THE FINAL NIE IN AFRIKAANS NEGATIVE SENTENCES*

An interesting syntactic property of Afrikaans is the use of the so-called double negative in sentences which express a negative proposition by means of a negation word like geen ('no', 'none', 'not (any)'); geeneen ('no one'); geensins ('by no means', 'in no way'); g'n ('never', 'not'); nerens ('nowhere'); nie ('not'); niemand ('nobody'); niks ('nothing'); nooil ('never'). In terms of this property such a negative sentence contains, as a general rule, a second 'negation word' in final position, viz. the item nie. The phenomenon may be illustrated with the foUowing examples. l

In terms of this property such a negative sentence contains, as a general rule, a second 'negation word' in final position, viz. the item nie.The phenomenon may be illustrated with the foUowing examples.l (1) Subject initial main clauses (a) Sy sluil nooit die deur NJE she locks never the door not 'She never locks the door' (b) Bulle is g'n so arm NJE they are not so poor not 'They aren't all that poor' (2) Subordinate clauses With the exception of a few well-known works such as (Klima 1964), (Kraak 1966) and (Jackendoff 1972), the description of sentential negation received relatively little attention in the early versions of Chomskyan generative grammar.And it was not until the publication of especially (pollock 1989) that the syntax of negation became a topic of serious research within the Principles and Parameters framework; this renewed interest is still evident in the Minimalist Program, the most recent development within the Principles and Parameters model. 2 As far as negation in Afrikaans is concerned, (Waher 1978) still represents the only detailed description within the broad generative approach. 3  It is not the aim of this paper to give an overview of the various proposals in the literature regarding the description of sentential negation.Also, no attempt will be made to give either a detailed description of sentential negation in Afrikaans or a comparative analysis of this phenomenon in Afrikaans and any other language.The aim of the paper is much more modest: it will examine the possibilities which the Minimalist Program presents for the syntactic description of the final nie in Afrikaans negative sentences, and more specifically subject initial clauses like those in (I) and (2).The discussion will focus on two general questions: (a) what is the categorial status of the final nie, and (b) where and how is it generated in sentence structure?The rest of the paper is organised as follows.By way of background, section 2 gives a brief overview of some of the relevant assumptions and mechanisms of the Minimalist Program.In section 3 various possible descriptions of the final nie are critically examined, after which an analysis is outlined which appears not only to express the relevant facts, but to be compatible also with the assumptions and mechanisms of the Minimalist Program.In section 4 a brief summary is given of the major findings, and some potential pFoblems are also noted for further investigation.

Some minimalist assumptions and mecbanisms
The organisation of the grammar within the Minimalist Program may be represented schematically as in (6) below. 4Spell-Out in (6) marks the (arbitrary) point at which the derivation of a sentence is split into two separate parts, respectively yielding its PF (sound) and LF (meaning) representations.Operations which take place before SpeU:.outform part of the overt syntax and are reflected in the perceptible PF representation of the sentence.Operations

(6) LEXICON
COMPUTATIONAL SYSTEM -J, overt operations which take place after Spell-Out, and which lead to LF, form part of the covert syntax, hence their effects are not phonologically visible.
The derivation of a sentence starts with the selection of substantive items from the lexicon, each item consisting of a set of features.Two general types of features are distinguished, viz.lerical-categorial (Le) features and formal (F) features, which are interpreted at the two interface levels PF and LF.LC-features consist of semantic features; categorial features like [nominal], [verbal]; and (presumably) phonological features.! The F-features of substantive items relate to, amongst others, morphological properties such as tense, case and agreement (person, number, gender), and are each selected with a particular value (+1-).The substantive categories are selected independently of each other, and are subsequently projected and merged with one another through the operations of the Generalised Transformation, the only structure-building mechanism within the Minimalist Program.
For a well-formed sentence to be derived, the various substantive items must be licensed for interpretation at the PF and LF levels.To this end a further system of junctional categories is postulated, generated above and to the left of the system of substantive categories.Functional categories include, amongst others, AgrS (subject agreement), AgrO (object agreement), and T (Tense), each consisting of a set ofF-features, the same F-features that are associated with the substantive categories.In contrast to those of the substantive categories, however, the Ffeatures of the functional categories are not specified for particular values.Licensing of a substantive category is then effected by moving its F-features upwards and to the left, into positions where they can be checked against the corresponding features of a functional category.During checking the F-feature of a substantive category supplies a particular feature value to the relevant F -feature of a functional category, provided that the two categories (or more specifically, their F-features) are structural sisters.The movement of F-features is effected by Move-F, one of the operations of the Generalised Transformation (GT).
The F-features of the functional categories belong to two types, viz.V-features and N-features.
A V-feature, on the one hand, must agree with the corresponding F-feature of a substantive head.Since feature checking is only possible in a sisterhood relationship, it follows that the Ffeature of the substantive head must be adjoined to the relevant functional head.N-features, on the other hand, must agree with the corresponding features associated with phrases.In this case the relevant phrase moves to the Specifier (Spec) position of the functional category X so that it forms the sister of the first projection xpl above the functional head. 6Give~ that the features of a functional head X are, via percolation; also available at its projections, hence also at xpl, checking can proceed in accordance with the sisterhood condition.In short, V-features are checked in head-head configurations, and N-features in Spec-head configurations.
Two further assumptions regarding F-features should be noted here.The first concerns the question of feature strength.The F-features (i.e. both the V~ and N-features) of ajunctional head may be either strong or weak, with the possibility of parametric variation between languages.Zwart (1997) proposes, for example, that the V-and N-features of Agr are both strong in Dutch, a proposal that will be accepted for Afrikaans as well.'Strong features must be checked (i.e.supplied with a value) in the overt syntax, before Spell-Out, otherwise the derivation will crash at PF.The checking of weak features, by contrast, can be 'postponed' until the covert syntax, after the point of Spell-Out; weak features are not visible at PF and can occur unchecked at that level.
The second assumption concerns interpretation at the PF level.To be interpreted (i.e.lexically realised) at PF, the F-features of a substantive head must be combined with LC-features within a categorial head.Suppose for instance that the F-features of a substantive head Yare adjoined to a functional head X to form the morphosyntactic complex XI If Xl does not contain any LC-features, it will not constitute a legitimate (interpretable) PF-object, which means that it will not be lexically realised.In such a case one of the following operations may be performed.
(I) If there is a further functional head Z higher up in the structure which does contain LCfeatures, Xl can be adjoined to Z by means of Move-F to form the morpho syntactic complex Zl ZI will then qualify as an interpretable PF-object, since it contains both Fand LC-features.The operation is only pennissible, however, if it will result in an Ffeature of Z being supplied with a value; in other words, Z will attract Xl only if Z can gain an F-feature value in the process.
(II) Suppose there is not an appropriate functional head Z higher up in the structure.In such a case the LC-features of the substantive head Y must be moved overtly to Xl --i.e.before Spell-Ou~ --just like its F-features. to form a lelritimate PF-obiect.This is effected by Move-LC, a futher movement operation of GT.It should be noted, though, that the overt movement of LC-features represents a 'costly' operation in terms of grammatical .computation, one that is available only as a 'last resort'.
The preceding overview of minimalist assumptions and mechanisms can be made concrete with reference to the examples in ( 7) and ( 8).The structure underlying the embedded sentence in (7) can be represented roughly as in (9) below8 This structure has been derived by Lexical Selection and the GT-operations Project and Merge.The subject .ry. and the direct object die deur in (8) have already been moved overtly to [Spec, AgrS] and [Spec, AgrO), respectively; these two operations are required so that the strong N-features of AgrS and AgrO can be checked before Spell-Out.Notice that (9) exhibits an initial SVO-order (or in more general tenns, Spec-head-complement), the only underlying word order that is provided for within the Minimalist Program. 9

NP
Given that the V-features of Agr (and probably those of T as well) are strong in Afrikaans, it follows that the F-features of the V sluit --indicated as F(v) in ( 9) -must also move before Spell-Out.This involves at least three operations.Firstly, F(v) is adjoined to AgrO, yielding the two-segment category AgrO l with F(v) and AgrO as its daughter-constituents; the strong V.features of AgrO can be checked in this configuration.Secondly.AgrOl is adjoined to T to fonn the two-segment category Tl with AgrOl and T as its daughters, a configuration in which the V -feature of T can be checked.Thirdly, Tl is adjoined to AgrS yielding the two-segment category AgrSl, which provides the configuration for checking the strong V-features of AgrS.
The problem, however, is that AgrS} does not contain any LC-features, which means that it does not constitute an interpretable object at PF.This problem can be overcome in terms of the proposal (1) above, that is, by adjoining AgrS 1 to the functional head C, a further instance of Move-F.C obviously contains LC-features, since it can be spelled out as the complementiser dal.Hence, by adjoining AgrSI to C, a two-segment category C I can be fonned in which F(v) is combined with LC-features.The important question, of course, is whether C has anything to gain by such an operation, that is, whether C contains a V -feature which can be supplied with a value via AgrSI-to-c.Zwart (1997) argues on the basis of agreement facts that C must indeed •contain such a V-feature: in various dialects ofDutcb, Frisian and Gennan the complementiser agrees in person andlor number with the subject and the inflected verb, a phenomenon which can only be accounted for in tenns of Move-F to c. IO Given Zwarts' analysis, C thus attracts AgrSI in order to obtain an F-feature value; as a consequence, a legitimate PF-object --the two-segment head C I --is created, one which contains both F-and LC-features.
The effect of the four overt operations involving F(v) can be illustrated in the structure (10) below.Since the LC-features of the V sJuil --indicated as LC(v) --do not have to be moved in the overt syntax, the V is spelled out in its initial position under the VP; (10) thus reflects the surface SOy word order of the embedded sentence in (7).11 Consider next the main clause in (8).The structure underlying ( 8) is almost identical to that presented as (9) above for the embedded sentence in (7).The only difference is that (9), in the case of the main clause (8), does not contain a CP dominating AgrSP2, since (8) is not introduced by a complementiser (e.g.dal).In short.the embedded sentence in (7) represents a CP, and the main clause in (8) an AgrSP.As in the case of ( 7). the F-features of the V sJuil in (8) must be moved overtly so that the strong V -features of Agr and T can be checked before /"--.... I~ /'-....
Spell-Out.Three operations are involved in this, viz.(i) F(v)-to-AgrO, (ii) AgrOl_to_ T and (iii) T'-to-AgrS, with F(v) eventually forming part of the two-segment category AgrSI.However, AgrS I does not contain any LC-features, which means that it does not constitute a legitimate PF-object.The derived structure moreover lacks a higher functional category with LC-features --like C, as in the case of the embedded sentence (7) --to which AgrSI can be adjoined by means of Move-F.Since F(v) cannot be interpreted without LC-features, the derivation will therefore crash at PF.The only solution to this problem is to adjoin the LC-features of the V sluit to AgrS I before Spell-Out (cf the last resort proposal (II) above).This will then yield the two-segment category AgrS2, a legitimate PF-object containing both F-and LC-features.One of the consequences of moving the LC-features of the V sluit to AgrS in the overt syntax is that sluit will be lexically spelled out in the second structural position of the sentence.This is in accordance with the surface SVO word order of the main clause (8).The structure (11) below illustrates the various overt operations involving F(v) and LC(v) in the derivation of ( 8).The examples in ( 7) and ( 8) are both positive declarative sentences.In the next section we will examine whether the assumptions and mechanisms outlined above can provide a framework for the description of negative declarative sentences in Afrikaans, and more specifically of the structural position of the final nie.

3.
Negative sentences and the syntax ofthe final nie Dit hlyk riat 5)' hykans 1/00it die geTeg mook nie it seems that she almost never the dish makes not 'It seems that she almost never makes the dish' As mentioned above, it is not the aim of this paper to give a detailed analysis of the syntax of sentential negation in Afrikaans.The following assumptions about the selection and licensing of the relevant negation words, for example, will be accepted here without further discussion. ( Negation words are selected in the form ofF-and LC-features.
A negation word contains an F-feature [+ neg] that has to be checked against the corresponding F-feature of a Junctional head Neg, where checking implies that the functional head is supplied with an F-feature value. (c) The functional head Neg has a strong N1eatuTe; this means that a substantive phrase with the F-feature [+ neg] must be moved to [Spec, Neg] before Spell-Out, where feature checking can then take place in a Spec-head configuration.

(d)
The functional head Neg occurs in a structural position between T and AgrO.13 These assumptions can be made concrete with reference to the embedded sentence in (15).The structure underlying (l 5) may be represented roughly as in ( 16).In this structure the subject .1)1 and the direct object die deur have already been moved overtly to the Specifier positions of AgrS and AgrO, respectively; for ease of exposition the overt movements involving the Ffeatures of the V sluit are not indicated (cf structure (10) above).The AP nooit is represented in ( 16) as an adverbial that is adjoined to the VP; the final nie is ignored for present purposes (17) dal.sy die deur nooil sluil nie It is not clear exactly how the word order in a sentence like (17) can be accounted for.One possibility might be that Neg does not have a fixed hierarchical position between T and AgIO, but that it can also be projected between AgrO and VP.The word order difference between sentences like those in (IS) and ( 17) could then be ascribed to the variable position of Neg.
Another possibility might be to postulate a further functional category above Neg in structures of the type ( 16), one to which the object could be moved overtly to derive the word order in (17).14The merits of these suggestions, and for that matter the whole question of word order variation in Afrikaans negative sentences, fall outside the scope of this paper and wiD not be investigated further here.
This brings us to the analysis of the final nie in negative sentences (henceforth, NIE it seems that she never the dish makes (•nearly) not 'It seems that she never makes the dish' Apparently, then, NTE is not a substantive item, which leaves only one other possibility, viz that it represents a functional item.Given this conclusion, the question arises to which specific functional category NlE belongs.An obvious proposal would be to analyse NTE as a phonetic realisation of the functional category Neg, that is, as the head ofNegP in a structure like (16).
On this proposal the structure underlying the embedded sentence in (15) will take roughly the fonn in (20).The subject sy, the object die deur, and the adverbial nooit have already been moved overtly in (20); the various overt movements involving F(v), i.e. the F-features of the V s/uit, are not indicated.V NP However, an analysis along the lines in (20), with NIE representing the head of NegP, is problematic in various respects.Firstly, NIE does not appear in sentence-final position, but to the left of the direct object die deur and the V s/uit.Since only leftward, upward movements are provided for within the Minimalist Program, NIE cannot be moved to the right into final position.To derive the surface word order of (15), both the V and the object would therefore have to move overtly to positions to the left of NIE.However, it is not at all clear which positions, if any, could serve as possible landing sites for these two constituents.In the case of the object die deur, (20) does not contain an apparent Specifier position to which the NP could be moved; and even if such a position were postulated, it is not clear how/whether the relevant operation could be motivated in terms of strong N-feature checking.In the case of the V sluit, it could be argued that both F(v) and LC(v) are adjoined to Neg.But this then raises the question why LC(v) has to be moved overtly.This cannot be because F(v) will be stranded in Neg without LC-features: Neg already contains such features (spelled out as NIB), and F(v) will in any case be combined in the course of the derivation with the LC-features in C (cf. the representation in (1 0) above).
A second problem with the analysis in ( 20  It thus seems reasonable to take NIEl (like the sentence-final nie) to be a functional item, rather than a substantive item like the negation words nooit, niemand, nie, etc.Within the Minimalist Program a functional category is projected only if it is necessary for the licensing of a substantive item X, in other words, if it can be checked against a particular F-feature of X.
Each F(x) will thus result in the projection of one, and only one, corresponding functional category.This explains, for example, why a structure with only one finite, intransitive verb cannot have more than one AgrS or T. Consider now the assumption ( 14   A similar problem is found with sentences like those in (24) below, each of which contains a NIE without any accompanying negation word.Here, too, the non-occurrence of a negation word implies the non-occurrence of Neg, leaving NIE without a structural position.It should be noted, however, that the acceptability judgements of native speakers vary considerably with regard to sentences of the type in (24); some speakers find such sentences unacceptable or at best marginally acceptable.IS derivation of an embedded sentence like the one in (IS), dat ~ nooil die deur sluit NIE.The structure underlying (15) would then roughly take the form in (25).In this structure the subject ~, the object die deur and the adverbial nooit have already been moved overtly; the movement operations involving F(v) are not indicated. ( In structural terms the VP in (25) represents the complement of the functional head X.Notice that the VP forms a left-complement of X, thereby expressing the fact that NIE occupies the sentence-final position in the surface word order.In section 2, however, it was pointed out that the Minimalist Program provides for only one general, underlying word order, viz.Spec-headcomplement.Thus, even though the complement-head order in the structure (25) can describe the sentence-final position of NIE in (15), it is unacceptable on general theoretical grounds, which raises serious doubts about the merit of an analysis along the lines in (25).1l We have now examined various proposals for the analysis of NIB, and it was argued in each case that there are empirical and/or theoretical considerations which reflect negatively on the merit of the proposal.We tum now to a possible alternative which is apparently not subject to the same objections.The central hypotheses of this proposal may be formulated as follows: (26)(a) NIB is the phonetic realisation of a functional head which can be indicated with the category label Pol(arity).
(b) Pol is projected as the topmost functional category in sentence structure, e.g.
above AgrSP in sentence-initial main clauses and above CP in embedded sentences.
(c) Pol is projected in accordance with the universal underlying word order postulated within the Minimalist Program, viz.Spec-head-complement; in other words, Pol is projected upwards and to the left in sentence structure, taking its relevant sisterconstituent (e.g.AgrSP or CP) as a right-complement.
In tenns of these hypotheses, the structure underlying the embedded sentence in (l 5) may be represented as in ( 27) below.The subject sy, the object die deur and the adverbial nooit have already been moved overtly; the movement operations involving F(v) are not indicated (cf the representation in (1 0) above).The question now is how the surface word order of (15), with N1E in sentence-final position, can be derived from the underlying structure (27).Within the Minimalist Program there seems to be only one possibility, viz. to move the entire CP overtly to the specifIer position of Pol.
Overt movement of a phrase yP to the Spec of a functional head X is only possible, however, if yP contains a particular F-feature that must be checked before Spell-Out against the corresponding strong N-jealure of X.In other words, X will attract yP only if X can gain a feature value in the process.IfCP-to-Pol is the only way in which N1E can end up in sentencefinal position, it thus follows that CP contains an F-feature that must be checked against the corresponding F(= N)-feature of Pol.And since the operation is an overt one, it also follows that the relevant feature of Pol must be strong.Given these conclusions, the obvious question is which F-feature is involved in CP-to-Pof.Apparently such a feature cannot be associated with one of the substantive phrases in (27), i.e. with the VP, the subject.ry, the object die deur, or the AP nooi!.If this were the case, Pol would simply attract one of these phrases, and NlE would not end up in sentence-final position.The feature also cannot be exclusively associated with C (and via percolation, with CP) since Pol, according to the hypothesis (26)(b), is postulated in subject-initial main clauses as wel~ which lack a (C)P.This then suggests the following working hypothesis, in terms of which the relevant feature is associated with V: (28) Verbs have an F-feature [ poll with a particular value in 'negative polarity' sentences (e.g.sentences containing negation words like niemand, nooil, nie, etc.); this feature is checked against the corresponding strong N-jea/ure of the functional category Pol, which means that the N-feature is supplied with a value.18 Given (28), it could be argued that [ poll forms part of the F -features, F(v), of the V sluit in (27) (though cf.note 18).It was explained in section 2 that F(v) is involved in at least four overt movement operations in such an embedded sentence structure, viz.(i) F(v)-to-AgrO, (ii) AgrOI-to-T, (iii) TI-to-AgrS, and (iv) AgrSI_to~C (cf. the representation in (10)).In the case of( 27) the effect of these operations may be illustrated as follows: /"'--..
The last operation in (29), i.e.AgrSl_to-C, results in F(v) being incorporated into the complex.
functional head C i F(v) is furthermore available via percolation at the projections ofC i , in this case the maximal projection CPo In terms of the hypothesis (28) [ pol] is associated with F(v); at this stage of the derivation it is also the only F-feature that is still unchecked.Hence, since the CP contains an F-feature that can supply a value to the corresponding strong N-feature of the functional head Pol, it follows that the CP is attracted to [Spec, Pol] in the overt syntax.In this way, then, the surface word order of the embedded sentence in (15) can be derived, with NIB in sentence-final position.
The above analysis of embedded negative sentences like the one in (15) also holds for subjectinitial main clauses with NIE, except of course that such main clauses do not contain a CPo Consider again the example in (1 )(a), repeated here as (30).In the derivation of this sentence the F-features of the V are moved to the functional head AgrS via at least three operations, viz.
F(v)-to-AgrO, AgrOl_to_ T, and T1-to-AgrS (31) illustrates the effect of these operations.Oosthuizen 1996 for an application of this proposal in the analysis of word order phenomena in Afrikaans wh-questions.It is also generally accepted in the literature that NP forms part of the functional category DP (Determiner Phrase).For ease of exposition these two proposals will not be implemented here.

9.
See Kayne 1993;Chomsky 1994: 25-31;Bennis 1995. 10. See Zwart 1997: 136-146,204-205. 11 15.Although the examples in (24) do not contain any negation words, they all entail negative expressions.For example, (c) implies that he hasn't got much experience and (d) that I will not help you.In each case the negative entailment can be traced to a specific item, viz.onmoontiik, nou/iles, weinig, vemiet, kwaiik, and skaars, respectively.For ease of reference these items may be termed 'negative-entailment items' (or 'entailmentreversing items').The differences/similarities between negative-entailment items and negation words such as nooit, nie, niles, geen, etc. will not be investigated in this paper.See e.g.Hoeksema (n.d.) and the references cited there for discussion of entailmentreversing items in Dutch and English.
16.A functional category like DP (see note 8) can of course occur to the right under the VP as the complement of a transitive V.But such a structure is the result of Merge, whereas the type of structure under discussion concerns the projection of a functional category on the basis of the F-feature composition of a V (or possibly a negation word).DP is projected on the basis of the F-features of a substantive category lower down in the structure (i.e.contained within the OP); after which the OP is merged with a categorial head, e.g.V.
17.This conclusion follows irrespective of where in the structure X is projected, that is, it also holds if X appears above T(P), Neg(P), Agr(P) or C(P).
18.This hypothesis raises two important questions.First, which substantive category serves as the initial 'source' of the F-feature [ pol], in other words, as part of which category's feature composition is [ poll introduced into the computational system?One possibility might be that [ pol] represents an F(v), so that it enters the derivation via the lexical selection of a verb; (28) is formulated in terms of this possibility.Another possibility might be for [ pol] to be introduced via the• selection of a negation word (or a negativeentailment item like nou/iks, weinig, kwa/ik, etc. as in (24); see note 15); [ pol] could then be 'picked up' at the appropriate functional category during overt head-to-head movement ofF(v), e.g. by adjoining F(v) to the functional head Neg in the course of the derivation.Although [ pol] is referred to as an F-feature that is associated with verbs in (28) and in the rest of the discussion (Le.the first possibility mentioned above), this is done purely for the sake of convenience.The essential aspects of the analysis below would also hold if [ pol] is combined with F(v) at a later stage of the derivation (as expressed by the second possibility).The second question raised by the hypothesis (28) concerns the exact content of the notion 'negative polarity sentence' on the one hand, and the relationship, if any, between sentential negation and 'negative polarity' on the other hand.These questions will not be investigated further here.

( a )
Jan beweer dot hy niks onlhou NJE lohn claims that he nothing remembers not 'lohn claims that he remembers nothing' (b) Ek twyfel of Imlle regtig geen kontant het NIE I doubt whether they really no cash have not 'I doubt whether they really have no cash' (3) Topicalisation constructions (a) Nerem voel ,ry vei/ig NIE nowhere feels she safe not 'Nowhere does she feel safe' (b) Met niemand anders het dit gebeur NIE with nobody else has it happened not 'It happened to nobody else' (4) Interrogative constructions (a) Wiljy dan geensins hetrokke raak NIE? want you then in no way involved become not 'Don't you want to become involved at all?' (b) Wie hel nie opgedang NIE? who has not arrived not? 'Who hasn't arrived?' (5) Imperative constructions (a) Moet geeneen vertrou NIE must no-one trust not 'Don't trust anyone' (b) Moenie dit doen NJE' must-not it do not 'Don't do it!'
Ek sal jou noDi! (*glad) NIEI vergeet NIE2 Niemand anders (*hykans) NIEI het opgedaag NIE2Di!blyk dat sy niks (*absoluut) NIEI /can onthou NIE2 )(b) above.In terms of (14)(b) the functional category Neg is projected only if the structure contains a substantive category with a [+ neg] F-feature.It furthermore follows from the general assumption about the occurrence of functional categories that the selection of such a substantive category cannot result in the projection of more than one Neg.The problem which sentences like those in (21)pose for the analysis in (20) should now be obvious: each of these sentences contains two functional items NIE, which implies that the structure must contain two functional categories Neg.In fact, though, each sentence contains only one substantive item with a [+ neg] feature, which means that only one Neg can be projected.IfNIEl is taken to be the phonetic realisation of the single projected Neg, NIE2 would thus be left without a structural position, and vice versa.A third problem for the analysis in (20) concerns sentences like the foUowing: Ek was nie seker NlE of jy ook wou saamkom NIE I was not sure not whether you also wanted-to along-come not 'I wasn't sure whether you would have liked to come along as well' The sentences in (23) each contain two NIE's, one in the main clause and one in the subordinate clause.If NIB is the phonetic realisation of the functional category Neg, it follows that every main and subordinate clause in these examples must contain a functional head Neg.In terms of the assumption (14)(b), however, a Neg can only be projected if the relevant structure contains a substantive item with the F-feature [+ neg].The main clauses in (23) each contain such an item (viz.the negation words geen, nie, niks), but not the subordinate clauses.Since the non-occurrence of a negation word implies the non-occurrence of Neg, the final nie in the subordinate clauses in (23) is therefore left without a structural position.
19.As shown by the examples in (i), sentences like those in (35) can also occur with afinal NIE, that is, with two NIB's (see also (23) for similar examples).It is not clear whether/ how this phenomenon relates to the problem under discussion.(i)(a) Niemand is afgestuur NIE in daardie wedstryd NIE (b) Hy wi/ nerens heen gaan NIE in die aand NIE ( c) Ek kon nie 'n op/ossing lay NIE vir die prob/eem NIE ) concerns sentences containing two NIE's, one associated with sentential negation (NIE2), and the other with constituent negation (NIEI; cf.
secondly, NIE! cannot be modified: Baie mense sal nie die grondwet lees NIE as dit net in een taal geskryf is NIE many people will not the constitution read not if it only in one language written is not 'Many people won't read the constitution ifit is only written in one language' Sy wil niks doen NlE as jy vir hoar goon lag NlE she will nothing do not if you for her go laugh not 'She doesn't want to do anything if you're going to laugh at her' •'It's not clear whether he will come' (c) Zwart 1994for the proposal that C(P) is split into the two functional categories Wh(P) and Top(P); see Ek kon me 'n oplossing kry NIE vir die probleemSy het nih gese NIE op die vergaderingNiemanti is beseer NIE in die aanvalThe examples in (35) each have an 'extraposed' PP in sentence-final position, which is clearly probJematical for the proposed analysis of NIE (cf the hypotheses in (32)(b, C».19This problem, like the questions in (33), is left here as a topic for further investigation.It should however be noted that the problem posed by the examples in (35) is not confined to negative sentences.As the examples in (36) show, PP's can apparently occur freely in different structural positions in positive sentences as well.As far as could be ascertained, a proper account of this phenomenon has not yet been presented within the Minimalist Program, at least . It is possible in non-standard varieties of Afrikaans (including Colloquial Afrikaans) for an embedded sentence that is introduced by an overt complementiser to have the surface SVO order associated with subject initial main clauses.For examples and discussion see Robbers 1997 and the references cited there.It is not clear exactly how this phenomenon can be accounted for within the minimalist framework outlined in this section.See Zwart  1997: 234-241for an analysis of similar embedded verb second phenomena in Yiddish, Icelandic, Mainland Scandinavian and Frisian. .A possibility that could be considered here is to postulate a functional category Focus above AgrOP, which overtly attracts a substantive category with the F-feature [+ focus].See Langer 1995 for an analysis of 'scrambling' phenomena in German in terms of the feature [ focus ].
12. A sentence like (12)(a) Sy siuit (nooit) die deur nie is acceptable without the negation word (Le. the time adverbial nooit), but then the nie functions as a negation word, with the sentence expressing a meaning that can be paraphrased roughly as 'She definitely doesn't lock the door', that is, without a time indication.13.See e.g.Pollock 1989.14