Navigation – Plan du site
I. Descriptions et analyses de langues berbères particulières

Copulative Predication in Tarifit Berber

Abdelhak El Hankari
p. 81-113


Prédication copulative en berbère tarifit

Cet article explore la typologie de la prédication copulative en berbère tarifit. On identifie 3 copules principales : (1) verbale, (2) nominale et (3) locative. Etant donné que tous ces éléments peuvent être employés comme prédicats, l’article propose une configuration uniforme qui rend compte de leur dérivation. Cette structure consiste en un niveau lexical inférieur occupé par le prédicat (VP, NP etc) et une projection fonctionnelle plus haute représentée par le Predicate Phrase (PredP). La tête Pred entre ensuite dans une relation d’accord avec la tête basse située en position de complément, créant une phrase à copule prédicative. Comme ces constructions sont toutes marquées pour le temps/aspect, la dérivation est étendue à une projection TP, où le trait pertinent est valué. L’analyse syntaxique étant unifiée, la différence entre les divers prédicats copulatifs est réduite de façon optimale au statut catégoriel du prédicat de base occupant la projection lexicale basse. Cet article identifie une autre construction copulative correspondant au locatif proxémique. L’examen précis de cette configuration révèle que ce locatif particulier occupe la position de sujet tandis que la tête Pred est occupée par un trait syntaxique seulement, mais n’encode aucune information phonologique.

Haut de page

Texte intégral

1. Introduction

  • 1  The following abbreviations are used for the representation of the data: 1, 2, 3 → ‘1st, 2nd, 3rd (...)

1The copula in English is a verbal category, which is expressed by the invariable ‘be’. This lexical element co-occurs with a predicate DP (John is a doctor), AP (John is sick) or PP (John is on the roof). By contrast, Tarifit Berber like many other languages has a much more productive copula system displaying a correlation between the morphological form of the copula and its syntactic structure. So, the choice among various forms is mainly dependent on the categorial status of the predicate (VP, DP etc.). The highlighted copula in (1)1 inflects for tense/aspect and subject-agreement, which suggests that it is a verbal category. Syntactically, iri is the head of the clause and used as an intransitive verb with no internal argument. Its interpretation in that sentence is existential.

(1) i-srma-n t-iri-n.


‘Fish exists.’

2Unlike (1), the copula below in (2) is exclusive to a predicate that is nominal. The optionality of the lexical subject is an instantiation of pro-drop, which is evidence that the construction is a clause.

(2) argaz-a ð- a-zʤið.

man-that be sg-king

‘That man is a king.’

3Another interesting element, which can also be used as a copula, is aqa in (3). The interest of this copula comes from the fact that it hosts the object clitic, but the latter pronoun refers to the subject of the sentence at Logical Form. In terms of its distribution in the syntax, aqa co-occurs with a locative predicate PP. As for its categorial status, it does not appear to belong to any of the two major word classes. It cannot be nominal since it does not inflect for number or gender and cannot be verbal since it does not inflect for tense and subject-agreement.

(3) aman aqa-θn gi ð-ə-nda.

waterpl in f-cs-lake

‘Water is in the lake.’

4Another element which is used as a copula is the adverbial proxemics locative in (4). Like aqa, this copula also implies location but differs in that it occurs in the initial position of the clause that contains it.

(4) ðin ʃi (n-) i-xddam-n gi ð-addar-θ.

there some of pl-worker-pl in f-house-f

‘There are some workers in the house.’

5The data discussed in (1)-(4) show that the four copulas have the same grammatical function in that they all connect the predicate and the subject. However, they differ in terms of their categorial property. The copula iri (1) is a verbal category in that it inflects for tense/aspect and subject-agreement. The copula ð- (2) is nominal in that it only co-occurs with a nominal predicate. Similarly, the copulas aqa (3) and ðin (4) are adverbial locatives in that they co-occur with a locative PP. So, the variation in form between the four copulas is contingent on the categorial property of the predicate. After this brief survey, the rest of the paper looks more closely at these copula constructions separately and also examines the syntactic implications responsible for their derivation.

2. The verbal copula: iri

  • 2 As will be discussed below, the copula has various forms depending on its aspectual marking. I cho (...)
  • 3  Note the allomorphic variation of the copula in (6) and (7), which arises from the gemination of / (...)

6As pointed out in (1), the verbal copula in Tarifit is represented by iri 2. Its verbal status comes from the fact that it behaves like any other verbs. Morphologically, it inflects for tense and subject-agreement, which are typically verbal inflections (5)-(8)3. In (5), iri is marked for future tense and subject-agreement. Syntactically, the copula is found in a negative (6) and interrogative clause (7). Constructions like these are typically verbal. Due to the fact that the subordinate complementiser in (8) always selects a verbal clause, it follows that iri must belong to the verbal category for it to be the complement of the subordinator qa.

(5) i-fʤah-n að- iri-n g- iɣa.

pl-farmer-n in field

‘The farmers will be in the field.’

(6) u- ʤi-n ʃi gi iɣa.

neg1 neg2 in field

‘They are not in the field.’

(7) mani ʤa-n.


‘Where are they?’

(8) ð-nna-y qa Comp

t-iri-n g- iɣar. in field

‘She told me that they are (usually) in the field.’

  • 4  The /l/ has re-emerged, again, as a result of lexical borrowing and found mainly with borrowed nou (...)
  • 5  Traditionally, it is argued that verbs are marked for aspect but no tense. The main aspectual form (...)

7To the best of my knowledge, Tarifit is the only variety whose verbal copula is realised as iri. Instead, other varieties use ili. This variation in form is due to a phonological innovation whereby lateral consonants are rhoticised in Tarifit and realised as /r/4. The table below in (9) illustrates this phonological difference between Tarifit and other varieties using the four aspectual markers generally found in Berber5. If the forms included below appear with verbs, the fact that they also appear with iri is further evidence that this copula is a verb.





Other varieties





rra → /ʤa/





Perf. Neg

rri → /ʤi/


9Although iri in the data above has a locative PP as the predicate, which may suggest that it is a locative copula identical to aqa, I show in section three that the latter is the true locative. The correlation of iri with locative constructions is simply as a last resort when reference is made to particular tense and aspectual situations that are not compatible with aqa. The primary meaning of the copula is existential. This interpretation is much clearer in sentences where iri is used as a bare intransitive verb with no complement, as seen in (1). A similar use is repeated as in (10). In that sentence, the copula as the head of an embedded clause does not select anything so it can only have an existential meaning.

(10) ð-arzzu að- ð-iri = /atiri/. fut

‘She wants to exist.’

  • 6  The tense and aspectual marking of ara may vary, depending on the form of the main verb and also o (...)

10In Tarifit, at least, a distinction must be drawn between the copula iri as the main verb and a similar form, which is used as an auxiliary verb and expressed by the invariable form: ara similar to the English ‘be’. The functional element encodes grammatical features only, mainly tense and aspect, but has no semantic meaning as can be seen from (11). In that case, ara does not inflect for subject-agreement and acts only as a modifier of the main verb it selects and subsequently marks it for past tense. The fact that subject-agreement in (12) is only marked on the main lexical verb suggests that we are dealing with a simple clause. Furthermore, ara cannot stand alone in the clause but must be supported by a proper lexical verb, as can be seen from the ungrammaticality of (12). Unlike its use as a copula main verb in (5)-(8), ara in (11) is semantically empty. This is typical of functional elements, which encode only formal/grammatical feature but have no semantic meaning6.

(11) ara ð-ss-awar a(r)- mmi-s.

pst to son-her

‘She was talking to her son.’

(12) *ara ag- mmi-s.

pst with son-her

2.1 The derivation of the verbal copula

11Before looking at the structure and derivation of the copula iri, let me outline the basic tenets of my analysis. I argue that the structure of the predicate is built from a basic lexical category, which spells out the predicate as in (13). This structure should generalise to all other copula constructions. As a head, X is a variable representing categories like N, V etc. depending on the grammatical category of the predicate. In the case of the verbal copula, X should be understood as V.

12For this basic category to be a predicate, it is introduced by a Predicate Phrase (PredP) (Bowers 1993, 2001, Svenonious 1994, Ouhalla 2013 etc.), as in (14). This functional head has a category-changing role, which turns the head of the XP into a copulative predicate. Following the merging of the PredP, its functional head enters into a syntactic relation with the predicate in the complement position yielding a predicate construction.


13Due to the fact that all the copula sentences are marked for tense, an additional TP projection above the PredP is therefore justified and represented as in (15). This structure, which I take to be representative of the copula sentence in Tarifit, consists of three layers: (1) a lexical layer which encodes lexical information regarding the categorial status of the predicate, (2) a PredP layer whose head contains the [+Pred] copular feature and (3) a TP layer where the tense/aspect feature is valued.


14Although some constructions appear to be associated with particular aspects, I show that this is simply a prototypical reading and that there are independent reasons which suggest that these features are encoded in the head of TP. The structure in (15) is uniform for all the copulative predicates. The derivation starts at the bottom of the structure. The predicate (complement of PredP) may move to the Pred head above it to value the [+Pred] feature and the subject merges in Spec,PredP, as a requirement for the predicate to have a subject. The predicate may move higher to T to value tense.

15With this in mind, the derivation of a verbal copula sentence like (16a) is schematised as in (16b). In view of the fact the copula has the properties of a main verb and therefore lexical, it should then be base-generated as the head of the VP. To build a copulative predication, iri undergoes movement from V-to-Pred to value the [Pred] feature. As a requirement for the sentence to have a subject, the external argument is merged in Spec,PredP and enters into agreement with the Pred-head. The copula may move to T to value the tense feature there, followed by the movement of the subject to Spec,TP to value the EPP feature.

(16) a. ð-i-kttuf-i-n t-iri-nd g- w-ɣndu. in cs-hole

‘Ants are being in the hole.’


17As will be discussed in the next section, the notion of ‘adjective’ in the attributive case in English is realised in Berber by the nominal copula selecting a nominal predicate. In the predicative case, however, it is realised using a stative verb as in (17). So, sentences like these have no copula. Under the present analysis, their derivation should be the same as iri. The verb as the head of the VP raises to T via Pred, followed by the movement of the subject from Spec,PredP to Spec,TP yielding the surface representation in (17). A language like English would have the copula ‘be’ occupying the Pred and the adjective would be lower in the predicate/complement position.

(17) a-kʃʃuð -a i-fsus.

sg-wood dem

‘This wood is light.’

2.2 The copula ‘iri’ and complex clauses

18There appears to be some parametric variations regarding the use of iri in complex clauses. So, I decided to devote a separate section to this issue. Ouali (2011) reports that iri in Tamazight may select another lexical verb, as can be seen from (18). In that sentence the copula is marked for future and subject-agreement, which selects the verb tǝddun ‘go’. That verb, in turn, is marked for present tense and subject-agreement. The fact that the copula and the second verb it selects inflect for different tenses and are also marked for subject-agreement led Ouali to conclude that these sentences involve two TP projections and therefore complex clauses.

(18) dað ilin la tǝddun (Tamazight)

Fut Be-AOR.3p Pres go-IMP.3p

aðay nawǝð.

when arrive.1p

“They will be leaving when we arrive.”

(Ouali 2011: 53)

19In his review of Ouali (2011), El Hankari (2013) demonstrates that iri in Tarifit behaves slightly different. A similar sentence like the one in Tamazight is ruled out in Tarifit, as can be seen from (19). The major problem with this sentence has to do with the use of an additional lexical verb in the complement position of iri. To salvage (19), only one verb must be used at a time. In (20), iri is used as the only verb in this particular clause and the sentence is grammatical. Similarly, in (21), the only verb used in the clause is uɣur ‘go’ and the sentence is grammatical. If iri cannot select another main verb, unlike Tamazight, this is an indication that the double TP hypothesis may not apply to Tarifit.

(19) *að- iri-n uɣur-n wami n-xðər.

fut. when 1pl-arrive.perf

‘They will be leaving when we arrive.’

(20) að- iri-n gi ð-addar-θ

fut. in f-house-f

aʧmi ʁa n-xðər.

when fut.wh 1pl-arrive

‘They will be at home when we arrive.’

(21) að- uɣur-n aʧmi ʁa n-xðər. when fut.wh 1pl-arrive.perf

‘They will leave when we arrive.’

20However, Tarifit still allows some particular verbs in the complement position of iri. Consider the data below in (22)-(23). The verbal copula is used as a main verb, which inflects for tense and subject-agreement in the usual fashion. The copula then selects another lexical verb, which also inflects for subject-agreement. So, the natural question that arises from this is: why is it that iri can select a verb in (22)-(23) but this option is not available to (19)? An examination of the data from a close range reveals that this discrepancy lies with the kind of verb the copula selects. The ungrammaticality of (19) is due to the fact that the verb which co-occurs with iri is eventive. When the verb in the embedded clause is stative, the sentence is grammatical (22)-(23). In other words, iri only allows stative verbs in the complement clause but does not tolerate action/eventive verbs.

(22) i-sərma-n t-iri-n arxs-n g- w-nbðu.

pl-fish-pl in cs-summer

‘Fish is (usually) cheap in summer.’

(23) aman ʤa-n səmð-n gi ðara. in f-spring

‘Water is cold in the spring.’

21If we take constructions, which involve stative verbs to be small clauses (SCs) similar to English it can then be argued that iri in Tarifit selects only SCs. Conversely, the same verbal element in Tamazight may select either. The claim that stative verbs are heads of SC amounts to the fact that these clauses may not be tensed. There is evidence, which suggests that these verbs are less likely to be tensed when used in combination with iri. Instead, they always default to some kind of neutral/unmarked form and any other tensed form makes the sentence ungrammatical as can be seen from (24)-(25). In these sentences, iri selects a stative verb yet the sentences are ruled out due to the tense/aspect marking on the stative verb. It is the highlighted markers that make the sentences ungrammatical. This tense-marking issue is not limited to the forms used with these two sentences below but applies across the board. That is, any tense or aspect form other than the neutral form in (22)-(23) makes the sentence ungrammatical. Although what we refer to here as the neutral/unmarked form of the stative verb appears somewhat homophonous with the perfective form, the fact that it is the default invariable form suggests that the state denoted by the verb makes no reference to any particular time.

(24) *t-iri-n t-arxis-n.

‘They are cheap.’

(25) *að- iri-n að- arxs-n.

fut. fut.

‘They will be cheap.’

  • 7 A reminder that subject-agreement is an inherent property of Berber, in that all verbs must displa (...)

22If clauses which involve stative verbs are un-tensed when selected by iri, as we argue, and following our general analysis whereby tense information is encoded in TP, it follows that these clauses have the PredP only but no TP projection. But this claim leaves us with another problem having to do with the fact that the stative verbs in (22)-(23) are marked for subject-agreement, which generally correlates with tense7. Furthermore, postulating a structure without TP for the lower clause headed by the stative verb may prove problematic since subject-agreement is arguably valued in T.

  • 8  As for the common view where subject-agreement is generally assumed to go hand-in-hand with tense, (...)

23Despite the lack of correlation between tense and subject-agreement within the lower clause, I believe that the structure proposed can still account for this typology if the agreement associated with the stative verb is taken to be a property of the Pred – head instead of T. Note that agreement relation in the clause is not necessarily an exclusive property of T. In fact, agreement has been associated with various functional categories (AgrP, IP, TP etc.). So, there is no reason why agreement should not be valued in the head of the PredP8. In view of these facts, the derivation of the sentence involving ‘iri + stative verb’ should look like (26) below. Assuming that agreement holds under asymmetric c-command, the structure contains one TP as the main clause headed by iri and two PredP projections. The higher PredP is part of the verbal copula and the lower PredP belongs to the lower SC headed by the stative verb. In accordance with the structure proposed for the copula predicate in (15), the verbal copula and the stative verb are derived in the usual fashion. For iri, this verb originates in the higher V and then moves to value the [Pred] feature, which assigns it a predicational interpretation. The copula then moves further to T to value tense. Similarly, the stative verb originates in the lower V and its subject in Spec,SC. On the assumption that the agreement of the stative verb is located in the Pred[Agr] above it, this head can c-command the subject inside the SC and subsequently agrees with it. The subject can then raise to Spec,PredP to value the EPP feature. From there, the subject has the advantage of bringing the agreement of the copula iri under T. This agreement relation can now value the Case feature of the subject, yielding agreement on the copula: iri-n ‘’. The subject may then undergo movement to Spec,TP (step-wise via Spec,VP and Spec,PredP of the higher clause) to value the EPP feature there. The stative verb may merge with the Pred head possibly at PF, so that it can be spelt out together with its agreement-marking: arxs-n ‘’. As can be seen, the advantage of the analysis is that it accounts for the same subject-agreement on two different verbs. Furthermore, postulating a PredP that takes care of agreement within the SC solves the dichotomy between subject-agreement on the stative verb and the lack of tense on that same verb. The advantage of the proposed analysis is that it can also be extended to the Tamazight sentence in (18), since iri and the verb in that sentence share the same subject. The only possible difference is that agreement in the lower clause in Tamazight is valued in T, in that (18) clearly shows that the verb in that clause is tensed and therefore requires a TP projection.


24Without pre-empting my discussion of the nominal copula in the next section I wish to briefly discuss some particular aspects of this predicate, which provide further support to the argument that subject-agreement within the SC takes place in the PredP. Consider the sentence below in (27a), which is similar to (26) except that the SC (complement of iri) is a nominal predicate. The same sentence with a SC that is verbal is included in (27b). Despite the categorial difference regarding the predicate inside the SC, these two sentences have the same interpretation and can be used interchangeably in Tarifit. Other similarities include the fact that the nominal copula manifests pro-drop, exactly like verbal clauses. If pro-drop correlates with subject-agreement, and if this phenomenon is linked to the presence of the nominal copula ð-, it can then be argued that this morpheme is a manifestation of agreement between the Pred and the DP in Spec,SC when the clause is nominal. Note that the nominal (SC) has no time reference so it is not marked for tense, which implies that this clause has no TP. Under this approach, we have evidence that subject-agreement with the nominal predicate takes place without the presence of tense and that this agreement has no position where it can be valued other than the PredP. A parallelism can then be established between a SC clause that is nominal and another one that is verbal. In (27a), subject-agreement is spelt out as ð- in the lower PredP because the predicate is nominal, but the same agreement in the higher PredP is spelt out as -n because the predicate is verbal (iri-n).

(27) a. i-srma-n t- iri-n [ð- i-məʁʁran-n]SC.

pl-fish-pl be- pl-big-pl

‘Fish is big.’

b. i-srma-n t- iri-n [mʁar-n]SC.


‘Fish is big.’

25Going back to the difference between Tarifit and its Tamazight counterpart regarding clauses which involve iri selecting another verb, and if the second verb in Tamazight is tensed as sentence (18) appears to suggest, a parameter-setting can then be established between the two varieties. The main clause headed by iri in Tarifit selects a PredP (28) whereas the same clause in Tamazight selects another TP (29).

(28) [TP T] [PredP Pred] [VP V] [PredP Pred] [VP V]]]]].


(29) [TP T] [PredP Pred] [VP V] [TP T] [PredP Pred] [VP V]]]]].


26However, it is still not clear why iri in Tarifit allows stative verbs but not eventive verbs. A likely possibility could be due to its use as a locative copula. As mentioned in the previous section, the locative copula aqa is not morphologically marked for tense/aspect since it is not a verbal category. So, its unmarked form defaults to the present tense. When the locative predicate involves particular tense or aspectual situations, iri must be used as an alternative to aqa. In section four, dealing with the locative copula, I show that the predicate/complement of the locative is also a SC. So, iri selecting a SC could be due to the fact that it has acquired some syntactic peculiarities of the locative predicate despite the fact that its categorial status as a verb is maintained.

3. The nominal copula: ð-

  • 9  In Zenaga Berber, spoken in Mauretania, this copula is apparently found as ad- (Taine-Cheikh 2010)

27Earlier in section one, it was shown that the nominal copula in Tarifit is realised as ð-. The same element is generally found in other varieties as d- (Chaker 1983, El Moujahid 1997, Kossmann 1997, Galland 1988)9. As a functional element, this copula is semantically empty. Its role in the clause is mainly grammatical, in that it connects a DP – subject and a nominal predicate with an attributive property (30)-(32). The predicate can be a DP (30) or a nominal modifier (31). Note that a DP with a nominal modifier can occur without the copula ð-, as in (32). However, this construction is a simple noun phrase with a modifying adjunct and cannot be equated with the structure of predication. The DP in (32) must be present and cannot be dropped since it is the head of the phrase. By contrast, the presence of the copula in predicative constructions like (30) and (31) allows the first DP to be dropped freely.

  • 10  While this copula is attested with all the major studied Berber varieties, some parametric variati (...)

28With respect to its form, the copula remains invariable regardless of gender and number of the subject. For instance, the nominal predicate in (30)-(31) agrees with the subject – DP in number and gender but this agreement has no impact on the copula. Due to its affixal nature, ð- always procliticises to the predicate it selects triggering stress on the first syllable of the noun: [ðá]σ. In this sense, it behaves more like a nominal affix and should therefore be classified as a nominal category10.

(30) (said) ð- a-ðβiβ.

said be sg-king

‘Said is a doctor.’

(31) (said) ð- a-mzian.

said be sg-young

‘Said is young.’

(32) ð-addar-θ ð-a-ʃmra-ʧ.

f-house-f f-sg-white-f

‘The white house.’

  • 11  Some nominal predicates may allow temporary reading under particular discourse contexts. In (i), t (...)

29In terms of its tense-marking, the nominal copula in basic sentences always defaults to present tense (30)-(31). However, the focus in these sentences is more on the generic attribute of the subject than tense. So, the property attributed to the subject by the predicate in these sentences is inherent and permanent. Nominal predicates (and adjectives) having a permanent property, unlike verbal predicates, is not unique to Berber but appears to be cross-linguistically common (Milsark 1974, Carlson 1977, Baker 2013). It is this generic feature, which clashes with the presence of temporal adverbs making the sentence ungrammatical (33)11. The generic versus specific reading, which follows from Individual-level (ILP) versus Stage-level predicate (SLP) (Carlson 1977, Kratzer 1996), will be revisited in the next section. After this descriptive survey, the derivation of the nominal predication is examined next.

(33) *ð- a-wssar nhara.

be sg-old today

‘He is/looks old today.’

3.1 The derivation of the nominal copula

30As we have seen in (30)-(31), nominal predication consists of the predicate DP, the copula ð- and the subject. Its derivation, using (30) as an example, is schematised as in (34). The predicate is base-generated lower in the DP. To form a predicate out of this, the PredP is merged above it and its head enters into agreement with the noun: aðβiβ ‘doctor’. The realisation of the copula ð- under the Pred node is arguably the outcome of this agreement, as discussed in the previous section. The subject of the predicate is inserted in Spec,PredP. The derivation may project higher into a TP so that the sentence can be marked for tense (present) and the subject may undergo further movement to Spec,TP to value the EPP feature there. In view of the affixal nature of the copula, the morpheme ð- procliticises onto the DP at PF yielding the nominal predicate complex: ð-aðβiβ ‘be-doctor’.


31While the data in (30)-(31) display the subject-predicate order, the predicate-subject order is equally possible as can be seen from (35). The two alternations were referred to by Ouhalla (1988) as subject and predicate preposing, respectively. In his investigation of the word order of Tarifit, El Hankari (2010) demonstrates that this Berber variety has now developed a topic-comment system. In a transitive clause where all arguments are lexical, the subject is topicalised in Spec,TP. When the object is a clitic, the topic position is filled by the V + Obj.CL yielding the predicate[]-subject order. With this in mind, the two orders in (35)-(36) arguably reflect this alternation. If we assume that the nominal copula is some kind of clitic since it procliticises to the predicate DP, similar to the predicate V + Obj.CL in the verbal clause, it can then be argued that (35) is an instance of the topicalisation of the predicate: copula + DP in Spec,TP. On the other hand, (36) is representative of the topicalisation of the lexical subject (SVO). This alternation between the two orders may be taken as evidence that the structure of the nominal predicate behaves like any other clause, regardless of the categorial status of the actual predicate. This, in itself, lends further support to the unified analysis of the syntax of copulative predication.

(35) ð- a-ðβiβ said.

be sg-doctor said

‘Said is a doctor.’

(36) said ð- a-ðβiβ.

said be sg-doctor

‘Said is a doctor.’

  • 12  The incompatibility of the nominal copula with negation was previously noted by Ouhalla (1988) and (...)

32Also important is that the nominal predicative copula is not compatible with negation, which explains the ungrammaticality of (37). Earlier in (9), we showed that negation correlates with a special aspectual form marked on the main verb and referred to in the Berber linguistic tradition as irrealis. So, the ungrammaticality of (37) may be due to this aspectual feature rather than negation as such. This sensitivity to particular aspect/tense features can also be noticed from the presence of the highlighted imperfective morpheme in (38), which is not compatible with the nominal predication. Similar sensitivity is also found with the locative copula, as will be seen in the next section. This follows from the morphosyntax of Berber whereby only verbal categories may inflect for the four aspectual forms discussed in (9). So, the fact that non-verbal copulas may not be compatible with some of these aspectual features is expected12.

(37) *u- ð- a-ðβiβ ʃi said.

neg1 be sg-doctor neg2 said

‘Said is not a doctor.’

(38) mi-s i-*t-ð-mʁur.


‘His/her son is growing.’

  • 13  One of the languages that are widely discussed, where this binary distinction is marked using morp (...)
  • 14  This hypothesis was suggested by Kratzer (1995), who argued that this aspectual information is the (...)

33Another issue, which needs to be addressed, has to do with the aspectual property of the predicate. In our discussion of this particular copulative clause it was shown that it has a generic reading, which makes the property attributed to the subject permanent and cannot be temporary. This is reminiscent of the traditional distinction between the ILP and SLP (Carlson 1977)13. The generic reading of the nominal copula raises the question as to whether this property is inherent to the predicate. This, in itself, would raise additional issues such as whether this feature is a lexical property of the nominal predicate14 or possibly associated with the copula ð-. If these possibilities are proven to be true, they would be at odds with our earlier assumption. In the structure proposed to account for the copulative predication in (15), it was claimed that the tense and aspect features are the property of T. So, the possibility that permanent versus temporary reading is associated with the Pred or the DP would be problematic.

34Despite appearance, there are independent reasons to rule out the possibility that these features are lexical. Consider the data below in (39a&b). These sentences are the same as (30)-(31), with an additional functional verb. In (39a), the PredP is selected by ara → past-imperfective. The same nominal predicate in (39b) is also selected by ataf → future-imperfective. The data point to the fact that the generic reading is not maintained when the nominal predicate is selected by tense/aspect elements, like the ones below. The two morphemes make the property attributed to the subject temporary, with a beginning and an end, and therefore not permanent. If tense and aspect markers occupy the T position, it follows that the aspectual reading is dependent on the TP projection, and not on the other lower projections. On the basis of these facts, it can then be argued that the prototypical reading that is generic is the covert tense feature in T, which is interpreted as the present tense. When other tense/aspect markers are used, like the ones in (39), this reading may change.

(39) a. ara ð- a-ðβiβ.

pst.prog be sg-doctor

‘He was a doctor.’

b. ataf ð- a-ðβiβ.

fut.prog be sg-doctor

‘He will be a doctor.’

4. The locative copula: aqa

35While aqa is also used to form a copula predication, this element displays a number of properties which set it apart from the previously discussed copulas. First, aqa is a locative copula in that it co-occurs with the locative PP, as can be seen from (40a). Evidence that we are dealing with a locative copula comes from the fact that the locative PP must be present for the sentence to be grammatical. The ungrammaticality of (41b) is due to the missing PP, which suggests that the locative is a predicate argument and not simply an adjunct. Other properties include the fact that aqa appears with an object-clitic. Important is that the clitic in (41a) is required, which suggests that it behaves more like an agreement marker than a pronominal argument similar to subject-agreement in verbal clauses. The absence of the clitic makes the sentence ungrammatical, as in (41c). The agreement-like property of the object-clitic in (41a) somewhat co-indexes with the lexical subject, which makes it receive the logical interpretation of the subject of the sentence. This co-indexation allows the lexical subject to be dropped freely. As for its categorial status, the locative does not belong to any of the two major word classes (i.e. nominal or verbal) in that it does not inflect for either nominal or verbal morphology. Note that aqa cannot be a preposition either since it selects an object-clitic whereas prepositions in Berber select dative-clitics, when the object is pronominal.

(40) a. (i-fəʤah-n) aqa-ðn g- iɣar.

pl-farmer-pl in field

‘The farmers are in the field.’

b. *(i-fəʤah-n) aqa-ðn.


c. *(i-fəʤah-n) aqa g- iɣar.

pl-farmer-pl be in field

36The predicate involving aqa always defaults to present tense. As for its aspectual reading, it is generally specific but the generic reading is not ruled out. In (41), the proposition may be interpreted as specific but the generic reading is also possible (i.e. ‘they live there permanently’).

(41) (i-mddukar inu) aqa-θn gi- Lhoceima.

pl-friend my in Alhoceima

‘My friends are in Alhoceima City.’

37Another slightly different syntactic environment in which aqa is found can be seen from (42a). The copula in that sentence has a DP as the predicate, and not a locative PP. Furthermore, the lexical argument has an object function. Evidence that this DP is the object and not the subject comes from (42b). When the lexical argument in that sentence is substituted for a pronoun, the latter has an accusative form. The alternation between the lexical DP and its pronominal counterpart also suggests that the clitic has an argument status, and does not behave like an agreement marker as in sentences whose predicate is a PP.

(42) a. aqa a-mddukər inu.

be sg-friend my

‘Here is my friend.’

b. aqa-θ.

‘Here is he.’

38It is important to note that there is another locative predicate with identical properties, except that it is used in the interrogative clause. This copula is realised as ka, as can be seen from (43) below. This sentence is the interrogative counterpart of (42), where the wh- operator mani ‘where’ refers to the locative PP. So, the natural question is whether aqa and ka are two different copulas or they are simply allomorphs of a single morpheme. There are a number of reasons, which appear to suggest that they are likely to be allomorphs of the same morpheme. First, aqa and ka are in complementary distribution; one form occurs in declarative clauses and the other occurs in interrogative clauses. Secondly, these forms display identical syntactic properties in that they both co-occur with a locative PP except that location with ka is expressed by the wh-. Thirdly, these elements are morphologically similar. Aside from the deletion of the initial vowel a, the two forms minimally differ in one single sound: /q/ versus /k/. The two consonants are similar in that they are both dorsal but the uvula /q/ appears to undergo assimilation becoming the velar /k/, due to the neighbouring /n/, which is part of the wh- mani ‘where’. Note that this is the only syntactic environment in which ka is found, so it is not possible to test it using other wh- words.

(43) mani ka-θn?


‘Where are they?’

39One last point has to do with the locative predicate seen in (40a), which takes a locative PP as its complement. In section two, it was pointed out that iri may also co-occur with the same locative PP. However, this option is only allowed when reference is made to particular tense and aspectual situations that are not compatible with aqa. In (40a), for instance, iri must be used as an alternative to aqa only if the sentence is in the imperfective or future. When the locative predicate is not marked for tense/aspect, which defaults to present or perfective as in (40a), aqa must be used.

4.1. The derivation of the locative copula

  • 15  While the locative appears to behave similar to the English existential ‘there’, aqa is not existe (...)

40Our discussion of the locative copula sentences in the previous section shows that aqa may co-occur with a predicate PP or DP. While (40a) is a clear locative predicate expressed by the PP, the question is whether (42) is a locative predicate at all since the copula in that sentence co-occurs with an DP at the exclusion of the locative PP. In this paper, I shall argue that there are reasons to suggest that (42) is also a locative predicate but this feature is expressed differently. More specifically, location in (42) is manifested by the highlighted morpheme in (44a&b) that is affixed to the copula. This deictic may be realised as a prefix (44a) or a suffix (44b). So, the locative aqa which we have been treating as a basic morpheme is in fact morphologically complex consisting of the invariable copula qa and a-/-ya. The latter morpheme encodes location and proximity between the speaker and addressee15.

(44) a. aqa Nunja.

here Nunja

‘Here is Nunja.’

b. qaya Nunja.

there Nunja

‘There is Nunja.’

41Although the two sentences above clearly show the compositional nature of the copula, the picture is not that straightforward when looking at the general behaviour of this locative. The de-compositional nature of the locative in (44) does not equally apply to sentences whose predicate is a locative PP. Consider the data below in (45). In (45a), the predicate of the sentence is the locative preposition with a pre-posed subject. As an alternative to the PP, location can be expressed using the deictic pronoun as in (45b). The inclusion of the PP in that case remains optional, which is an instance of locative-doubling. Of particular importance is the fact that aqa can co-occur with any of the three deictic elements, regardless of their degree of proximity. Conversely, the use of the deictic in the predicate that does not involve the locative PP (45c&d) is more constrained. In (45c), a-qa ‘here-be’ is only compatible with ða ‘here’. Similarly, qa-ya ‘be-there’ in (45d) is only compatible with ðiha ‘over there’. Other forms are ruled out simply because they do not agree with the deictic affix in proximity. The natural question that arises from this is, why is it that this agreement does not apply to sentences that take the locative PP as in (45b)? The fact that aqa can co-occur with any deictic pronoun is evidence that this form does not involve any location and this meaning is expressed by the locative PP (45a) or by the deictic pronoun (45b). In other words, aqa in (45a&b) is a bare copula which fills the Pred node and takes the locative PP or the deictic pronoun as a complement.

(45) a. nunja aqa-t gi ð-hanut.

nunja in f-shop

‘Nunja is in the shop.’

b. nunja aqa-t ða/ðin/ðiha

nunja here/there/over there

(gi ð-hanut).

(in f-shop)

‘Nunja is here/there/over there (in the shop).’

c. a-qa nunja, ða/*ðin/*ðiha.

here-be nunja here/there/over.there

‘Here is Nunja.’

d. qa-ya nunja ðiha/*ðin/*ða.

be-over.there nunja over.there/there/here

‘There is Nunja (over there).’

42Further evidence of the compositional nature of the locative element used in (45c&d) can be seen from (45e). In that sentence, the presence of the locative -ya as a suffix to the copula qa makes the locative PP optional. Conversely, the same locative PP is required (i.e. complement) when it co-occurs with a bare locative predicate in (45a). This is evidence that this copula does not involve any locative morpheme and location is expressed by the predicate PP. There is no doubt that the two forms (aqa versus a-qa) are diachronically related, in view of their similarities but the initial vowel in aqa that co-occurs with a locative PP might have been grammatically frozen in time and is now part of the copula root. This would be expected considering that Tarifit is one of the most innovative Berber varieties. So, the question as to why one predicate involves the locative PP and the other does not can be accounted for straightforwardly. Only sentences, which involve a bare copula take the locative PP as a complement (45a&b) whereas sentences whose copula involve the locative morpheme do not require the locative PP (45c,d&e), since location is expressed by that morpheme.

e. nunja qa-ya-t (gi ð-hanut).

nunja in f-shop

‘Nunja is there (in the shop).’

43Before examining the derivation of the two configurations in (46)-(47), the kind of typology discussed whereby location may be expressed by the locative PP or the deictic is not exclusive to Tarifit but was previously proposed for English and many other languages (Freeze 1992, Kayne 2008). In his study of existential and locative predication across a wide range of languages, Freeze argues that the English sentences: (a) ‘the book is on the table’ versus (b) ‘there is a book on the table’ are derived from a basic structure which contains the locative (PP or ‘there’) and the DP in its specifier position. Following the merging of the copula, English has the option of moving the DP to the specifier of the functional category that contains ‘be’ yielding (a) or by moving ‘there’ yielding (b). However, Freeze further shows that this typology may be subject to parametric variation. For instance, a language like Russian does not have an equivalent of the English ‘there’, as can be seen from (46)-(47). In that case, Russian can either move the DP as the subject yielding (46) or the locative PP yielding (47). So, the presence of the locative morpheme affixed to the copula makes Tarifit more like English than Russian.

(46) kniga byla na stole.

book.nom.fem was on table.loc

‘The book was on the table.’

(47) na stole byla kniga.

on table.loc was book

‘There was a book on the table.’ (Freeze 1992: 553-554)

44If we take the locative morpheme a- to be more or less the equivalent of the English ‘there’, the alternation between the DP and the locative can then naturally be extended to Tarifit. Starting with the construction, which makes use of the locative seen in (45a) its derivation is schematised as in (48). Assuming that the predicate position is a SC, the DP and the PP are base-generated there. Following the insertion of aqa for the purpose of building a copula clause, the Pred functional head enters into agreement relation with the DP yielding object-agreement on the copula: aqa-t. The DP moves to Spec,PredP as a requirement for the sentence to have a subject. The predicate then gets marked for tense under T and the subject undergoes further movement to Spec,TP to value the EPP feature there. So, we can now have a better picture about what seemed to be a contradiction where the object-clitic has the logical interpretation of the subject. In a sense, object-agreement reflects the underlying representation of the DP as the object of the copula. But this DP becomes the subject of the sentence at the surface following its movement to Spec,PredP. As for the object-clitic, which shows up as an agreement on the copula and also refers to the subject of the sentence, it can be argued that it receives this logical interpretation through its co-indexation with the lexical DP in Spec,PredP (i.e. the surface subject). It is more likely that the overt agreement displayed on aqa is possibly spelt out following this co-indexation (not before) since object agreement in Berber is not overtly realised on the predicate, but subject-agreement is.


  • 16  In addition to Freeze (1993) and Kayne (2008), see also den Dikken (2006) for a theory that makes (...)

45With respect to the predicate, which makes use of the locative morpheme as an alternative to the PP (45c-e) its derivation is represented as in (49). Like the PP predicate, the SC involves DP and location. However, the latter feature is expressed by the deictic a- under Locative Phrase (LocP). Following the merging of the Pred, the locative morpheme undergoes inversion to Spec,TP via Spec,PredP, similar to the English existential locative16. Since the predicate is marked for tense, the copula qa may move to T to value the [+present] feature there. Due to the fact that the locative a- has a clitic-like property in that it shows up as a pro-clitic to the copula, the deictic morpheme may merge at PF yielding the surface order: [a-qa>DP]. This PF operation may be argued to be motivated by Ouhalla’s (2005a) clitic constraint, according to which clitics are banned from occurring at the beginning of the clause and should therefore merge with an adjacent functional category. Unlike (48), the copula qa in (49) does not bear overt (object) agreement in that the clitic rather alternates with the lexical DP. This behaviour may have to do with the position of the DP. The latter argument remains in-situ, since the Spec,PredP is occupied by the locative morpheme. So, agreement between the Pred – head and its internal argument proceeds in the usual fashion. However, this agreement is not fleshed out on the copula since no co-indexation takes place between the copula and the DP in Spec,PredP in that this position is now filled with the locative morpheme. So, the argument in this particular case is an object of the copula at both underlying and surface representation. Under the proposed analysis, the object clitic being an agreement marker arises only when the underlying object becomes the subject at the surface representation.


5. The proxemics copula

  • 17  There are three locatives of this kind, which mark proximity vis-à-vis the speaker and addressee: (...)

47Another set of elements, which can be used as copula predicates are the proxemics. In terms of their semantic property, they have a locative meaning equivalent to the English ‘here/there’17. As non-copulas, these locatives are generally found in clause-final position with an adverbial function as in (50). The locative in that case co-occurs with a lexical verb. In copula predicate sentences, like the one in (52), the locative has a different distribution in that it must be in the initial position of the clause. This syntactic position makes it behave like a verbal predicate. In (52), ðin selects an argument – DP and can also co-occur with a temporal adverb allowing the clause to be marked for (present) tense. Additional evidence which suggests that ðin is a predicate clause comes from the fact that it behaves like any other verbal clauses. It can be negated (52), used in the interrogative clause (53) and may also be used as an embedded clause selected by a complementiser (54). The locative ðin in all these sentences has a predicative function, except for (50) where it is used as a simple adverbial locative.

(50) zrin-θ ðin. there

‘They saw him there.’

(51) ðin i-nβʒiw-n (nhara).

there pl-guest-pl (today)

‘There are guests (today).’

(52) u- ðin həd.

neg- there

‘There is no one.’

(53) wi i(g)- ðin?

who- Comp. there

‘Who is there?’

(54) ð-nnay qa ðin ʃi (n) i-wssura. Comp there some of pl-old

‘She told me that there are some old men.’

48One last remark, which was pointed out earlier, has to do with the position of the proxemics in the sentence. In a basic declarative clause, like the one in (51), the locative must always be in the initial position of the clause. The SV order is not allowed as in (55), unlike other copula constructions. This issue is addressed next.

(55) *i-nβʒiw-n ðin.

pl-guest-pl there

5.1 The derivation of the proxemics copula

49As we have seen from the discussion in the previous section, this element also has a locative meaning since it corresponds to the English ‘here/there’. Further evidence that this particular predicate is locative comes from its co-occurrence with the locative-PP, as in (56). The proxemics and the PP are similar in that they both have locative meaning. Note also that this co-occurrence implies locative-doubling, which explains the optional presence of the PP.

(56) ðin i-nβjiw-n (g- w-xxam).

there pl-guest-pl in cs-room

‘There are guests (in the room).’

50If this predicate is locative, how is it then different from a-qa/aqa? One possibility is to argue that it is existential, similar to the English existential ‘there’. Although the Tarifit sentences involving ðin look fairly similar to the English existential and also locative-existentials in Romance (Freeze 1999, Kayne 2006), this possibility is unlikely in view of the fact that the predicate DP following ðin may be definite. Conversely, the predicate DP in existentials is non-definite. In (57), the predicate – DP may only have a definite interpretation. The proper noun in that sentence clearly makes reference to a particular identifiable individual. If ðin ‘there’ is not existential, it can then be argued that this property in Tarifit is expressed by the verbal copula iri as pointed out in section two.

(57) ðin nunja (g- w-xxam).

there nunja in cs-room

‘There is Nunja (in the room).’

51The main property mentioned above, which sets ðin ‘there’ apart from all other copulas is that it must occur in clause-initial position. In (55), we showed that the preposing of the predicate – DP is not allowed. Within the proposed analysis, the position preceding the Pred – head is reserved for the subject, which is merged in Spec,PredP and then moves higher to Spec,TP. In view of this fact, I shall argue that the predicate – DP cannot be the subject since that position is already filled. The more likely candidate is the locative itself, since it is the one that consistently appears in the initial position. This would explain the ban of the lexical DP from that position and therefore accounting for (55). The hypothesis makes the prediction that the position of the locative ðin ‘there’ is the same as the English locative ‘there’. As for the Pred – head, it can be argued that it is filled with a phonetically null copula. Note that the copula not having a phonetic realisation is not uncommon. Benmamoun (2008) discusses several instances from Moroccan Arabic, where the copula has no overt representation. He refers to these copulative constructions as ‘verbless sentences’. In his cross-linguistic survey of existentials and locatives, Freeze also shows that Tongan is one the languages which has a phonetically null copula.

52The status of the locative predication may be subject to cross-linguistic variation. For instance, Romance locatives are not subjects (Freeze 1992, Kayne 2006). The evidence that is usually used to justify the claim is that locatives co-occur with another subject, as can be seen from French in (58). The highlighted locative-clitic y cannot be the subject, since that position is occupied by the dummy il. However, this behaviour cannot be extended to Tarifit since the latter has no subject other than the actual locative. So, it can be argued that Tarifit differs from its Romance counterparts in that the locative is in the subject position, similar to English. This would also explain the reason why Tarifit has morphologically different locatives. Under the proposed analysis, the locative (a)qa is a copula – head whereas the locative ðin is an XP/subject.

(58) il y a une voiture dans le parking.

it loc. has a car in the carpark

‘There is a car in the carpark.’

53If the proxemics is the subject of a null copula as we argue, the derivation of a sentence like (57) should look like (59) below. The predicate is a SC, which consists of the DP and the locative ðin ‘there’. When the Pred – head is merged to build a predicate clause, the locative undergoes movement to the subject position in Spec,PredP whereas the DP remains in the complement of the phonetically empty copula. The locative undergoes final movement to Spec,TP to value the EPP feature there, similar to the previously discussed subjects. Like the locative-clitic in (49), ðin ‘there’ may also undergo PF merger under adjacency due to the clitic constraint pointed out in the previous section. The clitic property of the proxemics comes from the fact that they undergo movement to a second position like other pronominal clitics (Dell & Elmedlaoui 1989, Ouhalla 2005). The proxemics-locative being the subject would be similar to the English ‘there’. Conversely, Romance locatives (the Catalan hi, French y and Italian ci) being copula heads should move to T via the Pred. These options are both discussed by Freeze (1992).


6. Concluding Remarks

55This paper examined the typology of copulative predication in Tarifit Berber. Three major categories were identified: (1) the verbal-existential copula is spelt out by iri, (2) nominal by ð-, and (3) locative by aqa. The proxemics locatives were also shown to take part in this typology. However, these are subjects with a phonetically null copula occupying the Pred functional head.

56I proposed a unified structure, which captures all these copulative predicates under the PredP. This projection has the main function of turning the lexical representation of the predicates into copulative sentences. As for their tense and aspect features, these are argued to be the properties of TP. This amounts to the claim that these features are syntactic and not inherent to these copulas.

57Of particular importance is the use of the verbal copula iri in complex clauses. When it is the head of a main clause, iri was shown to select only a small clause but cannot select a main clause, unlike Tamazight.


59I am grateful to Jamal Ouhalla for useful discussions and comments on an earlier version of this paper.


Haut de page


Baker M. (2013). Lexical Categories: Verbs, Nouns and Adjectives, Cambridge Studies in Linguistics, Cambridge University Press.

Basset A. (1952). La langue berbère. Handbook of African Languages 1, ser. Ed. Daryll Forde. London: Oxford University Press.

Benmamoun E. (2008). “Clause structure and the syntax of verbless sentences”, in Foundational Issues in Linguistic Theory: Essays in Honor of J., R. Vergnaud, R. Freidin, C., P. Otero & M., L. Zubizarreta (eds), 105–131. Cambridge MA: The MIT Press.

Bowers J. (1993). “The syntax of predication”, Linguistic Inquiry 24: 59-656.

Bowers J. (2001). “Predication”, in M. Baltin and C. Collins (ed.) The Handbook of Contemporary Syntactic Theory, 299-333. Oxford: Blackwell.

Borsley R., D. Tallerman M. and Willis D. (2007). The Syntax of Welsh, Cambridge University Press.

Carlson G. N. (1977). Reference to Kinds in English, PhD dissertation, University of Massachusetts, Amherst.

Chaker S. (1983). Un Parler Berbère D’Algerie (Kabylie), Doctorat D’état, Université de Provence, Marseille, France.

Chomsky N. (2001). “Derivation by Phase”, in Ken Hale: A Life in Language, in M. Kenstowicz [ed], Cambridge: Massachusetts Instit Technology, 1-52.

Chomsky N. (2004). “Beyond Explanatory Adequacy”, in A. Belletti (ed.) The Cartography of Syntactic Structures, Volume 3. Oxford University Press.

Dell F. & Elmedlaoui M. (1989). “Clitic ordering, morphology and phonology in the verbal complex of Imdlawn Tashelhiyt Berber”, Langues Orientales Anciennes Phonologie et Linguistique 2, 165-194.

El Hankari A. (2013). “Agreement, Pronominal Clitics and Negation in Berber”, Brill’s Annual of Afroasiatic Languages and Linguistics 5, 227–254. A review of Ouali H. (2011). “Agreement, Pronominal Clitics and Negation in Berber”, London/New York: Continuum International Publishing Group.

El Hankari A. (2010). The Morphosyntax of Tharfith Berber, PhD dissertation, the University of Queensland, Australia.

El Moujahid E. (1997). Grammaire Générative Du Berbère. Morphologie et Syntaxe du Nom en Tachelhit, publications de la faculté des Lettres et des Sciences Humaines, Rabat, Maroc.

Freeze R. (1992). “Existentials and other Locatives”, Language: 553-595.

Galland L. (1988). “Le berbère”, in Les langues dans le monde ancient et moderne, III: Les langues chamito-semitiques, J. Perrot et D. Cohen, 207-242. Paris: CNRS.

Guerssel M. (1986b). “On Berber Verbs of Change: A Study of Transitivity Alternations”, Lexicon Project Working Papers. MIT Press, Cambridge, Massachusetts.

Hale K. and Guerssel M. (1987). “Studies in Berber Syntax”, Lexicon Project Working Papers 14, 159-190 Cambridge, MA: The MIT Press.

Hale K. and Keyser J. (2002). Prolegomenon to a Theory of Argument Structure, The MIT Press, Cambridge, Massachusetts.

Halle M. and Marantz A. (1993). “Distributed morphology and the pieces of inflection”, in K. Hale and, Keyser J. (eds.) The View from Building 20, 111-176, MIT Press, Cambridge, Massachusetts.

Kayne R. (2008). “Expletives, Datives, and the Tension between Morphology and Syntax”, T. Biberauer (ed.) The Limits of Syntactic Variation, John Benjamins, Amsterdam.

Kayne R. (1994). “The Antisymmetry of Syntax”, Linguistic Inquiry Monograph, MIT Press, Cambridge, Massachusetts.

Kossmann M. (1997). Grammaire du parler Berbère de Feguig (Maroc oriental), Paris: Peeters.

Kratzer A. (1996). “Stage-level and individual-level predicates”, in G. Carlson & J. Pelletier (eds.) The Generic Book 125-175, Chicago: Chicago University Press.

Laoust E. (1932). Siwa: son Parler, Paris. Laibrairie Ernest Leroux.

Luján M. (1981). “The Spanish Copula as Aspectual Indicators”, Lingua (45) 165-210.

Marantz A. (1997). “No Escape from Syntax: Don’t Try Morphological Analysis in the Privacy of Your Own Lexicon”, University of Pennsylvania Working Papers in Linguistics 4 201-225.

Milsark G. (1974). Existential sentences in English, PhD dissertation, MIT.

Ouali H. (2011). Agreement, Pronominal Clitics and Negation in Berber, London/New York: Continuum International Publishing Group.

Ouhalla J. (2013). “Agreement Unified”, in L. Cheng, and N. Corver (eds) Diagnosing Syntax 314-333, Oxford: Oxford University Press.

Ouhalla J. (2005). “Agreement Features, Agreement and Anti- Agreement”, Natural Language and Linguistic Theory 23, 655-686, Springer, Netherlands, Dordrecht.

Ouhalla J. (1988). The Syntax of Head Movement: A Study of Berber, Ph.D dissertation, UCL, London.

Penchoen T. (1973). Tamazight of the Ayt Ndhir, Los Angeles: Udena Publications.

Schmitt C. (1993). “Ser and estar: Amatter of aspect”, North East Linguistic Society (NELS) 22, 411-426.

Schmitt C. (2005). “Semi-copulas: Event and aspectual composition”, in P. Kempchinsky and R. Slabakova (ed.) Aspectual inquiries, 121–45. Dordrecht: Kluwer.

Svenonious P. (1994). “Predication and Functional Heads”, in J. È. Camacho, L. Choueri, & M. Watanabe (ed.) The Proceedings of the Fourteenth West Coast Conference on Formal Linguistics, 493-507, CSLI, Stanford, California.

Taine-Cheikh C. (2010). “The role of the Berber deictic and TAM markers in dependent clauses in Zenaga”, in I. Bril (ed.) Clause Linking and Clause Hierarchy, Syntax and Pragmatics. Studies in Language Companion series. John Benjamins Publishing Company. Amsterdam/Philadelphia.

Haut de page


1  The following abbreviations are used for the representation of the data: 1, 2, 3 → ‘1st, 2nd, 3rd person’, aor → ‘aorist’, caus → ‘causative’, cl → ‘clitic’, Comp → ‘complementiser’, cs → ‘Construct State’, dat → ‘dative’, dem → ‘demonstrative’, f → ‘feminine, fut → ‘future’, imperf → ‘imperfective’, m → ‘masculine’, neg → ‘negation’, neut → ‘neutral’, perf → ‘perfective’, prog → ‘progressive’, pst → ‘past’, obj → ‘object’, pl → ‘plural’, sing → ‘singular’.

2 As will be discussed below, the copula has various forms depending on its aspectual marking. I choose here iri as the basic form, which is found in the imperative and the aorist being the neutral/unmarked form of the verb. Cadi (2006) also uses iri as the unmarked form, in Tarifit, corresponding to the French être.

3  Note the allomorphic variation of the copula in (6) and (7), which arises from the gemination of /r/: /r/ + /r/ = /ʤ/.

4  The /l/ has re-emerged, again, as a result of lexical borrowing and found mainly with borrowed nouns.

5  Traditionally, it is argued that verbs are marked for aspect but no tense. The main aspectual forms, which alternate on the verb, are perfective and imperfective. These are interpreted as past and present, respectively. Two other forms are identified in the Berber linguistic tradition. The irrealis/perfective form which is exclusive to negation and the aorist form which is the neutral/unmarked form, generally associated with the imperative (Laoust 1932, Basset 1952, Penchoen 1973, Hale and Guerssel 1987, Ouhalla 1988).

6  The tense and aspectual marking of ara may vary, depending on the form of the main verb and also on whether the verb is eventive or stative. This is summarised in the table below in (i). With respect to eventive verbs, the combination of ara with the perfective form yields past-perfective and the imperfective form yields past-imperfective or progressive. As for stative verbs, the combination of ara with the perfective form yields past-tense and imperfective form yields past-imperfective or progressive. It should be noted though that ara always marks the verb for past-tense, regardless, but the feature that varies is aspect. It is important to note that this same element, which is realised in other varieties as ala, is a present marker in Tamazight (Ouali 2011).


Eventive VerbsStative Verbsara + PerfPast – Perf.Pastara + ImperfPast – Imperf/Progr.

7 A reminder that subject-agreement is an inherent property of Berber, in that all verbs must display subject-agreement when used in the clause regardless.

8  As for the common view where subject-agreement is generally assumed to go hand-in-hand with tense, this correlation is not always true cross-linguistically. Welsh, another VSO language, has some embedded clauses (referred to as ‘i-clauses’) that are not tensed yet display subject-agreement (Borsley, Tallerman and Willis 2007).

9  In Zenaga Berber, spoken in Mauretania, this copula is apparently found as ad- (Taine-Cheikh 2010).

10  While this copula is attested with all the major studied Berber varieties, some parametric variations may still be found. Chaker (1983) reports that Taqbaylit allows this copula to be used with “weather expressions”: ‘it is cold, hot etc.’ Unlike Taqbaylit, Tarifit does not allow the nominal copula in this particular context as in (i). Instead, this predication is simply realised by a DP as in (ii). This suggests that nominal predication in Tarifit can also be realised by a bare DP without the need for the nominal copula.

(i) *ð- a-smmið.

be sg-cold

‘It is cold.’

(ii) a-smmið.


‘It is cold.’

11  Some nominal predicates may allow temporary reading under particular discourse contexts. In (i), the property of being ‘pale’ attributed to the subject may be temporary if the person is sick and does not look well that day.

(i) ð- a-wraʁ nhara.

be sg-yellow today

‘He is pale today.’

12  The incompatibility of the nominal copula with negation was previously noted by Ouhalla (1988) and Ouali (2011). Ouhalla accounts for this by selection, arguing that the first negative particle u- must select the negative form displayed on the verb. Ouali deals with this by feature-pairing where negation enters into agreement relation with T, using the theory of Agree (Chomsky 2001, 2004).

13  One of the languages that are widely discussed, where this binary distinction is marked using morphology, is Spanish. Spanish has two copulas: estar, which is generally argued to correlate with ILP and ser with SLP (Luján 1981, Schmitt 1993; 2005). If the nominal copula is generic, this would be evidence that Berber also displays morphological distinction regarding these features, similar to Spanish.

14  This hypothesis was suggested by Kratzer (1995), who argued that this aspectual information is the property of the lexical predicate.

15  While the locative appears to behave similar to the English existential ‘there’, aqa is not existential in Tarifit in that the lexical DP that follows may be definite. This can be seen from the doubling case in (i).

(i) aqa-t, Nunja

here-3f.obj Nunja

‘Here is she, Nunja.’

16  In addition to Freeze (1993) and Kayne (2008), see also den Dikken (2006) for a theory that makes use of similar inversion.

17  There are three locatives of this kind, which mark proximity vis-à-vis the speaker and addressee: ða ‘here’ (i.e. close to speaker), ðin ‘there’ (i.e. far from speaker but close to addressee) and ðiha ‘over there’ (i.e. far from both speaker and addressee).

Haut de page

Pour citer cet article

Référence papier

Abdelhak El Hankari, « Copulative Predication in Tarifit Berber », Corpus, 14 | 2015, 81-113.

Référence électronique

Abdelhak El Hankari, « Copulative Predication in Tarifit Berber », Corpus [En ligne], 14 | 2015, mis en ligne le 29 août 2017, consulté le 08 août 2020. URL :

Haut de page


Abdelhak El Hankari

The University of Newcastle, Australia

Articles du même auteur

Haut de page

Droits d’auteur

© Tous droits réservés

Haut de page
  • Logo Revues électroniques de l’université de Nice
  • OpenEdition Journals