Interaction of negation with tense, modality and information structure in Standard Arabic *

The aim of this paper is to consider the interaction of tense, mood and focus with negation in Standard Arabic. This interaction can be observed via marking the tense and mood of the sentence, or via selecting a particular type of tense, or being associated with Information Structure. Building on this fact, the current paper provides a unified analysis, in which negation in Arabic can be accounted for without a NegP projection.

The goal of this paper is to provide a typological sketch to the negation system of these three broad classes, focusing on considering the interaction of tense, mood and focus with negation in Arabic.The current paper also aims at providing a unified analysis, in which negation in Arabic can be accounted for without a NegP projection.
(4) a. ya-ðhabu T-Taalib-u ʔila l-madrasat-i Kula yawm-in.3m-go-Pres the-student-Nom to the-school-Gen every-Acc day-Gen "The student goes to school every day."b. laa ya-ðhab-u T-Taalib-u ʔila l-madrasat-i.neg.Pres 3m-go-IND the-student.Nom to the-school-Gen Kula yawm-in every.Acc day.Gen "The student does not go to school every day." As the preceding examples show, the verb after these four negative markers must be in an imperfective form.Otherwise, the derivation crashes: (5) ʔar-rjal-u lam *(takalam-uu).the-men-Nom neg.past 3m-past talk.mp(6) laamma *(kataba) zaid-un l-qiSSat-a.neg.pf wrote.pastZaid-Nom the-story-Acc (7) ʔaT-Tullaab-u lan *(daraus-uu).the-students-Nom neg.fut 3m-past study.mp(8) laa *(ðhahaba) T-Taalib-u ʔila l-madrasat-I kul-a neg.Pres went-Past the-student to the-school-Gen every.Acc yawm-in.day.Gen The interaction between tense and negation with these negative markers is highly remarkable.Each negative marker in (1-4) inflects for tense.As a result, we find lam and lamma make a morphological alternation in the verb form.Lam is associated with the past tense interpretation, whereas laamma is associated with the perfect tense interpretation.Accordingly, lamma in (2b) negates the occurrence of the situation up to a given reference point.Thus, (2b) indicates that 'Zaid has not written the story but there was an intention to bring about the situation'.As for lan, it allows only future tense interpretation.Finally, laa occurs in sentences with present tense interpretation.Consequently, each negative collocates with a certain temporal adverb as the following table indicates: (9) The reason for these different tense interpretations with these negatives may come from the fact that "tensed verbs are in complementary distribution with tensed negatives.When the negative inflects for tense the verb cannot do so" (Benmamoun 2000: 96).
The last interesting fact to note at this point is that each negative marker in (1-4) occurs with a particular variant of the imperfective form: lam and lamma occurs with jussive, laa with indicative and lan with subjunctive.
The interaction of negation with tense and modality, however, leads Fassi Fehri (1993: 163) to propose that the negatives lam, lammaa, lan and laa should be treated syntactically as Modal negatives1 since they entertain a specific governing relation with (inflected) verbal projections and can assign Mood to the imperfect verb that follow them, and make its function as a carrier of mood, without any particular tense specification as they are inflected for tense as illustrated above in (9).In addition to being inflected for tense and able to assign mood, these negatives behave like modals in that they must be adjacent to the verb that follows them (Fassi Fehri 1993: 163).Accordingly, it is unacceptable to insert any element between the verb and the negative marker: This being the case, these tensed modal negatives, as proposed by Fassi Fehri (1993), select a clausal inflectional projection, not a bare VP as in English and French.That is to say, modal negatives are in a head position, and hence they can attribute the right selectional properties, and their complement is TP, not VP (contra Benmamoun 2000, Ouhalla 1994, andBahloul 1996) as can be shown by the simplified derivation in (14) below: (14) Note that the derivation in ( 14) further illustrates that the verb moves from V to T achieving the VSO order, and then moves further into Neg to become adjacent to the head Neg satisfying requirements of mood assignment.
Benmamoun (2000: 99) suggests that there is a formal reason that requires the verb to move to the negative and merge with it other than satisfying requirements of mood assignment.As a starting point, he proposes, in line with Ouhalla (1994) and Bahloul (1996), that NegP in Arabic is always located between TP and VP.Adopting Aoun's (1981) proposal that in Standard Arabic tense is expressed independently of the verb, he proposes that the merger or incorporation between negation and the verb relates to the tense feature carried by the negatives laa, lam, lammaa and lan.This feature only forces the presence of the verb.To formally account for the merger between the verb and the negative marker, Benmamoun assumes that T with verbal sentences can be divided into two types.The first type of T is that of past and future verbal sentences.And it is valued with two sorts of features: a [+V] feature and a [+D] feature (i.e., nominal feature).The second type of T is specified only for a [+D] feature and can be found only in sentences with present tense interpretations.
It is natural that the V feature is satisfied by verbal heads.But the question that should arise now is that since the [+D] feature is a nominal feature, what features on the verb are capable of checking this nominal feature?Following Ritter (1995), Benmamoun notes that one feature on the verb that is able to check [+D] is person agreement, and that in a null subject language such as Arabic, agreement is generally assumed to be a nominal element and therefore it enables the verb to check the [+D] features of functional heads.
Since T with the tensed negatives lam, lammaa and lan is specified for a [+V] feature, then these negatives cannot by themselves check that feature.This is due to the fact that the tensed negatives, by virtue of carrying the temporal features of tense, are specified for [+D] feature, but not for [+V] feature (Benmamoun 2000: 100).Thus, the adjacency and merger between the two lexical heads are straightforwardly explained as follows.To satisfy the [+D] feature of Neg, the verb must merge with Neg, via move (internal merge, in the sense of Chomsky 2001Chomsky , 2008)).Then the whole complex moves into T to have its [+V] and [+D] features checked as represented by the derivation in ( 15). (15) In the present tense sentences headed by the negative laa, the case is slightly different.According to Benmamoun's proposal, T in present tense lacks the [+V] feature and therefore it cannot attract the verb.However, the merger between the negative laa and the verb, carrying subject agreement, is justified to satisfy the [+D] feature of Neg.Such a satisfaction requires the verb to move to Neg to have the [+D] feature of Neg checked2 as shown by ( 16) below.( 16) As evidence for the proposal that the categorical features of the elements in T are not uniform, Benmamoun cites the well-known phenomenon in Arabic of the absence of an overt verbal copula in the present tense vs. its obligatory presence in the past tense (17).This lends, according to him, direct evidence that T in present tense lacks the [+V] feature, and therefore it does not force the presence of the copula.

The fundamental role of Laa
Following Benmamoun (2000: 95), I assume that laa is the default negative marker in Arabic.This is not surprising since laa is the original source of all negative markers, apart from maa.That is, lam, lammaa, lan and laysa are derived from laa as indicated in (18) below, taken from Al--khawalda (1997: 253).
by the Agree theory (Chomsky 2000(Chomsky , 2001)).Second and more importantly, it is not clear why the [+D] feature of Neg in verbal sentences can only be checked by the verb not by the subject which is a nominal element and closer to Neg than the verb.Such an analysis leads to a complex derivation by further assuming that the whole complex Neg and V moves into T as illustrated above in (15).Later in section 5, I argue, following Fassi Fehri (1993), that the merger between the verb and the tensed negatives can be convincingly explained if we consider these negatives as modals.
(18) a. laa + m =lam b. laa + mmaa =lammaa c. laa + ʔan = lan d. laa + ʔysa = laysa Moreover, two distinguishing features, according to Benmamoun, make laa the default negative maker in Arabic.(i) laa is the only negative that can be used as the negator in answers to questions with all types of tenses.(ii) Unlike the others negative markers; laa, like the typical constituent negative ghayr, can occur in constituent negation.Laa in this case, as the example in (22) shows, is always followed by an indefinite subject in the accusative case without nunation,3 negating the attribution between the subject and the predicate by denying the existence of the subject.As laa specifically negates the existence of the subject, the subject and predicate cannot be reordered (Hasan 1970: 1/688).This can be illustrated by the following ill-formed sentences derived from the example in ( 23).

Word order on Laysa
Because laysa is a free morpheme, unlike the other Arabic negative particles, it can be located at the beginning of the sentence as in (32) above or directly before the predicate as in (33) or between the predicate and the subject as in ( 34): (33) ʔT-Taqs-u laysa jamiil-an the-weather-Nom neg-3ms nice-Acc "The weather is not nice."(34) Jamiil-an laysa T-Taqs-u nice-Acc neg-3ms the-weather-Nom "The weather is not nice."But laysa does not permit both the subject and predicate to precede it as illustrated in (35).
(35) * ʔl-walad-u tawil-an laysa the-boy-Nom tall-Acc neg-3ms The reason for this can be attributed to the fact that Arabic is a head first language in which heads are canonically positioned before their complements.Therefore, Arabic neg morphemes are treated as heads of syntactic Neg phrases.Being heads, they can attribute the right selectional properties, and their headness is strongly supported by its governing and Case/Mood assigning properties.Consequently, they must precede what should be negated.

A free negative class: Maa
The last negative marker that negates both verbal and verbless sentences is instantiated by maa.As for verbal sentences, this negative negates mostly verbal sentences in the context of past tense as in (36) and rarely of the present tense as in (37), but in the context of the future tense, maa is ruled out (38). ( 36 This interesting negative marker shares with modal negatives, discussed earlier in section 2.1, in the property of negating verbal sentences, but differs from them in that it is not merged with inflection and hence does not have any Case marking or Mood marking properties (Fassi Fehri 1993: 162).More interestingly, maa can be adjacent to the verb it negates as shown in ( 36) and (37) above and (40a) below, or it might be separated from it by the subject as (39) and (40b) show us.
(39) maa zaid-un jaaʔ-a.neg Zaid-Nom came.past.3ms"Zaid did not come."(40) a. maa qaala maher-un l-haqiqat-a.neg said.past.3msMaher-Nom the-truth-Acc "Maher did not say the truth."b. maa maher-un qaala l-haqiqat-a.neg Maher-Nom said.past.3ms the-truth-Acc "Maher did not say the truth."(Onizan 2005: 31) Turning now to its status with nominal sentences, maa can negate nominal sentences 'with non-verbal predicates verbless sentences', but they must have the following properties: (i) maa must precede both the subject and the predicate; (ii) their subject must be definite; (iii) their subject and predicate can be reordered (Hasan 1970: 1\594) Back to Benmamoun's analysis that negation in Arabic is specified for a [+D] feature and how this feature can be checked with maa, Benmamoun assumes first that maa is in the spec of NegP since it does not have any Case marking or Mood marking properties.Its [+D] feature can be checked as follows.In the past tense, as in (36) above, the verb must move to T to support tense and check its [+V] feature.Since the verb carries agreement features that can check the [+D] feature of negation, it moves through the negative projection and merges with its Spec.In present tense, as in (37) above, maa's [+D] feature can be checked by the agreement features on the verb without moving the whole complex Neg+V to T (since T lacks V features) or can be checked by an immediate merge between the negative and the subject in case of nominal sentences with non-verbal predicates such as (41a) above.The three different ways to check [+D] feature of the negative maa can be represented in the following derivations respectively:

Distinguishing features of Maa
Maa not being associated with a particular tense and hence co-occurring with all predicate types -leads Fassi Fehri (1993:162)  Zaid-Nom writer.Acc "Zaid is not a writer."d. maa zaid-un ʔaxi-ii / ħaDir-an/ fii l-bayt-i.Neg Zaid-Nom brother-Nom.my-Gen/present-Nom/ in the-house "Zaid is not my brother/ present/in the house." Maa is also distinctive in that it differs from other negatives in the extraction properties.Maa cannot be preceded by any element.Conversely, extraction over other negatives is possible.This can be seen by the contrast in ( 46) and ( 47 c. zaid-an lam ʔ-ara.Zaid-Acc nge.past 1s-see-JUS "Zaid, I did not see."(Fassi Fehri 1993: 167) d. zaid-an lammaa ʔ-ara.Zaud-Acc nge.Pf 1s-see-JUS "Zaid, I have not seen."f. zaid-an lan ʔ-araa.Zaid-Acc nge.past 1s-see-SUB "Zaid, I will not see." Relying upon the important insight of Fassi Fehri (2005), the reason why maa has this distinguishing feature is that maa is selected only by C and therefore it must be adjacent to C, meaning that the negative occupies a position higher in the structure, compared to laa or its variants.That is, maa "is higher than the subject of predication (or Focus) and can have no Spec, it is not located in TP, but higher, although it is lower than C" (Fassi Fehri 2005: 15).This is evidenced by the occurrence of maa before the tensed auxiliary in complex tense structures, and the impossibility of appearing on the lower verb, unlike laa.In what follows, I conclude the discussion about maa by providing a suggestive argument indicating that the above features make maa different from other negatives not only in terms of its syntactic behavior, but also in its position in the derivation and in its association with Information Structure.

The derivational position of maa
Fassi Fehri (2005) opts for an analysis whereby maa projects its own projection NegP located above TP since maa is selected only by C, as shown by the simplified representation in (50).( 50) Although the analysis in (50) is possible, there is another analysis that is more convincing and explains better why maa is selected by C. As a starting point, we should recall that maa and laa are the major negation elements in Arabic.As a result, they should have different functional properties and distribution.As demonstrated in section 2.2, laa and its temporal variants interact with tense in the sense that they mark the tense of the sentence.The case with maa, by contrast, is completely different.Maa does have temporal interpretation.Instead, following Ouhalla (1993Ouhalla ( , 1997)), maa differs from the inflected negatives in having a focus feature, in addition to the feature of NEG.This is supported by the fact that maa is considered as a stronger denial, acting as morphological marker of negative contrastive focus and hence does not create its own NegP projection as in (51). (51) Analyzing maa as a mere negative marker is problematic because it fails to explain the semantic difference between lam and maa.Under the assumption that maa is just a negative marker, there is in fact no difference in meaning between the sentences negated by lam and those negated by maa as illustrated by following two examples ( 53) is repeated again from (40a)): (52) a. lam ya-qul maher-un l-ħaqiqat-a neg.past 3s-say-JUS Maher-Nom the-truth-Acc "Maher did not say the truth."(53) maa qaala maher-un l-ħaqiqat-a Neg said.past.3msMaher-Nom the-truth-Acc "Maher did not say the truth." The syntactic difference between the two negatives is obvious.Lam inflects for tense, which precludes the verb from doing so, while maa cannot carry tense, which allows it to be spelled-out on the verb.However, there is no semantic difference as can be seen from the translations in the two examples above.The same situation can be also observed by making the same contrast in (54-57) between maa and laa on one hand and maa and laysa on the other.The translations that conclude each example below indicate that there is no semantic difference between the two negatives although the difference in syntactic structure can be noticed (i.e., laysa assigns case, but maa does not).( 54 The conclusion that can be drawn from the above data is that it is inconceivable that maa functions only as a negative marker and optionally can replace the above certain negatives.There must be a reason to force the speaker of the language to use maa instead of other negatives.Otherwise, if maa and the inflected negatives lam, laa, laysa convey an identical meaning, maa would be redundant in the language. While accepting Ouhalla's hypothesis that maa is singled out from laa and its variants by virtue of having a focus feature, I argue contra Ohualla's (1993) idea that maa can occupy the spec position of FocP, forming a constituent with a phrase which is negatively focused. 4Instead, following Fassi Fehri (2005), I assume that maa does not have a Spec since it does not allow any element to occur on its left (see the example in ( 45)).It is, as first proposed by Fassi Fehri (2005), selected directly by C. From this it follows that occupying the head of FocP, as represented in (51) above, is the only option left to maa (whether it is followed directly by a verb or a DP).

Negation Formation without NegP
Pollock (1989, and much subsequent work) has developed the well-known idea that sentential negation receives its own functional projection, NegP.That is, neg is an element which is syntactically encoded as a functional head, on a par with tense, agreement, mood and aspect.Although the NegP approach has been adopted and is widely accepted in most current generative studies, some recent studies have attempted to argue against the universality of this approach for all languages sentential negation (see for instance, Webelhuth 1989, Zimmermann & Stromswold 1999 for German and Kim 2002 for Korean). 5 Negation in Arabic can add supportive evidence in favor of these studies.As argued in the previous section, negation with the negative maa can be 4 Briefly, Ouhalla's (1993) analysis can be summarized as follows.Maa marks negative contrastive focus, and can either occupy the head position of FocP, or form a constituent with a phrase which is negatively focused.The former happens when maa is followed by the verb, and the latter occurs when maa precedes the DP, assuming this is an instance of constituent negation. 5The main conceptual problem caused by the postulation of a functional phrase NegP is that it does not behave like other phrases in terms of blocking movement across its head as the case with the Arabic tensed negatives, viz, laa and its temporal variants.To avoid this problem, others, notably Belletti (1990), suggest that the negator must occupy the Spec of NegP, instead of its head.But such a suggestion, as pointed out by Zimmermann & Stromswold (1999, citing Ernst 1992) assigns to the negator two properties usually reserved for heads: the negator is compulsory, unlike all the other specifiers, and it is the semantic centre of the projection.Furthermore, Zimmermann & Stromswold (1999) provide acquisitional evidence against the NegP hypothesis.They found no evidence of an early stage of neg-initial sentential negation (as predicted if there is a VP internal subject stage), no evidence of correct placement of negation during all stages of acquisition (as predicted under a Full Competence approach), and no evidence that correct placement of negation only occurs once agreement has been acquired (as predicted under an Incrementalist approach).