INTERPRETATION AND GRAMMAR INTERACTION IN THE SPANISH SUBJUNCTIVE ADJUNCTS

This article proposes a multifactor analysis of the Spanish subjunctive adjuncts. By examining the features that govern Spanish mood and their interaction, I present a complete overview of diverse accounts on the Spanish subjunctive. I outline the general principles of those explanations and comment on their advantages and disadvantages in formalizing an explanation concerning the interaction of mood with syntax, semantics, and pragmatics. Next, from mood contribution to the flow in discourse, based on Quer’s (1998, 2001) approach, I provide a role for the subjunctive mood in the discourse, which can be established by a variety of factors. Instead of assigning rigid meanings to the subjunctive, this approach allows the speaker to relate the proposition to the actual world and provide an evaluation of it. In addition to this, I present a proposal to explain mood alternation in adjuncts considering the interaction of different components of the grammar in the process of interpretation. This contribution offers a dynamic view of meaning in which context and individuals are crucial for the interpretation of the Spanish subjunctive.


Introduction
Mood is a morpho-syntactic category that serves to express the evaluation of a proposition.In Spanish, there have been numerous studies associating the subjunctive mood with invariant meanings, or typically following the rough realis/irrealis opposition (Givón 1994;Picallo 1985;Kempchinsky 1986;Rochette 1988;Giorgi & Pianessi 1997;Rivero 1971;Goldin 1974;Lavandera 1983;andLunn 1989a, 1995, among others).In this article, I take a different approach by demonstrating that these explanations prove to be an inaccurate framework for the analysis of mood distribution in Spanish.I aim at characterizing mood in terms of model shift and contributing to the flow in discourse from a dynamic view of context change.
The structure of the article is as follows.I first present a basic description of the Spanish subjunctive and its alternation with the indicative.Next I review the different analyses of the Spanish subjunctive.The focus will be on the problems that these approaches present and the advantages of adopting a multi-factor analysis.I begin with a discussion of the single-factor analyses that deal with the Spanish mood choice, including syntactic and semantic accounts, as well as more recent pragmatic analyses.Then I present a multi-factor analysis for the Spanish mood alternation in which the notion of model evaluation, that is, contexts of interpretation, is a central one.I describe it and show how speakers evaluate propositions.Finally, I turn to morpho-syntactic, semantic and pragmatic considerations of the models of interpretation, presenting those features necessary for the Spanish mood choice in those least-studied contexts: Spanish subjunctive adjuncts (temporal, concessive, and conditional clauses).

The Spanish Subjunctive: A basic description of the Spanish mood alternation
There are two main moods in Spanish: the indicative (IND) and the subjunctive (SUBJ).The subjunctive mood is used differently across contexts (Lozano1995).Subjunctive selection may be obligatory in some grammatical contexts, or it may be optional, used as a marker to indicate a specific meaning.Complement clauses are typically 'fixed contexts': the subjunctive is licensed by lexical selection from the main verb (as in ( 1) and ( 2)): (1) El profesor piensa que pasamos el examen the teacher think-3SG-PRES-IND that pass-1PL-PRES-IND the exam 'The teacher thinks we passed the exam' (2) El profesor quiere que pasemos el examen the teacher want-3SG-PRES-IND that pass-1PL-PRES-SUBJ the exam 'The teacher wants us to pass the exam' Contexts such as relative clauses, (3) and ( 4), are considered free-choice subjunctive where mood selection is not governed by the syntactic context of the embedded clause.It independently contributes to the semantic composition and the mood marker only needs to be compatible with the intended semantics.In the examples below, if the relative clause describes a specific referent that the speaker has in mind (the secretary who speaks Spanish), the clause is marked with the indicative as in (3).In contrast, a non-specific referent (any possible secretary that may speak Spanish) is described by means of a subjunctive relative clause as in (4).
(3) Una secretaria que habla español a secretary that speak-3SG-PRES-IND Spanish same terms as the 'free' subjunctive contexts.However, they cannot be considered as a single category.In these contexts, specificity (whether time or place is specific or not), temporality (whether the event has occurred or not), and the meaning of the clausal connectors jointly determine subjunctive use.For instance, in ( 5) and ( 6), mood selection depends on the specificity of the time indicated (Zagona, 2002).
(5) Tomó esa clase cuando su profesor lo autorizó take-3SG-PAST that class when her/his teacher it approve-3SG-PAST-IND 'S/he took that class when her/his teacher approved it' (6) Tomará esa clase cuando su profesor lo autorice take-3SG-FUT that class when her/his teacher it approve-3SG-PRES-SUBJ 'S/he will take that class when(ever) her/his teacher approves it' In (5) a specific time is in mind.In ( 6) there is no reference to a particular time.The event of 'being authorized' may happen any time.In ( 7) and ( 8), on the other hand, the subjunctive is selected if the event is going to occur in the future (7), and the indicative is used when the event has already happened or is a habitual occurrence (8).Similarly, purpose clauses (9) are subjunctive while adjuncts expressing cause are indicative (10), even in the future (11).( 7) Mi mamá nos dará un premio cuando pasemos el examen my mom us give-3SG-FUT-IND a prize when pass-3PL-PRES-SUBJ the exam 'My mom will give us a prize when we pass the exam' (8) Mi mamá nos dio un premio cuando pasamos el examen my mom us give-3SG-PAST-IND a prize when pass-3PL-PAST-IND the exam 'My mom gave us a prize when we passed the exam' (9) Vino para que le den dinero come-3SG-PAST-IND to that him give-3PL-PRES-SUBJ money 'He came in order to be given some money' (10) Vino porque le dieron dinero come-3SG-PAST-IND because him give-3PL-PAST-IND money 'He came because they gave him some money' (11) Vendrá porque le darán dinero come-3SG-FUT-IND because him give-3PL-FUT-IND money 'He will come because they will give him some money' The subjunctive in conditional clauses has also been described as free-choice subjunctive.Structurally speaking, conditional clauses have a two-part clause consisting of an antecedent clause, the protasis, and a consequent clause, the apodosis.Mood selection depends both on the meaning and on the time of the event.For example, in (12) the indicative is selected in the protasis to express a future result if a condition is met in the present.The subjunctive is used in the protasis to express a future or a past result of an unlikely or non-real condition in the present or in the past, as in ( 13) and ( 14) respectively.Two observations can be made from this general description.First, the subjunctive is used to express different meanings, despite the common perception that it has the same meaning across different clauses.Second, rather than a single feature characterizing the subjunctive selection, what these examples show is an interaction of different features intervening in the determination of mood.In section §3, I will outline the different approaches to the analysis of the Spanish subjunctive.I will also review their advantages and disadvantages by examining how they deal with the observations mentioned above.

Approaches to the analysis of the Spanish subjunctive
There have been several proposals in the literature that attempt to characterize the Spanish mood alternation.Some of them have suggested the indicative /subjunctive distribution is based on one single underlying feature (Givón 1994;Picallo 1985;Kempchinsky 1986;Rochette 1988;Giorgi & Pianessi 1997;Rivero 1971;Goldin 1974;Lavandera 1983;andLunn 1989a, 1995, among others).Other studies, on the contrary, have stated that it is impossible to propose a unified semantics for the subjunctive mood in Spanish (Bell 1980;Lozano 1995), or that it is important to take into account the interaction of syntactic, semantic and pragmatic features in selecting the mood (Quer 1998).In this section, I present some of the most important approaches, grouping them by the perspective they adopt in their explanations.

3.1.Single-factor analysis of the Spanish subjunctive
A single-factor analysis of the Spanish subjunctive allows the heterogeneous meanings of this structure to be explained and predicted as a class.The contrast between the indicative and the subjunctive signals a contrast between some common features defined from different perspectives.The traditional approach (Givon 1994) claims that the notion of realis/irrealis plays a decisive role in the distribution of the subjunctive and indicative moods.While the indicative would require a proposition that holds in the actual world, the subjunctive would be used to express the same proposition as if it holds in a non-actual world.In this sense, mood has usually been presented in binary terms: certainty vs. uncertainty, reality vs. irreality, factual vs. non-factual, main-clause-context occurrence vs. subordinated-clause-context occurrence.In other words, the indicative has been considered as the unmarked, default mood that is not restricted to any syntactic environment, and the subjunctive as the marked mood, largely restricted to embedded clauses (Bello 1847;Gili Gaya 1969;Alarcos-Llorach 1970;Real Academia Española 1973;Manteca Alonso-Cortés 1981;Porto Dapena 1991).
Nevertheless, this broad characterization of the subjunctive has been proven to be empirically incorrect: the subjunctive can be used to express factual events, as (15) illustrates, and can be used in main clauses, as in ( 16): (15) Me alegro de que hayas venido myself glad-1SG-PRES-IND about that have-2SG-PRES-SUBJ come-PPT 'I am glad that you have come' (16) Quisiera aprender alemán want-1SG-PAST-SUBJ learn German 'I would like to learn German' Given these facts, many researchers over the past decades have contributed to our understanding of the indicative/subjunctive alternation.In next sections, I will present the syntactic, semantic, and pragmatic approaches elaborated to explain the characteristics of the subjunctive and its alternation with the indicative.

Syntactic approaches
From a syntactic perspective, the Spanish mood alternation depends on the syntactic construction involved, specifically, on the matrix verb.The subjunctive is limited to the subordinate verb and its use is dependent on the main verb and its specific requirements.In this sense, the dependent character of the subjunctive has been linked to the tense properties of this mood: they are defective and have to rely on the full tense specification of the matrix verb.As a result, there is a sequence of tense and the impossibility of co-reference between the main and the embedded subjects.There is also an operator adopted as a special licensing mechanism, which is either supplied by the subjunctive or by the predicate that takes the subjunctive complement (Quer 1998).In contrast, the subordinate verb will be in the indicative when the main verb conveys the idea of 'factuality', whereas verbal matrices expressing wish, doubt, possibility or emotion condition the selection of the subjunctive in the subordinate clause (Ramsey & Spaulding 1956;Cressey 1971;Shawl 1975;Bosque 1989;Kempchinsky 1986Kempchinsky , 1998)).
In current grammatical theory, functional categories such as person, number, tense, aspect and mood are represented above the verb in a syntactic tree.The lexical verb moves to these categories in order to check the syntactic and semantic formal features through overt inflectional morphology (Chomsky, 1995).Giorgi & Pianesi (1997) and Cinque (1999) identify mood as a morpho-syntactic property encoded by means of a semantic feature called mood.This feature is encoded in the functional category MoodP, where the feature [+IRREALIS] is checked through the subjunctive morphology when the verb raises to that node in order to be interpreted.Indicative clauses do not project MoodP.Mood is also a manifestation of a binary classification since propositions need to be evaluated against the real world.If the complement proposition is asserted or at least evaluated with respect to the real world, indicative mood appears.The subjunctive is used when the complement is not supposed to be true in the real world, but in a different possible world.
However, there are contexts in which the same asserted matrix verb can be followed by subordinate verbs either in the indicative or the subjunctive.This choice has been recently explained by another mood licensing mechanism.It corresponds to the polarity subjunctive that can be licensed by the presence of an operator such as negation or interrogation that takes under its scope a verb that normally selects the indicative (Borgonovo, Bruhn De Garavito, & Prévost, 2005;Stowell, 1995).In these cases, both moods can alternate without creating an ungrammatical structure and the difference is solely interpretational.From a dialectal variation point of view, Blake (1983) and Jensen & Lathrop (1973) propose a different explanation for these uses.
While in different varieties of Spanish it is more frequent to find creer ('to think') followed by the indicative and no creer ('not to think/believe') followed by the subjunctive, in some varieties of Spanish it is common to find no creer followed by the indicative as well.
Despite the determination of many linguists to try to explain these alternations from a syntactic perspective, their efforts still fall somewhat short, and the accurate elucidation of the use of the Spanish mood remains an incomplete task.It is also clear that all of these accounts are mainly concerned with explaining mood alternation in complement clauses.In summary, syntactic approaches provide a better understanding of why subjunctive verb forms require special licensing conditions and why they are obligatory in certain contexts.They account for the distributional patterns of the subjunctive in complement clauses and subjunctive licensing mechanism such as polarity.However, in considering subjunctive complements as a purely syntactic phenomenon, these approaches do not entirely describe the complex nature of the Spanish mood alternation.They take into account specific syntactic structures and their characterization of mood is based on selection and licensing patterns.Although there is few researchers who have recognize the value of semantic features in their syntactic approach (Giorgi, 2009), most of these authors darken the acknowledgment of the relationship between mood and meaning.A fruitful view for an explanation should also take into consideration the semantic and pragmatic dimensions.The semantic dimension, however, should not be strictly considered as a result of the lexical semantics of the selecting predicates.In contexts such as relative and adverbial clauses, other important properties are involved as I show in the next section.

Semantic approaches
The semantic approach attempts to unify as much as possible all the different uses of the subjunctive mood in Spanish.Despite the fact that these studies take into account the interaction between syntactic and semantic features, they claim that semantic considerations such as presupposition, assertion and specificity, among others, play the most important role in the distribution of the Spanish subjunctive (Bolinger 1959;Rivero 1971Rivero , 1977Rivero , 1990;;Terrel 1976;Goldin 1974;Terrell & Hooper 1974;Rojas 1977;Solé & Solé 1977;Bybee & Terrell 1990;Gonzalo 1990;King 1992;Farkas 1985Farkas , 1992Farkas , 2003)).Differing from the syntactic approaches, these accounts aim to explain not only complement clauses but also contexts such as relative and adverbial clauses.
Semantic approaches generally consider that the meaning of the subjunctive is related to the speaker's attitude toward the information being reported.However, there are different perspectives about how this attitude is defined.The introduction of concepts such as assertion and presupposition offers the first important elements in the semantic characterization of mood.Using some tools provided by possible world semantics in order to explain the Spanish mood alternation, Farkas (1985Farkas ( ,1992)), in her very influential analysis for recent studies on mood in Romance languages, suggests a classification into weak and strong intensional predicates.She proposes that the indicative is selected when the proposition is true in the real world as far as the subject is concerned; otherwise, the subjunctive is chosen.Farkas uses the notion of the truth of a proposition relativized to individuals, that is, the speaker's beliefs need to be considered.Villalta (2007) argues in favor of a semantics based on the comparison of contextually relevant alternatives.According to her, speakers compute alternative semantic values of the sentence in complement clauses, that is, two semantic values rather than a semantic value.She claims that, in complement clauses, the subjunctive mood is an operator to evaluate the alternatives for the matrix predicate.In this sense, although she points out that the weak/strong intensional contrast captures an important difference between epistemic and desire predicates, other predicates select the indicative mood even if they behave like desire predicates.This is the case with verbs such as adivinar 'guess', sospechar 'suspect' and concluir 'conclude' since one may guess, suspect or conclude contradictory propositions.Similarly, emotive factive predicates (lamentar, regret) select the subjunctive mood, although they presuppose the truth of the complement clause and the speaker believes the complement clause to be true in the actual world.
Other important concepts used by the semantic approaches are existentiality and specificity.They are exclusively used in order to explain the Spanish mood choice in relative clauses (Rojas 1977;Solé & Solé, 1977;King 1992;Rivero 1977Rivero , 1990;;Gonzalo 1990).Traditionally, existence is associated to the antecedent's presence in the actual world and, mainly, the speaker's belief about this presence as ( 17) and ( 18) show.King (1992) adds that the affirmation expressed by the indicative can be seen as fact, that is, [+objective], and the lack of it conveyed by the subjunctive as [objective].
find-2SG-PAST something that to you like-2SG-PAST-SUBJ 'Did you find anything you liked?' However, the notion of existentiality poses some problems.There are some descriptions with relative clauses in the indicative and the subjunctive and no existential claims.These clauses, however, are not semantically identical.Since existential claims are inapplicable in both cases, their difference in meaning is not related to existence, as sentences ( 19) and (20) illustrate: 132 (19) Voy a comprar una blusa que no sea tan elegante como ésta going-1PL-PRES to buy a blouse that not be-3SG-PRES-SUBJ so elegant as this 'I am going to buy a blouse that may not be as elegant as this one' (20) Voy a comprar una blusa que no es tan elegante como ésta going-1PL-PRES to buy a blouse that not be-3SG-PRES-IND so elegant as this 'I am going to buy a blouse that is not as elegant as this one' In ( 19) the use of the subjunctive does not deny the existence of the blouse, but rather is related to its identification, that is, the speaker does not know which blouse in particular s/he is going to buy.In (20) the speaker is asserting both the existence and the identification of the blouse s/he is going to buy.Rivero (1977Rivero ( , 1990) ) suggests the term specificity in order to explain examples such as the previous ones.The notion of specificity, on the other hand, is related to definiteness and cannot be compared to the notion of existentiality (Rivero 1977(Rivero , 1990;;Gonzalo 1990).Specificity deals with the pairing up of an individual or an object and its characteristics, and, according to Rivero (1990), it offers a clear explanation for the selection of mood in relative clauses: the indicative refers to a specific description, the subjunctive to a nonspecific one.
Broadly speaking, from the semantic approaches, the Spanish mood alternation is no longer seen as a procedure wherein the main verb exclusively determines the use of the subjunctive or the indicative.Most of the authors explain the selection as a semantic notion manifested in syntactic terms.This notion is related to different definitions that could be grouped into three main categories: propositions referring to something that is realized/unrealized at the moment of speech; propositions referring to something that is true/untrue or certain/doubtful; and propositions referring to something that is known/ unknown or identified/unidentified to the speakers.The major differences lie in how these meanings are realized in the choice of mood.
It seems to be clear that a combination of syntactic and semantic factors is required to deal with the Spanish mood alternation.However, there are still some issues to be resolved.First, semantic approaches do not completely explain the problem of factive predicates.The factive reading may result not only from the matrix verb in combination with the complement clause, but also from other factors such as the pragmatics of the sentence and its context (Rullmann 1991 as cited in Schulz, 1999: 29).Second, although semantic accounts aim to explain the use of the subjunctive in different contexts, their focus is only on complement and relative clauses.King (1992) attempts to present a complete description of mood choice based on the [±objective] feature.However, in attempting to summarize the semantics of this feature, he ends up with a list that corresponds to the notion of presupposition and existence as explained above.As Palmer (2001) has noted, it is problematic to try to cover all the semantic factors that condition the use of the subjunctive under one term.
In general, neither syntactic nor semantic accounts offer an explanation of the use of the subjunctive in these contexts.Other studies of Spanish mood selection have indicated that much of the variation among native speakers in their choice of mood should be based on pragmatic considerations and the examination of the discourse context.Accordingly, in the following subsection, I review the most important pragmatic studies.

Pragmatic approaches
Under pragmatic accounts, researchers examine more than the speaker's attitude toward the information expressed.These studies seek to describe how the information encoded by mood markers interacts with contextual factors to express the meaning intended by the speaker.From this perspective, it is clear that the tradition of the Spanish subjunctive analysis based on the sentence has changed.Discourse, this new unit of study, is the background that allows researchers to examine how native speakers construct meanings through mood choice.Some of these studies formalize their proposal either in terms of the discourse (Lavandera 1983(Lavandera , 1984)), or within the framework of relevance theory (Lunn 1989a, 1995, Ahern 2004, Ahern & Leonetti 2004).Others look for a redefinition of the term assertion based on speakers' intentions in order to explain Spanish mood choice (Mejías-Bikandi 1994, 1996).Although these researchers may have a different concept of what pragmatics is, they all consider contextual elements such as the discourse and speakers' intentions in their analysis of the subjunctive.Guitart (1991) and Mejías-Bikandi (1994, 1998) provide examples mainly with complement clauses using factive, semifactives, or non-factive verbs (Guitart 1991, Mejías-Bikandi 1994, 1998).Lavandera (1983Lavandera ( , 1984)), Lunn (1989aLunn ( , 1995) ) and Ahern (2004) also consider certain adverbial contexts in which the notions of assertion and relevancy seem to explain the speaker's intentions.These elements add different and additional issues to the complexity of the Spanish mood choice that are relevant for subjunctive adjuncts.In general, these studies highlight the speaker's point of view and the ways in which s/he uses mood alternation to convey additional information; this is not possible to explain under a syntactic or semantic account.
From the complete overview, it should be noticed that although most of the accounts of the use of the Spanish subjunctive are of great significance for the understanding of the role of the mood choice in grammar, no explanation can give a complete description of the Spanish subjunctive mood.The syntactic accounts, whose explanations are based on the notion of the dependency and independency of the two clauses, cannot explain the free-choice use of the subjunctive in certain contexts.The semantic explanations, using notions such as assertion or presupposition, hardly explain the use of the subjunctive with factive predicates.Finally, pragmatic approaches have only proved to be accurate in describing certain adverbial clauses.In sum, it seems to be clear that a single-factor analysis of the Spanish subjunctive is difficult, if not impossible, to achieve.An accurate theoretical account should involve syntactic, semantic, and pragmatic factors as well as their interaction in order to describe the complexity of the Spanish mood system.

A multi-factor analysis of the Spanish subjunctive
The characterizing complexity of Spanish mood choice seems to be quite resistant to a unified linguistic analysis and description.Bell (1980) points out that the great semantic flexibility inherent in the Spanish mood system cannot be captured using such a restricted approach.As we mentioned above, the all-in-one explanations of the occurrence of the Spanish subjunctive provide clear descriptions for certain uses such as the lexically-selected subjunctive or free-choice subjunctive in relative clauses, but they fail to elucidate the use of the subjunctive with factive predicates or some adverbial clauses.According to Bell, most studies try to find semantic similarities to parallel the syntactic phenomena, that is, they are grouping together constructions that have more semantic dissimilarities than similarities.In this sense, it is important to recognize that not just one semantic feature is involved, but rather a range of features with a range of semantic functions (1980: 389).Although his study does not aim to develop a complete analysis of the Spanish subjunctive but rather some tools to be used in its pedagogy, Bell specifically mentions the need for finding a theory of semantic-syntactic structure.
Similarly, Lozano (1995) suggests the use of a system of features to account for the use of mood in Spanish.His study not only follows a new direction in the analysis of the Spanish subjunctive but also incorporates perspectives coming from child language development and U.S. Spanish spoken by bilinguals.Arguing against a single generalization, he proposes a multifaceted subjunctive that involves different grammatical meanings and pragmatic uses.By associating different types of modalities with the utterances used to express them, Lozano suggests that there is a deontic-imperative subjunctive as in ( 21), a deontic-evaluative subjunctive as in ( 22), and a dubitative subjunctive as in ( 23): (21) El guardia mandó que el chofer parara el coche the guard send-3SG PAST that the driver stop-3SG PAST-SUBJ the car 'The policeman ordered the driver to stop the car' ( 22) Lástima que lo haya leído en este invernizo febrero pity that it have-3SG-PRESPERF-SUBJ read in this wintery February 'It is a shame that you have read it in this wintery February' (23) Dudo que se lo pidan a ella doubt-1SG-PRES that herself it ask-3PL-PRES-SUBJ to her 'I doubt that they ask her about it' In (21), a command is expressed.In ( 22), the speaker is expressing an evaluation of the complement proposition.In (23), the speaker expresses indetermination or doubt regarding the event indicated in the complement proposition.According to Lozano, these subjunctive meanings are arranged in a network of contrasts and syntactic contexts in which they can appear, and temporally constrained in the real world.This is the case for the imperatives and imperative-like sentences in Spanish whose events can only be intended to happen immediately or in the future.This author states that the semantics of the Spanish subjunctive not only enables speakers "to express clear-cut directives, desires, evaluations and judgments, but also to voice uncertainties, doubts, half-truth and tenuous thoughts" (Lozano 1995: 113).In spite of his contribution to the analysis of the Spanish subjunctive, it is not quite clear how to explain the use of the subjunctive in relative clauses, which involve considerations such as specificity in mood selection.Similar to the previous attempts, Lozano ends up assigning meanings to the subjunctive.

3.3.Mood and interpetation
Endorsing the multifactor analysis of the Spanish subjunctive and avoiding assigning specific meanings to the subjunctive, Quer (1998) discusses different factors contributing to the Spanish subjunctive selection.From a different perspective than that of Lozano (1995), Quer takes into account the interaction of syntactic, semantic, and pragmatic features as well as the evaluation that the speaker does in selecting the mood.
In contrast to previous studies, this author provides a role for this mood in the discourse.Quer (1998) proposes that mood has no inherent semantic content itself, but rather it functions as a mark of speaker distancing, that is, it reflects a change in the interpretation model in which the proposition is being evaluated.In this sense, different levels of the language system (morpho-syntactic, semantic and pragmatic) participate in the determination of mood by interacting among themselves.In this interaction, it is possible to observe different levels of priority between the components of the language because one may precede another or all of the others.Utterances are interpreted by taking into account specific aspects of the context and the individual anchors that represent models of evaluation.
Based on a dynamic theory of meaning, Quer (1998) proposes a concept of model that incorporates into the semantics certain elements of interpretation that have been restricted to the domain of pragmatics and language use.An utterance is interpreted taking into account a set of referents -entities that have been introduced into the context by the proposition-and a set of conditions -predicates that are known to hold of these entities.Meaning is not only what words mean in isolation; it is also determined by context, variation in truth conditions, and individuals.For example, the model for a sentence like a dog barked is represented in (24), a notational variant of a predicate calculus formula.
(24) A dog barked X dog (x) barked (x) Sentences could also have presuppositions about the types of context in which they can be interpreted.For instance, the dog chased a cat presupposes that there is a dog in the discourse context.The general representation of a sentence is a single assertion and a set of presuppositions as seen in ( 25).Assertions change the context by adding information provided by their content.In this sense, they take place within a context in which the speakers share some beliefs and assumptions, and they also affect the speakers' background assumptions about what the actual world is like because they update their common ground.Presuppositions, on the other hand, describe a set of worlds in which all of the propositions assumed to be taken for granted are true.The update operation consists of removing the worlds in which these propositions are false while keeping the worlds in which the proposition is true.In other words, a presuppositional context changes systematically with the utterance of a new sentence.Its interpretation results in a new informative context that involves the new information (Stalnaker 1973(Stalnaker , 1974(Stalnaker , 1978)).In general, under a dynamic perspective, utterances are interpreted against the background of specific contexts.They play an important role because, different from utterance contexts (time of the utterance, the speakers, their intentions), this dynamic notion allows the introduction of the discourse as a relevant feature in defining a sentence meaning.According to Quer (citing Giannakidou 1998 andCondoravdi 1994), a context can be defined as ( 26): (26) c = <cg(c),W(c), s, h, w 0 , f… > It involves both Kaplanian and informational parameters. 1 Among the first ones s corresponds to the speaker, h to the hearer, w 0 to the world where the utterance takes place, and f to the function that assigns values to variables.The informational parameters of the context are the common ground cg(c), the information shared by the illocutionary agents at the point where an assertion is uttered, and the context set W(c), that is, the set of worlds compatible with the informational state stored as the common ground.As new assertions are added, the common ground is incremented and the context set is reduced.From this perspective, meaning involves more than semantics.If we examine an expression such as "Lovely weather we're having", the actual weather at the time and place where the statement is produced is the most important element to determine the meaning.If the weather is fine, this statement will likely be expressing a necessary truth about it as well as an expression of delight at this fact.It also can convey new information to the hearer.Nevertheless, if the weather is not fine, the sentence must be understood to be a sarcastic comment about the awful weather.
In this way, Quer considers that the notion of truth must be relativized to models within a context and to individuals.Following Giannakidou (1998), he defines models of individuals as in ( 27). ( 27) Let c = cg(c), W(c), M, s, h, w 0 , f, be a context A model M(x) ∈ M is a set of worlds in c associated with an individual x x is called the individual anchor 1 Kaplan's theory evaluates linguistic expressions in contexts.Content is examined in, or with respect to, contexts.Each context c has at least an agent, time, location, and possible world associated with it.Thus, the content of a sentence corresponds to a proposition that have individuals, properties, and relations as constituents.The content of a sentence S with respect to c is made up of the contents of the words in S with respect to c.
The term individual anchor is borrowed from Farkas (1985Farkas ( , 1992)).It proposes sentences are not true or false in isolation but always with respect to some individual.In this sense, models of individuals are defined as collections of worlds c and represent distinct sets of worlds.For sentences embedded under epistemic verbs, for example, M(x) stands for some individual's belief state: it represents the epistemic status of that individual and includes worlds compatible with what x believes in the actual world.Its definition corresponds to (28) where M(x) is indexed with B: It is important to notice that the worlds compatible with one's beliefs are not identical to the worlds compatible with one's dreams.M B (x) and M D (x) refer to different but possibly intersecting sets of worlds with respect to the same individual.In the case of assertives, the model is treated as conceptualizing the context of reported conversation.It includes worlds compatible with what x takes the reported conversation to be, as illustrated in (30).M RC (x) should include worlds different from the ones in M B (x) and M D (x).
(30) Model of reported conversation Let c = cg(c), W(c), M, s, h, w 0 , f,…> be a context A model M RC (x) ∈ M is a set of worlds in c associated with an individual x, representing worlds compatible with what x takes the reported conversation to be.
All these models M(x) are basically epistemic.According to Giannakidou (1998), what one dreams represents the belief state of an individual while (s)he is dreaming, and what one takes the reported conversation to be represents the belief state of an individual as regards to the reported conversation.Therefore, as I mentioned above, meaning is not only what words mean in isolation; it is also determined by context, variation in truth conditions, and individuals.Quer (1998) underlines the variation of model type according to the worlds introduced by the matrix predicate.According to him, this conception is essential for the interpretation of embedded propositions, since they introduce specific types of models into the context.Quer proposes that mood signals the kind of model a given proposition must be interpreted in, that is, mood morphology is "the overt marking of a change in the model for the evaluation of the proposition or property expressed by the embedded clause."(Quer 2001: 81).For complement clauses, two types of subjunctives have been proposed: intensional and polarity subjunctive.Although the properties of the selecting element are different, the general subjunctive category is not.Thus, instead of having two different subjunctive complements attributed to differentiating properties of the embedded clauses, Quer suggests that the fixed subjunctive should be interpreted as an s-selectional fact related to lexico-semantic properties of the subordinate clause.The characteristics of the selecting head correspond to the element that selects the embedded clause mood.Following Farkas (1992), Quer points out that the triggering property of the subjunctive is that these predicates (volitional, directives, permissives, among others) are intensionally anchored to the relevant individual, that is, they allow the introduction of a set of future alternative worlds as a lexical property.Thus, epistemic verbs (such as creer), although intensional, generally do not appear with subjunctive complement clauses because they introduce only a particular world.In sum, "lexical selection is what determines mood in intensional domains given that the computational system requires morphological encoding of such a dependency on economy grounds" (Quer 1998: 90).On the other hand, when there is a choice between the subjunctive and indicative, the shift from one mood to the other signals a shift in the model of evaluation for that proposition, that is, mood overtly marks information about the models where clauses are to be interpreted by speakers.
In relative clauses, lexical selection is not an issue.They offer a descriptive condition of a noun that corresponds to an argument of the main predicate.In this sense, the mood selection signals the model where this condition is evaluated.According to Quer, the mood shift to the subjunctive indicates that the property expressed by the relative clause has to be evaluated in the model introduced by the intensional verb buscar 'look for'.By contrast, there is no mood shift with the indicative because the condition is evaluated in the epistemic model of the speaker, or in the model of evaluation anchored to the referent of the main clause.This evaluation creates the effect of 'wide scope' of the property.In adjuncts such as concessive conditionals where the subjunctive mood is displayed, Quer proposes that the evaluation model is an 'expanded' one.Speakers must go through different worlds or situations in order to evaluate the concessive antecedent.With the indicative, concessives are evaluated in the base world in the epistemic model of the speaker where both propositions (the adjunct and the main proposition) are asserted.With the subjunctive, only the proposition in the main clause is.
In summary, Quer's contribution provides a role for the subjunctive mood in the discourse.Although Villalta (2007) points out that it remains unclear what exact property of a model triggers mood shift in complement clauses, this explanation for the Spanish mood alternation is accurate in describing an impressive array of constructions in which the subjunctive mood may appear.It also involves the interaction of different components of the grammar in the process of interpretation.This contribution offers a dynamic view of meaning in which context and individuals are crucial for semantic interpretation.

Morpho-syntactic, semantic, and pragmatic considerations in Spanish
subjunctive adjuncts From Quer's perspective (1998, 2001), the Spanish subjunctive mood signals a change in the context of the evaluation of a proposition, that is, a 'model shift.'This approach recognizes that mood contributes no inherent meaning itself.The speaker directly relates the proposition to the actual world, and the evaluation of the proposition requires considerations of other worlds and criteria, with the relevant modal base defined in different terms depending on the context (Ahern 2005).Unlike complement clauses, the use of the subjunctive in adjunct clauses do not follow fixed distributional patterns.In determining mood in adjuncts, the speaker considers different components of the language system.At the interpretive level, in addition to the evaluation of the model, speakers must also evaluate and identify the role played by temporal determination of the status of the clause, co-occurrence or sequence-oftense-phenomena, and sensitivity to semantic and pragmatic factors, contextualized meanings.
In this section, I will concentrate on discussing mood selection in adjunct clauses.I will only focus on those features such as tense, specificity, habituality, among others, which are involved in mood selection.In particular, I will discuss each of the main contrasts in temporal, concessive and conditional clauses.My aim is to accurately describe them and to illustrate the interaction of morpho-syntactic, semantic, and pragmatic considerations involve in the speaker's evaluation of the propositions.

4.1.Temporal clauses
Simple finite clauses present a relationship between a state or event and the deictic moment of speech, or speech time (following Reichenbach, 1947).Temporal adjunct clauses do not alter this relationship, but instead situate the event in relation to other events (Zagona 2002: 61); or they delimit the temporality of the event in the main clause (García-Fernández 1999), as shown in ( 31) and ( 32): (31) Sandra llegó antes de que saliéramos Sandra arrive-3SG-PAST-IND before of that leave-1PL-PAST-SUBJ 'Sandra arrived before we left' (32) Sandra llegó cuando salimos

Sandra arrive-3SG-PAST-IND when leave-1PL-PAST-IND
'Sandra arrived when we left' The type or relationship between those two events is conditioned by the subordinating conjunction; thus, they can take place either with one before/after the other, or they can occur simultaneously.

Clauses with cuando
Clauses introduced by cuando 'when' exhibit some interesting patterns.Reference to future events is marked by the subjunctive.This appears as an instance of sequence-of-tense phenomena, in that both the main and the embedded clause bear a [±PAST] tense (Zagona 2002 The cuando clause in (33) refers to a particular moment and to a known fact in the past.It specifies the temporal setting of the event expressed by the matrix clause.In this utterance, the speaker states that the main event took place in that moment and assigns a factual reading to the meaning.On the contrary, in (34), the embedded event is referred as a virtual event that supposedly takes place at a non-specific moment in the future.This basic alignment of mood and temporality gives way to an additional semantic contrast, based on the ambiguity of the Spanish imperfective present tense.Cuando clauses in the present contribute information about the epistemic status of the clause, by disambiguating between a habitual interpretation of the present tense ('I usually give him a present'), and a future reading of the present tense ('I will give him a present').A habitual situation, as in ( 35), represents a fact for the speaker (it has happened and will happen again), but a unique event projected into the future, as in (36), is merely construed as possible, not actual.An additional fact underscores the aspectual considerations of mood selection: in certain cuando clauses, mood contrasts involve the specificity of place or time, as shown in ( 5) and ( 6) repeated here as (37) and ( 38): (37) Tomó esa clase cuando su profesor lo autorizó take-3SG-PAST that class when her/his teacher it approve-3SG-PAST-IND 'S/he took that class when her/his teacher approved it' (i) Juan se fue antes de que María llegara.'Juan left before she arrived'.(ii) Si Juan se fue antes de que María llegara, no pudieron conocerse.
'If Juan left before she arrived, they were not able to meet' (38) Tomará esa clase cuando su profesor lo autorice take-3SG-FUT that class when her/his teacher it approve-3SG-PRES-SUBJ 'S/he will take that class when(ever) her/his teacher approves it' Example (37) is said with a specific time in mind, that is, the moment when the teacher gave permission to the subject to take the class.In this case, that time is emphasized by the indicative.Example (38) does not refer to a particular time.The event of 'being authorized' may happen any time.
I will concentrate in this article on the distinctions resulting from the sequence-oftense phenomenon and the epistemic status of the temporal clauses with cuando.Temporal clauses that refer to situations following the reference time (matrix clause event) are used solely in the subjunctive.Antes (de) que 'before' clauses are always followed by a subjunctive verb, because they refer to an action that follows the main event time as ( 39) and (40) illustrate: (39) El vuelo salió, antes de que llegara José the plane leave-3SG-PAST-IND before of that arrive-3SG-PAST-SUBJ Jose 'The plane took off before Jose arrived' (40) El vuelo saldrá, antes de que llegue José the plane leave-3SG-FUT-IND before of that arrive-3SG-PRES-SUBJ Jose 'The plane will take off before Jose arrives' It is important to note that the subordinate event is unspecified with regards to utterance time.According to García-Fernández (1999), antes (de) que clauses contribute to factual, non-factual, and counterfactual readings of the subordinate verb.However, these readings do not affect the use of the subjunctive.Antes (de) que clauses also exhibit the sequence-of-tense phenomena as mentioned above (Zagona 2002).Table 2 shows their main characteristics.Temporal clauses with mientras 'while' involve a simultaneous relationship between the event expressed in the main clause and the one expressed in the subordinate clause.Both events take place at the same time in the present or in the past, "which means that reference is made to experienced or factual states of affairs" (Haverkate 2002: 135).Sentences ( 41) and ( 42) present some examples: (41) Hago el té mientras haces la lista del mercado make-1SG-PRES the tea while make-2SG-PRES-IND the list of groceries 'I make the tea while you are making the grocery list' (42) Hacía/hice el té mientras hacías/hiciste la lista del mercado make-1SG-IMP/PRET the tea while make-2SG-IMP/PRET-IND the list of groceries 'I was making/made the tea while you were making/made the grocery list' In ( 42) there is a difference in meaning caused by an aspectual distinction between the verb tenses.With the imperfect the event has no end point.This tense contributes a habitual interpretation in the past.The preterite, on the other hand, refers to a completed action with a specific duration in the past.Although this distinction is very important in aspect-oriented studies, I will only focus on the issue related to the use of the indicative mood in this work.Table 3 presents the main distinctions:

4.2.Concessive clauses
Concession is a complex operation.König (1988) and König & Siemund (2000) observed that a concessive construction has the logical form as in ( 43) and ( 44): (43) Although p, q (44) Although it is raining, I'm going out for a walk A connection is implied between the two propositions (p, q) as expressed in (43).In the two related clauses, "the speaker asserts the propositions against the background assumption that the two types of situations which p and q describe are generally incompatible" (Crevels 2000: 313).This can be schematically summarized as in ( 45)-( 46).
(45) Although p, q implicates a presupposition if p', then normally ¬q' (46) Although it is raining, I'm going out for a walk [if it is raining] then normally [I am not going out for a walk] Spanish expresses concessive relations by specialized connectives such as aún cuando 'even though', aún si 'even if', and aunque 'although', among others.In this study I will concentrate on concessive constructions with aunque 'although' as illustrated in ( 47): (47) El vuelo salió aunque estaba/estuviera lloviendo the plane leave-3SG-PAST-IND although be-3SG-PAST-IND/PAST-SUBJ raining 'The plane took off even though it was raining' In (40) the main clause is asserted and it is implied, as mentioned above, that the two propositions are in some sense incompatible.In this particular example, the incompatibility is based on the common assumption that a plane does not take off when it is raining.Regarding the mood, both the indicative and the subjunctive are possible, as shown in (40).However, although the opposition still holds, they do not mean the same.
On the one hand, mood selection must show understanding of the role played by temporal determination and factuality.Constructions with aunque + IND involve a factual interpretation with the present or past as exemplified in (48).The speaker states that both events took place.With the future tense, these constructions contribute a non-factual reading, as shown in (49).
(48) Aunque está/estaba lloviendo, José sale/salió a caminar although be-3SG-PRES-IND/PAST-IND raining Jose leave-3SG-PRES/PAST to walk 'Even though it is/was raining, José is going/went for a walk' (49) Aunque ya habrán empezado la reunión, hago la llamada although already have-3PL-FUT PERF-IND started the meeting make-1SG-PRES the call 'Although they already have started the meeting, I am making the phone call' Constructions with aunque + SUBJ involve a non-factual interpretation with the past as exemplified in (50).Additionally, the embedded event is believed to be false by the speaker, which contributes to a counterfactual reading.
(50) Aunque ellos los apoyaran/hubieran apoyado, perderían/habrían perdido el partido although they them support-3PL-IMP-SUBJ/PLUPERF-SUBJ lose-3PL-COND/PERFCOND the game 'Although they supported/had supported them, they would lose/have lost the the proposition in the main clause and the embedded event may be referred to either as a factual event, or a virtual event, as illustrated in ( 54) and ( 55) respectively.
(54) Aunque te caigan mal mis padres, debes intentar ser más amable although you fall-3PL-PRES-SUBJ bad my parents must-2SG-PRES attempt be more nice 'Even though you don't get along with my parents, you must try to be nicer' (55) Aunque llegaran a un acuerdo, los principales problemas seguirían sin solución although arrive-3PL-PAST-SUBJ to an agreement the main problems continue-3SG-COND without solution 'Even if they made a deal, the main problems would remain unsolved' In (54), both the speaker and the hearer know the event expressed in the adjunct clause to be an actual event.Accordingly, this information has little or no relevance for them and so the subjunctive is used.In (55), the embedded event indicates nonfactuality.In these types of constructions, mood and temporality are aligned, i.e., the imperfect subjunctive of the aunque clause correlates with the conditional of the main clause, and the pluperfect subjunctive are associated with the perfect conditional.These patterns are very similar to those of the conditional clauses and are related to counterfactuality.
In summary,

4.3.Conditional clauses
Conditional constructions, as shown in (56), involve two clauses in a particular relationship whose logical representation has been formulated as if p, q.The main clause, formally described as consequent or apodosis (q), establishes a condition, and the adjunct clause, referred to as antecedent or protasis (p), is dependent on this condition. 456) Si nos dan el préstamo, iremos a Londres if us give-2SG-PRES-IND the loan go-3PL-FUT to London 'If we get the loan, we'll go to London' In Spanish, conditional clauses may be linked with a small set of conjunctions such as con tal (de) que 'provided that', en caso (de) que 'in case of', siempre que 'provided that', mientras 'as long as', and si 'if', among others.They may precede or follow the main clause as ( 57)-( 60) exemplify.Si-clauses, as opposed to the other above conjunctions, admit the selection of the indicative and the subjunctive mood.
(57) Si/en caso de que tuviera el dinero, compraría el auto If/in case that have-1/3SG-PAST-SUBJ the money buy-1/3SG-COND the car 'If/in case I/s/he had the money, I/s/he would buy the car' (58) Compraría el auto, si/en caso de que tuviera el dinero, buy-1/3SG-COND the car if/in case that have-1/3SG-PAST-SUBJ the money 'I/s/he would buy the car, if/in the case that I/s/he had the money' (59) Con tal de que/con la condición de que/siempre que/mientras escribas, todos estaremos felices provided that/with the condition that/always that/as long write-2SG-PRES-SUBJ all be-1PL-FUT happy 'Provided that/on the condition that/provided that/as long as you write, all of us will be happy' (60) Todos estaremos felices, con tal de que/con la condición de que/siempre que/mientras escribas all be-1PL-FUT happy provided that/with the condition that/always that/as long as write-2SG-PRES-SUBJ 'All of us will be happy, provided that/on the condition that/provided that/as long as you write' In this article I will concentrate on two main conditional constructions: clauses with mientras and clauses with si.

Clauses with mientras
As mentioned above, mientras 'while' clauses with indicative contribute to a temporal reading of the events.On the other hand, the mientras clauses with subjunctive involve a non-factual interpretation of the events.The speaker does not determine the duration of the event that s/he is referring to as shown in ( 61) and ( 62).
(61) Usarás/usas el auto mientras seas cuidadoso use-2SG-FUT/PRES the car while be-2SG-PRES-SUBJ 'You'll drive the car as long as you are careful' (62) Usabas/*usaste el auto mientras fueras cuidadoso use-2SG-IMP/*PRET the car while be-2SG-IMP-SUBJ 'You'd drive the car as long as you were careful' Once again, the alignment between mood and temporality is present in mientras clauses.However, one fact draws attention to this relation as illustrated in (61): the subjunctive cannot be used with the preterite in the main event.Since this tense marks a completed event, the meaning is linked to an actual event, not to a non-factual interpretation; and the indicative is used instead.Mientras clauses contrasts are summarized in Table 5.

Conclusion
In this article I have sought to explain those relevant features that the speaker needs to take into consideration to evaluate the contextual alternatives introduced by temporal, concessive, and conditional clauses.Table 7 shows those contexts that exhibit co-ocurrence or sequence-of-tense phenomena.Table 8 illustrates those contexts that involve semantic and pragmatic contrasts, that is, contextualized meanings.I have argued for a multifactor perspective in the analysis of the Spanish subjunctive.After examining different approaches and their implications and shortcomings, I have presented a point of view based on sets of worlds also called "models" (Giannakidou 1998;Quer 1998).This perspective proposes contexts of evaluation that are different for the main and the subordinate clauses and link mood and the interpretative effects on propositions.Speakers evaluate propositions taking into account the model relativized to a context and an individual.Instead of attributing meanings to mood categories, speakers consider them by assessing different levels of the language system.As mentioned above, this approach considers the Spanish mood alternation as a complex and systematic process of interpretation rather than a simple and unmotivated distribution of meanings.
Finally, I have presented morpho-syntactic, semantic and pragmatic considerations of the models of interpretation in temporal, concessive and conditional clauses.In selecting the appropriate Spanish mood, speakers must show understanding of the role played by temporal determination of the status of the clause and sensitivity to semantic and pragmatic factors.This approach, in general, is a dynamic view of meaning and interpretation.

(
28) Belief model Let c = cg(c), W(c), M, s, h, w 0 , f,…> be a context A model M B (x) ∈ M is a set of worlds in c associated with an individual x, representing worlds compatible with what x believes Speakers may also want to understand M(x) as representing a fictional reality according to some individual; M(x) in this case involves worlds compatible with what x dreams or imagines.Its definition corresponds to (29): (29) Dream model Let c = cg(c), W(c), M, s, h, w 0 , f,…> be a context A model M D (x) ∈ M is a set of worlds in c associated with an individual x, representing worlds compatible with what x dreams 2

( 35 )
Le doy un regalo cuando viene a verme him give-1SG-PRES-IND a present when come-3SG-PRES-IND to see-me'I am giving him a present when he comes to see me' (36) Le doy un regalo cuando venga a verme him give-1SG-PRES-IND a present when come-3SG-PRES-SUBJ to see-me 'I will give him a present whenever he comes to visit me'

Table 1 .
Table 1 presents these contrasts: Cuando 'when' clauses Main Clause Temporal Clause

Table 3 .
Mientras 'while' clauses Main Clause Temporal Clause Table 4 presents these contrasts: Aunque hubieran llegado a tiempo, no habrían conseguido entradas even though have-3PL-PAST-SUBJ arrived on time no have-3PL-COND got tickets 'Even if they had arrived on time, they would not have gotten tickets'

Table 6 .
Si 'if' clauses Main Clause Conditional Clause

Table 7 .
Summary table: Subjunctive adjuncts with sequence-of-tense phenomena

Table 8 .
Summary table: Subjunctive adjuncts with semantic and pragmatic contrasts