IMPERATIVE ACROSS ROMANCE LANGUAGES
EMIL IONESCU1
Abstract. Negative imperatives in Romance languages are expressed in different ways, but variation seems to be controlled by the distinction between ‘true’ and ‘surrogate’ imperatives. In this paper I attempt to formulate a hypothesis which accounts for the nature of this distinction. The structure of the paper is the following. In the first section I present relevant data, then I describe three attempts that cope with this issue and I give a critical evaluation of these attempts; in the last section I propose my own hypothesis.
Keywords: Romance languages, imperative, negation, illocutionary force, scope.
1. VARIATION IN THE RELATIONSHIP BETWEEN POSITIVE AND NEGATIVE IMPERATIVES IN ROMANCE LANGUAGES
The relationships between negative imperatives and their corresponding positive forms in Romance languages show a considerable amount of diversity. One may identify four distinct types of such relationships.
Type I: asymmetry between 2nd person singular and plural, in the sense that singular negative imperative are not obtained from the positive form, whereas plural negatives are.
This is the case of standard Italian. 2nd person singular2:
(1) a. Telefona!
call (sg, imper) ‘Call!’
b. *Non telefona!
neg call (sg, imper) ‘Don’t call!’
c. Non telefonare!
neg call (sg, inf) ‘Don’t call!’
1 Department of Linguistics, University of Bucharest, [email protected].
I express my gratitude to the anonymous reviewer of this paper. She/he identified several errors and made substantive suggestions that contributed to improve my study. I owe the revision of the English version of this paper to my colleague Mihaela Cristescu whom I express my warmest thanks, too. I remain solely responsible for all kinds of errors that may appear in the present version.
2 The varieties of imperatives from Italian and Western Romance area that are used in this paper are borrowed from the comprehensive corpus of Rafaella Zanuttini (Zanuttini 1997).
RRL, LXIII, 3, p. 241–252, Bucureşti, 2018
2nd person plural:
(2) a. Telefonate! b. Non telefonate!
call (pl, ind) neg call (pl, ind)
‘Call!’ ‘Don’t call!’
Other languages illustrating the same situation are several Italian dialects (Rome, Naples, Sicily, Padua) and standard Daco-Romanian. A sample from this latter idiom in the 2nd person singular looks as follows:
(3) a. Telefonează!
call (sg, imper) ‘Call!’
b. *Nu telefonează!
neg call (sg, imper) ‘Don’t call!’
c. Nu telefona!
neg call (sg, inf) ‘Don’t call!’
2nd person plural:
(4) a. Telefonaţi! b. Nu telefonaţi!
call (pl, ind) neg call (pl, ind)
‘Call!’ ‘Don’t call!’
Type II: asymmetry between 2nd person singular and plural. In this situation, it is the plural negative imperative which is not obtained from the positive imperative. The Northern Italian dialect from Cortina D’Ampezzo is a case in point:
(5) a. Lourà!
work (pl, imper) ‘Work!’
b. *No lourà!
neg work (pl, imper) ‘Don’t work!’
c. No lourade!
neg work (pl, subj) ‘Don’t work!’
As for the singular, the negative form results from the positive one:
(6) a. Laóra (tu)! b. No laóra!
work (sg, imper) neg work (sg, imper) ‘Work!’ ‘Don’t work!’
Type III: In the case of this type, a symmetry may be noticed between 2nd person singular and plural. Negative imperative cannot be obtained from the positive imperative.
Thus, in Spanish both the singular and plural negative imperatives differ from their corresponding positive forms. 2nd person singular:
(7) a. Habla!
speak (sg, imper) ‘Speak!’
b. *No habla!
neg speak (sg, imper) ‘Don’t speak!’
c. No hables!
neg speak (sg, subj) ‘Don’t speak!’
2nd person plural:
(8) a. Hablad!
speak (pl, imper) ‘Speak!’
b. *No hablad!
neg speak (pl, imper) ‘Don’t speak!’
c. No habléis!
neg speak (pl, subj) ‘Don’t speak!’
Finally, the type IV also displays a symmetry, this time, though, between negative and positive 2nd person singular imperatives; the negative is obtained from the positive.
This is the case of Old Milanese. 2nd person singular:
(9) a. Guarda! b. No guarda!
look (sg, imper) neg look (sg, imper) ‘Look!’ ‘Don’t look!’
The situation may also be observed in some other Romance languages, such as Old Italian and two dialects of Romanian, Aromanian and Megleno-Romanian3. Here is the 2nd person singular of the oldest forms of imperatives and their negated form in Aromanian:
(10) a. Dî! a’. Nu dî!
give (sg, imper) neg give (sg, imper ‘Give (it to me)!’ ‘Don’t give (it to me)!’
3 The same situation may be found in non-standard Daco-Romanian: the negative imperative forms of the verbs of the third conjugation class a duce, a face and a zice are obtained by prefixing negation to the positive form. However, the difference from Megleno-Romanian and Aromanian is important: in these two latter dialects, this mode of composition is a rule, whereas in Daco-Romanian prescriptive grammars forbid it.
b. Du! b’. Nu du!
bring (sg, imper) neg bring (sg, imper) ‘Bring (it)!’ ‘Don’t bring (it)!’
c. Fă! c’. Nu fă!
do (sg, imper) neg do (sg, imper) ‘Do (it)!’ ‘Don’t do (it)!’
d. Vin! d’. Nu vin!
come (sg, imper) neg come (sg, imper) ‘Come!’ ‘Don’t come!’
2. APPROACHES TO VARIATION
This diversity may be characterized in a simpler way, thanks to two patterns concerning the relationship between negation and imperative forms. One may thus identify a case of compatibility between negation and imperative form, as in the type IV (Aromanian), where any form of imperative may be prefixed by negation. This, however, is a rare situation. Much more frequently, one finds both compatibility and incompatibility, which means that certain imperative forms may be negated, while others may not. The situation is illustrated by the rest of the other types – type I (Italian, Daco-Romanian), type II (the Cortina D’Ampezzo dialect) and type III (Spanish). The problem that raises now before the approach is accounting for these complex situations of compatibility and incompatibility. In this respect there are mainly three main approaches, the ones of Rivero, Zanuttini and Han4. All of them are developed within the Minimalist Programme.
2.1. Rivero (1994)
According to Rivero’s analysis (Rivero 1994, Rivero and Terzi 1995), the imperative mood encodes an illocutionary feature which must be checked through movement, the checking site being C0, a position that is not the base locus of an imperative. When moving to C0, imperative verbs leave behind pronominal clitics (if any), and this accounts for the post-verbal clitic placement in imperative structures (for instance, in Italian, Fallo! (‘Do it’)). If the verb is not in the imperative mood but is used with directive force, it will not encode the directive force and, consequently, it will not target C05. According to Rivero, in this latter case its appropriate landing site is I0.
It is essential to notice that the Neg0 node (which is responsible for negated imperatives) lies in the tree somewhere between C0 and I0. So, if the verb is in the imperative mood and the Neg0 node is filled with lexical material, Neg0 prevents the imperative verb from moving to C0. This explains the ungrammaticality of a structure like the Italian imperative construction *Non telefona!. On the contrary, if the verb is in a distinct mood (for instance, infinitive) but is used with directive force, it will move up to
4 The approaches of Rivero and Han are not particularily interested in Romance languages.
They aim at a more comprehensive, crosslinguistic account. However, facts from Romance languages are present in their analyses.
5 A directive is a type of illocutionary force that consists of the attempt of the speaker to determine the hearer to do something (Searle 1982: 53).
the I0 node, a position which is lower than Neg0. This movement is licit and is instantiated, for example, by the Italian construction Non telefonare! (‘Don’t call!’).
According to Rivero then, compatibility with negation is a property of a certain class of imperatives, those which are not in the imperative mood but are used with directive force. These are, following Rivero’s terms, surrogate imperatives. Incompatibility with negation, on the other hand, is a property of another class of imperatives, those which, due to their directive feature, are forced to move to C0. These are true imperatives.
2.2. Zanuttini (1997)
Zanuttini (1997) proposes a different view. She considers that the compatibility/incompatibility of imperatives with negation is a matter of subcategorization.
The elements of subcategorization that Zanuttini has in view are negation (as a head) and the verb (as a complement). In the languages which fall in the range of Zanuttini’s analysis (mainly, historical and dialectal varieties of Italian and, more generally, languages from Western Romance area), negation used in imperative sentences and negation used in the rest of the sentences are phonetically identical. Nevertheless, according to Zanuttini, the subcategorization requirement of a negation which combines with an imperative is different from the requirement that regards the combination between negation and a verb with a different illocutionary force. This contrast surfaces for instance in the Italian pair * Non canta!/ (Lui) non canta, (‘Don’t sing!’/ ‘He is not singing’) where the same verb form can illustrate the (etymological) imperative and the third person singular of the present indicative, respectively. As it may be noticed, combinations with negation do not have the same grammaticality effect, despite the fact that the forms involved in them are the same.
This is because, according to Zanuttini, in the imperative ungrammatical construction *Non canta! non requires a certain verb form type, which cannot be illustrated by canta (but can be and is illustrated by cantare). On the contrary, (Lui) non canta is fine, just because negation in this (non-imperative) construction does not require the same type of verb form as in imperatives. Thus the ultimate claim on which Zanuttini’s analysis rests is that Italian reconstructs (or, perhaps, inherits) in a hidden form the (relatively) old Latin distinction between negation nōn and negation nē. Recall that the former one was used, roughly speaking, for non-imperatives, whereas the latter one was used for imperatives and subjunctives of volition (Ernout and Thomas 1964: 148, 230).
Due to this hypothesis, negation of imperatives is defined in Zanuttini’s analysis as requiring that the selected verb form be morphologically rich from the point of view of the mood information. Imperatives that do not observe this requirement create ungrammaticality. Zanuttini shows that imperatives which fail to obey this restriction are in fact true imperatives. Etymologically, Zanuttini says, true imperatives are root forms with no mood information. ‘Suppletive’ imperatives instead (that is, subjunctives, infinitives, etc.) do comply with the negation requirement6.
6 As the anonymous reviewer of this paper notices, the term ‘suppletive imperative’ is not the most appropriate to describe the fact that some imperatives are morphologically identical to other moods (for instance, infinitive, subjunctive or even indicative). This is why, throughout this paper I prefer Rivero’s term ‘surrogate imperative’.
2.3. Han (2001)
Han’s analysis (Han 2001) is the most semantically oriented approach. With Han, the incompatibility between negation and imperative verbs is the consequence of a certain syntactic configuration, one in which negation C-commands the imperative7. This syntactic configuration is the result of the fact that the imperative verb is constrained to move to C0, a position which in Han’s analysis represents the universal checking node for the directive feature encoded by imperatives.
Movement to C0 of the imperative may also involve the negation movement to the same position. This may happen, if negation is a preverbal clitic item. In this case, negation reaches the C0 position, and this in turn leads to the final syntactic configuration, one in which negation C-commands the verb. This configuration, though, is not semantically acceptable, because a negative imperative construction does not express its meaning by means of a negation outscoping the directive force. In other words, a negative imperative construction also expresses a directive, and this means that in fact it is the force which has the widest scope in the clause, not the negation.
Han’s analysis therefore shows why positive imperatives are well-formed and semantically acceptable constructions: they are so because the imperative verb moves to C0, to check its directive feature. No syntactic constraint is thus violated.
In the case of ‘failed’ negative imperatives, the verb moves to the same position, this time, though, along with the clitic preverbal negation. Again, from a syntactic point of view no constraint is violated. However, this syntactic configuration is semantically ruled out, just because of the wrong scope relationship between negation and directive. Negation now outscopes the directive but it is not allowed to.
2.4. Some critical remarks
There are various degrees to which these proposals successfully cope with the compatibility/incompatibility between negation and imperative forms across Romance languages. These degrees actually define the very adequacy of each approach.
In the case of Rivero’s analysis, what is worth emphasizing is the fact that the languages which allow for the combination of negation with true imperatives are considered to locate the checking node of the directive force in I0, not in C0. A striking consequence of this hypothesis is the fact that I0 as a checking node is lower than Neg0. This means that in the L(ogical) F(orm) representation negation scopes over the directive force. An unintuitive semantic perspective on negative imperatives obtains consequently, because this scope relation actually means the cancellation of the directive force expressed by the imperative form.
Zanuttini’s analysis avoids this criticism, because of her conclusion that the negation which is applied to imperatives is not the same as the negation applied to non-imperatives.
7 Following Reinhart (1976), Chomsky (1986), and Kayne (1994), the C-command relation between two constituents X and Y may be defined as follows:
X c-commands Y iff X and Y are categories and X excludes Y (i.e., no segment of X dominates Y) and every category that dominates X dominatesY (see also Han 2001: 309).
As already shown, this amounts to say that imperative negation has a supplementary feature, the directive one. So, illocutionary force scopes over negation.
Zanuttini’s hypothesis has good empirical coverage in the area of Western Romance languages. However, her hypothesis fails to cover facts from Eastern ones. Thus, in Daco- Romanian there are etymologically true imperatives which are rejected by negation even if they are mood-inflected. For example, the form cântă in the imperative ill-formed construction *Nu cântă! (‘Don’t sing!’) comes from the Latin form canta and has been reanalysed as composed of a root cânt- and a morpheme -ă carrying mood, tense and person information (Cuniţă 1989: 148). At the same time, in Aromanian, imperatives which, just like in Italian, are root forms, may be negated. The most prominent examples are four old irregular forms: Nu dă! (‘Don’t give!’), Nu du! (‘Don’t bring!’), Nu fă! (‘Don’t do!’), Nu vin! (‘Don’t come!’). Both these cases represent disconfirmations of Zanuttini’s hypothesis.
Finally, in Han’s approach the central point is that the syntax may supply well- formed structures which semantics cannot interpret. There is, however, a certain syntactic condition involved in Han’s account: imperatives carrying the feature of directive force have to move to C0 together with negation, and this may occur only if negation is a preverbal clitic. Thus, the scope relationship between verbs with directive force inherently encoded and negation is determined by what Han calls the V-to-C0 movement.
Problems of empirical coverage appear in the case of this explanatory hypothesis, too, and they come from two distinct directions. On the one hand, in Aromanian, negation is a pre-verbal clitic and, etymologically, one may find true imperatives which, according to Han, carry the directive feature. These are just the four old irregular imperatives (already mentioned above). In their case, the V-to-C0 movement is expected to take place along with negation and to produce the same semantic effect of non-interpretation. However, the real semantic effect is an interpretable construction and this fact remains unexplained.
On the other hand, a more striking counter-example comes from Modern Central Occitan. This dialect is particularly important because it has postverbal negation. When imperatives that carry the directive feature occur with negation, the construction is ruled out: Canta! (‘Sing!’) but *Canta pas! (‘Don’t sing!’). Its well-formed alternative has to be a surrogate imperative (a subjunctive): Cantes pas!. Obviously, in both cases described above additional devices of explanation have to be put forward.
3. AN ALTERNATIVE APPROACH
The history of the approaches to negative imperatives seems to attest a progressive orientation towards semantics. Rivero’s analysis ignores the scope relationship between force and negation. Zanuttini does not ignore it but sees it as a consequence of the fact that negation for imperatives encodes the directive feature. Finally Han’s approach is of a particular interest, in that it shows that, as long as the syntax does its job and supplies well- formed structures the problem of the compatibility/incompatibility between imperatives and negation locates at the interface between semantics and illocutionary force.
I think that Han’s option represents the right strategy in the attempt of coping with the problem of the negative imperatives. This is why, in what follows, I propose an account which also relies on the scope relations between operators. However, unlike Han’s analysis,
this approach makes no special appeal to syntax. It consists of three steps: specifying a principle of well-formedness which regards the relation between the scope of the operators in an imperative sentence; describing a method of determining these scope relations; and applying these tools to imperative sentences in a given language.
3.1. A scope-based principle of well-formedness for imperative sentences
The starting point in the construction of the analytical frame is the trivial requirement that sentences have illocutionary force. It is not easy to imagine what a sentence with no illocutionary force would look like (because such sentences are rather sentences of logic systems), but it is certainly easier to imagine how speakers of a natural language react when they are not sure what illocutionary force is precisely associated with the sentence they hear. This fact is a testimony that the connection between sentence and illocutionary force is essential in natural languages.
“Sentence” means here main sentence (for instance, It is raining). As for illocutionary force, the view adopted is the standard one, namely that illocutionary force is an operator. Operators have scope, and, roughly speaking, in our case the scope of an illocutionary force cannot be but the whole propositional content to which the illocutionary force in question applies.
Coupled with the requirement that sentences have force, the view that force is an operator has an interesting consequence. It entails that force must be the operator with the widest scope in the sentence, because otherwise it could not mark the sentence as a whole8. In the case of imperative sentences, for instance, the directive force must outscope any other operator.
More generally, then, the widest scope of the illocutionary force appears to be a principle of (semantic and pragmatic) well-formedness for a certain category of sentences which will be here the object of investigation. At issue are sentences that use functional elements (intonation, mood morphology, particles etc.) to encode force. It is this principle that we propose to adopt in the analysis of imperative sentences in Romance languages.
However, before doing this, we need a means of determining the scope relations between force and other operators, especially negation.
3.2. Determining scope relations
The phenomenon that helps to determine the relationship between scope operators in sentences with functionally-encoded force is the semantic parallelism/lack of parallelism of these sentences with sentences in which illocutionary forces are expressed lexically.
Romance languages analysed here all illustrate the distinction between lexical and functional means in the expression of the illocutionary force. This distinction permits the identification of a rough meaning equivalence between a sentence with illocutionary force expressed through functional elements and a sentence in which the same illocutionary force is expressed by means of a performative verb used as the main verb of the clause (see also
8 In the Minimalist literature this requirement is expressed by the statement that illocutionary force is uniformely associated with Co.
Han 2001: 307–308). This is true for (at least) the following illocutionary forces: order (directive), question and assertion. As an illustrative language, we use here English. Order:
(11) Don’t go home! I order you not to go home.
Question:
(12) Don’t you go home (at five o’clock)? I am asking you whether you don’t go home at five o’clock.
Assertion:
(13) You don’t go home (at five o’clock) I state that you don’t go home (at five o’clock).
In each second member of the pairs above (i.e. I order you not to go home (at five o’clock), I am asking you whether you don’t go home (at five o’clock), I state that you don’t go home (at five o’clock)), the syntactic order between the expression of the illocutionary force and the expression of negation mirrors the semantic order between the force operator and the negation operator: indeed, the expressor of the illocutionary force precedes (and C- commands) the expressor of negation. Symmetrically, illocutionary force outscopes negation. Thus, sentences with lexical encoding of the force help to disclose the scope relations between force and negation in sentences where force is encoded through functional means. The scope relation in these latter sentences must be the same: force outscopes negation.
Situation changes when we wonder whether sentences with functional encoding of the force may also express a reading determined by the reversed scope relation between force and negation The reversed scope relation means that negation outscopes force. Do then sentences Don’t go home!, Don’t you go home (at five o’clock)? and You don’t go home (at five o’clock) express a meaning determined by the wide scope of negation with respect to the force ?
As far as our sample language, English, is concerned, the answer is no. A reading of the sentence Don’t go home! which is determined by the scope of negation over the directive force would have to mean the following paraphrase expressed by a corresponding sentence with lexical encoding of the force:
(14) I don’t order you to go home.
Obviously, this is not the actual paraphrase, because Don’t go home! does not mean I don’t order you to go home. The same holds for the two other sentences with force functionally encoded. So, in these cases a lack of semantic parallelism arises. It shows that in our language sample, sentences having functional encoding of the force cannot express a meaning resulting from the wide scope of negation with respect to the force.
3.3. One prediction and four types of imperatives in Daco-Romanian
Under the hypothesis that the widest scope principle of the illocutionary force is active in a given language for sentences with functional encoding of the illocutionary force,
the principle makes a certain prediction. It predicts that in the language in question there must be one of the following two situations, as far as imperative forms are concerned:
either all the types of imperatives express a directive with the widest scope in the sentence, or there are types of imperatives whose directive is (inherently) narrow in scope. These latter imperatives may then occur with operators which outscope them. This, however, means violation of the principle and, consequently, ill-formedness.
The two situations predicted by the principle are actually illustrated by Aromanian and Daco-Romanian. Whereas in Aromanian all the types of imperatives receive a lexical paraphrase (with the verb of illocutionary force taking scope over negation), in Daco- Romanian a certain type of imperative expresses a directive force whose scope is systematically within the scope of negation. I will describe more closely the case of Daco- Romanian.
The typology of the imperative forms in Daco-Romanian looks as follows:
I. Plural imperatives homonymous with corresponding present plural indicatives: Protestaţi! (‘Protest!’ (pl))/ Voi protestaţi (‘You are protesting’ (pl))
II. Singular imperatives homonymous with bare infinitive but non-homonymous with corresponding indicatives: Nu protesta! (‘Don’t protest!’ (sg)) = protesta (bare infinitive) ≠ El nu protestează (third person singular present indicative)
III. Singular imperatives non-homonymous with bare infinitive but homonymous with (the third or second person) singular present indicative:
Protestează! (‘Protest!’(sg))/ El protestează (‘He is protesting’); Fugi! (‘Run!’ (sg))/
Tu fugi (‘You are running’)
IV. Singular imperatives homonymous both with bare infinitive and (third person) singular present indicative: (Nu) bea! (‘(Don’t) drink!’ (sg))/ El (nu) bea (‘He (does not) drink’); (Nu) scrie! (‘(Don’t) write!’ (sg))/ El (nu) scrie (‘He (does not) write’)
This typology is morphological and it is important to say that it is independent of the behaviour of these forms with respect to negation. Nevertheless, it is relevant to negation:
the first class of forms are used to express both positive and negative imperatives, the second is exclusively used to express negative imperatives, the third class is, on the contrary, uniquely specialized in expressing positive imperatives, whereas the fourth one is like the first (the members of this class only differ in their grammatical number from those of the first class).
The distinct properties of the Romanian imperatives with respect to negation may now be accounted for by means of the widest scope principle of the directive force. Thus, imperatives of classes I and IV may be paraphrased by sentences in which the lexically expressed directive takes scope over the positive or negative propositional content:
(15) (Nu) protestaţi! Vă cer să (nu) protestaţi.
‘(Don’t) protest!’ (pl) ‘I am asking you (not) to protest.’
(16) (Nu) bea! Îţi cer să (nu) bei.
‘(Don’t) drink!’(sg) ‘I am asking you (not) to drink.‘
Imperatives of class II may be translated by sentences with directive force lexically expressed, only if they have negative form. In this case, too, force outscopes negation, as the translation shows:
(17) Nu protesta! Îţi cer să nu protestezi.
‘Don’t protest! ‘(sg) ‘I am asking you not to protest.’
However, if one associates a positive form of class II (i.e. a bare infinitive) with directive force, the association fails, which means that the form (or perhaps the type of content illustrated by the form) is not proper for that force9. So, one cannot obtain an imperative and translation also fails, accordingly.
(18) *Protesta!
protest (sg, inf)
Finally, verb forms of class III express imperatives only in positive polarity:
(19) Protestează! Îţi cer să protestezi.
‘Protest! ‘(sg) ‘I am asking you to protest.’
If the same form is used in negative polarity with directive force, the result is out:
*Nu protestează!. Given the positive pair Protestează! which is correct, one may draw the conclusion that in *Nu protestează! the positive form protestează is associated with the directive force. However, the whole nu protestează is not. This may show that directive force remains within the scope of negation. Translation proceeds as follows:
(20) *Nu protestează! Nu îţi cer să protestezi.
neg protest (sg, imper) ‘I am not asking you to protest.’
3.4. Back to Italian
The above analysis may extend the use of the hypothesis regarding the illocutionary force as the widest scope operator to the case of the other Romance languages, as well, and, of course, to the case of Italian. Italian dialects with preverbal negation show that what led Zanuttini to assume the existence of two negation markers is a plausible reason: it is the systematic correlation between negation and mood marking in the case of imperatives, and, also, the absence of this correlation in the case of the other verb forms. The problem is, though, that whereas Zanuttini’s explanation cannot be adopted for Daco-Romanian and Aromanian, the hypothesis proposed here for imperatives in this area is applicable to the rest of Romance languages. For instance, what has been said about the Daco-Romanian
9 One cannot ignore, as a principle of formation of a speech act, the association between an illocutionary force and a verb form. The association implies that not any verb form may be associated with any force. This seems to be the explanation of failed imperatives, such as *Protesta!. I am not dealing with this issue in this paper.
form protestează may also be said about a form like the Italian parla (‘speak’); when used with directive force, parla is a fine positive imperative (Parla!), but in its negative form, non parla, it cannot express the corresponding negative imperative: *Non parla!. One cannot see why in Non parla!, parla could not have the same force, once it has it independently, in Parla!. And with this remark in mind, ill-formedness receives the natural explanation that the directive expressed by parla in *Non parla! inherently has scope under negation.
4. CONCLUSION
The conclusion reached by the present analysis is therefore that imperatives which are incompatible with negation in Daco-Romanian express a directive force inherently dominated by negation. This, though, does not seem to have any correlation with syntactic phenomena. Actually, the only connection with syntax seems to be the fact that in the majority of Romance languages with postverbal negation (standard French, Perigord Occitan, Northern Italian dialects – Piemontese, Milanese, Valdostain – see Zanuttini 1997: 60-98), the incompatibility between negation and true imperatives does not exist. In all these languages, the C-command relation between imperative and negation (imperative C-commands negation) seems to mirror a scope semantic relation between the two operators. However, as seen above in the case of Modern Central Occitan, even in these languages there are exceptions which show that syntax and semantics can diverge.
Under these conditions, the scope idiosyncrasy of a certain class of forms – in Daco- Romanian, imperatives homonymous with the third person singular indicative, but non- homonymous with bare infinitive – rather appears to be a lexical curiosity. However, as long as this scope idiosyncrasy manifests in other languages, as well, it is certain that at issue is not just a curiosity, but, it seems, one of the phenomena that cannot be accounted for by means of structural factors.
REFERENCES Chomsky, N., 1986, Barriers, Cambridge, MA, MIT Press.
Cuniţă, A., 1989, “Imperativul”, in: M. Sala (coord.), Enciclopedia limbilor romanice, Bucureşti, Editura Ştiinţifică şi Enciclopedică, 148–149.
Ernout, A., Fr. Thomas, 1964, Syntaxe latine, Librairie Klinksieck, Paris.
Han, Ch-H., 2001, “Force, Negation and Imperatives”, The Linguistic Review, 18, 289–325.
Kayne, R., 1994, The Antisymmetry of Syntax, volume 25 of Linguistic Inquiry Monograph.
Cambridge, London, The MIT Press
Reinhart, T., 1976, The Syntactic Domain of Anaphora, MIT dissertation.
Rivero, M.L., 1994, “Clause Structure and V Movement in the Languages of Balkans”, Natural Language and Linguistic Theory, 12, 1, 63–120.
Rivero, M.-L., A. Terzi, 1995, “Imperatives, V movement and Logical Mood”, Journal of Linguistics, 31, 301–332.
Searle, J. R., 1969, Sens et expression, traduit de l’anglais par Joelle Proust, Paris, Les Editions de Minuit.
Zanuttini, R., 1997, Negation and Clausal Structure, Oxford, Oxford University Press.
