OMISSION AND DOUBLING OF THE TEMPORAL AUXILIARY het IN THE AFRIKAANS OF AN AGRAMMATIC APHASIC

Two types of grammatical errors most frequently characterize the language of agrammatic aphasics, viz. the omission of syntactic and morphological elements, and incorrect word order. These phenomena are well-docwnented in several languages (Lapointe 1983a,b; Caramazza and Berndt 1985; Miceli and Caramazza 1988; Miceli et aI. 1989; Ouhalla 1993; Hagiwara 1995; Cahana-Amitay 1997; Friedmann and Grodzinsky 1997; Bastiaanse and Van Zonneveld 1998; De Roo 1999 inter alia). Data from the aphasic, PE, provide further evidence for both types of grammatical errors. Word order errors are not discussed in detail here. In this paper I focus specifically on irregularities in the production of het in past tense constructions.


University of Stellenbosch
In this paper, I examine both the inconsistent omission and the doubling of the past tense auxiliary het 'have' observed in spontaneous data obtained from an Afrikaans-.speaking agrammatic aphasic.)Two types of grammatical errors most frequently characterize the language of agrammatic aphasics, viz. the omission of syntactic and morphological elements, and incorrect word order.These phenomena are well-docwnented in several languages (Lapointe 1983a,b; Caramazza and Berndt 1985; Miceli and Caramazza 1988; Miceli   et aI.1989; Ouhalla 1993; Hagiwara 1995; Cahana-Amitay 1997; Friedmann and   Grodzinsky 1997; Bastiaanse and Van Zonneveld 1998; De Roo 1999 inter alia).
Data from the aphasic, PE, provide further evidence for both types of grammatical errors.Word order errors are not discussed in detail here.In this paper I focus specifically on irregularities in the production of het in past tense constructions.
Several hypotheses about the syntax of agrammatism have been proposed.Within the framework of generative grammar, agrammatism has been characterized as impairment of functional categories (FCs), for instance the impairment of Tense (T), Agreement (Agr) and Complementiser (C) systems (Ouhalla 1993;Hagiwara 1995;Friedmann and Grodzinsky 1997; Platzack in press\ It has been suggested that in agrammatism, functional heads are either missing because the grammatical knowledge is somehow lost, or that the grammatical knowledge is retained, but that the processes that use this knowledge are disrupted (Grodzinsky 1990;Ouhalla 1993).
More recently, it has been suggested that FCs are not entirely missing in aphasic agrammatism, but rather that they are selectively impaired, a proposal known as the "Tree Pruning Hypothesis" (Hagiwara 1995;Friedmann and Grodzinsky 1997).
Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 In this paper, I show that for the aphasic PE, impainnent in the verbal domain in Afrikaans manifests in irregularities in the Tense rather than the Agreement system.
More specifically, I show that there are irregularities in the production of auxiliaries, as evidenced by the omission and doubling of the temporal auxiliary he! in past tense constructions.What is particularly interesting is that these irregularities in the agrammatic version of standard Afrikaans reflect grammatical constructions found in other varieties of Afrikaans, synchronic as well as diachronic.It would seem then that these irregularities do not constitute "abnormal" language outside of the constraints of Universal Grammar (UG).
Investigating the irregularities in the production of auxiliaries in the Afrikaans aphasic data may lead to interesting comparisons with aphasic studies in other Gennanic languages.Further, it allows for speculation regarding the nature of syntactic problems observed in the language of aphasics.One such speculation is that these irregularities may reflect shifts in parameter settings rather than the selective deletion of or damage to FCs.
The analysis of the data in this study was conducted within the specific framework of the Minimalist Program (Chomsky 1995(Chomsky ,1998(Chomsky ,1999)), the most recent version of the generative theory of Universal Grammar.In this framework, movement results from the need to license features that are interpretable at the logical form (LF) and the phonetic fonn (PF) interfaces, and to eliminate those features that are not interpretable at the relevant interface.The data presented here suggest that there is at least some fonn of impairment in the subject's FCs, although the evidence does not support the Tree Pruning Hypothesis.I explore the possibility that the FCs are neither destroyed nor selectively pruned, but rather, that problems arise in the spelling out of FCs at PF.

2.
An analysis of the temporal auxiliary het in standard Afrikaans I assume that Afrikaans, like Dutch, is underlyingly an SOV language that exhibits the surface V2 phenomenon in main clauses  On the basis of these assumptions, the derivation of the sentence in (I)(b), dat hy gelag het, is as follows: i.
Merge the functional head AUX and the VP, hy gelag, to form AUXP. The DP hy is the external argument (or subject) of gelag.

II.
Merge the functional head PART and the AUXP to form PARTPI. Merge the functional head Complementiser (C) dat and Tp 2 , to form CP.
AUX is left in sentence-fInal position, and the relevant features associated with it are spelled out as het in Afrikaans.
The resulting structure is represented in (2).

C~2
~at DP~TPI The derivation of the sentence in (\)(c), Hy het gelag, is as follows: i.
Merge the functional head AUX and the VP, hy gelag, to form AUXP. The DP hy is the external argument of gelag. ii.
Merge the functional head PART and the AUXP to form PARTPI.   spontaneous sample occurred.However, the het is not phonetically realised or spelled out.The derivation of example (l.l), Hy dit gerun, is as follows: 1.
Merge the functional head AUX and the VP, dit gerun, to fonn AUXP, The OP hy is the external argument (or subject) of gerun. ii.
Merge the functional head PART and the AUXP to form PAR TP '.
iii vii.
Move the relevant tense features from AUX to T where they are interpreted; however, the het is not spelled out phonetically.
The resulting structure is represented in (4) below: In summary, then, in example (II.I) hel is not spelled out in the initial position under AUX, but at both landing sites under T and C. In example (II.2) hel is spelled out at T, and also at the foot of the chain in the initial position under AUX  As far as could be ascertained, the phenomenon of auxiliary doubling has not been systematically described in the literature on generative syntax: It is unclear, for example, under which conditions her may/must be phonetically spelled out.It is also not clear whether or not this phenomenon is indicative of parametric variation.The phenomenon of doubling poses interesting questions for the copy theory of movement, and the spelling out of traces.

S. Doubling of other syntadic elements
Apparent doubling of syntactic elements that do not change the interpretation of a sentence is not restricted to the past tense auxiliary her in Afrikaans.For example, this phenomenon is found in negation constructions in standard Afrikaans, as illustrated in IV.2 Die vet kos nou R4000 nOll vi,.'n kat wat no Die vet kos nou R4000 vi,.'n kat wat no personality her nie persoTUllity het nie the vet cost now R4000 now for a cat which the vet cost now R4000 for a cat which no no personality have not personality have not 'The vet now costs R4000 for a cat which has 'The vet now costs R4000 for a cat which has no no personality' personality'

Word order
As stated earlier, impairment in the verb phrase results in aberrant word orders.This has been shown in studies of agrarnmatic aphasia in other Germanic languages (Cabana-Amitay 1997; Bastiaanse and Van ZOIUleveld 1998; De Roo 1999 inter alia).
Incorrect word order was also observed in the aphasic speech of PE.
Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 Below are examples of aphasic utterances where there is V-to-I (or T) movement in a main clause, but not raising to C; the result is a construction in which the word order resembles that associated with an SVO language.
Type V : SVO-like word order  In conclusion, I argue that the term "agrammatic" is misleading since the language produced by the aphasic PE is not agrammatic although it is not standard Afrikaans.
Sometimes the Afrikaans produced by this aphasic differs from standard Afrikaans, which was the pre-stroke language, but the "irregular" structures that its containsspecifically, the doubling of syntactic elements -are nevertheless found in other varieties of Afrikaans as wel1. 14

NOTES
The term "agrammatic" is generally used to describe the grammatical phenomena typical of the language output of Broca's aphasics."Broca's aphasia" is the tenn traditionally used to describe the neurogenic condition that results from a cerebra-vascular insult in Broca's area of the brain (Grodzinsky 1990;Lesser and Milroy 1993).However, the involvement of Broca's area in Broca's aphasia is controversial, hence the use of the term "agrammatic aphasia" (Murdoch 1990).
These assumptions are also based on arguments provided by Robbers (1997).
xpl corresponds to what is generally referred to as X' (or X-bar) in the literature.
The term "standard Afrikaans" is used here to refer to the morphosyntactic and not the lexcial aspects.This applies to all the aphasic data presented below.
I assume that the relevant tense features of hel move to T but that its phonetic features stay behind in their initial position under the AUX.Chomsky (1995: 200-212) has suggested that the notion 'trace' be replaced by the notion 'copy'.In this paper the term "trace" is used throughout to refer, like the term "copy", to a remnant of the category that has been moved.
The item want is omitted from the analysis since its status is unclear: want could be C or a connective.

10.
Given the spelling out of traces illustrated in examples (11.1) and (11.2), it is conceivable that the following pattern could occur: (i) Hier het baie dinge het gebe1.lrhel here has many things has happened has 'Here many things have happened' An example from Malay Afrikaans provided by Kotze (\985) shows precisely this phenomenon: dat my overleede man nooyt meer als van de een Flaa!s zo hy heeft op that my late husband never more than of the one farm that he has on Ordinantie gehad heeft met zyn vee betrokken ordinance past-have with his cattle grazed 'that my late husband never had more than one farm on which he was allowed by ordinance to graze cattle' 12. See e.g.Oosthuizen (1998) and Nienaber (1965).
I was unable to find examples illustrating the omission of he! in the works consulted for this paper.Interestingly, howeyer, the omission of the past tense auxiliary is apparently found in some other Germanic languages, for example in embedded clauses in certain varieties of Swedish (Tarald Taraldsen, Anders Holmberg, lohan Rooryck -personal communications).
Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 3 .Verbs in Afrikaans do not overtly inflect for person, number or gender.Past tense is primarily encoded in the auxiliary he! 'have', as a separate lexical item, and there is morphological affixation of the past Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 participle prefix geto the main verb.Consider, for example, sentences (I)(a) to (c) below.Sentence (1)(a) illustrates the simple declarative, (1)(b) the simple past in a subordinate clause, and (1)(c) the past tense in a main clause where the auxiliary occurs in second position.

111 .
Move the whole VP to[Spec, PARTPI].The reason for this is to check the participle feature of PART via spec-head agreement.The result is P ARTP2.Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 iv.Merge the functional head Tense (T) and PARTp 2 to form Tpl.v.Move the DP hy to [Spec, TplJ to form TP2vi.
iii.Move the whole VP to[Spec, PARTp l ].The reason for this is to check the participle feature of PART via spec-head agreement.The result is PARTp 2IV.Merge the functional head Tense (T) and PARTp 2 to form Tpl.v.Move the DP hy to [Spec, Tpl] to form TP2. vi.Move the AUX to T where it is spelled out as het, leaving the trace under AUX.The resulting structure is represented in (3) below.Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49I 3.The agrammatic data Spontaneous samples were obtained from an agramrnatic aphasic, PE, who suffered a stroke in 1994.His auditory comprehension is largely intact.Prior to the stroke, he spoke standard Afrikaans 6 , and was resident in Cape Town.The examples of the type in (l)(a) to (c) above occur in his speech.In addition, the following examples from the spontaneous data have been selected for analysis: Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 Type I : Omission of het

4_
An analysis of Type I and Type II data 4_1 Type I : Omission of het In (Ll) to (1.4) of the aphasic data above, the obligatory temporal auxiliary her is omitted.I assume that the past tense is interpreted because of the morphological inflection geon the main verb, and also from the discourse context in which the Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 Thus, on the basis of the word order facts and the presence of ge-, I claim that past tense interpretation at LF has taken place, but that there is a problem at the PF interface resulting in het not being phonetical.!y realised or spelled out 7 • 4.2 Type n : Doubling of het In example (11.1) of the aphasic data, het is spelled out twice.The double het does not occur in standard Afrikaans.The derivation of example (11.1),Rier het baie dingf/ het gebeur, is as follows: i. Merge the functional head AUX and the VP, baie dinge gebeur, to form AUXP. The ADV hier is adjoined to form vp2. ii.Merge the functional head PART and the AUXP to form PARTPI. iii.Move the whole vp2 to [Spec, PARTp l ].The reason for this is to check the participle feature of PART via spec-head agreement.The result is PARW.iv.Merge the functional head T and P ARTP2 to form TPI. v. Move the DP, baie dinge, to [Spec, TPI] to form TP2. vi.Move the ADV hier to [Spec, Cpl] to form Cp2. vii.Move the relevant tense features from AUX to T, and from T to C. The features are interpreted at both landing sites, and phonetically spelled out as het at T and at C. The resulting structure may be represented as in (5).The het is spelled out in the second structural position, as weI.! as in one of the trace positions in the chain. 8Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0example (IL2) of the aphasic data., Want sulke goed soos wasmasjien het gebreek het, is as follows: 9 i.Merge the functional head AUX and the VP, sulke dinge soos wasmasjien gebreek, to form AUXP. ii.Merge the functional head PART and the AUXP to form PARTPI. iii.Move the whole VP to [Spec, PARTp l ].The reason for this is to check the pariciple feature of PART via spec-head agreement.The result is PARTP2.iv.Merge the functional head T and PARTp 2 to form TPI. v. Move the DP, sulke dinge soos wasmasjien, to [Spec, TPI] to form TP2vi.Move het from its initial position under the AUX to T, where it is phonetically spelled out.In this example, the trace of het under the AUX is also spelled out, which results in the doubling of het.The resulting structure is represented in (6).Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0 God alone know what have there happened have 'God alone knows what happened there' (from a report on the SAFM radio news programme" AM live", 29/4/1998) Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49

( 9 )
.12 Other varieties of Afrikaans, among others Cape Afrikaans and Malay Afrikaans, show doubling of prepositions, as illustrated in (10) and (1 I).and doubling of the copular verb is, as illustrated in (12)Y (9) Ek sal dit nie doen nie I will it not do not 'I will not do it' (10) Met wie her jy mee gesels?with whom have you with talked 'With whom did you talk?' (11) Hulle het teruggelwm uit Holland uit.they have back+past+come from Holland from 'They came back from Holland' (Kotze 1985) (12) Ek hoor ook dar da veele ouwers is war tevreede is met so e school is I hear also that there many parents are who satisfied are with such a school are 'I also hear that there are many parents who are satisfied with such a school' (Adhikari (ed.) 1996: 96) Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49I 60 As illustrated below, doubling of syntactic elements other than the past tense auxiliary het is also found in PE's spontaneous language; these instances of doubling do not occur in standard Afrikaans.The first of these is the doubling of the auxiliary is in passive constructions, as in the following example: Type III : Doubling of is APHASIC DATA STANDARD AFRIKAANS III.! Nou's my kombuis is nou updated Nou's my !cnmbuis updated now is my kitchen is now updated now is my kitchen updated 'Now my kitchen is updated' 'Now my kitchen is updated'The second of the doubling phenomena in the aphasic data involves adverbs, as in the Nou's my kombuis is nou updated Nou's my kombuis updated now is my kitchen is now updated now is my kitchen updated 'Now my kitchen is updated' 'Now my kitchen is updated' patterns observed in the production of the temporal auxiliary het in past tense constructions in the Afrikaans produced by the aphasic PE, previously a speaker of standard Afrikaans.The first of these is the correct production of the auxiliary as illustrated in examples (1)(a) to (c).The second is the intennittent omission of the obligatory het where het is semantically interpreted, but not spelled out phonetically.The third is the doubling of het which is asswned to be the result of the spelling out offeatures at some or all of the landing sites, including the head and the tail of chains.The second and third of these patterns seem to indicate that the affected individual has problems with the spelling out at PF of temporal auxiliariesthere is either no spell-out or there is excessive speH-out at more than one position in the chain.The aphasic data presented here provides independent evidence for verb movement that can be separated from the phonetic realisation of features.It appears as though interpretation and spell-out are two different processes.In the examples where het is Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 omitted in the aphasic speech' of PE, spell-out at PF may be impaired although interpretation at LF occurs.This may provide support for the independence of the PF and LF interfaces and for the claim that interpretation at these interfaces are separate processes.It is also clear that the theoretical framework can account for the aphasic data and that the language produced by the aphasic PE is constrained by the principles of UO.The patterns observed in the aphasic data are not random as regards the positions where spell-out is possible: the temporal auxiliary can only be spelled out in a possible head position, tail position or an intermediate landing site of a chain.What is unclear is under what conditions spell-out may/must occur in one or more of these positions.
(ii) Dan vertel die man nou waf het gebe1.lrhet no1.l daai dag het then tell the man now what have happened have now that day have 'Then the man told what happened that day' II.See also the example from Malay Afrikaans in note 10.The phenomenon of past tense auxiliary doubling is also found in older varieties of Afrikaans, as in Stellenbosch Papers in Linguistics, Vol.33, 2000, 49-66 doi: 10.5774/33-0-49 the following example from a letter written in 1783 (Van Oordt, L.e. Die Kaapse Taalargief Letter 65): Ek nog nie Irons gehad hierdie jaar nie Ek hel nog nie kans gehod hierdie jaar nie I yet not chance past+have this year not I have yet not chance past+have this year not 'I have not yet had a chance this year' 'I have not yet had a chance this year' Type U : Doubling of hel . Move the whole VP to [Spec, PARTP'].The reason for this is to check the participle feature ofPART via spec-head agreement.The result is PARTP2.