D ESCRIPTION OF WH-CONSTRUCTIONS AND PP IN W ELSH
2.2.3 Movement in object position
This subsection considers wh-constructions where a variable is identified in object position of a verb. As already shown in 1.1, Welsh has two verbal constructions as in the other Celtic languages: synthetic and periphrastic constructions (see Borsley et al. 2007: 38). I will first consider object wh-constructions where the verb is synthetic, then turn to the case of periphrastic verbs.
Examples of synthetic verbs are given below, with a relative clause in (17) and a wh- question in (18). The complementizer aS is used in Literary Welsh.
(17) y ffrwydrad (a) glywais i __ wedyn the explosion C hear.PAST.1S I then
‘the explosion that I heard then’ (Borsley et al. 2007: 119) (18) Beth (a) glywaist ti __ wedyn?
what C hear.PAST.2S you then
‘What did you hear then?’ (Borsley et al. 2007: 106)
A verb may show agreement with a subject but not with an object in Welsh (i.e. an accusative language), so there is no way to check the agreement effect in this context. However, overt pronouns are unavailable in this position.17
16
Borsley et al. (2007: 141-44) point out that the resumptive strategy may also be used in embedded subject wh-constructions. They suggest that both movement and resumptive strategies are possible in embedded clauses, although the movement strategy is more usual in colloquial Welsh. See also Willis (2000: 553-57) for more detailed discussion.
17
48 (19) * y ffrwydrad (a) glywais i e wedyn
the explosion C hear.PAST.1S I it then
‘the explosion that I heard then’ (Borsley et al. 2007: 119) (20) * Beth (a) glywaist ti e wedyn?
what C hear.PAST.2S you it then
‘What did you hear (it) then?’ (Borsley et al. 2007: 108)
The unavailability of resumptive pronouns suggests that a trace left by movement is posited in object position of synthetic (i.e. finite) verbs.
We now turn to object wh-constructions of periphrastic verbs. In the literary language, an object-agreement clitic precedes the verb, as illustrated in (21a). This clitic may be dropped in speech, but it still triggers mutation on the non-finite verb (dwyn > ddwyn), as in (21b). In the most colloquial variety, both object-agreement clitic and its mutation are absent, as in (21c).
(21) a. y car mae ’r lladron wedi ei ddwyn __ Lit. W. the car be.PRES.1S the thieves PERF 3MS steal.INF
b. y car mae ’r lladron wedi ddwyn __ the car be.PRES.1S the thieves PERF steal.INF
c. y car mae ’r lladron wedi dwyn __ Col. W. the car be.PRES.1S the thieves PERF steal.INF
‘the car that the thieves have stolen’ (Borsley et al. 2007: 120)
This stylistic variation is essentially the same in wh-questions.
(22) a. Beth ydych chi ’n ei fwyta __? Lit. W. what be.PRES.2P you PROG 3MS eat.INF
b. Beth ydych chi ’n fwyta __? what be.PRES.2P you PROG eat.INF
c. Be’ {’dych / ’dach} chi ’n bwyta __? Col. W. what be.PRES.2P you PROG eat.INF
49 In Literary Welsh, a complementizer may be used in front of auxiliary verbs in both relatives and wh-questions. The complementiser y(r) is used if the auxiliary is bod ‘be’, as illustrated in (23). Either aS or y(r) is used if the auxiliary is modal such as gallu ‘be able’, cael ‘be allowed’, etc. as in (24) (Borsley et al 2007: 120).
(23) Beth yr ydych yn ei fwyta? what C be.PRES.2PPROG 3MS eat.INF
‘What are you eating?’ (Borsley et al. 2007: 111) (24) a. Beth a allaf ei wneud?
what C can.1S 3MS do.INF (Borsley et al. 2007: 111)
b. Beth y gallaf ei wneud?
what C can.PRES.1S 3MS do.INF (Borsley et al. 2007: 112)
‘What can I do?’
Let us examine the two criteria to determine whether the case of the periphrastic construction involves movement or resumption. The first criterion is agreement. The clitic agrees with its antecedent in Literary Welsh. The example (25a) demonstrates agreement between the clitic eu18 and the plural antecedent y dynion. If the singular clitic ei is used, it turns out to be ungrammatical in Literary Welsh, as in (25b). However, Willis (2000) points out that (25b) which does not show agreement is observed colloquially.19
(25) a. y dynion y byddwch yn eu cwrdd Lit. W. the men C be.FUT.2PPROG 3P meet.INF
‘the men that you’ll be meeting’
b. * y dynion y byddwch yn ei gwrdd Lit. W. the men C be.FUT.2PPROG 3MS meet.INF (Willis 2000: 535)
18
The plural clitics ein ‘our’, eich ‘your’ and eu ‘their’ do not trigger mutation in principle (see footnote 23 below).
19
The acceptability of strict agreement between a clitic and an antecedent varies between speakers. In my judgement tests, the sentence without the object-agreement clitic is the most acceptable, see [11] in appendix.
[11c] Dyna ’r llyfrau mae Ieuan wedi prynu. that’s the books be.PRES.3S Ieuan PERF buy
50 The same situation is obseved in wh-questions. The clitic normally agrees in person and number with the wh-object, as in (26). However, Borsley et al. (2007) also observe that this strict agreement is not required in speech, instead, as in (25b) above the clitic in the default third-person masculine singular form ei or its mutation alone can be found regardless of person and number.
(26) Pa rai wyt ti wedi (eu) clywed o’r blaen? which ones be.PRES.2S you PERF 3P hear before
‘Which ones have you heard before?’ (Borsley et al. 2007: 110)
Rich agreement on clitics in Literary Welsh seems to suggest that this is a reflex of the resumptive strategy, that is, there is a null pronoun in postverbal object position. This position is taken in Awbery (1977) and Rouveret (2002). However, the crucial fact is that an overt pronoun cannot sit in this position even in the literary variety. In non-wh-environments, agreement clitics allow both an overt pronoun and a null object after the non-finite verb (Borsley et al. 2007).
(27) Mae Ifan yn ei fwyta (e). be.PRES.3S Ifan PROG 3MS eat.INF it
‘Ifan is eating it.’ (Borsley et al. 2007: 110)
In wh-constructions, however, an overt pronoun is never possible in object position of non- finite verbs (Tallerman 1990: 310).
(28) * y car mae ’r lladron wedi (ei) ddwyn e the car be.PRES.3S the thieves PERF 3MS steal.INF it
‘the car that the thieves have stolen’ (Borsley et al. 2007: 120) (29) * Beth ydych chi ’n (ei) fwyta e?
what be.PRES.2P you PROG 3MS eat it
51 The unavailability of overt pronouns seems to suggest that a trace left by movement is posited in object position of the periphrastic verbs.
Wh-constructions formed in object of periphrastic verbs present a two-way difficulty (see Borsley et al. 2007: 112-14 and Willis 2011 for more discussion). If we assume that the resumptive strategy is involved, we need separate treatments for synthetic and periphrastic verbal constructions, that is, object wh-constructions of the synthetic verbs use the movement strategy whereas the periphrastic verbal constructions use the resumptive strategy. On the other hand, if we assume that movement is involved in object wh-constructions of the periphrastic verbs, we need to treat wh-construction and non-wh-construction differently. In other words, periphrastic verbs can take a pronoun as their object in non-wh-context, but in
wh-context they take a trace derived by movement rather than a resumptive pro despite of the
presence of agreement clitics in the literary variety. However, as we saw above, an overt pronoun is never possible in object position in wh-constructions. Furthermore, the strict agreement between the clitic and the antecedent is not always observed in Colloquial Welsh, and the agreement clitics are often omitted in this variety. Therefore, following Borsley et al. (2007), we can conclude that at least in Colloquial Welsh the movement strategy is involved in object position regardless of whether the verb is synthetic or periphrastic.