Writing Samples

1 A Case of Grammaticalization in the Use of the Perfect for the Preterite in Bilbao Spanish. Introduction There is a general tendency in Indo-European languages for the Present Perfect (Auxiliar + Participle) to expand its contexts of use at the expense of the Preterite (Comrie, 1976; Harris, 1982; Fleischman, 1983, among others). However, this phenomenon does not take place in all of these languages and, especially, in all dialects to the same extent. The focus of this article is a variationist study of the increasing use of the PP instead of the Preterite in the Peninsular Spanish dialect spoken in Bilbao, in specific contexts where the Preterite was previously favored. This dialect merits study because it seems to follow the same diachronic processes as others that have already completed this change such as French or Italian. This phenomenon is of interest specifically to variationists because even though the PP is very frequently used in Peninsular Spanish in hodiernal (today) contexts (Schwenter, 1994), as Kempas (2006) points out it has not yet completely substituted the Preterite in all of them. In other words, the PP and the Preterite alternate with one another and they can be considered variants of a linguistic variable: the range of grammatical functions along the perfect-to-perfective path following Schwenter and Cacoullos (2008). Therefore, the main research question that will be answered at the end of this article is the following: What linguistic factors favor the use of the innovation (PP) instead of the Preterite?

The Spanish Case: Grammaticalization of an Anterior into a Perfective

2 It seems that the PP in some Spanish dialects is following the same path as other Romance languages (e.g. French, Italian and Romanian). It has already encroached on the semantic domain of the Preterite (Harris, 1982; Fleischman, 1983). Schwenter (1994) even predicted that “Alicante Spanish, like French, will continue to extend the semantic domain of the PP until it ultimately displaces the Preterite in spoken discourse.” (p. 80) He categorized the developmental stages of Romance languages proposed by Harris (1982) and Fleischman (1983) in the following way1: Stage 1: Present Perfect (PP) used as a resultative (Sicilian, Calabrian) Stage 2: PP begins to assume limited anterior functions (continuing or habitual). This is the case of Galician, American Spanish and Portuguese. Stage 3: PP for past actions with current relevance (Castilian Spanish, Catalan) Stage 4: PP for all past perfectives (French, Northern Italian, Romanian) But what are these specific uses of the Preterite that the Present Perfect is usurping? As Howe and Schwenter (2003) describe, the Preterite in Spanish describes punctual situations in the past. In other words, it presents past events as bounded situations. Moreover, it does not have any connection to the present by itself. This is why the Preterite is considered to be a perfective. The perfective function has recently been defined as a variety of grammatical aspect that involves the beginning and the end of an event (García-Fernández and Martínez-Atienza, 2003). In addition, the Preterite is the most appropriate verbal form for narratives since it refers to sequences of past events (Bybee et al. 1994). The PP, however, is a past tense still connected to the Present from 1

For more details, see Schwenter (1994, p.77).

3 stage 1 up to 4 (see above). In Peninsular Spanish, it encodes the anterior aspectual function since the past action is still currently relevant (the results of the past event may still be in force). When the PP advances to stage 4, the current relevant effects are missing and it can be said that it has adopted perfective (or aorist) functions with no connections to the present (Schwenter, 1994b, García Fernández, 2000). Therefore, this change is mainly aspectual. The use of the PP for the Preterite is a diachronic innovation of a form that is going through a process of grammaticalization. Hopper and Traugott (2003, p.1) explain that grammaticalization refers to the part of the study of language change that is concerned with such questions as how grammatical items develop new grammatical functions. That is, the PP does not disappear as a form but becomes something else: it incorporates the new functions to the previous ones (Lindstedt, 2000, p. 366). Serrano (1994) adds that the interaction between discourse and grammar is essential since the role that context plays is that of generating the appropriate frame for the form to adopt more than one grammatical function. According to her, what is happening here is actually a double grammaticalization process: the grammaticalization of the functions of the auxiliary and the compound form to acquire a perfective or aorist meaning. Therefore, she considers it a syntactic (aux + perfect) –semantic (anterior into perfective) change. According to her, this replacement of the Preterite by the PP takes place in order to present as relevant in the present a past action that the Preterite only shows as concluded. The PP can do that since it has the ability to relate the past and the present. Schwenter & Cacoullos (2008) propose an innovative hypothesis regarding this grammaticalization process in terms of “default status”. Their results suggest that the PP

4 is becoming the default expression of past perfective in Peninsular Spanish instead of the typical interpretation of remoteness, where the PP gradually extends into more remote past contexts. They argue that this shift takes place to a higher extent in temporally nonspecific or indeterminate past situations than in specific or definite temporal ones. Regarding the verbal forms, grammaticalization processes occur in the change of the following categories: mood, tense or aspect. In many languages and some Spanish dialects, the PP might be grammaticalized into a perfective or aorist. The coexistence of anterior and perfective forms is due to the fact that both functions can express the same event or state in the same way, they show the same way of conceptualizing situations. The verbal tenses that have developed these functions are selected by the speaker to express a kind of eventuality, to express their viewpoint about the structure of a situation based on these two uses (Alturo, 1999). Schwenter (1994b) argued that the “hot news” or immediate past use forms a bridge to advanced grammaticalization processes such as perfectives. This is what happens in Madrid, where the PP is used as a perfective or aorist in addition to the anterior or perfect function (a past situation is extended to the current time of communication). This is also the case of other cities such as Alicante (Schwenter, 1994) and Seville as reported by Squartini and Bertinetto (2000). Moreover, it is starting to emerge in the Canary Islands (Serrano, 1994; Piñero, 1998). Serrano argued that in the Canary Islands the anterior use of the PPs is the most frequent and the perfective was rarely being used. Madrid is supposed to be at an advanced stage of grammaticalization in Peninsular Spanish (Serrano, 1994). Within the framework of the gravity model of

5 linguistic diffusion of Trudgill (1974), one might infer that the change has spread out from Madrid because of its central location. This model proposes that change spreads from the largest to the next largest city, in a predictable order; the influence of one city on another is proportional to the size of the city and inversely proportional to the distance between them. This is why we can find a more advanced process of grammaticalization (similar to that of French, for example) compared to that of the Canary Islands speech (an island variety that obviously has been more isolated from the Peninsula). In Madrid, the use of the PP is categorical for past events occurring “today”(96%, data by Serrano, 1994, p. 48). Ex: Hoy he venido a dar un beso a mi abuelo (p. 49) Lit: “Today I have come to give a kiss to my grandfather”. It is frequently found in past events occurring “yesterday” (70%) but the Preterite is preferred for “before yesterday” situations (82%). Ex: Hace cosa de dos meses, J. y yo nos encontramos en una discoteca (p. 49) Lit. “About two months‟ ago, J. and me met in a club.” This indicates that the change is more advanced here since it does not distinguish the hodiernal-prehodiernal or „today-before today‟ distinction anymore (the norm for the standard Castilian Spanish, see Schwenter, 1994). These studies show overall frequencies so a multivariate statistical technique is necessary to get more accurate results. The substitution of the Preterite by the PP is an increasing change in some Spanish dialects. But what motivated this change in the first place? Serrano (1994) proposes that the motivation of this change was to relate past situations that speakers want to present as relevant in the time of communication in order to highlight and bring into the present the concluded event. That is, to make relevant in the present an action that the Preterite presents as concluded and perfective or aorist. Therefore, I am dealing

6 with a functionally oriented explanation and a process that responds to a communicative need of the speakers. According to what it has been explained so far, it may be tempting to think that the grammaticalization of the PP is very advanced in all Peninsular Spain but it is not. It is well known that there are still regions where the Preterite is used more than the PP for recent past events or within “today”: Asturias, Cantabria, León and Galicia. Their linguistic behavior regarding this phenomenon resembles more that of American Spanish (the PP is used for certain anterior functions but the Preterite is preferred in most past perfectives, see table 1, by Schwenter (1994, p. 77). In fact, according to Alarcos-Llorach (1994), hypercorrection is commonly found when they attempt to substitute the Preterite for the PP in spoken speech. This section intended to provide an overview on the grammaticalization of the PP across Peninsular Spanish dialects and point out several gaps in the literature. Next, we will proceed to the definition of the linguistic variable that will be the focus of this study to be able to respond to the research question that will addressed in this article: what internal factors favor the use of the PP in Bilbao Spanish.

Syntactic Variation: the PP vs. the Preterite In variationist sociolinguistics, most of the work has been done in phonology. As well known, two variants of a phonological variable have the same communicative purpose and the same truth value. In syntactic variation, however, there are usually slight differences in meaning. Therefore, there has been controversy about doing variationist studies with syntactic variables in the same way as phonological variables (see

7 Lavandera, 1978). Sankoff and Thibault (1981) propose the notion of weak complementary distribution as an alternative of truth-conditional equivalents between morphosyntactic variables. The two forms can either have the same function or different functions that alternate. Here the distinction between the two forms would not be relevant for the speakers so they could be considered two variants of the same variable. This is the case of the Preterite and the PP in Peninsular Spanish. In fact, Schwenter and Cacoullos (2008) claim that the PP is becoming the default expression in this dialect Serrano (1997) proposes to define meaning according to communicative purpose: “If the same communicative purpose can be expressed through different linguistic meanings that have no phonological, lexical or semantic similarity, variation is clearly being produced” (p. 1066-1067). Following Sankoff and Thibault (1981), she claims that we have to take into account the pragmatic purpose in which discourse is framed, not the truth values. In any case, it is crucial to delimit the context of variation in order to make the study of true syntactic variants possible. Following all this, Serrano explains that in Canarian Spanish the contexts in which the PP and the Preterite alternate occur when the event time and utterance time are very close. The Preterite fits in this context by assuming the pragmatic function of relevance in the present where “being relevant” means to have contextual effects. The definition of this variable that she proposes would be “reference close to the time of communication expression”. In Alicante, there is hardly individual variation for hodiernal past events. The loss of the Preterite here corroborates that the PP has grammaticalized into a true hodiernal perfective (Schwenter, 1994). This means, the PP is used for perfective functions when the temporal frame is today. Overextensions of the PP into

8 prehodiernal situations were found, which indicates a possible advance of this form into a prehodiernal perfective since younger speakers are the ones that use more the PP for situations occurring before “today”. Therefore, all this supports Schwenter‟s claim that the Alicante dialect is entering its final stage (stage 4) of grammaticalization (p. 78).

Recent Studies in Spain There are many recent studies in the literature of the grammaticalization of the PP into a perfective5 in Spanish (Bartens and Kempas, 2007; Copple, 2005; Howe and Schwenter, 2003; Kempas, 2005, 2006, 2009; Schwenter and Cacoullos, 2005; Serrano, 1998; among others). In Spain, there have been studies of this phenomenon mainly in Madrid (DeMello, 1994; Serrano (1994, 1995, 1998) and the Canary Islands (Herrera & Medina, 1994; Piñero, 1998, 2000; Serrano, 1994, 1995, 1998) in Alicante (Schwenter, 1994, Howe & Schwenter, 2003) and in Seville (DeMello, 1994). Recently, Kempas (2005, 2006) has also investigated the use of the two tenses in hodiernal contexts among students in several Spanish cities (Bilbao, Santander, León, Oviedo, Madrid and Granada) and Copple (2005) performed a diachronic study of these tenses with a corpus of written Peninsular texts from the 15th and 19th century and studied a corpus of the 20th century oral Peninsular speech. Schwenter (1994) shows that what is characteristic of Alicante Spanish is the hodiernal/prehodiernal distinction (the PP is used in hodiernal situations and the Preterite in prehodiernal situations), which even takes place in narratives. Squartini (1995)

5

See García-Fernández and Martínez-Atienza (2003) and Schwenter (1994b) for definitions of perfectives.

9 mentions that in Seville, even though this distinction occurs, the Preterite is used for narratives. Piñero (2000) claimed that Las Palmas de Gran Canaria‟s Spanish can be considered a transition area regarding the use of these two tenses since the PP is more frequently used than in varieties of American Spanish but less than in the Peninsular Spanish norm. In general, the grammaticalized hodiernal PP is still rare in the Canary Islands and it is frequently used in formal situations, especially in writing (Serrano, 1998). In contrast to the Canary Islands, the “perfective” PP is used in Madrid “mostly for expressing events close in time to the moment of communication” (Serrano, 1998, p. 127). It is completely grammaticalized for hodiernal past events and is starting to spread over prehodiernal contexts. In fact, Serrano‟s data show that the PP is still preferred in “yesterday” contexts but as long as the temporal frame increases in distance from the time of communication, the Preterite usage increases. This leads her to support the hypothesis that the PP will end up usurping completely the semantic domain of the Preterite following other Romance languages like French and Italian. Nevertheless, there is counterevidence for this. As Berschin (1976) showed, in Peninsular Spanish, between “ahora mismo” (right now) and “hace poco” (a short time ago) there is a difference in the use of the two tenses; that is, the PP is preferred for the first one since the adverbial contains the moment of speech and the Preterite for the second one. On the other hand, there was not a difference between “hace dos horas” (two hours ago) and “hace un mes” (a month ago); the Preterite is more used in both cases. This happens because the opposition between a close past and a remote one is established in subjective temporal

10 interpretations according to the communicative intentions of the speaker (Bustos Gisbert, 1995). Kempas‟ (2005) study supports this claim. He analyzed his data using a fill the blanks written task, in which the participants had to select PP/ Preterite according to the adverbial phrase of the utterance only in hodiernal contexts. With “hace dos horas” (two hours ago), the Preterite is prevalent in all the Spanish regions studied, being Bilbao the city with the highest percentage of occurrences of the PP and León and Oviedo with the fewest. The general use of the Preterite in Asturias and León for hodiernal past contexts is explained by the influence of the Galician-Portuguese domain. These two areas have been considered to be very conservative linguistically speaking (Kempas, 2006, p. 53). Among the verbs that were used, “estar” (to be) was the most frequent in all the cities and again Bilbao had the highest number of occurrences. In fact, the PP appears in Bilbao with all the verbs of the sample (“estar” to be, “llegar” to arrive and “venir” to come), when most of them only used it with “estar”. Kempas, then, concludes that Bilbao has the highest percentage of grammaticalized PPs in hodiernal contexts than the rest of the cities. This is also confirmed with the adverb “hoy” (today) followed by a subordinated clause starting with “luego” (afterwards). As hoy can be aspectually ambiguous, he introduced the adverb luego to make sure the only interpretation he is reading is the perfective one since the anterior does not allow a sequence of past events. When the action takes place in the immediate past, all the cities used the PP in all the occurrences, except for León and Oviedo that presented alternation with the Preterite (with predominance of the PP though). Kempas argues that what determined the selection of one form over the other was the temporal frame of the verbal form and the

11 geographical origin of the speaker. The fact that Bilbao data shows more occurrences with the PP than the rest of the dialect samples across the temporal contexts Kempas studied made him propose that the grammaticalization is more advanced there. His first hypothesis was that the reason might be that this city was the only one (within his sample), in which Spanish was in contact with another language, Basque. He discarded this hypothesis later since in Basque the contexts were PP and Preterite are used, are similar to those of Peninsular Spanish. In fact, the responses of people that spoke Basque as their first language did not significantly differ from those that spoke Spanish as their first language or those of Spanish monolinguals. In addition, we cannot forget that in Bilbao, Basque is a minority language. At the end of the 19th century and beginning of the 20th, Bilbao received a lot of immigration from other regions of Spain. Finally, he concludes that a better explanation would be that this is a grammaticalization phenomenon that might be followed by other Spanish regions later on as predicted by Trudgill‟s model. In the same way, Almgren and Idiazabal (2001) claim that the Basque PP fulfills similar functions to the Spanish one. Among all the Peninsular dialects studied in the literature, the frequent use of the PP in prehodiernal contexts (as we know, it is already predominant in hodiernal contexts in most of the country) has often been associated with Madrid. As the capital city of Spain and the influences it receives from other speech communities from within and outside the country, it is logical to expect this to be the case following Trudgill‟s linguistic diffusion model. In any case, Kempas (2006) suspects that the prehodiernal PP is preferably used in oral speech, it has not been yet extended to writing. Copple (2005), nevertheless, found that 25.2 % of the prehodiernal contexts (overtly modified and

12 understood frames) were already encoded in the PP in her study of written texts of the 19th century. In her study of the 20th century oral Peninsular speech, this percentage increased into 45% for “yesterday” and “last year” contexts confirming once more the extension of this tense into prehodiernal contexts as predicted by several researchers such as Schwenter (1994) and Serrano (1994). Despite this, other researchers like Howe and Schwenter (2003) consider this use to be “odd” in Peninsular Spanish but they do not claim it is agrammatical. This is due to pragmatic differences cross-dialectally. However, as Kempas explains, nobody would doubt saying that the hodiernal PP is grammatical in Peninsular Spanish. According to him, the explanation for the use of the prehodiernal PP might be found in the association of the time of the event with the day of communication (p. 271). This association would be based on the emotions of the speaker. It is a mental association since emotionally, this event is as present in his life as if it had happened during the day of communication. In other words, it is a temporal identification between the today and the yesterday of the speaker (p. 272). He suggests this might be interpreted as current relevance. Even though there were not a lot of examples of prehodiernal PPs in his data (Oviedo had the highest frequency though), he argued that their occurrence indicated that this use was in process of grammaticalization in these speakers‟ dialects. He finally concluded that the main area of the prehodiernal PP use can be located in Asturias and Cantabria (Santander has the second highest frequency) despite the still predominant use of the Preterite in these regions. In other words, even though the Preterite is preferred in prehodiernal contexts, the frequencies of the PP are higher in these two regions than in the rest. He claims that this use has gone into further grammaticalization in Asturias than

13 in the other regions since to them, due to the hypercorrection (the overuse of the PP trying to follow the standard norm), it sounds as grammatical as the hodiernal PP. In any case, this is something local. At the same time, Granada and Alicante show the lowest frequency in Kempas‟ data. This actually contradicts what Schwenter (1994) and Serrano (1994) found for Alicante and Madrid respectively. Regarding the word order, the prehodiernal PP is favored by the postverbal position of the adverb. What seems to be determinant in the use of the PP in this context is where the temporal adverb occurs. If it occurs before the predicate, the PP will not arise very frequently since the adverb already defined the time of the event but if it occurs after the predicate, it might do it more often because of this lack of specification (Brugger, 2001, Kempas, 2006). So far, it does not seem that there is an agreement about the motivation of this change across these studies. Reyes (1990) claims that there has been a desubjectivization of the perfect, the “past” value has imposed over the current relevance one. That is, the PP has substituted the Preterite in order to specify a past action close to the time of the communication. For Schwenter (1994), the explanation would be that the boundary between the “anterior” function with current relevance and the “perfective” one has disappeared. That is, “gradual relaxation of the current relevance requirement is the key to this change" (Schwenter 1994:78). Finally, Serrano (1995) proposes that PP “achieves sufficient communicative efficiency to acquire a new grammatical function.” (p. 59) She offers a function-oriented explanation that responds to the speakers‟ communicative needs. Apparently, there is a difference regarding the use of the PP and the Preterite in the East Basque Country (the French provinces) and in the West (the Spanish provinces).

14 In the East, the PP is used for prehodiernal and hodiernal situations due to the influence of French whereas in the West, there is a prehodiernal-hodiernal distinction following the Spanish tendency (Rotaetxe, 1988). However, what is really relevant for this work regarding language-contact in this language is what Haase (1992) explained about young people replacing the Preterite with the PP. In fact, he suggests that this contact-induced change does not come from Spanish but from French, where the grammaticalization of the PP is already completed. This section has provided an overview of the recent findings on the PP across Peninsular dialects included the target dialect (Bilbao Spanish). As seen, this dialect has only been studied by Kempas but he used a written task to collect his data, which could be problematic since the choices subjects make could reflect more their attitude towards this phenomenon or the impression they want to make on the researcher rather than what they would actually say if they did not pay attention to the way they speak. That is why, I consider essential to conduct the study using sociolinguistic interviews and running a multivariate statistical technique to get a probabilities, which would provide more accurate results than frequencies. Therefore, I proceed now to explain my methodology to carry this study.

Methodology

The Community According to the Spanish National Institute of Statistics (INE), there are 20,000 more women than men living in Bilbao, making it a total of around 350,000 citizens. If we take into account all the surrounding towns on both sides of the river that separates

15 the city, there are more than 500,000 people who travel to Bilbao on a daily basis either to work or to study. In Bilbao, Basque is a minority language. Due to its industrial development at the end of the 19th century and at the beginning of the 20th, it has received many immigrants from other Spanish regions, which makes its Spanish a very appealing focus of study. Therefore, the subjects that I considered for my study were monolingual and bilingual speakers, who have spent (lived, studied or worked) in Bilbao all or most of their lives. According to Fernández Ulloa (2001), in the Biscay province 0.68% of the population is monolingual in Basque, 15% are bilingual, 7.5% are passive bilingual (those who understand or have some knowledge of Basque) and 76% are Spanish monolinguals. I conducted around 49 interviews among people who have lived, studied or worked in Bilbao most of their lives. The interview has been semi-directed due to the facility of obtaining occurrences in these two past tenses in a spontaneous conversation. However, at the end of around 45 minutes of interviews, they were asked to narrate what they did that same day, the previous day and two months ago. This was done in a way in which the interviewer omitted the use of a verb to avoid influencing the choice of a tense by the interviewee following Schwenter (1994) and Serrano (1994): “Cuéntame tu día hoy, ayer, hace dos semanas.” Tell me about your day today, yesterday, two weeks ago. Once I collected all the data, I classified the tokens of PP or Preterite modified by temporal adverbs taking into account several internal constraints: temporal modification in hodiernal and prehodiernal contexts, the absence of temporal modification, the occurrence of the adverb before or after the verbal form or the use or not of the first person in the verb.

16 I recorded 49 interviews, out of which 36 were finally used. Each interview was at least 45-minutes long and a minimum of 50 tokens with any of the two variants (PP/ Preterite) and a maximum of 200 were extracted from every interview. In fact, 5520 tokens have been used for this project. I considered 20 linguistic factor groups and tested them on VARBRUL (Sankoff, Tagliamonte and Smith, 2005) in order to get frequencies of use of PP/ Preterite for every linguistic factor group. Then, I proceeded to obtain probabilities in order to be able to get more accurate results. Lastly, the interpretation of the results has taken place.

The linguistic variable (the dependent variable) and the internal constraints (independent variables). The linguistic variable that is going to be studied in this project is a morphosyntactic one: the change from perfect to perfective (Schwenter and Cacoullos, 2008). The PP and the Preterite are variants of the same variable. Both forms serve the same communicative purpose as Serrano (1997) stated. The 49 interviews were recorded from 2004 up to 2006 for this project. I excluded the tokens that were within incomplete sentences or when the sentences or the clauses were repeated, I only counted the first one. Within these interviews, I only considered those that had a minimum of 50 tokens and a maximum of 200, based on Cameron (2005). These tokens were encoded for the following factor groups or internal constraints. I include here examples taken from my data for every factor group (due to space limitations, only one per factor group; also, the word that is relevant for every factor group is underlined). In the following section, predictions will be explained after the factor groups are presented:

17 1. Temporal Reference: Today, Yesterday, 2 days ago, more than 2 days up to 7, more than 7 days up to a month, more than a month up to 6 months, more than 6 months up to a year, more than a year up to 5, more than 5, not specific and no temporal reference at all. The temporal reference was determined from the discourse and from the previous knowledge of the interviewer regarding when the events took place due to the fact that most speakers belonged to the social network of the interviewer. Adverbs that cannot be placed in any other category such as frequency adverbs were included in the not-specific one. Copple (2005) found, as it is expected since it is its prototypical context, that the PP is more frequently used with this temporal reference. Schwenter and Cacoullos (2008) also found this group to be significant. The PP was the most frequent variant in today, irrelevant and indeterminate contexts and it drastically drops in yesterday and before contexts. What is interesting about this factor group is its magnitude as they indicated, the effect of the temporal reference overshadows other effects. Schwenter and Cacoullos claim in Peninsular Spanish there is a true temporal effect, “with hodiernal contexts strongly favoring the PP, with a probability of .93 and before today contexts strongly disfavoring the PP with a probability of .13.” (p. 10) so I expect to find similar results. Ex: “Esta mañana no he ido a trabajar.” Lit. “This morning I have not gone to work.” 2. Temporal Adverb or Clause: Yes, No. I expect the PP to occur more often without a temporal adverb.

18 Ex: “Eso es algo personal que he tenido yo con ella.” Lit. “This is something personal that I have had with her” 3. Place of the Adverb or Clause: Today before, Today after, Yesterday before, Yesterday after, Two days ago before, Two days‟ ago after, Others before, Others after, Does not apply. My expectation was to find a lower probability of PPs when the adverb occurs before the tense due to the influence of the temporal specification in before-today contexts. Adverbs of today are considered all that occur within today. The same happens with the rest of the categories, others being considered previous to two days ago. Ex: “A Alex lo he visto hace una semana.” Lit. “Alex I have seen a week ago”. 4. Distance of the Adverb or Clause from the verb: Before within the same clause, Before within the preceding clause, Before within 2-3 preceding clauses, Before 4 or beyond, After within the same clause, After within the next clause, After within 2-3 next clauses, After 4 or beyond, Does not apply. Ex: “He tenido todo el día dolor de garganta” Lit. “I have had throat sore all day” This was coded as “after within the same clause”. I would expect to find the PP more often where the structural distance is bigger. 5. Person and Number: 1s, 1p, 2s specific 2s nonspecific, 2p, 3s, 3p. This category is taken from Copple (2005). Schwenter and Cacoullos (2008) expected to find the PP more used with the first person for subjectivity (deictic) matters that are associated to this tense. Kempas (2009), however, contradicts this. He found that in prehodiernal (before-today) PPs, personal deictic words play a marginal role but I

19 was unable to account for this matter since I included all PPs in one category. In any case, the second person was more favored by the PP than any other sets in my data. Ex: “No le has devuelto la llamada.” Lit. “You have not returned his call.” 6. Verb Semantics: Movement, Stative, Cognitive, Emotions, Communication, Perception, Others. These categories are based on Downing (1996) and Copple (2005). Copple found that emotion verbs are the group that has a higher frequency of occurrences in PP since these kind of verbs express feelings that tend to continue into the present as it generally happens with stative predicates. There has also been a shift from the Preterite to PP in the Movement verbs even though they occurred in a lower frequency than other categories such as perception, communicative, stative or cognitive verbs. Ex: “No he ido allí.” Lit. “I have not gone there.” This is an example of a movement verb. 7. Transitivity: Transitive, Intransitive. In order to give account for transitivity, I took into account whether the verb was used in a transitive or intransitive way (whether they had a direct object or not). Those verbs that admit indirect object but do not admit direct objects such as gustar (to like) verbs are, therefore, considered intransitive. Following Copple‟s (2005) categorization of object pronouns, I considered reflexive verbs as intransitive as can be seen in the following example: Ex: “Me he levantado a las 8.” Lit. “I have got up at 8.” 8. Mood: Subjunctive, Not in Subjunctive.

20 This category and the next two are intended to test whether mood somehow influences the use of one tense or the other. In the following example, we find the occurrence of the PP in subjunctive but it is important to mention that in Spanish, the subjunctive takes place within a subordinate clause most of the times. Therefore, it is predictable to think there might be an interaction between main and subordinate clauses when it comes to subjunctive. However, this goes beyond the scope of this article. Ex: “No creo que haya estado aquí antes.” Lit. “I don‟t think I have been here before.” 9. Mood of the Previous Verb: Subjunctive, Not a Subjunctive. In this category and the following one, my purpose was to look for priming effects regarding mood. Ex: “Esa película no me ha gustado mucho. Me he reído muy poco.” Lit. “I have not liked that movie that much. I have hardly laughed. 10. Mood of the Following Verb: Subjunctive, Not a Subjunctive. Ex: “Le he recomendado que no vaya a ver esa película.” Lit. “I have recommended him not to go to see that movie.” 11. Tense of the Previous Verb: Infinitive, Present, Preterite, Present Perfect, Imperfect, Periphrastic Future, Future, Past Perfect, Other. 11 and 12 categories are designed to look for priming effects (repetition of a previously used form). Copple (2005) found that the Present and the PP favor the PP and the Imperfect favors the Preterite. Schwenter and Cacoullos (2008) found evidence of priming effects too.

21 Ex: “Eso me ha marcado, claro, mi vida ha cambiado radicalmente.” Lit. “That has affected me, of course, my life has changed radically.” 12. Tense of the Following Verb: Infinitive, Present, Preterite, Present Perfect, Imperfect, Periphrastic Future, Future, Past Perfect, Other. These last 4 categories are designed to detect priming effects. Even though a structure (or structures) that primes another, would presumably come before the target structure, in this case the PP or the Preterite, I considered it would be interesting to take a look at the following verb to see if the target token had an influence on it. I would expect so. Ex: “Hoy he quedado con mi madre para tomar un café.” Lit. “Today I have met my mother to have a coffee.” 13. Narratives: Verb within a narrative, verb not within a narrative. The definition of narrative is taken from Labov (1972) referring to past events within clauses ordered in a temporal sequence. Narratives are generally associated with the perfective aspect since they involve temporal bounded past actions and consequently to the Preterite tense. However, the emerging occurrence of the PP within narratives in Peninsular dialects (Schwenter, 1994; Serrano, 1994; Kempas, 2006) accounts for the semantic bleaching of the PP in these dialects, assuming the perfective function previously relegated to the Preterite. This compatibility of the PP with narrative sequencing is a feature of the development of grammaticalization (Howe and Schwenter, 2008). Nevertheless, there are differences across dialects (Howe, 2006). Ex: “He tenido que levantarme temprano porque tenía una reunion. Después, he dado mis clases y me he ido a comer con Ana.” Lit. “I have had to get up early

22 because I had a meeting. Afterwards, I have taught my classes and I have gone to lunch with Ana.” 14. Clause type: Main, Subordinate. According to Schwenter and Cacoullos (2008), the PP is more expected in relative clauses since they encode background information which is the function perfects are supposed to fulfill in narratives. Ex: “Creo que ha estado un mes en Londres.” Lit. “I think that she has been in London for a month.” 15. Animacy of the object pronoun: Participant, Close relationship, Other humans, Inanimate, No pronoun used, Does not apply. This categorization is taken from Copple (2005). She did not find this factor group to be statistically significant but it has been tested again in this study. Double objects were excluded from the analysis so I placed them within “does not apply”. Interestingly, Copple did not mention anything regarding this. She did not consider the reflexive se as a direct object pronoun either since it was not in the chart she used so as I followed the same categorization, I considered reflexive verbs to be intransitive. Within the no pronoun used category, I considered not only transitive verbs that did not have an object pronoun but intransitive verbs as well, this is, all the cases where there was no pronoun. Ex: “He ido al dentista esta tarde.” Lit. “I have gone to the dentist in the afternoon.” 16. Telicity of the verb: Telic, Atelic. Telic verbs are typically associated with the perfective aspect since what defines them is that they have an endpoint and perfective aspect encodes completed events.

23 Therefore, telic verbs are expected to be found in Preterite tense. Atelic verbs, however, refer to states and activities so they are usually encoded in the PP. This is why, the occurrence of the PP with telic verbs would point more strongly towards the perfectivity of this tense (see Vendler, 1967). As expected, Copple (2005) found atelic verbs are used more in the PP tense and telic verbs in the Preterite. Ex: “Mi madre ha estado aquí esta tarde.” Lit. “My mother has been here in the afternoon.” 17. Polarity and clause type: - Declarative, + Declarative, - Interrogative, + Interrogative. In this category, - Declarative would stand for a negative declarative clause vs. + Declarative that would for an affirmative declarative clause. The same applies to interrogative clauses. This category was found not to be significant by Copple (2005) but it has been tested again in this study. She found a tendency though towards the use of the PP more for questions and in negative assertions since most of the negative clauses had an indefinite temporal reference which is the typical context of occurrence of the PP. Ex: “Nunca he estado en Italia.” Lit. “I have never been to Italy.” 18. Irregular Form in the Preterite: Yes, No. Verbs whose form in the Preterite are irregular (those that no follow the regular rule established for –ar, -er, or –ir verbs) are expected to occur more in the PP. If this factor turns out to be significant, this would be an original contribution to the field since it is not usually treated in the literature on the Spanish PP. Ex: “Ha traído mucha comida del super.” Lit. “She has brought plenty of food from the grocery store.”

24 19. Information status of the subjects: Discourse- Old, Discourse- New, Does not apply. This category is based on Prince (1992), who claimed that Discourse status accounts for the subject-non subject distribution: “ Thus, of the three factors studied- Discourse status, Hearer- status and Definiteness, one was found by the VARBRUL program to account all by itself for the subject- non subject distribution: Discourse-status.” (p. 316) If the subject can be inferred from the discourse, I would consider it discourse old. If Not, discourse-new. The “does not apply” option refers to impersonal verbs, those verbs that do not have a subject such as haber (there is, there are) or weather verbs or expressions such as llover (to rain). Ex: “Ha hecho mal tiempo.” Lit. “There‟s been bad weather.” 20. Temporal Adverb: Location, Durative, Frequency, Does not apply. This classification is based on Smith‟s (1991) and Copple (2005). We might expect the Preterite to occur more with temporal adverbs of location (such as ayer “yesterday”) and the PP with durative (the focus is on the length of the action) and frequency (how often the event happened) adverbs. Schwenter and Cacoullos (2008) found this factor group to be significant in the variable-rule analyses. Ex: “Ha sido siempre así.” Lit. “It has been always like this.”

25 General Predictions What do I expect to find in my data? Based on previous research, these are my predictions: Prediction 1: In prehodiernal contexts, the use of the PP is almost non-existent as Kempas (2006) found in his Bilbao sample. In fact, when it does occur, it will mostly when it is not modified by a temporal adverb (Copple, 2005; Kempas, 2006). The goal of this prediction is to corroborate their findings. Prediction 2: Certain semantic verb classes such as Perception, Stative, Cognitive and Emotion verbs are more likely to occur in the PP since they allow a fluid conception of time, a conception of the past in present; this is, these verbs do not have a clear beginning or endpoint (Downing, 1996). Prediction 3: There will be priming effects (the occurrence of a particular form primes a subsequent form) so the Present and the PP will favor the PP and the Imperfect and the Preterite will favor the Preterite as claimed by Copple (2005). Prediction 4: Even though narratives will be more frequent in the Preterite, PP will also be found in narratives as in previous research such as Schwenter (1994) or Serrano (1994). I will attempt to corroborate these findings. Prediction 5: Verbs with an irregular form in the Preterite are more frequently encoded in the PP due to the difficulty of accessing this irregular form. To sum up my predictions, the factor groups that are expected to be statistically significant and, therefore, contribute to the selection of the PP over the Preterite are the

26 form of the previous tense, the temporal reference, the temporal adverbs and the telicity of the verb (Copple, 2005). Results In order to present a multi-variate analysis, a binomial varbrul (Goldvarb X) was run with step up and down: The following groups were considered to be significant: temporal reference, person and number, mood, following verb‟s mood, previous verb‟s tense, following verb‟s tense, narratives, telicity, polarity and clause type and type of temporal adverb. Therefore, the groups that were eliminated were: the presence or absence of a temporal adverb, the temporal distance of the adverb, verb semantics, transitivity, previous verb‟s mood, clause type, animacy of object pronoun, irregular form in the Preterite and information status of subjects. Now, a one level analysis is run with the all the groups that were significant in order to get probabilities. These groups are ranked in order of importance (range): Group

Range

1) Temporal Reference

96

2) Previous Verb‟s Tense

65

3) Person and Number

51

4) Following Verb‟s Tense

46

5) Mood

44

6) Temporal Adverb Type

36

7) Polarity and Clause Type

30

8) Following Verb‟s Mood

25

27 9) Telicity

18

10) Narrative

17

N= 1900 (PP), Total Chi-square = 5746.9725, Chi-square/cell = 1.2142, Log likelihood = -1233.757 The probabilities are the following:

Temporal Reference Table 1 Temporal Reference ____________________________________________________________________ Probability % PP Total N % of data __________________________________________________________________ today .99 98% 836 44% not temporally modified

.76

63%

497

26%

not specific

.64

50%

350

18%

+ 1 month- 6 months

.44

15%

15

0.7%

+ 6 months- 1 year

.42

15%

7

0.3%

+ 1 year- 5 years

.24

11%

25

1.3%

+1 week- 1 month

.23

7%

6

0.3%

+ 5 years

.18

8%

140

7.3%

yesterday

.10

3%

16

0.8%

2 days‟ ago

.07

2%

6

0.3%

3 days- 1 week

.03

1%

2

0.1%

Range 96 _______________________________________________________________________

28 The temporal frame of today is the one that favors the innovation the most (.99) When there is no temporal frame that can be inferred from the discourse (.76) and the temporal frame is not specific (.64), the PP is still favored. These results are similar to those of Schwenter and Cacoullos (2008). However, in the rest of the cases the innovation is disfavored. There is not a correspondence between the variant and the distance of the temporal frame from the present: in yesterday contexts (.10), the innovation is more disfavored than in the following intervals: from a week into a month (.23), from a month into 6 months (.44) and from a month up to a year (.42). Surprisingly, in the intervals from a month up to 6 months (.44) and from 6 months into a year (.42), the innovation is slightly disfavored. Finally, the PP is the most disfavored variant in the temporal frame of two days up to a week. (.03)

Previous Verb‟s Tense Table 2 Previous verb’s tense ______________________________________________________________________ Probability % PP Total N % of data ______________________________________________________________________ PP

.85

82%

691

36%

Future

.66

57%

12

0.6%

Present

.63

46%

564

30%

Infinitive

.56

34%

229

12%

Other

.55

35%

149

8%

Periphrastic Future

.49

55%

16

0.8%

29

Imperfect

.35

18%

116

6%

Preterite

.23

6%

113

6%

Past Perfect

.20

17%

10

0.5 %

Range 65 _______________________________________________________________________ The PP is more favored when the previous verb is also a PP (.85), what constitutes priming. It is still favored after a future (.66), a present tense (.63), an infinitive (.56) and other tenses that were not listed (.55). It is disfavored in the rest of the cases, above all after a Past Perfect (.20) and a Preterite (.23) as expected. Person and Number Table 3 Person and Number _____________________________________________________________________ Probability % PP Total N % of data _____________________________________________________________________ 2 s non specific

.89

83%

5

0.02%

2 s specific

.74

74%

94

0.5%

3p

.55

32%

152

1%

1s

.54

37%

871

46%

3s

.46

31%

564

30%

2p

.46

57%

15

0.1%

1p

.38

24%

199

11%

Range 51 ________________________________________________________________________

30 The PP is more favored when it is used in the second singular form in specific (.74) or nonspecific (.89) “you”. In the first person singular (.54) and third person plural (.55) the variant is very lightly favored. On the other hand, in the rest of the persons it is disfavored, specially in the first person plural form (.38).

Following verb‟s tense Table 4 Following verb’s tense ______________________________________________________________________ Probability % PP Total N % of data _______________________________________________________________________ PP

.86

84%

657

35%

Future

.72

58%

7

0.3%

Periphrastic Future

.62

50%

22

1%

Present

.59

46%

551

29%

Infinitive

.49

29%

205

11%

Other

.47

33%

149

8%

Imperfect

.43

22%

158

8%

Past Perfect

.39

20%

11

0.5%

Preterite

.26

8%

140

7%

Range 46 ________________________________________________________________________

The PP is favored when the following verb is also a PP (.86) and a future

31 (.72) or a periphrastic future (.62) and very slightly favored it is followed by a present (.59). When the following verb is an infinitive the PP is not clearly favored or disfavored (.49). With other tenses, the PP is lightly disfavored (.47) and the same happens with the imperfect (.43). With the Past Perfect (.39), and the Preterite (.26), it is clearly disfavored. Since the previous verb is the one that primes on the following verb, if the latter is the PP, it is expected that another PP occurred before it. The same happens with the Preterite, obviously.

Mood Table 5 Mood _____________________________________________________________________ Probability % PP Total N % of data _____________________________________________________________________ Not in subjunctive

.51

34%

1876

97%

Subjunctive

.07

20%

24

2%

Range 44 _____________________________________________________________________

When the verb is not in subjunctive, the PP is lightly favored (.51). However, when it is, the PP is very strongly disfavored (.07); there is hardly variation.

32 Temporal Adverb Table 6 Temporal adverb ________________________________________________________________________ Probability % PP Total N % of data ________________________________________________________________________ Frequency

.78

54%

253

11%

Duration

.67

34%

154

11%

Location

.42

26%

771

76%

Range 36 ______________________________________________________________________

The PP is favored when the temporal adverb is of frequency (.78) or duration (.67) and disfavored when it is of location (.42). Polarity and Clause Type Table 7 Polarity and clause type _______________________________________________________________________ Probability

% PP

Total N

% of data

______________________________________________________________________

Negative interrogative

.78

83%

15

0%

Negative declarative

.66

49%

185

6%

Affirmative interrogative

.57

45%

92

3%

Affirmative declarative

.48

32%

1608

89%

Range 30 _______________________________________________________________________

33 The PP is favored in negative clauses, especially in the interrogative ones (.78). It is lightly favored in the interrogative affirmative clauses (.57) and slightly disfavored in the affirmative declarative ones (.48).

Following Verb‟s Mood Table 8 Following verb’s mood _____________________________________________________________________ Probability % PP Total N % of data _____________________________________________________________________ Subjunctive

.74

54%

38

1%

Not in subjunctive

.49

34%

1862

98%

Range 25 _____________________________________________________________________

The PP is favored when the following verb is in subjunctive (.74) and disfavored when it is not (.49). The fact that the verb‟s mood and the following verb‟s mood are significant factor groups is a very interesting issue that is beyond the scope of this article.

Telicity

34 Table 9 Telicity ____________________________________________________________________ Probability % PP Total N % of data ____________________________________________________________________

Atelic

.59

29%

874

53%

Telic

.41

39%

1026

46%

Range 18 ____________________________________________________________________ With atelic verbs, the PP is favored (.59) as expected and with telic verbs is not. (.41)

Narratives

Table 10 Narratives _______________________________________________________________________ Probability % PP Total N % of data _______________________________________________________________________

Not narratives

.63

54%

663

21%

Narratives

.46

28%

1237

78%

Range 17 _______________________________________________________________________ In narratives, the PP is lightly disfavored (.46) while in non-narrative genres it is not. (.63). These results provide evidence of the perfective function that PP

has started to adopt since narratives are the typical genre, in which the Preterite is expected to occur.

35

Discussion The discussion will be carried out by testing the predictions: Prediction 1: In prehodiernal contexts, the use of the PP will be almost nonexistent in keeping with the findings of Kempas (2006) in his Bilbao sample. In fact, when it does occur, the majority of the PPs will occur when it is not modified by a temporal adverb (Copple, 2005). This prediction does not hold for the most part since the PP can be found to different extents in all prehodiernal contexts as it is shown in the differences in probabilities across the prehodiernal categories. In some cases like (1 month- 6 months) and (6 months-1 year), the probabilities of occurrence of the PP are worth of further observation (.44 and .42 respectively). What is more, it is favored in not specific and not temporal modified clauses as expected from the literature. Prediction 2: Certain semantic verb classes such as Perception, Stative, Cognitive and Emotion verbs are more likely to occur in the PP since they allow a fluid conception of time, a conception of the past in present (Downing, 1996). This prediction holds for the most part. Even though the Preterite is the most preferred variant in all cases, it is true that the percentages of the PP in some types of verbs are higher than other verbs as it is the case of those of emotion, perception and cognition. Apparently, this is not the case of stative verbs. However, if we look closer, we can notice that despite the fact that the percentage is somewhat low, the number of occurrences is not. In any case, this group was found to be significant by chi-squares but eliminated under the one-level analysis of VARBRUL.

36 Prediction 3: Atelic verbs (stative and activity) are more frequently encoded in the PP than the telic verbs as found in Copple (2005). However, an increase in the use of the PP with telic verbs is noticed, which are the result of the erosion of the anteriority of the PP. The prediction holds: atelic verbs have a higher probability of being encoded as a PP but telic verbs are really close. This group was considered to be significant by chisquare and one-level analysis of VARBRUL. Prediction 4: There will be priming effects so the Present and the PP will favor the PP and the Imperfect and the Preterite will favor the Preterite as claimed by Copple (2005). The prediction holds: the PP and the Present favor the PP and the Preterite and Imperfect disfavor it. It is important to point out that there are other tenses that favor the PP. These are the Future, Infinitive and Other categories. On the other hand, the Past Perfect favors the Preterite. Prediction 5: Even though narratives will be more frequent in the Preterite, PP will also be found in narratives, as it is noted in previous research such as Schwenter (1994) or Serrano (1994). The prediction holds: even though the PP is disfavored within narratives, it is very slightly disfavored. This group was significant by chi-square and one-level analysis. Despite what happens in the vast majority of other varieties of Spanish, I actually do see instances of PPs in narrative contexts. Prediction 6: Verbs with an irregular form in the Preterite are more frequently encoded in the PP due to the difficulty of accessing the irregular form.

37 The prediction does not hold: Verbs with an irregular form in the Preterite are not encoded in the PP more frequently than those with regular forms. This factor group was not significant in either chi-squares or the probability based model. To summarize, my results support Copple‟s (2005) and Schwenter and Cacoullos‟ (2008) findings. Moreover, in my project I discovered other linguistic factors that also contribute to the selection of the innovation over the Preterite. These factors are: person and number, mood, following verb‟s mood, following verb‟s tense, narratives and polarity and clause type.

Answering the Research Question Previously, I asked the main research question of this article: What internal factors favor the use of the innovation (PP)?

After all discussed above, this is the list of the internal factors that favor the use of the PP over the Preterite: temporal reference, person and number, mood, following verb‟s mood, previous verb‟s tense, following verb‟s tense, narratives, telicity, polarity and clause type and type of temporal adverb. These results support previous research and provide a new contribution to the field. Some interactions are suggested such as polarity and clause type and mood. It would be interesting to test the significance of these interactions in future research.

Conclusion

38 This study was a challenge for me since no research had been done in Bilbao, a bilingual city (Spanish and Basque) even though it had the highest frequency (out of a sample of several Spanish cities) of grammaticalized hodiernal PPs according to Kempas (2006). The influence of Basque on the Spanish spoken in that area was later discarded as an explanation for this. The fact that I am a native speaker of Bilbao Spanish and that I also speak Basque made me more curious about this issue. Previous research on the topic opened the path to me, especially for selecting linguistic factor groups that favored the change (Copple, 2005, Schwenter and Cacoullos, 2008) and regarding methodology. I finally adapted mine from Schwenter (1994) and Serrano (1994). I decided to record semi-elicited sociolinguistic interviews to get vernacular speech: they were open interviews at the beginning based on Labovian modules and I asked the interviewees to narrate me the present day, the previous one and the day before to get enough data. Factor groups are run under a one-level analysis to get a model based on probabilities which is more reliable than frequencies and eliminates the factors that are not significant under this model. A Step-Up and a Step-Down are run and eleven factors (ten of them linguistic) are selected for the model. All of these linguistic factors were considered significant: temporal reference, person and number, mood, following verb‟s mood, previous verb‟s tense, following verb‟s tense, narratives, telicity, polarity and clause type and type of temporal adverb. These findings support previous research by Copple (2005) and Schwenter and Cacoullos (2008) except for the case of telicity, polarity and clause type that were not found significant earlier. In addition, they provide new evidence of significance in the following cases: mood, following verb‟s mood,

39 following verb‟s tense and narratives. Hence, this article provides additional descriptive support for the claim that the PP/ Preterite grammaticalization pathway is widespread in several varieties of Peninsular Spanish.

REFERENCES Alarcos Llorach, E. (1994). Gramática funcional del español. Real Academia Española Espasa Calpe. Almgren, M. and Idiazabal, I. (2001). Past tense verb forms, discourse context and input features in bilingual and monolingual acquisition of Basque and Spanish. In Cenoz, J. & Genesee, F. (eds.), Trends in Bilingual Acquisition, 107-130, Amsterdam: John Benjamins. Alturo, N. (1999). El papel de la anterioridad y de la perfectividad en la representación de estados y eventos. In Serrano, M. J., Estudios de variación sintáctica, 143-172. Frankfurt, Germany; Madrid, Spain: Vervuert; Iberoamericana. Bartens, A. and Kempas, I. (2007) Sobre el valor aspectual del Pretérito Perfecto en el español peninsular. Revista de investigación lingüística, 151-171. Berschin, H . (1976). Praeteritum und Perfektgebrauch im heutigen Spanisch. Tuebingen: Niemeyer. Brugger, Gerhard. 2001. Temporal Modification, the 24-hour rule and the location of reference time. In Current Studies in Spanish Syntax and Semantics, Luis SilvaVillar and Javier Gutiérrez-Rexach (eds.), 243-270. Berlin/New York: Mouton de Gruyter. Bustos Gisbert, J. M. (1995). La temporalidad en español: análisis intencional. Lingüística Española Actual, 17, 143-66. Bybee, J., Perkins, R. and Pagliuca, W. (1994). The evolution of grammar:tense, aspect, modality in the languages of the world. Chicago: University of Chicago Press. Cameron, R. (2005). Aging and Gendering. Language in Society, 34 (1) 23-61.

40 Comrie, B. (1976). Aspect. Cambridge: Cambridge University Press. Copple, M. (2005). A diachronic study of Peninsular Spanish Present Perfect. Paper presented in the Hispanic Linguistics Symposium. DeMello, G. (1994). Pretérito compuesto para indicar acción con límite en el pasado: Ayer he visto a Juan. Boletín de la Real Academia Española, 611-633. Detges, U. (2001). Tiempo, retórica y cambio funcional. La evolución del perfecto compuesto español desde la Edad Media hasta el siglo XX. Downing, A. (1996). Discourse- pragmatic distinctions of the past-in-present in English and Spanish. In Berry, Margaret, Butler, Christopher, S., Fawcett, Robin, P., and Guowen Huang (eds.) Meaning and Form: Systemic functional interpretations: Meaning and choice in language: Studies for Michael Halliday, 509-531. Norwood, NJ: Ablex. Fernández-Ulloa, T. (2001). Linguistic attitudes and beliefs: definition and measurement. The case of Bermeo (Bizkaia). Letras de Deusto, 31, (90), 153-184. Fleischman, S. (1983). From pragmatics to grammar: Diachronic reflections on complex past and futures in Romance. Lingua, 60: 183-214. García-Fernández, L. (2000). La gramática de los complementos temporales. Madrid: Visor. _________ and Martínez-Atienza, M. (2003). La expresión de los eventos inconclusos en español. Revista española de lingüística, 33, (1), 29-67. Haase, M. (1992). Resultative in Basque. Función, 11-12, 225-256. ________ (1994). Tense and Aspect in Basque. In Thieroff, R.; Ballweg, J (ed.), Tense Systems in European Languages, 279-92. Tübingen: Niemeyer, VI. Hammond, R. M. (1997). Spanish and Basque contact and linguistic change. In Hammond, R.M.; MacDonald, M. G. (eds.) Linguistic Studies in honor of Bohdan Saciuk. West Lafayette, IN: Learning Systems, XI.

41 Harris, M. (1982). The “past simple” and the “present perfect” in Romance. In Studies in the Romance Verb, ed. Nigel Vincent et al., 42-70. London: Croom Helm. Herrera-Santana, J. and Medina-López, J. (1994). Sobre los usos de las formas de perfecto en el español atlántico, Anuario de las letras, 291-301. Hopper, PC and Traugott, EC (2003). Grammaticalization. Cambridge: Cambridge University Press. Howe, C. (2006). Cross-dialectal features of the Spanish present perfect: A typological analysis of form and function. Ph.D. dissertation, Ohio State University. _______ and Schwenter, S. (2003). Present Perfect for Preterite across Spanish. Selected Papers from NWAV 31. University of Pennsylvania Working Papers in Linguistics, 61-75. _______________________ (2008). Variable Constraints on Past Reference in Dialects of Spanish.In Selected Proceedings of the 4th Workshop on Spanish Sociolinguistics, ed. Maurice Westmoreland and Juan Antonio Thomas, 100-108. Somerville, MA: Cascadilla Proceedings Project. Inoue, K. (1979). An analysis of the English present perfect. Linguistics, 17, 561-589. Kempas, I. (2002). Sobre las actitudes de estudiantes españoles hacia el uso del pretérito perfecto prehodiernal en comparación con los estudiantes santiagueños (Argentina). Neuphilologische Mitteilungen: Bulletin de la Societé, Neophilologique/ Bulletin of the Modern Language Society, 103, 435-47. _________ (2005). Sobre el uso del Pretérito Indefinido y el Pretérito Perfecto en el español peninsular en acciones producidas durante el día de habla, Revista Española de Lingüística.

42 _________ (2006). Estudio sobre el uso del pretérito perfecto prehodiernal en el español peninsular y en comparación con la variedad del español argentino hablada en Santiago del Estero. Unpublished doctoral dissertation. University of Helsinki. __________ (2009). El uso prehodiernal del Pretérito Perfecto desde el punto de vista de la deixis personal. Neuphilologische Mitteilungen,177-196. Lavandera, B (1978). Where does the sociolinguistic variables stop? Language in society, 7, 171-82. Labov, W. (1972). The transformation of experience in narrative syntax. Language in the inner city, 354-396. Lindstedt, J. (2000). The perfect- aspectual, temporal and evidential. in Oesten Dahl, Tense and Aspect in the languages of Europe. Berlin ; New York : Mouton de Gruyter. Piñero, G. (1998). El uso del perfecto simple y compuesto en combinación con unidades de Tiempo que incluyen el ahora de la enunciación en la norma culta de Las Palmas de Gran Canaria. LEA XX, 109-127. ________ (2000). Perfecto simple y Perfecto compuesto en la norma culta de Las Palmas de Gran Canaria. Frankfurt am Main : Vervuert ; Madrid : Iberoamericana. Prince, E. (1992). The ZPG Letter: Subjects, Definiteness, and Information-Status. In Mann, W.C. and S. A. Thompson (eds.) Discourse Description: Diverse linguistic Analyses of a fund-raising text, Philadelphia: John Benjamins, 295-325. Reyes, G. (1990). La pragmática lingüística: el uso del lenguaje. Barcelona. Rotaetxe, K. (1988). Approche axiologique et semantique de quelques temps verbaux Basques. Cill 14.1-2, 263-267. Sankoff, D, and Thibault, P. (1981). Weak complementarity: tense and aspect in Montreal French. In B.B. Johns & D.R. Strong (eds.), Syntactic Change.

43 Natural Language Studies 25:205-216. __________ , Tagliamonte, S. and Smith. E. (2005). Goldvarb X. Computer program. Department of Linguistics, University of Toronto, Canada. Schwenter, S. (1994a). The grammaticalization of an anterior in progress: Evidence from a peninsular Spanish dialect. Studies in Language. 18, 71-111. __________ (1994b). “Hot news” and the grammaticalization of perfects. Linguistics. 32, 995- 1028. __________ and Torres Cacoullos, R. (2008). Defaults and Indeterminacy in temporal grammaticalization: the “perfect” road to perfective. Language Variation and Change, 40.1. Serrano, M. J. (1994). Del pretérito indefinido al pretérito perfecto: un caso de cambio y gramaticalización en el español de Canarias y Madrid. Lingüística Española Actual. 16, 37-57. ____________ (1997). On the variability of syntax: some theoretical remarks. Cauce, Revista de Filología y su Didáctica, 21, 1053-1073. ____________ (1998). Grammaticalization and change in progress: Present Perfect in Spanish. In Silva-Villar, L. and Gutiérrez-Rexach, J. (eds.), Perspectives on Spanish Linguistics, 2, 117-136. ____________ (1999). Nuevas perspectivas en variación sintáctica. In Serrano, M.J. (ed.) Estudios de variación sintáctica. 11-49. Frankfurt, Germany; Madrid, Spain: Vervuert; Iberoamericana. Silva-Corvalán, C. (2001). Sociolíngüistica y pragmática del español. Georgetown University Press: Washington, DC. Smith, C. (1991). The parameter of aspect. Dordrecht, The Netherlands: Kluwer

44 Academic Publishers. Squartini, M. (1995). The Simple and Compound Past in Romance Languages. Quaderni del Laboratorio di Linguistica, 9, 219-240. __________ and Bertinetto, P. M. (2000). The Simple and Compound past in Romance languages. in Oesten Dahl, Tense and Aspect in the languages of Europe. Berlin; New York : Mouton de Gruyter. Trudgill, P. (1974). The social differentiation of English in Norwich. CUP. Vendler, Z. (1967). Verbs and times. Linguistics and philosophy. In Z. Vendler. 97- 121. Ithaca, NY: Cornell University Press. Wilson, J. and Henry, A. (1998). Parameter setting within a socially realistic linguistics, Language in society, 27, 1, 1-21.

45 Social influence in the use of the Present Perfect for the Preterite in Bilbao Spanish

Introduction It seems that the PP in some Peninsular Spanish dialects is following the same path as other Romance languages (French, Italian, Rumanian…). It has already encroached on the semantic domain of the Preterite (Harris, 1982; Fleischman, 1983). Schwenter (1994) predicted that Spanish is following the same path as French so the PP will eventually supplant the Preterite. The use of the PP for the Preterite is a diachronic innovation of a form that is going through a grammaticalization process. It does not disappear as a form but becomes something else. It incorporates the new functions to the previous ones (Lindstedt, 2000, p. 366). In Madrid, the PP is used as a perfective in addition to the anterior function (a past situation is extended to the current time of communication), as well as in Alicante (Schwenter, 1994) and Seville as reported by Squartini and Bertinetto (2000). Moreover, it is starting to emerge in the Canary Islands (Serrano, 1994; Piñero, 1998). Serrano argued that in the Canary Islands the anterior use is the most frequent and the perfective was rarely being used. Social issues are contributing to this change as a consequence of the linguistic and cultural exchange between the Canary Islands and the Peninsula, the media and the attitude towards this phenomenon. In contrast to Kempas (2002), who argued that Spanish speakers did not have a positive attitude towards this change, Serrano (1995a) found that in the Canary Islands, there is a positive attitude considering this tense more appropriate for formal situations and describing it as “more correct” and “more educated”. This positive

46 judgment value is probably promoting the change above all in the middle socioeconomic class and the second generation, that tries to follow the Peninsular norm.In Madrid, the use of the PP is categorical for past events occurring “today”, it is frequently found in past events occurring “yesterday” but the Preterite is preferred for “before yesterday” situations. This indicates that the change is more advanced here since it does not distinguish the hodiernal-prehodiernal (today vs. pre-today) distinction anymore (the norm for the standard Castilian Spanish). The leaders of this change seem to be the second generation (35-55) of the middle socio-economic class since they have a special interest in adopting more standard and prestigious variants to compensate for their social status. There were not significant results regarding gender. On the other hand, the fact that the Preterite is more used by the third generation makes Serrano (1994) predict a future advancement of the grammaticalization process since it might be an indication of a change in progress. There was no negative attitude towards the change. On the contrary, it was considered even more positive for the first generation and the middle social class. Schwenter (1994) indicated that in Alicante, where the grammaticalization is more advanced than the Castilian norm, the PP has become the educated norm for many past situations. Kubarth (1992) also found that the PP was preferred for literary uses in Buenos Aires. This change is taking place in most Peninsular dialects. The dialect under study here is that of Bilbao, a dialect that Kempas (2005) presents as the one with the highest use of the hodiernal PP in comparison to other Spanish cities, what in his opinion might work as a pioneer for other Spanish regions.

47 According to what it has been explained so far, it may be tempting to think that this change of the PP into a perfective is very advanced in all Peninsular Spain but it is not. It is well known that there are still regions where the Preterite is used more than the PP for recent past events or within “today”: Asturias, Cantabria, León and Galicia. Their linguistic behavior regarding this phenomenon resembles more that of American Spanish. In fact, according to Alarcos-Llorach (1994), hypercorrection is commonly found when they attempt to substitute the Preterite for the PP in spoken speech.It is necessary in variationist sociolinguistics to study this change in progress like several researchers have claimed (Schwenter, 1994, Serrano, 1994) taking into account a set of social variables (like age, social class and gender). Schwenter (1994) only took into account that of age, Serrano (1994) took the three of them but she did not find significant differences across gender and she selected a combination of level of education, occupation and income as the factors for assigning social class. It is very optimistic that her data showed a curvilinear pattern since females and people from the middle group produced higher occurrences of the innovative PP. Kempas (2005, 2006) only studied a social variable: city of origin so he was not able to predict a change in progress. All this suggests that more research is needed that can account for this change. I studied the same social variables than Serrano did but using only occupation for social class for reasons that is explained in the methodology of the study.

Syntactic Variation: the PP vs. the Preterite We might think that the Preterite and the PP with a perfective meaning are variants of the same variable: they have the same truth conditions and as Havu (1984)

48 claims, they are synonymous tenses from the functional point of view (the speaker can use any of them without changing the meaning of the sentence, or the logical implications derived from it). He studied the linguistic interpretations of three Spanish informants towards the substitution of the Preterite for the PP in an excerpt and they did not find any difference; they found the text completely grammatical. He took this a proof to validate his argument. Following all this, Serrano (1997) explains that in Canarian Spanish the contexts, in which the PP and the Preterite alternate, are when the event time and utterance time are very close. The Preterite fits in this context by assuming the pragmatic function of Relevance in the present where “being relevant” means to have contextual effects. The definition of this variable that she proposes would be “reference close to the time of communication expression”. In Alicante, the degree is grammaticalization is bigger and there is hardly individual variation for hodiernal past events. There are few recorded examples of a Preterite in hodiernal contexts. The loss of this option here corroborates that the PP has grammaticalized into a true hodiernal perfective (Schwenter, 1994). Overextensions of the PP into prehodiernal situations were found, what indicates a possible advance of this form into a prehodiernal perfective since younger speakers are the ones that use more the PP for situations occurring before “today”. Therefore, all this supports Schwenter‟s claim that the Alicante dialect is entering its final stage of grammaticalization.

49 Change in Progress or Age-Grading?

Schwenter (1994) studied for the first time this grammaticalization process in progress in Alicante Spanish. He found that older speakers have a tendency to use the Preterite more for hodiernal contexts than younger speakers. The almost categorical use of the PP by the younger generations leads him to claim that we are facing a change in progress instead of age grading (Howe & Schwenter, 2003). However, it is difficult to predict a change in progress taking into account only one social factor, even if it is age, which is considered to be the most important one (Guy et al.,1986). There must be some degree of age stratification (young people using the innovation more than older people do). Actually, among young people, teenagers tend to use more the innovative forms more frequently than other age groups (Guy et al. 1986; Cameron, 2000). Regarding social classes, a change from below may show a curvilinear pattern where the middle social classes are more advanced in the use of this innovation. Moreover, in terms of gender, females lead the change (Guy, 1990; Labov, 2001). These are the typical social factors used to study change in progress, to which Cameron (2000) added a mismatch between stylistic and social stratification. Guy et al. (1986) mention ethnicity as another social indicator of change in progress and other researchers have also considered other social constraints like social networks (Bailey et al, 1991, Milroy & Gordon, 2003).Schwenter (1994a) only took into account one social variable as indicator of change in progress: age. As it was mentioned before, age can be the most important factor but not sufficient to give evidence of a change in progress since we can

50 be facing age-grading. This is why, I took into account other social factors like gender and class to have a set of social variables.

Research Questions Socially speaking, there are several quantitative sub-questions that need to be solved a priori in order to be able to answer the main research question: Do women significantly use more the PP than men? Does the middle class significantly use the PP more than the other classes? Do the younger generations significantly use the PP more than the older ones? Then, the main research question of this article can be stated in the following way: Is the PP undergoing an actual change in progress from an anterior aspect into a perfective one in Bilbao?

General Predictions The purpose of this study is to test whether the perfective PP instead of the Preterite in hodiernal and prehodiernal contexts is actually facing a change in progress in Bilbao as Schwenter (1994) and Serrano (1994) predicted for Alicante and Madrid respectively. The perfective PP, characteristic of the Peninsular Spanish, marks temporal distance rather than current relevance (Copple, 2005b). Before analyzing the data, it is

51 necessary to have some predictions for each social variable. What do I expect to find in my data? Based on previous research and my own suspicions, these are my predictions: Prediction 1: The first generation and the second generation will show more occurrences of PP than the third one in both hodiernal and prehodiernal contexts like Herrera and Medina (1994) found in the Canary Islands and Schwenter (1994) did in Alicante. Prediction 2: Women will use the PP more than men in all the contexts like Medina and Corbella (1996) found in the Canary Islands. Prediction 3: The intermediate class group (technical, sales, secretarial) will show a higher percentage of PP than the rest of the groups like Herrera and Medina (1994) found in the Canary Islands.

The Community According to the Spanish National Institute of Statistics (INE), there are 20,000 more women than men living in Bilbao, making it a total of around 350,000 citizens. If we take into account all the surrounding towns on both sides of the river that separates the city, there are more than 500,000 people who travel to Bilbao on a daily basis either to work or to study. Even though it is a bilingual city, Basque is a minority language. Due to its industrial development at the end of the 19th century and at the beginning of the 20th, it has received many immigrants from other Spanish regions, which makes its Spanish a very appealing focus of study. Therefore, the subjects that I considered for my

52 study were monolingual and bilingual speakers, who have spent (lived, studied or worked) in Bilbao all or most of their lives. According to Fernández Ulloa (2001), in the Biscay province, there is a 0.68% of the population that is monolingual in Basque, 15% are bilingual, 7.5% are passive bilingual (those who understand or have some knowledge of Basque) and 76% are Spanish monolinguals. As we have seen, Peninsular Spanish is characterized by the preferred use of the hodiernal perfective PP and several studies have taken place regarding this issue in a variety of Spanish cities (e.g. Madrid or Alicante). As an explanation of the still scarce preference for the PP in prehodiernal contexts in Peninsular Spanish, Kempas (2002) argued that the advancement of the form is restrained by the literary norm that is supported by the educational system and the media.In a later study, Kempas (2005) found that Bilbao speakers are the only ones of his sample that admit not only L1 Spanish but also L1 Basque. He refuted the hypothesis of the influence of Basque in the use of the prehodiernal PP since he did not find significant differences in the amount of occurrences of this type of PP between L1 or L2 Spanish speakers. Even though Kempas‟ data seems interesting, he collected his data through an written elicitation task, in which informants had to fill in gaps with the appropriate verb. Therefore, this might indicate more what they think it is appropriate (their linguistic attitudes) or the tense they think they would use rather than what they actually say. It is well known in sociolinguistics that speakers‟ vernacular speech (the way they speak when they do not pay attention to their speech) does not always coincide with the way they think they speak. Labov (1996) provides some examples of this mismatch. This is why, a study is necessary, in which we can count on vernacular speech to make sure these results are realistic. “The vernacular, in which the minimum attention is paid to speech,

53 provides the most systematic data for linguistic analysis.” (Labov, 1984, p. 29) The best way to obtain this is through a sociolinguistic interview: “Face-to-face interviews are the only means of obtaining the volume and the quality of recorded speech that is needed for quantitative analysis."”(Labov, 1984, p. 29)

External constraints: The social variables In order to predict a change in progress, several social factors have to be considered. It is the combination of these factors that indicates whether we are facing one or not (Cameron, 2000). Age. As Guy et al. (1986) states, the most important social factor is age. There must be some degree of age stratification (young people using the innovation more than older people do). Adults are more conservative due to the pressure of using standard language in the workplace. Like it was mentioned, considering “age” is necessary but not sufficient. Within age, I classified my informants into three generations following Serrano (1994): first generation (18-34), second generation (35-54) and third generation (55-on). I expected to find higher uses of PP in the first age group in both hodiernal and prehodiernal contexts like Schwenter did (even though he only used two age groups: 1825 and 40 on). Gender. As it is well-known in the variationist theory, in changes from below females are the leaders in the use of the innovation and men use more the conservative form. As innovators, they are the ones that differentiate themselves from men responding first to the social status of linguistic variables (Labov, 2001). Regarding the leaders of linguistic change, Labov found that these were women who had earned a respectable

54 socio-economic status and had a nonconformity history and this was shown in their sociolinguistic position. Actually, Serrano (1995) proved this right in her research in the Canary Islands regarding the use of the PP and the Preterite. In her data, while there was no significant differences among sexes, within the group of men, those of the middle social class, were the ones that used more the innovation. In the case of women, those who did it more belonged to the upper social classes. We cannot forget that this innovation is considered the standard norm in the Peninsular dialect, which is considered prestigious by the Canary Island speakers. Hence, the greater motivation to adopt it by the women of the upper classes. Social Class. Social class has been one of the most frequently used social variables in sociolinguistics. Generally, results obtained in other social factors such as gender are interpreted in terms of social class. Occupation is the social class factor that is considered to correlate more with linguistic variation, at some points even more than the combined index of occupation, education and house value (Labov, 1966). Labov found that occupation was more sensitive to gender differences. Eckert (2000) explains that occupation indicates the way speakers participate in their linguistic market rather than education. This is why, I classified my interviewees into three occupational categories: Professionals (such as business executives, accountants or teachers), Technical/Sales/Secretarial (such as sales personnel, artists, physical therapists or secretaries) and Skilled Labor (such as carpenters, plumbers or hairdressers). As Cameron (2000) did, I classified housewives who did not work outside the home by their husbands‟ occupation. In the same line, I classified adult students who still live at home with their parents according to their parents‟ occupation. A curvilinear social class

55 distribution, where the intermediate social groups are the ones who use the innovation more in contrast to the highest or the lowest groups, who use it less, is typical of changes in progress (Labov, 1966). Labov‟s hypothesis was that the intermediate groups had a motivation for adopting the innovation. This might have a “covert” prestige as a marker of local identity and by using a more prestigious form, they try to compensate for their lack of social status while the upper groups try to resist the change.

Methodology

Previous research on the topic opened the path to me, especially for selecting linguistic factor groups that favored the change (Copple, 2005, Schwenter and Cacoullos, 2008) and regarding methodology. I finally adapted mine from Schwenter (1994) and Serrano (1994). In this article, I will just focus on the social factors that influence on the use of the PP instead of the Preterite in Bilbao Spanish. I decided to record semi-elicited sociolinguistic interviews to get vernacular speech: they were open interviews at the beginning based on Labovian modules and I asked the interviewees to narrate me the present day, the previous one and the day before to get enough data. As it was mentioned earlier, my main focus was taking into account 3 social variables that were only considered by Serrano: age, sex and class. I adapted her age categories (20-34, 35-54, 55on) and regarding class, she used a combined index of education, income and occupation. I only took into account occupation following Cameron (2000) and Labov (2001) since occupation correlates more with linguistic variation than any other class marker. I adapted Cameron‟s (2000) occupational categories into three groups: high class, middle

56 class and low class mainly based on the role of language to perform their job. 49 interviews were recorded, out of which 36 were finally used. Each interview was at least 45-minutes long and a minimum of 50 tokens with any of the two variants (PP/ Preterite) and a maximum of 200 were extracted from every interview. In fact, 5520 tokens have been used for this project. 7 social factor groups have been considered. The first step was to get frequencies though chi-squares. Then, I proceeded to a run a probability based model.

The analysis of the data (VARBRUL) The statistical method that I used for analyzing the data was GOLDVARB X (2005). I conducted 49 sociolinguistic interviews among people who have lived, studied or worked in Bilbao most of their lives. These people form part of three generations (from 18-34, 35-54 or 55 on), two sexes (women and men) and three social classes (upper working class, lower middle class and middle middle class). The sociolinguistic interview has been pretty open due to the facility of obtaining occurrences in these two past tenses. However, in any case, at the end of around 45 minutes of interviews, they were asked to narrate what they did that same day, the previous day and two months ago. This was done in a way in which the interviewer omitted the use of a verb to avoid influencing the choice of a tense by the interviewee following Schwenter (1994) and Serrano (1994): “Cuéntame tu día hoy, ayer, hace dos semanas.” Tell me about your day today, yesterday, two weeks ago. I introduced the numbers into VARBRUL to get frequencies of use of every tense for each social variable. Then, I proceeded to obtain probabilities in

57 order to be able to get more accurate results. Lastly, the interpretation of the results took place: Sex The first external factor group that is being examined here is sex: Table 1 Sex _____________________________________________________________ PP P Total % _____________________________________________________________ Females

N %

1217 33.7

2396 66.3

3613

65.5

Males

N %

683 35.8

1224 64.2

1907

34.5

Total

N 1900 3620 5520 % 34.4 65.6 ______________________________________________________________

The Preterite is the most recurrent variant in each gender but males use the PP more frequently than females. However, Chi-square did not find this group to be significant.

Class The second social factor group is class. There are three main social classes: high, middle and low:

58 Table 2 Class _____________________________________________________________ PP P Total % _____________________________________________________________ High

N %

514 30.0

1201 70.0

1715

31.1

Middle

N %

718 37.4

1201 62.6

1919

34.8

Low

N %

668 35.4

1218 64.6

1886

34.2

Total

N 1900 3620 5520 % 34.4 65.6 ______________________________________________________________

The Preterite is the most used variant in any class but the middle class is the one with the highest percentage of PPs as expected, followed by the low class. This time, Chisquare found this group to be significant: Degrees of freedom: 2 Chi-square = 23.5006205807781 p is less than or equal to 0.001

59 Class and Sex Class and sex are combined to test for significance: Table 3 Class and Sex _______________________________________________________________ PP P Total % _______________________________________________________________ High Females

N %

399 32.6

826 67.4

1225

22.2

High Males

N %

193 29.2

467 70.8

660

12.0

Middle Females

N %

370 36.1

655 63.9

1025

18.6

Middle Males

N %

270 37.3

454 62.7

724

13.1

Low Females

N %

448 32.9

915 67.1

1363

24.7

Low Males

N %

220 42.1

303 57.9

523

9.5

Total

N 1900 3620 5520 % 34.4 65.6 ________________________________________________________________ The Preterite is preferred in all cases but this percentage is higher in the case of high males and that of PP is higher in low males. Chi-square found this group to be significant: Degrees of freedom: 5 Chi-square = 28.6125237950482 106 p is less than or equal to 0.001.

60

Age The fourth social group is only age. There are three groups: Table 4 Age ____________________________________________________________________ PP

P

Total

%

____________________________________________________________________

20-34

35-54

55-on

Total

N

895

1433

%

38.4

61.6

N

450

943

%

32.3

67.7

N

555

1244

%

30.9

69.1

N

1900

3620

%

34.4

65.6

2328

42.2

1393

25.2

1799

32.6

5520

_____________________________________________________________________

61 The Preterite is the most frequent variant across ages but the percentage of PPs is higher in the youngest group and it goes down as the age increases, as expected. Chisquare found this group to be significant:

Degrees of freedom: 2 Chi-square = 29.6252406592714 p is less than or equal to 0.001.

Age and Gender

The fifth factor is age and gender combined:

Table 5 Age and Gender _____________________________________________________________________ PP P Total % _____________________________________________________________________

20-34 Females

N %

571 39.7

866 60.3

1437

26.0

20-34 Males

N %

324 36.4

567 63.6

891

16.1

35-54 Females

N %

300 30.5

682 69.5

982

17.8

35-54 Males

N %

150 36.5

261 63.5

411

7.4

62 55-on Females

N %

346 29.0

848 71.0

1194

21.6

55- on Males

N %

209 34.5

396 65.5

605

11.0

Total

N %

1900 34.4

3620 65.6

5520

______________________________________________________________________ The Preterite is the most frequent variant in all cases but the percentage is higher in the 55-on Females and the highest percentage of PPs occurs with the group of females of 20-34, as expected.

Chi-square found this group to be significant:

Degrees of freedom: 5 Chi-square = 42.4476858235725 p is less than or equal to 0.001. The distribution is significant.

Class and Age The sixth group combines class and age: Table 6 Class and Age _____________________________________________________________________ PP P Total % _____________________________________________________________________ High 20-34 N 234 563 797 14.4 % 29.4 70.6 High 35-54

N %

149 29.7

353 70.3

502

9.1

63

High 55- on

N %

131 31.5

285 68.5

416

7.5

Middle 20-34 N %

390 42.2

534 57.8

924

16.7

Middle 35-54 N %

191 36.7

330 63.3

521

9.4

Middle 55-on N %

137 28.9

337 71.1

474

8.6

Low 20-34

N %

271 44.6

336 55.4

607

11.0

Low 35-54

N %

110 29.7

260 70.3

370

6.7

Low 55-on

N %

287 31.6

622 68.4

909

16.5

Total

N 1900 3620 5520 % 34.4 65.6 ______________________________________________________________________ As usual, the Preterite is the most used variant but the PP is higher with the Low class and the 20-34 age-range. The middle group of the same age group follows them and the oldest group of the middle class is the one with the lowest percentage of PPs. Chi-square showed this group is significant:

Degrees of freedom: 8 Chi-square = 82.9804942746102 p is less than or equal to 0.001. The distribution is significant.

64 Age, Gender and Class The final social factor group is the combination of the three social aspects under study: Table 7 Age, Gender and Class ______________________________________________________________________ PP P Total % ______________________________________________________________________ 20-34 High Females N 129 332 461 8.4 % 28.0 72.0 20-34 High Males

N %

105 31.2

231 68.8

336

6.1

20-34 Middle Females

N %

265 46.5

305 53.5

570

10.3

20-34 Middle Males

N %

125 35.3

229 64.7

354

6.4

20-34 Low Females

N %

177 43.6

229 56.4

406

7.4

20-34 Low Males

N %

94 46.8

107 53.2

201

3.6

35-54 High Females

N %

97 28.4

245 71.6

342

6.2

35-54 High Males

N %

52 32.5

108 67.5

160

2.9

35-54 Middle Females

N %

127 33.4

253 66.6

380

6.9

35-54 Middle Males

N %

64 45.4

77 54.6

141

2.6

35-54 Low Females

N %

76 29.2

184 70.8

260

4.7

35-54 Low Males

N %

34 30.9

76 69.1

110

2.0

65

55-on High Females

N %

95 37.7

157 62.3

252

4.6

55-on High Males

N %

36 22.0

128 78.0

164

3.0

55-on Middle Females

N %

56 22.9

189 77.1

245

4.4

55-on Middle Males

N %

81 35.4

148 64.6

229

4.1

55-on Low Females

N %

195 28.0

502 72.0

697

12.6

55-on Low Males

N %

92 43.4

120 56.6

212

3.8

Total

N 1900 3620 5520 % 34.4 65.6 ________________________________________________________________________

Again, in all the cases the Preterite has the highest percentage but it is most notorius in the case of 55-on high male group. The PP is in the 20-34 low males and after this group, 20-34 middle females. Chi-square found this group to be significant:

Degrees of freedom: 17 Chi-square = 140.288359919278 p is less than or equal to 0.001. The distribution is significant.

Most of the significant groups in the Step Up also occur in the Step Down and the other way around except for two groups:

66 a) class and age b) age, sex and class

The factor group “class and age” was eliminated in the Step Down so it was automatically eliminated from further discussion as the other eliminated groups. However, the factor group “age, sex and class” (the combination of the three social factors under study) was not eliminated while stepping down even though it was not considered significant in the Step Up. The one social constraint that survives within the probabilistic model, and does so weakly, is of interest for the following. Notice that this social constraint combines age, sex and class. That the model prefers this combination over selecting each factor separately, supports a commonsense claim in research in language and gender that gender as social practice is not easily separated from other social categories of experience. Thus, we find a social factor selected by the model in which categories of experience, age, sex and class, are not treated separately, and this lack of distinction supports the general claim in the literature (Cameron, 2005, p. 28). Now, a one level analysis is run with the all the groups that were not eliminated while stepping down (age, sex and class included since it was considered significant in the Chisquare). Total Chi-square = 5746.9725 Chi-square/cell = 1.2142 Log likelihood = -1233.757 The probabilities are the following:

67 Age, Class and Sex

Table 8 Age, class and sex _______________________________________________________________________ Probability % PP Total N % of data ________________________________________________________________________ 55-on high females

.73

37%

95

4%

55-on low males

.66

43%

92

3%

35-54 middle males

.63

45%

64

2%

55-on low females

.60

27%

195

12%

35-54 high males

.60

32%

52

2%

55-on middle males

.59

35%

81

4%

35-54 middle females

.55

33%

127

6%

55-on middle females

.53

22%

56

4%

55-on high males

.49

21%

36

2%

25-34 low females

.48

43%

177

7%

25-34 middle females

.46

46%

265

10%

25-34 middle males

.44

35%

125

6%

35-54 high females

.42

28%

97

6%

25-34 high females

.40

27%

129

8%

35-54 low females

.39

29%

76

4%

25-34 high males

.34

31%

105

6%

25-34 low males

.32

46%

94

3%

68 35-54 low males

.31

30%

34

1%

The PP is more favored with the oldest group of high class females (.73). It is not found any case of the youngest age group that was favoring the innovation. 55-on low class males (.66) and females favor the innovation (.60) and 35-54 high class males (.60) and middle class males (.63). 55-on middle class females (.53) lightly favor it and 55-on high class males (.49) lightly disfavor it. Finally, the groups that disfavor it more are low class males from the younger age groups: 20-34 (.32) and 35-54 (.31). Seen this way, it is hard to draw some conclusions based on this social pattern. However, if we rearrange the categories, this is what we can find:

Table 9 20-34 age group ________________________________________________________________________ Sex Class Weight ________________________________________________________________________ Female High .40 Female Middle .46 Female Low .48 Male High .34 Male Middle .44 Male Low .32 ________________________________________________________________________

In the lowest age group, we can see a linear pattern in the female group (the lowest class has the biggest probability use of the innovation and this decreases as the social class increases). In the male group, there is a curvilinear pattern where the middle class has a bigger probability value than the rest. This is an indicator of a change in progress following Guy et al. (1986, p.37). In his Figure 2, they do find a curvilinear

69 pattern for the males, which indicates the change in progress. But they also find a linear pattern for the females. The linear pattern was predicted by Tony Kroch previously. So, what Guy and others found was evidence for both types of class patterns, but that the patterns may play out differently according to gender. The key point, however, is finding the curvilinear pattern somewhere and finding it more than once. We have a total of 3 groups of males per class and 3 groups of females per class. This is a total of 6. Out of the 6, there is a curvilinear pattern 3 times, a linear pattern twice, and a u-shaped pattern in the oldest female speakers once. Thus, we find a curvilinear pattern the most.

Table 10 35-54 age group _______________________________________________________________________ Sex Class Weight _______________________________________________________________________ Female High .42 Female Middle .55 Female Low .39 Male High .60 Male Middle .63 Male Low .31 _______________________________________________________________________

Across the middle age group, both in female and male groups there is evidence for a curvilinear distribution. Table 11 55-on age group ________________________________________________________________________ Sex Class Weight ________________________________________________________________________ Female High .73 Female Middle .53 Female Low .60 Male High .49 Male Middle .59

70 Male

Low

.66

In the male group, a linear pattern is shown. Therefore, it seems that evidence of a change in progress is shown in this dialect.

Discussion 3 predictions were postulated based on previous research. Now, we can proceed to test them: Prediction 1: The first generation and the second generation will show more occurrences of PP than the third one in both hodiernal and prehodiernal contexts in keeping with the findings of Herrera and Medina (1994) in the Canary Islands and Schwenter (1994) in Alicante (see Table 4). Even though this factor group was considered to be significant by chi-squares, it was eliminated in one-level analysis. The prediction can be considered to be partially supported since the youngest group has more occurrences of the PP than the others, but the middle group has less than the oldest one. Therefore, this evidence supports previous research.

Prediction 2: Women will use the PP more than men in all contexts in keeping with the findings of Medina and Corbella (1996) in the Canary Islands (see Table 1). Although the number of occurrences of the PP is higher within the female group, the percentage is higher within the male one. However, this group was neither considered

71 significant by chi-squares nor by the one-level analysis which agrees with Serrano‟s (1994) findings.

Prediction 3: The intermediate class group (technical, sales, secretarial) will show a higher percentage of PP than the rest of the groups in keeping with the findings of Herrera and Medina (1994) in the Canary Islands. Middle class has a higher percentage of the PP than the other classes like previous literature suggests. This factor group was considered significant by chi-squares but not by the one-level analysis of VARBRUL (see Table 2).

Answering the Research Questions 1- Do women significantly use more the innovation than men? No, men use the innovation more than women. However, this percentage is not significant either by chi-squares or one-level analysis. This pattern does not follow what is expected from the variationist literature regarding changes from below: women tend to be the leaders of the linguistic innovation. However, it is consistent with Serrano‟s (1994) findings: there were no significant differences across gender. 2- Does the middle class significantly use the PP more than the other classes? Yes. The middle class significantly uses (by chi-squares) the PP more than the others. The low class follows it. This curvilinear pattern is what is expected in a change from below: the lower social groups introduce the change (Labov, 2001). Serrano (1994) was the only Spanish study on this linguistic variable that took into account class and this result supports her findings.

72 3- Do the younger generations significantly use the PP more than the older ones? Yes. The first and second generations significantly (only by chi-squares) use the PP more than the oldest one. This factor group was, nevertheless, eliminated by the probability model of VARBRUL. This coincides with Schwenter (1994) and Serrano (1994). However, Serrano found a higher percentage of use in the second generation. In general, this is what is expected: younger generations use the innovation more than the older ones; it is an increasing phenomenon. 4- Is the PP undergoing an actual change in progress from an anterior aspect into a perfective one in Bilbao? Yes. These social categories individually were found non- significant (the factor group “sex”, for example, was neither significant under chi-squares nor under probabilities of VARBRUL) or only significant by chi-squares since they were eliminated a posteriori in the one-level analysis (for example, the factor groups “age” and “class”).The combination of the three of them, however, was not eliminated by the step-down process. This shows this combination has some significance. If we look closer, linear (following Guy, 1986 and al.) and curvilinear patterns can be found. Then, it can be stated that there is evidence for an actual change in progress (see Tables 9, 10 and 11).

Conclusions Most of these factor groups were considered to be significant by chi-squares. Only gender was not found significant by chi-squares. This factor was also eliminated later in the probability based model. The only social factor group that was not eliminated by the probability based model is the combination of age, sex and class. It was not found

73 significant either, but since it is still present, this can be interpreted as a sign of some significance. In order to be able to predict a change in progress, a curvilinear pattern has to be found. After dividing this group into three small groups stratified by age, a symmetrical pattern is observed: In the youngest groups, a linear pattern is followed by several curvilinear ones until the last age group that is a linear pattern again. Therefore, evidence for the fact that the PP is undergoing an actual change in progress can now be claimed.In future research, it would be interesting to extend the age-groups into teenagers and children since this has not been investigated before. In addition, teenagers are expected to be the leaders of the linguistic change so this group would be of special interest. In this study, the main three social factors in variationist research are considered. However, it would be useful to add more social variables such as bilingualism (SpanishBasque) for future studies on the matter.

REFERENCES: Alarcos Llorach, E. (1994). Gramática funcional del español. Real Academia Española Espasa Calpe. Bailey, G., Wikle, T., Tillery, J., and Sand, L. (1991). The apparent time construct. Language Variation and Change. 3, 241-264. Cameron, R. (2000). Language change or changing selves?: Direct quotation strategies in the Spanish of San Juan, Puerto Rico. Diachronica. 17, 249-292. __________ (2005). Aging and Gendering. Language in Society, 34 (1) 23-61. Copple, M. (2005). A diachronic study of Peninsular Spanish Present Perfect. Paper presented in the Hispanic Linguistics Symposium.

74 Eckert, P. (2000). Linguistic variation as social practice: the linguistic construction of identity in Belten High. Blackwell Publishers. Fernández-Ulloa, T. (2001). Linguistic attitudes and beliefs: definition and measurement. The case of Bermeo (Bizkaia). Letras de Deusto, 31, (90), 153-184. Fleischman, S. (1983). From pragmatics to grammar: Diachronic reflections on complex past and futures in Romance. Lingua, 60: 183-214. Guy, G. (1990). The sociolinguistic types of language change. Diachronic, 7, 4767. Guy, G., Horvath, B., Vonwiller, J., Daisley, E., and Rogers, I. (1986). An intonational change in progress in Australian English. Language in Society. 15, 23-52. Harris, M. (1982). The “past simple” and the “present perfect” in Romance. In Studies in the Romance Verb, ed. Nigel Vincent et al., 42-70. London: Croom Helm. Havu, J. (1984). ¿Perfecto simple o Perfecto compuesto: oposición o libre variación?. In Suomela- Härmä, E.; Välikangas, O. (ed.), Actes du 9e Congres des Romanistes Scandinaves: Helsinki: Soc. Neophilol. Herrera-Santana, J. and Medina-López (1994). Sobre los usos de las formas de perfecto en el español atlántico, Anuario de las letras, 291-301. Howe, C. and Schwenter, S. (2003). Present Perfect for Preterite across Spanish. Selected Papers from NWAV 31. University of Pennsylvania Working Papers in Linguistics, 61-75. Kempas, I. (2002). Sobre las actitudes de estudiantes españoles hacia el uso del pretérito perfecto prehodiernal en comparación con los estudiantes sanguiagueños

75 (Argentina). Neuphilologische Mitteilungen: Bulletin de la Societé, Neophilologique/ Bulletin of the Modern Language Society, 103, 435-47. _________ (2005). Sobre el uso del Pretérito Indefinido y el Pretérito Perfecto en el español peninsular en acciones producidas durante el día de habla, Revista Española de Lingüística. _________ (2006). Estudio sobre el uso del pretérito perfecto prehodiernal en el español peninsular y en comparación con la variedad del español argentino hablada en Santiago del Estero. Unpublished doctoral dissertation. University of Helsinki. Kubarth, H. (1992). El uso del pretérito simple y compuesto en el español hablado de Buenos Aires. In Luna Traill, E. Scripta Philologica in Honorem Juan M. Lope Blanch a los 40 años de docencia en la UNAM y a los 65 años de vida, II: Lingüística española e iberoamericana; III: Lingüística indoamericana y estudios li. Mexico City, Mexico: Instituto de Investigaciones Filológicas, Universidad Nacional Autónoma de México. Labov, W. (1966). The social stratification of English in New York City. Washington, D.C.: Center for Applied Linguistics. ________ (1984). Intensity. Georgetown University Round Table on Languages and Linguistics, 43-70. ________ (1996). When intuitions fail. Papers from the Regional Meetings, Chicago Linguistic Society, 32, (2), 77- 105.

76 ________ (2001). Part C en Principles of Linguistic Change: Social Factors. (p. 325 – 411): Blackwell. Lindstedt, J. (2000). The perfect- aspectual, temporal and evidential. in Oesten Dahl, Tense and Aspect in the languages of Europe. Berlin ; New York : Mouton de Gruyter. Medina, J. and Corbella, D. (1996). El español de Canarias hoy: análisis y perspectivas. Frankfurt: Vervuert. Milroy, L. and Gordon, M. (2003). Sociolinguistics: Method and Interpretation: Blackwell Publishers, Malden, M.A. Piñero, G. (1998). El uso del perfecto simple y compuesto en combinación con unidades de Tiempo que incluyen el ahora de la enunciación en la norma culta de Las Palmas de Gran Canaria. LEA XX, 109-127. Schwenter, S. (1994a). The grammaticalization of an anterior in progress: Evidence from a peninsular Spanish dialect. Studies in Language. 18, 71-111. __________ and Howe, C. (2003). Present perfect for preterite across Spanish dialects.Selected Papers from NWAV 31. University of Pennsylvania Working Papers in Linguistics 9.2, ed. by Tara Sanchez & Uri Horesh, 61-75. __________ and Torres-Cacoullos, R (2008). Defaults and Indeterminacy in temporal grammaticalization: the “perfect” road to perfective. Language Variation and Change, 40.1. Serrano, M. J. (1994). Del pretérito indefinido al pretérito perfecto: un caso de cambio y gramaticalización en el español de Canarias y Madrid. Lingüística

77 Española Actual. 16, 37-57. ___________(1995a). Sobre el uso del pretérito perfecto y pretérito indefinido en el español de Canarias: pragmática y variación. Boletín de Filología de la Universidad de Chile, XXXV: 533-566. ____________ (1997, 1998). On the variability of syntax: some theoretical remarks. Cauce, Revista de Filología y su Didáctica, 21, 1053-1073. Squartini, M and Bertinetto, P. M. (2000). The Simple and Compound past in Romance languages. in Oesten Dahl, Tense and Aspect in the languages of Europe. Berlin; New York : Mouton de Gruyter. Young, R. and Bailey, R. (1996). VARBRUL analysis for second language acquisition research. Second language research and linguistic variation, ed. Robery Bayley and Dennis Preston, 253-306. Amsterdam/ Philadelphia: John Benjamins.

78