spanish varieties due to his frequent business trips to other parts of Spain, most notably Madrid. Thus, the exposure to the rising nuclear configuration in information-seeking interrogatives that are common in Castilian Spanish might be leading this subject to produce more interrogative utterances with this type of final contour. We leave this interesting and promising issue for further research.
Footnotes
Acknowledgements
The authors express their gratitude to the speakers who participated in the study for their generosity. This work is part of the research activities of HiTT (Basque Research Group in Theoretical Linguistics, University of the Basque Country UPV/EHU) and I-Communitas (Institute for Advanced Social Research, Public University of Navarre).
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by grants from the Spanish Ministry of Economy and Competitiveness (Grant reference FFI2012-38064-C02-01); the Spanish Ministry of Science and Innovation (Grant reference FFI2016-80021-P); the Basque Government (Grant reference IT1396-19); and the University of the Basque Country UPV/EHU (grant reference GIU18/221). These agencies are all non-profit organizations.
