CC BY
31
М. Амбар, М. Димитрова, Д. Амарал
Лиссабонский университет, 1600-414, Лиссабон, Португалия
Межъязыковая загадка сослагательного наклонения1
Цель этой статьи - исследовать дистрибуцию сослагательного наклонения в матричных и зависимых предикациях. Опираясь на недавние достижения в исследовании означивания временных признаков, мы предлагаем критическое обсуждение описанных в литературе характеристик этого наклонения. Основная часть работы посвящена межъязыковому варьированию в области кодирования сослагательности, наблюдаемому в романских, славянских и балканских языках. Особое внимание уделяется поведению сослагательного наклонения в матричных клаузах, которое существенно различается в исследуемых языках.
Ключевые слова: сослагательное наклонение, означивание временных признаков, пропозициональная установка.
M. Ambar, M. Dimitrova, D. Amaral
University of Lisbon, Lisbon 1600-414, Portugal
A crosslinguistic puzzle for Subjunctive2
The goal of the paper is to examine the intriguing behavior of the subjunctive mood in matrix and embedded clauses. Relying on recent studies on tense-features valuation, we will first discuss some well-known properties of this
1 Исследование проводилось при поддержке диссертационного гранта SFRH/ BD/99302/2013, предоставленного Фондом в поддержку науки и технология (Португалия).
2 This work has been developed under the PhD grant SFRH/BD/99302/2013 provided by Fundagaopara a Ciencia e a Tecnologia 'Foundation for the Science and the Technology'.
mood namely tense dependency, event and obviation effects. Then we will turn to crosslinguistic variation focusing on the way three groups of languages, namely Romance, Slavic and Balkan, codify subjunctive. Special attention will be paid to the selection of subjunctive in main clauses and to some discrepancies displayed by the languages under scrutiny.
Key words: subjunctive mood, tense-valuation, evaluation.
1. Introduction
The subjunctive mood has been subject to numerous discussions in the generative literature.3 However, some aspects of its characterization across languages remain unsettled.
As the label itself suggests, subjunctive encodes dependency. As a corollary it has been traditionally considered the mood of subordination par excellence. In the generative framework a now well-known line of inquiry has capitalized on the idea that in subjunctive clauses the tense of the embedded domain is anaphoric with respect to the tense of the matrix domain [Picallo, 1984; Raposo, 1985; Ambar, 1988; Borer, 1989; a.o.]. In a sense the selection of subjunctive in complements to volitional predicates has been therefore associated with tense defectiveness. We will refer to this line of inquiry as the anaphoric or dependent tense view, typing the C domain of subjunctive as [-T].
A different line of inquiry that may be extended to subjunctive clauses is advocated in Martin (1996) and Boskovic (1997), who argue that clausal complements to volitional predicates are [+T(ense)], inspired in Stowell (1982) and Eng (1991). Stowell observed that the temporal interpretation of Control infinitival structures under volitional predicates is independent with respect to the matrix tense; it is unrealized, future. Let's label this approach the non-anaphoric or independent tense view, typing the C domain of subjunctive as [+T].
Clearly, the two lines of inquiry conflict. Assuming that each of them captures different properties of the structures under study, it is desirable to solve that tension.
Our first goal in this paper is to examine to what extent a system designed to derive subjunctive clauses and to solve the tension between the dependency vs independency of the Tense analyses still keeps doing good predictions after broadening of its initial empirical domain. We are referring to the tense-features valuation system proposed in [Ambar, 1998, 2005, 2007, 2016] according to which t-features enter the derivation as bundles
3 For criticism on the realis-irrealis dichotomy in the study of indicative vs. subjunctive see [Farkas, 1992; Giannakidou, 1998], a.o.
[Chomsky, 2001].4 The distinction between tt-features, related to the highest morphological tense phrase (responsible for nominative), and tev-features, related to the lower tense object phrase5 (responsible for accusative), correctly captures the arguments of both views.
In addition, our second goal is to shed some light on the behavior of subjunctive in other contexts such as main clauses and unselected subjunctive clauses [Ambar, 2016]. Particularly important here will be the data from Russian and Balkan languages. As widely discussed in the literature [Dobrovie-Sorin, 1994, 2001; Giannakidou, 2009; a.o.] and differently from Portuguese, which we take as representative of Romance languages, Slavic and Balkan languages such as Russian, Greek and Bulgarian do not exhibit verbal subjunctive morphology and use special particles (by, na and da, respectively) for the expression of subjunctive. Still, Russian sharply diverges from Greek and Bulgarian with respect to the tense of the embedded verb and to obvia-tion. Thus, focusing on the three groups of languages (Romance vs Slavic vs Balkan) we aim at a better understanding of subjunctive by attempting to complete some puzzles concerning its expression across languages.
Crucially, the relation with evaluation and with the concept of (non)verid-icality [Giannakidou, 1998] will be accounted for under the speaker's projections EvaluativeP and AssertiveP, independently proposed in [Ambar, 2000, 2003] for other phenomena.
The paper is organized as follows. In section 2 we discuss data from subjunctive in embedded clauses referring to Ambar's system of tense-features valuation. In section 3 we focus on unselected subjunctive clauses and main subjunctive clauses pointing out to their relation with evaluation. In section 4 the data from Russian, Greek and Bulgarian will be discussed. Section 5 sets up the conclusion remarks.
2. Subjunctive embedded clauses, t-valuation
and obviation effects
The distribution of subjunctive in Romance and other European Languages has been thoroughly discussed in the literature [Picallo, 1984; Raposo, 1985; Ambar, 1988; Giannakidou, 1998; Quer, 1998, 2006; Kempchinsky, 2009; a.o.]. Taking European Portuguese as representative of the languages that display subjunctive morphology in the verb, examples (1a)-(1c) illustrate the selection of subjunctive vs. indicative in clausal complements to volitional (1a), directive (1b) or emotive factive (1c) predicates:
4 Pesetsky and Torrego (2004) also extend valuation to tense-features. For technical differences between their proposal and Ambar's one see the works cited.
5 As proposed in Ambar (1996, 1998) and in Pesetsky and Torrego (2001).
25
(1) Portuguese
a. A Maria quer que a filha fique the Mary wants that the daughter stays. subj em casa. at home
'Mary wants that her daughter stays at home.'
/ *fica / StayS.IND
b. O Joao pediu que a the John asked that the o relatyrio. the report
'John asked the secretary to write the report.'
secretâria escrevesse / *escreveu
secretary writes.suBj / wrote.iND
que that
os the
filhos children
c. O Pedro lamenta the Peter regrets tarde. late
'Peter regrets that his children will arrive late.'
cheguem /*chegam arrive.subj / arrive.ind
In (1) indicative is infelicitous under such predicates. The reverse scenario is observed with complements to epistemic verbs such as 'know'. In these contexts indicative but not subjunctive is selected:
Maria vem / *venha mais
Mary comes.ind / comes.subj more
(2) Portuguese
O Joao sabe que a the John knows that the tarde. late
'John knows that Mary is coming later.'
The indicative-subjunctive divide roughly illustrated above can be then related to the [±Tense] on C. Accordingly, on the one hand, volitional, directive and emotive factive predicates, as in (1), with [-T] on C, are only compatible with subjunctive which has been therefore associated with the defectiveness of the embedded tense. A note on complements to emotive factive verbs is mandatory here (even if limitations of space preclude a full discussion on the topic). The clausal complements of this class of predicates are tense domains [Ambar, 1998; Martins, 2001]. Thus, indicative, rather than subjunctive, should occur in this context. This is the case in languages like
Russian, Bulgarian or Greek, though not the case of Romance languages such as Portuguese, French or Spanish. Ambar (1998) proposed those verbs select DP whose D is [+T]; in turn D selects C [-T]. The generalization that subjunctive occurs in structures whose C is [-T] can then be maintained.
Epistemic predicates such as saber 'know' in (2), on the other hand, have [+T] on C and, consequently, select indicative which has been considered independent of the matrix tense. Note that the anaphoric or dependent tense analysis of subjunctive readily accounts for one of the central properties of this mood: obviation, exemplified in (3) below:
(3) Portuguese
O Joao . quer que ele,. va as aulas.
the John wants that he goes.suBj to-the classes 'John wants that he goes to the classes.'
The obviation effects straightforwardly follow from the anaphoric tense approach: with [-T] on C the binding domain of the embedded clause is extended to the matrix. Then Principle B of the Binding theory applies with the effect that the overt or null embedded subject pronoun (coindexed with Tense (or I), cf. Borer (1989)) cannot be bound by the matrix subject in its own binding domain.
The analysis of tense as anaphoric however conflicts with the non-anaphoric or independent tense view held by Boskovic (1997) and Martin (1996), as we saw above. This view is in line of Stowell's (1982) insight that control infinitives (though not ECM structures) are specified for tense. Eng (1991) supports Stowell's hypothesis by arguing that eventive verbs contain a temporal argument that needs to be bound. Tense is a binder for that argument, reason why eventive predicates can occur in these structures (though not in ECM: John tried to buy the book vs. John believed Peter to buy the book). Thus, just as the anaphoric or dependent tense view elegantly accounts for the obviation phenomenon, the non-anaphoric or independent tense view also accounts for other phenomena.
As mentioned above, we will adopt Ambar's system for t-features valuation which solves the apparent paradox between these two views. Her system shares with Pesetsky and Torrego's (2001, 2004) proposals the extension of valuation to t(ense)-features (for technical differences, namely on bundles of features and on the combinations [±interpretable] and [±valued] of features, see the works cited and also Duarte et al. (2005)). Adopting the probegoal system [Chomsky, 2001; Ambar, 1998, 2005, 2007, 2016] assumes that features enter the heads C, T, v, V as bundles. She distinguished between two types of tense-features - tt-features, (related to the high morphological
Tense (TP), responsible for nominative case) and tev-features (associated with the tense object (ToP), responsible for accusative and properties of event and Aktionsart). Much like Chomsky's (2001) interpretable vs uninterpret-able features, features in Ambar's system enter the derivation as valued or unvalued, the latter receiving a value by Agree. In subjunctive embedded clauses both t-features and t -features are unvalued in T and C. Howev-
t ev
er, only tev-features are valued in V; tt-features being unvalued. Therefore, the unvalued t -feature of v probes the t -feature of V and receives a value.
ev ev
Accordingly, the tev-feature of v is probed by T and later by C, which also get valued. The tt-feature however remains unvalued and, as a consequence, must be valued by the matrix verb or, for concreteness, by a given feature responsible for selection.
This system can be moreover extended to languages that exhibit other strategies for satisfying the selection-requirement. Russian is one such language. As the other Slavic and Balkan languages, Russian does not display subjunctive morphology on the verb. Instead, the verbal tense of subjunctive clauses is Past Indicative which is therefore specified for both tt-features and t -features. Since all features are valued, the domain is t-complete and cto
ev
is merged to satisfy EPP (as generally in indicative structures). The particle by is then merged in EvaluativeP, cto moves and adjoins to it (plausibly for clause typing):
(4) Russian
Иван хочет, чтобы он поцеловал Надю.
Ivan xocet ctoby on*i/j potseloval Nadju.
John wants that-suBj he kissed.past.3sG Nadja
'John wants that he kisses Nadja.'
According to the system we adopt here, by is the goal of the probe of the matrix verb which values the unvalued feature responsible for selection of subjunctive. The obviation effects in (4) are accounted for in a principled way: the embedded domain extends to the matrix one via incorporation of by into cto and the probe goal relation between the unvalued feature in the matrix verb and by.
Note that the system for t-features valuation adopted here solves the conflict between the dependency vs. independency views discussed above:
(i) tev-features, responsible for event, enter the derivation with a value which successfully accounts for Stowell's observations regarding the interpretation of unrealized future event;
(ii) the tt-features get valued by the matrix verb which captures the view according to which the embedded domain extends to the matrix one.
3. Unselected subjunctive clauses and main clauses
3.1. Unselected subjunctive clauses
Different works [Giorgi, Pianesi, 1997; Quer, 1998; Giannakidou, 1998, 2009, 2016; Ambar, 2005, 2007, 2016; Oliveira, 2008; Marques, 2009; a.o.] have pointed out that the selection of indicative and subjunctive is not so well-defined as it might seem at first glance when we consider verbs like acreditar 'believe' (5a) and imaginar 'imagine' (5b). As shown below, these verbs are compatible with both indicative and subjunctive:
(5) Portuguese
a. Acredito que ele consegue / consiga chegar believe.1sG that he manages.ind / manages.subj arrive.inf a tempo.
on time
'I believe that he will manage to arrive on time.'
b. Imagino que o Pedro está / esteja cansado. imagine.1sG that the Peter is.ind / is.subj tired
'I imagine that Peter is tired.'
According to Marques (2009) the subjunctive mood in (5a) and (5b) associates with the speaker's lower degree of commitment to the truth of the proposition. Ambar (2016) claims that the occurrence of subjunctive with such predicates is not a result of selection, hence the label unselected subjunctive clauses that we adopt. Thus, in contrast to the structures described by (1) and (2) above where subjunctive is selected by the matrix predicate which values the unvalued tt-features of the embedded verb, with predicates such as acreditar 'believe' or imaginar 'imagine', that typically select indicative, it is an Operator) merged in CP that does the job of assigning a value to the unvalued tt-features. As will be discussed below, the Op is an instantiation of some features of the Left Periphery in need of valuation, namely those of EvaluativeP.
3.2. Subjunctive main clauses
Let us now take a look at subjunctive main clauses. Here, we arrive at another controversy. Subjunctive has been considered the mood of subordination par excellence and therefore ruled out from assertions such as (6):
(6) Portuguese
Ele vai / *vá ao cinema.
he go.iND.3sG / go.suBj.3sG to movies
'He goes to the movies.'
However, it appears that it is productive in those main clauses in which the speaker's evaluation or kind of attitude are expressed, the indicative being ruled out here:
(7) Portuguese
Va / *Vai ele as aulas!
go.suBj.3sG / go.iND.3sG he to-the courses 'Let him go to the courses!'
Importantly, the data in (6) and (7) show that the indicative-subjunctive divide is more sophisticated: besides the divergences concerning tense-valuation in complements to given predicates, indicative and subjunctive differ with respect to the type of illocutionary force they associate with. Indicative is the mood of assertions, compatible with epistemic and factive predicates denoting 'knowledge', while subjunctive codifies the speaker's evaluations and kind of attitude [Ambar, 2016]. In Giannakidou's (1998) terms the properties of indicative and subjunctive are captured under the notion of (non) veridicality, the subjunctive being the mood of the nonveridical domain. In order to account for the interplay between indicative and subjunctive, we will consider two projections of the Left Periphery: the speaker's projections AssertiveP and EvaluativeP [Ambar, 2000, 2003, 2016]. We assume that these projections take part in the derivation of subjunctive:
(8) [EvaluativeP [AssertiveP [XP [WhP [FocP [XP [TP
The syntactic representation of subjunctive main clauses will then pattern with that of unselected subjunctive clauses, selection being a result of merging Op in EvaluativeP. In languages such as Portuguese, Op values the unvalued tt-features of V. In the next section we observe how the system proposed here extends to other languages, namely Russian and the Balkan languages.
4. More on the cross-linguistic puzzle
As shown in section 2, the system we adopt for Portuguese convincingly captures the data from Russian: since tt-features and tev-features are valued, the particle by is merged. By is the goal of the probe of the matrix verb responsible for selection and associated with evaluation. Moreover, Russian patterns with Portuguese by displaying obviation (see (4) above) which is derived accordingly.
Consider now Balkan languages such as Greek and Bulgarian. At first glance, they seem to pattern with Russian: a particle, na and da respectively, is inserted in subjunctive clauses. However, this similarity is only apparent. Balkan languages diverge from Russian in the following aspects:
(i) the subjunctive particles are obligatorily verb-adjacent (9); (ii) the tense of the subjunctive verb is Present Indicative (perfective non-past (PNP) in [Giannakidou, 2009]) (10); (iii) the obligatory obviation effects are voided (11a-b). Observe the following data from Bulgarian:
(9) Bulgarian
Iskam Ivan da (*Ivan) want.1sG John subj John 'I want John to buy the book.' (10) Bulgarian
Ivan da kupuvase
*Iskax wanted.lsG knigata. the-book
kupi
buy.PNP.3sG
/ bese
knigata. the-book
kupil
John subj buy.PAST.3sG / have.PAST.3sG bought
(11) Bulgarian
a. Ivan iska Peter da otide
John want.3sG Peter subj go.PNP.3sG
'John wants Peter to go to the movies.'
b. Ivan iska da otide
John want.2sG subj go.PNP.3sG 'John wants to go to the movies.'
na to
na kino. to movies
kino. movies
The structural position of Balkan subjunctive particles is a challenging matter. On the one hand, they behave like complementizers [Roussou, 2000], on the other, the strict verb-adjacency and the lack of obviation suggest that they are generated in a lower position, possibly MoodP as in Giannakidou (2009). Dobrovie-Sorin (1994, 2001) argues that Romanian subjunctive particle sг is part of the verbal morphology incorporating into the verbal cluster. This claim is in line with Giannakidou's (2009) observations regarding the defectiveness of tense compatible with na: PNP cannot occur on its own, therefore, na is merged in order to introduce the variable n (now) into the syntax. The dependency between na and the verb suggests that it is responsible for establishing a relation with given features of the matrix domain.
Here we will tentatively assume that Balkan subjunctive particles have a dual nature: they are generated in a lower position (within the TP field or below FinP) but they also establish a relation with the CP (triggered by given features of EvaluativeP). Therefore, the lack of obviation effects is derived according to the position in which the subjunctive particle
is interpreted: when the particle is in the TP zone, the domain of the embedded subject is transparent for binding which gives rise to the absence of obviation effects. Notice that Bulgarian da-structures in (11a-b) draw an ambiguity: they denote subjunctive (11a) but, in addition, they occur in contexts in which an infinitive would be selected in Romance languages (11b).6
Romanian is particularly revealing here. As shown in [Dobrovie-Sorin, 2001, p. 60] obligatory control structures are only possible when the complementizer ca does not intervene:
ajute pe help-suBj pe
(12) Romanian
a. Ion incepe s -o John starts sa her-cL.Acc 'John starts helping Mary.'
b. *Ion incepe ca pe Maria John begins that pe Mary sa her-cL.Acc help-3sG.suBj
Notice that the complementizer ca will be otherwise obligatory whenever a constituent such as pe Maria in (12b) precedes the sa-clause [Dobrovie-Sorin, 2001]. With obligatory control verbs, however, it blocks the relation between the matrix clause and the sa-clause.
Maria. Mary
ajute.
4.1. Some asymmetries in main subjunctive clauses
Coming back to subjunctive main clauses, we are now able to explain some intriguing asymmetries between the languages under investigation.
In section 3 we observed Portuguese main subjunctive clauses codify the speaker's evaluation of the state of affairs described. The evaluative reading confined to these structures seems to hold also for Greek (13), Russian (14) and Bulgarian (15). Giannakidou (2016) claims that the occurrence of Greek na in (13) is associated with epistemic modality:
(13) Greek
Ti na theli?
what subj want-3sG
'What might he want?' [Giannakidou, 2016, p. 183.]
6 Differently from Russian, Greek and Bulgarian have lost the verbal morphology for subjunctive and infinitive.
(14) Russian
Чтобы я такое сказал.
Ctoby ja takoe skazal!
that. subj I such said
'That I would say such thing!' [Baylin, 2012, р. 89.]
(15) Bulgarian
Kakvo da kupija? what subj buy.3sG
'What should I buy?'
A closer look at the data however suggests the languages under study display some restrictions regarding subjunctive main clauses. The Greek and Bulgarian examples in, respectively, (13) and (15), show subjunctive is felicitous in wh-questions. In Portuguese such structures are ruled out:
(16) Portuguese
*O que compre? what buy.subj
'What should I buy?'
Mezhevich discusses the occurrence of subjunctive in Russian main clauses [Mezhevich, 2006, р. 132-134]. In her terms by is a conditional particle:
(17) Russian
Я бы пропустила этот доклад. Ja by propusti-la etot doklad. I cond miss-PAST this talk 'I would skip / would have skipped this talk.'
(18) Что бы ты пропустила? Cto by ty propusti-la? what cond you-NOM miss-PAST 'What (e.g., which talk) would you skip?
However the data in (17) and (18) are distinct from (14) above. While in the former by functions as a marker of the so-called conditional mood,7
7 Counterparts of Russian by can be found in the other Slavic languages as a hallmark of conditional mood.
in the latter the occurrence of the complex ctoby underlies the evaluative reading of subjunctive main clauses. Our suspicion that by (a residue of the old aorist) functions as marker of modality in the examples in (17) and (18) is in conformity with the observations of Migdalski (2006) on Polish by. What is more, ctoby, parallel to Portuguese subjunctive, is infelicitous in wh-questions:
(19) Russian
*Что чтобы ты пропустила?
*Cto ctoby ty propusti-la?
what subj you-NOM miss-PAST
In the light of the data discussed so far, we can then distinguish between two groups of languages: (i) Greek and Bulgarian which permit subjunctive in questions and (ii) Russian and Portuguese which do not.
Let us start by (i). As discussed above, the subjunctive particles da and na appear to display an ambiguous behaviour being counterparts of the Romance subjunctive and infinitival structures. In the examples in (13) and (15) above, both da and na-clauses refer to future, unrealized event patterning with Stow-ell's (1982) observations on control infinitives. Moreover, the Bulgarian example in (15) can express both deontic modality and rhetorical or evaluative-like reading. The latter is not however a result of the properties of da but rather an outcome of the features of given projections in the Left Periphery activated by wh-movement.
Crucially, it seems to us that evaluation and true questions are generally incompatible. Evaluation, in the sense adopted here, implies knowledge or presupposition which are being object of the evaluation itself. Hence, a wh-question, being a request for information, seems to be incompatible with such a definition of evaluation. Interestingly, the Russian example in (14) consistently supports the assumption that evaluation implies knowledge. Observe that the evaluation reading confined to (14) is obtained by the combination of both cto and by and not by by itself. Cto is the indicative complementizer. According to Ambar (2016) it heads AssertiveP, the projection of indicative mood, codifying 'what the speaker knows' [Ambar, 2003]. The derivation of the subjunctive mood therefore involves raising of cto to by, the latter having been merged in Evaluativee [Ambar, 2016].
Following the claim that evaluation and questions are incompatible, we observe that in Portuguese and Russian wh-subjunctive questions are ruled out. Mezhevich's (2006) example in (18) is particularly important here: in structures in which by is merged in a lower position (Moode according to Migdalski (2006) on Polish by) nothing goes wrong for the interrogative interpretation. In (19) with ctoby, on the other hand, wh-movement is blocked.
As for Portuguese, besides the general incompatibility between neutral questions8 and evaluation, wh-subjunctive questions are infelicitous due to the unvalued tt-features whose valuation by the Op in EvaluativeP is blocked by the intervening wh-word.
5. Conclusions
Pursuing the twofold goal of this paper we: (i) adopted the system for tense-features valuation proposed in Ambar [2007, 2016] which captures both the dependency and the independency analyses of tense; and (ii) observed that main clauses denote evaluation (which has been extended to all occurrences of subjunctive mood) suggesting that the trigger for subjunctive is an Op merged in EvaluativeP. Arguing that the properties of subjunctive (namely evaluation) are incompatible with those of true questions, we observed that the apparent plausibility of subjunctive wh-questions in Greek and Bulgarian is actually a result from a consistent ambiguity between subjunctive and infinitive.
References
Ambar, 1988 - Ambar M. Para uma sintaxe da inversro sujeito-verbo em Portu-guKs. Ph.D. dis. University of Lisbon, 1988.
Ambar, 1996 - Ambar M. Infinitives vs Participles. Paper presented at LSRL Linguistic Symposium on Romance Languages XXVI. 1996. University of Mexico, City of Mexico.
Ambar, 1998 - Ambar M. Inflected infinitives revisited. Genericity and single event. Journal of Canadian Linguistics. 2000. Vol 43. Pp. 5-36.
Ambar, 2000 - Ambar M. Wh-questions and wh-exclamatives: Unifying mirror effects. Romance Languages and Linguistic Theory. C. Beyssade, R. Bok-Bennema, F. Drijkoningen, P. Monachesi (eds.). Amsterdam, 2000. Pp. 15-40.
Ambar, 2003 - Ambar M. Wh-asymmetries. Asymmetry in Grammar. A.M. Di Sci-ullo (ed.). Amsterdam, 2003. Pp. 209-259.
Ambar, 2005 - Ambar M. V-to-C, tense domains and conditions on merge and move. Paper presented at GLOW in Asia V Abstracts. Jawaharlal Nehru University, New Delhi, India, 2005.
Ambar, 2007 - Ambar M. Verb movement and tense. EPP and t-completeness. Proceedings of the XXXII Incontro di Grammatica Generativa. C. Picchi, A. Pona (eds.). Firenze, 2007. Pp. 1-20.
Ambar, 2016 - Ambar M. On Finiteness and the Left Periphery: Focusing on Subjunctive. Mood, Aspect, Modality Revisited: New Answers to Old Questions. J. Blaszack, D. Klimek-Jankowska, K. Mygdalski, A. Giannakidou (eds.). Chicago, 2016. Pp. 125-176.
8 As previously shown in the literature ([Ambar, 2000, 2003], a.o.) wh-questions can acquire rhetorical or echo-readings which result from wh-movement to the Left Periphery.
Boskovic, 1997 - Boskovic, Z. The Syntax of Nonfinite Complementation: An Economy Approach. Cambridge, MA, 1997.
Chomsky, 2001 - Chomsky N. Derivation by phase. Ken Hale: A Life in Language. M.J. Kenstowicz (ed.). Cambridge, MA, 2001. Pp. 1-52.
Dobrovie-Sorin, 1994 - Dobrovie-Sorin C. The Syntax of Romanian. Berlin, 1994.
Dobrovie-Sorin, 2001 - Dobrovie-Sorin C. Head-to-Head merge in Balkan subjunctives and locality. Comparative Syntax of Balkan languages. M.L. Rivero, A. Ralli (eds.). New York, 2001. Pp. 44-73.
Duarte et al., 2005 - Duarte I., Gonjalves A., Miguel M. Propriedades de C em frases completivas. Actas doXXEncontro da APL. Lisboa, 2005. Pp. 549-562.
Enj, 1991 - Enj M. On the absence of the present tense morpheme in English. University of Wisconsin, 1991.
Farkas, 1992 - Farkas D. On the semantics of subjunctive complements. Romance Languages and Modern Linguistic Theory. P. Hirschbuhler, K. Koerner (eds.). Current Issues in Linguistic Theory. Amsterdam, 1992. Pp. 71-104.
Giannakidou, 1998 - Giannakidou A. Polarity Sensitivity as (Non)veridical Dependency. Amsterdam, 1998.
Giannakidou, 2009 - Giannakidou A. The dependency of the subjunctive revisited: Temporal semantics and polarity. Lingua. 2009. Vol. 120. Pp. 1883-1908.
Giannakidou, 2016 - Giannakidou A. Evaluative subjunctive and nonveridicali-ty. Mood, Aspect, Modality revisited: New Answers to Old Questions. J. Blaszack, D. Klimek-Jankowska, K. Mygdalski, A. Giannakidou (eds.). Chicago, 2016. Pp. 177-217.
Giorgi, Pianesi, 1997 - Giorgi A., Pianesi F. Tense and Aspect: From Semantics to Morphosyntax. Oxford, 1997.
Kempchinsky, 2009 - Kempchinsky P. What can the subjunctive disjoint reference effect tell us about the subjunctive? Lingua. 2009. Vol. 119. Pp. 1788-1810.
Marques, 2009 - Marques R. On the selection of mood in complement clauses. Crosslinguistic Semantics of Tense, Aspect and Modality. L. Hogeweg, H. de Hoop, A. Malchukov (eds.). Amsterdam, 2009. Pp. 179-204.
Martin, 1996 - Martin A.R. A minimalist theory of PRO and control. PhD dis. University of Connecticut, 1996.
Martins, 2001 - Martins A.M. On the Origin of the Portuguese Inflected Infinitive: A new perspective on an enduring debate. Historical Linguistics. L.J. Brinton (ed.). Amsterdam, Philadelphia, 2001. Pp. 207-222.
Mezhevich, 2006 - Mezhevich I. Featuring Russian tense: A feature-theoretic account of the Russian tense system. Ph.D. dis. University of Calgary, 2006.
Migdalski, 2006 - Migdalski K. The Syntax of Compound Tenses in Slavic. Utrecht, 2006.
Oliveira, 2008 - Sobre os tempos do conjuntivo. O FascHnio da Linguagem -Actas do Colyquio de Homenagem a Fernanda Irene Fonseca. F. Oliveira, I.M. Duarte (eds.). Porto, 2008.
Pesetsky, Torrego, 2001 - Pesetsky D., Torrego E. T-t o-C movement: Causes and consequences. Ken Hale: A Life in Language. M. Kenstowicz (ed.). Cambridge, MA, 2001. Pp. 355-426.
Pesetsky, Torrego, 2001 - Pesetsky D., Torrego E. The syntax of valuation and the interpretability of features. Manuscript, University of Massachusetts-Boston, 2001.
Picallo, 1984 - Picallo C. The infl node and the null subject parameter. Linguistic Inquiry. 1984. Vol. 15. Pp. 75-102.
Quer, 2006 - Quer J. Subjunctives. The Blackwell Companion to Syntax. M. Eve-raert, H. Riemsdijk (eds.). Oxford, 2006. Pp. 660-684.
Raposo, 1985 - Raposo E.P. Some asymmetries in the binding theory in Romance. Linguistic Review. 1985. Vol. 5. Pp. 75-109.
Roussou, 2000 - Roussou A. On the left periphery; modal particles and complementizers. Journal of Greek Linguistics. 2000. Vol. 1. Pp. 63-93.
Stowell, 1982 - Stowell T. The tense of infinitives. Linguistic Inquiry. 1982.
Статья поступила в редакцию 03.09.2017 The article was received on 03.09.2017
Амбар Мануэла - Ph.D. (лингвистика); профессор кафедры общего языкознания, Лиссабонский университет, Португалия
Ambar Manuela - Ph.D. in Linguistics; Professor at the Department of General Linguistics, University of Lisbon, Lisbon, Portugal
E-mail: manuela.ambar.flul@gmail.com
Димитрова Маргарита - соискатель кафедры общего языкознания, Лиссабонский университет, Португалия
Dimitrova Margarita - Ph.D. candidate at the Department of General Linguistics, University of Lisbon, Lisbon, Portugal
E-mail: dimmargarita@gmail.com
Амарал Диана - соискатель кафедры общего языкознания, Лиссабонский университет, Португалия
Amaral Diana - Ph.D. candidate at the Department of General Linguistics, University of Lisbon, Lisbon, Portugal
E-mail: diana.flul@gmail.com
