OANA SĂVESCU1
Abstract. This paper provides an account for the grammaticality of multiple sluices in Romanian, which violate Superiority. It is shown that such cases arise if the order of the indefinite arguments in the antecedent clause parallels the order of the wh-phrases in the sluice. The paper argues that what accounts for the grammaticality of these sentences is a felicity condition which requires an identical partitioning of the set the leftmost indefinite and the leftmost wh-phrase range over. In the case in which an
‘apparently’ non D-linked wh-object in the sluice precedes a subject-wh phrase, this felicity requirement can only be attainted if the object indefinite in the antecedent of the sluice is topicalized over the subject indefinite to a position TopP, marked + distributive, which otherwise hosts the leftmost wh-phrase in a multiple constituent question.
Keywords: multiple sluicing, Parallelism, wh-movement, Romanian.
1. INTRODUCTION
Romanian multiple questions are similar to their counterparts in (Slavic) Balkan languages in that all the wh-phrases need to be fronted (with the exception of echo questions)2. This phenomenon also existed in Old Romanian, as shown by Stan (2012) and Gheorghe (2013). When several multiple wh-elements front, the order of the moved wh- phrases obeys (Anti) Superiority, with the subject wh phrase being the first one to front (1-4), again, a state of affairs that dates back to Old Romanian (Gheorghe 2013); if more than two wh-phrases are involved, then the order between the 2nd and 3rd argument is relaxed, as long as they follow the subject (5-6):
(1) Cine pe cine a lovit?
who PE who has hit
‘Who hit who(m)?’
(2) Cine cui a dat flori?
who who.dat has given flowers
‘Who gave flowers to whom?’
1 University of Bucharest, [email protected].
2 To be more precise, Romanian is actually closely similar to Bulgarian (cf Rudin 1988, Bošković 2002, Alboiu 2002) in that, unlike in Serbo-Croatian, for instance, nothing can intervene between the wh-phrases. This has been taken by Rudin (1988) to suggest that in both languages, the wh- phrases target one scopal position, which is SpecCP, but see below for arguments drawing from Alboiu (2002) that this position cannot be maintained. See also Vasilescu (2002) for a detailed description of Romanian multiple interrogatives.
RRL, LIX, 2, p. 179–188, Bucureşti, 2014
(3) *Pe cine cine a lovit?
PE who who has hit
‘Who hit whom?’
(4) *Cui cine a dat flori?
who.dat who has given flowers
‘Who gave flowers to whom?’
(5) Cine ce cui a dat?
who what who.dat has given
‘Who gave what to whom?’
(6) Cine cui ce a dat?
who who.dat gave has given
‘Who have what to whom?’
With multiple sluicing, similar Superiority effects can be observed: (7) and (9) below, in which the wh-sluices appear in the subject>object order, are perfectly grammatical, while (8) and (10), in which the subject wh-phrase follows an object wh- constituent are considerably worse for most speakers, if not ungrammatical:
(7) Au primit câţiva studenţi nişte lucrări, dar nu-mi amintesc cine ce.
have received some students some papers but not me remember who what
‘Some students got back some papers ,but I don’t remember who got back what’
(8) */? Au primit câţiva studenţi nişte lucrări, dar nu-mi amintesc ce cine.
have received some students some papers but not me remember what who
‘Some students got back some papers ,but I don’t remember who got back what’
(9) Cineva a promis fericire cuiva, dar nu ştiu cine cui.
somebody has promised happiness sombody.dat but not know who whom
‘Somebody promised happiness to somebody, but I don’t know who to whom’
(10) */?Cineva a promis fericire cuiva, dar nu ştiu cui cine.
somebody has promised happiness somebody.dat but not know whom who
‘Somebody promised happiness to somebody, but I don’t know who to whom’
Interestingly, however, Superiority with multiple sluicing can be violated in Romanian when the object indefinite in the antecedent clause is topicalized and thus precedes the subject indefinite. (11) below is perfectly acceptable:
(11) Nişte lucrări au primit câţiva studenţi, dar nu-mi amintesc ce cine.
some papers have received some students but not me remember what who
‘Some students got back some papers, but I don’t remember who got back what’
In this paper I will provide an account of what exactly accounts for the grammaticality of the Superiority violations in examples with multiple sluices like (11) above. My point of departure, which will constitute the basis for my analysis, will be a closer investigation of the behavior of multiple wh-questions, and I will suggest that wh- phrases target different functional projections in an “exploded” CP, along the lines of Rizzi (1997) (section 2). I thus depart from previous accounts in the literature on multiple wh phrases in Romanian (Alboiu 2002, Cornilescu 2000), which have the wh-phrases target
either Spec IP or the highest functional projection available (SpecMoodP). The analysis of multiple sluices which do not obey superiority will rely on some (modified) version of Parallelism, which requires the surface order of the arguments in the antecedent be identical with the order of the wh-phrases in the sluice (section 3).
2. ON (MULTIPLE) WH-QUESTIONS
Rudin 1988 (in Alboiu 2002, Bošković 2002) argues that Romanian, just like Bulgarian, is a CP-absorbtion language, in that wh-phrases in multiple questions all front to SpecCP. I follow Alboiu (2002) and argue that the position that multiple wh-phrases target cannot be SpecCP.
This can be immediately seen by examining the interaction of wh-phrases with topics in Romanian3.
As far as topicalization is concerned, there is no difference between the behavior of topics in matrix and in embedded clauses (in the examples below topics are underlined):
(12) Ion mâine va vorbi despre sluicing4. John tomorrow will speak about sluicing
‘John will speak about sluicing tomorrow’
(13) Am uitat că Ion mâine va vorbi despre sluicing.
have.I forgotten that John tomorrow will speak about sluicing
‘I forget that John will speak about sluicing tomorrow’
When topics interact with wh-phrases, the former obligatorily precede the latter:
(14) Mâine cine ce (*mâine) va face?
tomorrow who what (tomorrow) will do
‘Who will do what tomorrow?’
(15) La film Ion cu cine s- a dus?
at movie John with who when se has gone
‘To the movie, who did John go with?’
Given that topics are situated in a position above C0, as can be seen in (13) above, it is safe to conclude that wh-phrases cannot target SpecCP, but a position lower than C0.
The interaction between wh-phrases and focus phrases in Romanian suggests that the both types of constituents compete for the same position. This can be shown by the fact that (i) topics precede non-wh-focused constituents (16); and (ii) wh-phrases and non-wh focused phrases cannot co-occur (17):
3 I refer the reader to Alboiu (2002) for more detailed arguments against Rudin (1988), on the grounds of interactions of wh-phrases with topics and focus, lack of weak cross over effect with wh-movement, wh-islands and interacting wh-dependencies, etc.
4 Romanian is a VSO language, so whenever the subject appears preverbally, it is topicalized (Alboiu 2002, Motapanyane 1995, Cornilescu 2000)
(16) (*SCRISORI) mâine SCRISORI voi trimite.
letters tomorrow letters will.I send
‘It is letters that I will send tomorrow’
(17) (*SCRISORI) cine (*SCRISORI) va trimite (scrisori)?
letters who letters will send letters
I thus suggest, adopting Rizzi’s (1997) Split CP hypothesis that single wh-phrases target FocP in Romanian. This projection is a syncretic category, which has both a selectional +wh feature, and a +focus feature.
The question that arises at this point is what happens when two wh-phrases are fronted. Two possibilities arise: (i) both wh-phrases target SpecFocP, with Superiority being maintained by having the paths of the moved wh-phrases undergo crossing dependencies in a tucking in fashion à la Richards (1999) or (ii) one wh-phrase targets FocP and the other targets TopP. In what follows, I will argue for the latter approach.
2.1. Superiority Revisited
Comorovschi (1996) convincingly shows that “Superiority” effects in Romanian actually reduce to a prohibition against a non-D-linked wh-phrase being fronted over a wh-subject. While (18) below is ungrammatical, (19), in which a D-linked wh-phrase (care) precedes a wh-subject is perfectly fine:
(18) *Ce cine ţi- a dat ? what who you.dat has given
‘Who gave what to you?’
(19) Cu care fată cine a dansat?
with whoD-linked girl who has danced
‘Who danced with which girl?’
(adapted from Comorovschi 1996, ex. 114, p. 152)
This prohibition is to be understood against a more general condition that requires the leftmost wh-phrase in a pair-list (matching) question be D-linked and range over all the relevant members of a given, salient set. (19), for instance, can only be felicitously uttered in a context in which there is a contextually salient group of girls at a party and for each girl in that group there is somebody who danced with her. (19) cannot, however, be asked if we know that only some of the girls in the group actually danced.
This ‘universal’-like flavor of the left-most D-linked wh-phrase can be better seen if we examine the set of possible answers to a matching/pair-list question. Comorovschi (1996) observes that besides a pair-list interpretation, multiple constituent questions allow for a functional answer as well. She relates the question answer pairs in (20) with the ones in (21) (Comorovschi’s examples 95 and 96 respectively, p 51):
(20) Which student got back which paper?
A2: Every student got back his syntax paper
A1: John got back his paper on Romanian NP, Ken got back his paper on word order in Latin, and Ed got back his paper on Portuguese infinitives.
(21) Which book did every author recommend?
A1: Tom recommended book A and Jerry recommended book B A2 : (Every author) recommended his latest book.
Following previous work by Engdahl (1986) and Chierchia (1991, 1993), Comorovschi (1996) notes that the type of answer in (21) A1, consisting of a list of ordered pairs, is just a special case of the functional answer in A2. Functions can be specified din two ways: (i) by describing them, as in A2, or by their graph, i.e listing of all the argument value pairs, as in A1. Since answers in (20) are parallel to answers in (21), multiple questions like the ones in (20) appear to be interpreted as parallel with questions with universal quantifiers like (21). (22) below is the representation of the formalization of questions like (21) (quantifiers) and (23) represents the multiple question in (20) (both are built on Kartunnen’s (1977) semantics of questions):
(22) λ p ∃ f [∀x [book (f(x))]] & ∪p &p = ^ ∀x [author (x) → recommend (x, f(x)]
(23) λ p ∃x∃f [ student (x) & ∀x [paper (f(x))] & ∪p &p = ^ get back (x, f(x))], where f is a variable over functions from individuals to individuals.
Thus, the denotation of the multiple question in (20) is the set of true propositions of the form ‘x got back f(x)’ where x is a student and for all x, f(x) is one of x’s papers. Given that the referent of the answer to the rightmost interrogative phrase depends on the referent of the answer to the left-most, sentence initial one, the rightmost wh-phrase is interpreted as a function. Its domain is the set denoted by the 1st wh-constituent and the range is given by the set denoted by the dependent phrase.
Similarly, turning back to Romanian multiple constituent questions, (19) above will be interpreted on a par with (24), with a strong distributive quantifier fiecare ‘every/each’5: (24) Cu fiecare fată cine a dansat?
with every/each girl who has danced
‘Who danced with every girl?’
I thus suggest that the D-linked left most wh-phrase in (19), and in a multiple constituent question in general, targets a topic position immediately above FocP which is specified for (universal) distributivity and topichood (D-linking).
The current proposal is similar in spirit with the analysis of quantifier scope interactions in Beghelli (1997) and Beghelli and Stowell (1997), in which the distributor always needs to c-command the distributing share. Note that an alternative account, which would have the two wh-questions target the specifier of the same projection (Spec FocP) cannot straightforwardly account for the interpretational differences between the two wh- phrases; moreover, the required a-symmetric c-command (wide scope) relation between the distributor and the distributee wh-phrase cannot be met, since the two wh-phrases will c-command each other.6
Having shown that Superiority violations reduce to a requirement that the leftmost wh-phrase that crosses a subject be D-linked and interpreted distributively, we still have to account for multiple constituent questions in which nothing crosses the subject. Consider (25):
5 The terminology is borrowed from Szabolcsi (1996) in Beghelli (1997) and Beghelli and Stowell (1997).
6 This would also blatantly violate Antisymmetry, cf Kayne (1994).
(25) Cine ce ţi- a trimis?
wh what you.dat has sent
‘Who sent you what?’
We want to maintain the proposal that the two wh-phrases target two different functional projections, but the subject- wh-phrase in (25) has the form of a non-D-linked wh-phrase (cine ‘who’ vs care ‘which’), which may run against the hypothesis that it targets a position specified for D-linking.
Note, however, that the problem is only apparent. (25) cannot be felicitously asked in a situation in which no salient set of senders is available. A conjoined question would have to be asked instead:
(26) Cine si ce ţi- a trimis ? who and waht you.dat has sent
“Who did you receive and from who”
If, however, it is clear from the context of utterance who the potential senders are (say, my office mates), then (25) is a possible question. It thus follows that even though the wh-subject in a multiple constituent question may (morphologically) appear to be non D- linked, it is actually interpreted as D-linked, on a par with D-linked wh-objects that cross wh-subjects in ‘violation’ of Superiority. As such, it is reasonable to assume that the subject wh phrase also targets the SpecTop position above the FocP.
Comorovschi (1996) correctly notes that it is sometimes possible to ask a question like (25) and interpret the subject as D-linked even if no context of utterance is provided.
She suggests that in declarative sentences subjects generally carry old information, so they are more likely to be interpreted as topics than objects are. On the other hand, a constituent that answers the question is obligatorily focused. This, she argues, cannot preclude its functioning as a contrastive topic, so the answer to a question with a D-liked subject will pick an entity that is discourse old. Thus, a wh-subject of question will be more likely than a non subject to be interpreted as D-linked in the absence of a context, due to the high occurrence of subject topics in declaratives.
I remain uncommitted at this point as to whether the TopP position immediately above FocP is indeed a contrastive topic position. Based on our discussion above, I do, however, maintain the view that any leftmost wh-phrase in a multiple constituent question will target the same TopP position, which is specified for (universal) distributivity and which c-commands the position of the rightmost interrogative phrase. This will correctly capture the fact that this D-linked wh phrase has a ‘universal’ reading: all members of the set introduced by the denotation of the D-linked phrase will distribute over the members of the set denoted by the dependent, rightmost wh-constituent in SpecFocP.
This is potentially confirmed by a related set of data from Old Romanian, as noted by Gheorghe (to appear). Specifically, in certain contexts, it is possible to front multiple wh relative pronouns, with the relative/interrogative pronouns cine and care being interpreted as the universal quantifier fiecare/fiecine.
(27 ) a. Că de cuvinte are a întreba Domnulŭ [..], cine cum au lucratŭ that of words has to ask God who how have.3.pl worked
‘Because he will ask God how everybody worked’
(Coresi, Carte cu învăţătură, 539)
b. au luat drumul cel mare spre Scânteia, […] care încotro au putut”
have.3.pl taken road the big towards Scânteia who wherever could ‘They took the big road towards Scânteia, each of them wherever could go7.
(Costin, Letopisețul, 193) Before concluding this section, let us see how we can derive in our system the derivation of (25), repeated below as (28):
(28) Cine ce ţi- a trimis ? wh what you.dat has sent
‘Who sent you what?’
I have already suggested above that the FocP is a syncretic category, which has selectional +focus and + wh features. Following Alboiu (2002), I argue that wh feature is strong (selectional) on the wh –phrases themselves, which need to move to SpecFocP to get their wh-feature checked in a local, Spec-Head configuration. The system works as follows:
(i) first, the FocP, being selectional, attracts the closest wh phrase, the subject, which can thus check focus and wh-features; the wh-subject will then be attracted to SpecTopP in virtue of its being specified for D-liking and distributivity
(ii) next, the 2nd wh phrase, having a strong wh feature, undergoes movement to SpecFocP 8
3. BACK TO SLUICING
Having suggested an analysis of multiple wh-fronting in Romanian constituent questions, we are now in the position to account for the sluicing facts outlined in section 1 above.
Recall our examples in (8) and (11), which I repeat as (29) and (30):
(29) */? Au primit câţiva studenţi nişte lucrări, dar nu-mi amintesc ce cine.
have received some students some papers but not me remember what who ‘Some students got back some papers, but I don’t remember who got back what’’
(30) Nişte lucrări au primit câţiva studenţi, dar nu-mi amintesc ce cine.
some papers have received some students but not me remember what who ‘Some students got back some papers, but I don’t remember who got back what’
(29), with a non-D-linked wh-object preceding the wh-subject in the sluice in violation of Superiority is considered ungrammatical by most of my informants. In contrast, (30) is grammatical. The crucial difference between the two sentences is that in the antecedent of the sluice in (30), the direct object indefinite nişte lucrări ‘some papers’ has
7 In Old Romanian only the phrase care încotro has been preserved. See also Stan (2012).
8 When the multiple constituent question involves a D-linked object-wh-phrase preceding the subject, the derivation may involve an extra preliminary step which has the object first scramble over the subject; next, the object will move to SpecFocP, and the derivation will proceed as above.
undergone topicalization. Thus, in (30) the order of the object and subject indefinite matches the order of the object and subject wh-phrases in the sluice.
Recall from our discussion in section 2 above that in a grammatical or felicitous multiple constituent question it has to be the case that the leftmost wh-phrase be D-linked;
moreover, it is interpreted as a distributor, with all the members of the set it denotes being distributed over the members of the set denoted by the dependent (share)-the 2nd wh-phrase.
Now, in the sluices of both (29) and (30), the leftmost wh-phrase ce (what) is in its non-D- linked form9. The question is then why is the sluice in (30) ok, while the one in (29) is bad?
It seems to be the case that what accounts for the grammaticality of (30), with a superiority violation, as opposed to (29), is a certain condition that requires (some form of) parallelism between the order of the arguments in the antecedent and the order of the wh- phrases in the sluice. I argue that it is precisely the topicalization of the object in the antecedent clause that allows for a legitimate sluice “violating” superiority in (30).
The interpretation of the antecedent in (30) is that there is a contextually salient set of papers x, such that for every paper x in that set, there is a student y in the class such that that student got back that paper. In other words, all the members of the set denoted by the indefinite nişte lucrări ‘some papers’ are distributed over the set denoted by the subject indefinite câţiva studenţi ‘some/few students’. The fact that the sluice is grammatical, suggests that the antecedent creates the correct configuration which allows the leftmost wh- phrase ce ‘what’ to acquire a D-linked interpretation. Note that as the result of the topicalization of the direct object in the antecedent, the participants in the conversation partition the universe of discourse in such a way that the set denoted by the direct object indefinite is interpreted as contextually given. The fact that the this set is interpreted as a distributor over the set denoted by the subject indefinite suggests that the position the direct object targets as a result of topicalization is the same TopP position that otherwise hosts D- linked phrases in multiple wh-constituent questions.
In the antecedent of (29), on the other hand, the distributor is the subject indefinite;
as such, no context is created that would allow the object wh-phrase in the sluice be interpreted as D-linked, and the sentence is ungrammatical.
The discussion above regarding the possibility of an indefinite phrase to be interpreted as distributive recalls a proposal in Beghelli (1997) and Beghelli and Stowell (1997) who note that that certain indefinites, which Szabolcsi (1997) calls Group Denoting Quantifier Phrases (GQP) are weak distributors, or pseudo distributors, which only act as distributors in certain scopal configurations. As such, they are not inherently marked as +distributive, like strong distributive quantifiers such as every or each. Distributivity with these GQP arises only in certain scopal configurations, when the GQP c-commands another QP at Spell Out.
They also argue that objects GQP cannot distribute over subjects, in other words, GQPs which are c-commanded by a subject GQP cannot take inverse scope. Indeed, in (28) above, the only interpretation available is with the subject being interpreted as the distributor. Note, however, that the data which Beghelli (1997) and Beghelli and Stowel;
(1997) discuss involve configurations in which the direct object cannot distribute over the subject when the latter c-commands the former at Spell Out.
9 The D-linked counterpart of ce would be pe care.
In the Romanian example in (30), however, the topicalization of the direct object to a position which c-commands the subject at Spell Out creates the correct scopal configuration which would allow the object GQP to be interpreted as a distributor.
To conclude, successful sluicing with a Superiority violation can only take place in Romanian if the leftmost arguments in the two conjuncts have the same syntactic function.
This translates into the requirement that, in the case discussed here, the direct object be a distributor and hence c-command the distributee in both the antecedent and in the sluice.
This is possible if the participants in the conversation are able to partition identically the set that the leftmost indefinite in the antecedent and the leftmost wh-phrase in the sluice range over10.
4. CONCLUSION
In this paper I have provided an account of grammatical multiple wh-sluices which violate Superiority in certain configurations in Romanian.
We have seen that Superiority can be violated if the order of the indefinite arguments in the antecedent clause parallels the order of the wh-phrases in the sluice. I have argued that what accounts for the grammaticality of these sentences is a felicity condition which requires an identical partitioning of the set the leftmost indefinite and the leftmost wh- phrase range over. In the case in which an ‘apparently’ non D-linked wh-object in the sluice precedes a subject-wh phrase, this felicity requirement can only be attainted if the object indefinite in the antecedent of the sluice is topicalized over the subject indefinite to a position TopP, marked + distributive which otherwise hosts the leftmost wh-phrase in a multiple constituent question.
If the current proposal is on the right track, then one prediction that can be made is that the possibility of an ‘apparent’ non –D-linked wh-object to cross a subject-wh-phrase is not limited to sluicing configurations. In other words, since what allows the leftmost wh- phrase in the sluice to be interpreted as being D-linked and distributive is an identical partitioning of the set the leftmost indefinite in the antecedent and the highest wh-phrase in the sluice range over, then superiority violations are not ‘repaired’ by the deletion of the TP in any way. As such, it should be possible to obtain a grammatical sentence if the TP in (30) above is not deleted. This prediction is borne out, as the grammaticality of (31) shows:
(31) Nişte lucrări au primit câţiva studenţi, dar nu-mi amintesc ce cine a primit.
some papers received some students but not me remember what who has received
‘Some students got back some papers, but I don’t remember who got back what’
SOURCES
Coresi, Carte cu învăţătură = Coresi, Evanghelie cu învăţătură, Braşov, 1581; ed. S. Puşcariu, Al. Procopovici: Diaconul Coresi, Carte cu învăţătură, Bucureşti, 1914.
Costin, Letopisețul = Miron Costin, Opere, ed. P.P. Panaitescu, Editura de Stat pentru Literatură și Artă, București, 1958.
10 Note that this felicity requirement ensures that at LF, the antecedent and the deleted TP in the sluice will mutually entail each other, which recalls Merchant’s e-givenness condition on sluicing.
I am grateful to Mark Baltin (pc) for pointing this out to me.
REFERENCES
Alboiu, G., 2002, “The Features of Movement in Romanian”, București, Editura Universității din București.
Beghelli, F., 1997, “Distributivity and Pair List Readings”, in: A. Szabolcsi (ed.), Ways of Scope Taking, Dordrecht, Kluwer, 71−98.
Beghelli, F., T. Stowell, 1997, “Distributivity and Negation: The Syntax of Each and Every”, in:
A. Szabolcsi (ed.), Ways of Scope Taking, Dordrecht, Kluwer, 349−404.
Bošković, Z., 2002, “On Multiple Wh-Fronting”, Linguistic Inquiry, 33, 351−383.
Chierchia, G., 1991, “Functional WH and weak Crossover”, Proceedings of WCCFL 10, CSLI, Stanford, 75−90.
Chierchia, G., 1993, “Questions with Quantifiers”, Natural Language Semantics, 1, 181−234.
Comorovschi, I., 1996, Interrogative Phrases and the Syntax Semantics Interface. Dordrecht: Kluwer Academic Publishers.
Cornilescu, A., 2000, “Direct Object Movement and the Structure of The Romanian Left Periphery”, ms, University of Bucharest.
Engdahl, E. 1986, Constituent Questions, Reidel, Dordrecht.
Gheorghe, M., 2013, “Parametrul Deplasării multiple a elementului Wh (interogativ relativ).
Observații privind structurile cu grup interogativ/relativ în limba română veche”, Limba română, LXII, 2, 240−246.
Kartunnen, L., 1977, “Syntax and Seamntics of Questions”, Linguistics and Philosophy, 1, 3−44.
Kayne, R., 1994, The Antisymmetry of Syntax, Cambridge, Mass., MIT Press.
Motapanyane, V., 1995., Theoretical Implications of Complementation in Romanian. Padova:
Unipress (Rivista di Grammatica Generativa monograph series).
Richards, N., 1999, “Featural Cyclicity and the Ordering of Multiple Specifiers”, in: S. Estein, N. Hornstein (eds.), Working Minimalism, Cambridge, Mass., MIT Press, 127−158.
Rizzi, L., 1997, “The Fine Structure of the Left Periphery”, in: L. Haegeman (ed.), Elements of Grammar, Dordrecht, Kluwer Academic Publishers, 281−337.
Rudin, C., 1988, “On Multiple Questions and Multiple Wh-Fronting”, Natural Language and Linguistic Theory, 6, 445−501.
Stan C., 2012, “Elemente de sintaxă diacronică: structurile interogative/relative multiple”, in:
O. Chelaru-Murăruș, M. Cvasnâi Cătănescu, C. Ene, C. Ușurelu, R. Zafiu (eds), Text și discurs: omagiu Mihaelei Mancaș, București, Editura Universității din București, 433–437.
Szabolcsi, A., 1997, “Strategies for scope taking”, in: A. Szabolcsi (ed.), Ways of Scope Taking, Dordrecht, Kluwer, 109−155.
Vasilescu, A., 2002, Întrebarea. Teorie şi practică, Iaşi, Polirom.
