2
Tagalog ‘voice’ does not mark Case or valency: lessons from picture NP reflexives In Tagalog, a reflexive embedded inside a picture NP can be bound either by an agent or by a recipient in the same clause regardless of the voice type of the sentence (e.g. AV (1a), LV (1b), CV (1c)): (1) a. Nag-bigay av.prf-give si pn.pivot Joy Joy kay pn.acc Aya Aya ng acc sarili self niyang 3s.poss larawan. picture Actor voice (AV) b. B<in>igy-an <prf>-give-lv ni pn.gen Joy Joy si pn.pivot Aya Aya ng acc sarili self niyang 3s.poss larawan. Locative voice (LV) picture c. I-b<in>igay cv-give<prf> ni pn.gen Joy Joy kay pn.acc Aya Aya ang pivot sarili self niyang 3s.poss larawan. Circumstantial voice (CV) picture ‘Joy <k> gave Aya <j> a picture of herself <k/j> .’ Importantly, linear order does not play a role in the binding relations in (1a-c). Due to Tagalog’s flexible word order, a picture NP reflexive may precede its antecedent without aecting grammaticality (e.g. (1b’): B<in>igy-an ni Joy <k> ng sarili niyang larawan <j/k> si Aya <j> ). On the other hand, the reflexive must be properly bound by a c-commanding antecedent, as seen in (2a) vs. (2b): (2) a. P<um>atay <av>kill si pn.pivot Juan Juan ng acc sarili self niyang 3s.poss anak. child Actor voice (AV) ‘Juan killed the son of himself’. b. *P<um>atay <av>kill ang pivot sarili self niyang 3s.poss anak child kay pn.acc Juan. Juan Actor voice (AV) (intended: ‘The son of himself killed Juan.’) Theoretical implications. The grammaticality of (1a-c) has profound implications for the analysis of Tagalog voice. Specifically, the fact that the reflexive sarili niyang inside the Pivot-marked picture NP (‘picture of herself’) can be bound by the non-Pivot recipient (‘Aya’) in (1c) casts new light on two focal questions in Philippine-type syntax. First, is Pivot-marking (ang/si in Tagalog) a marker of subjecthood or topichood (e.g. Payne 1982; Shibatani 1988; Richards 2000; Aldridge 2004)? Second, does Tagalog voice morphology inflect for the Case of the Pivot phrase (AV: nom; pv: acc; LV: dat; CV: obl) (Rackowski & Richards 2005)? In this paper, we show that evidence from Tagalog picture NP constructions reveals that (i) the Pivot-marking (ang/si) is best analyzed as a marker of information structure status (e.g. topic), rather than a marker of nominative/absolutive Case (subjecthood), and (ii) the Case-agreement approach to Tagalog voice faces explanatory limitations when it comes to 3-place clauses such as ditransitives. Issues in previous analyses of Tagalog LV and CV axes. Existing approaches to Tagalog voice (Aldridge 2004; Rackowski & Richards 2005) share two core assumptions: (a) the Pivot phrase in LV/CV clauses is licensed as an applied object base-generated higher than any other internal arguments, and (b) v in LV/CV clauses bears an EPP feature, which attracts the closest DP (i.e., the alleged applied object Pivot in LV/CV clauses) to the outer Spec of vP, where the Pivot phrase is accessible to nominative/absolutive Case and A’-extraction (with assumption of an Attract Closest Condition). As illustrated with the CV example (3), however, both approaches wrongly predict the recipient ‘Aya’ in the CV sentence (1c) to be impossible to bind into the picture NP reflexive ‘picture of herself ’—as both assume the Pivot-marked picture NP to be licensed by a high applicative phrase that c-commands the recipient ‘Aya’, as in (3): (3) 1

CV, wh-cleft · ng cn.gen sarili self niyang 3s.poss inay mother si pivot Riza. Riza. Patient voice (PV) (marginal: ‘The mother of himself killed Riza .’) Claim

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: CV, wh-cleft · ng cn.gen sarili self niyang 3s.poss inay mother si pivot Riza. Riza. Patient voice (PV) (marginal: ‘The mother of himself  killed Riza .’) Claim

Tagalog ‘voice’ does not mark Case or valency: lessons from picture NP reflexives

In Tagalog, a reflexive embedded inside a picture NP can be bound either by an agent or by a recipient inthe same clause regardless of the voice type of the sentence (e.g. AV (1a), LV (1b), CV (1c)):

(1) a. Nag-bigayav.prf-give

sipn.pivot

JoyJoy

kaypn.acc

AyaAya

ngacc

sariliself

niyang3s.poss

larawan.picture

Actor voice (AV)

b. B<in>igy-an<prf>-give-lv

nipn.gen

JoyJoy

sipn.pivot

AyaAya

ngacc

sariliself

niyang3s.poss

larawan. Locative voice (LV)picture

c. I-b<in>igaycv-give<prf>

nipn.gen

JoyJoy

kaypn.acc

AyaAya

angpivot

sariliself

niyang3s.poss

larawan. Circumstantial voice (CV)picture

‘Joy<k> gave Aya<j> a picture of herself<k/j>.’

Importantly, linear order does not play a role in the binding relations in (1a-c). Due to Tagalog’s flexibleword order, a picture NP reflexive may precede its antecedent without affecting grammaticality (e.g.(1b’): B<in>igy-an ni Joy<k> ng sarili niyang larawan<j/k> si Aya<j>). On the other hand, the reflexivemust be properly bound by a c-commanding antecedent, as seen in (2a) vs. (2b):

(2) a. P<um>atay<av>kill

sipn.pivot

JuanJuan

ngacc

sariliself

niyang3s.poss

anak.child

Actor voice (AV)

‘Juan killed the son of himself’.

b. *P<um>atay<av>kill

angpivot

sariliself

niyang3s.poss

anakchild

kaypn.acc

Juan.Juan

Actor voice (AV)

(intended: ‘The son of himself killed Juan.’)

Theoretical implications. The grammaticality of (1a-c) has profound implications for the analysis ofTagalog voice. Specifically, the fact that the reflexive sarili niyang inside the Pivot-marked picture NP(‘picture of herself’) can be bound by the non-Pivot recipient (‘Aya’) in (1c) casts new light on two focalquestions in Philippine-type syntax. First, is Pivot-marking (ang/si in Tagalog) a marker of subjecthoodor topichood (e.g. Payne 1982; Shibatani 1988; Richards 2000; Aldridge 2004)? Second, does Tagalogvoice morphology inflect for the Case of the Pivot phrase (AV: nom; pv: acc; LV: dat; CV: obl) (Rackowski& Richards 2005)? In this paper, we show that evidence from Tagalog picture NP constructions revealsthat (i) the Pivot-marking (ang/si) is best analyzed as a marker of information structure status (e.g. topic),rather than amarker of nominative/absolutive Case (subjecthood), and (ii) the Case-agreement approachto Tagalog voice faces explanatory limitations when it comes to 3-place clauses such as ditransitives.

Issues in previous analyses of Tagalog LV andCVaffixes. Existing approaches to Tagalog voice (Aldridge2004; Rackowski & Richards 2005) share two core assumptions: (a) the Pivot phrase in LV/CV clausesis licensed as an applied object base-generated higher than any other internal arguments, and (b) v inLV/CV clauses bears an EPP feature, which attracts the closest DP (i.e., the alleged applied object Pivotin LV/CV clauses) to the outer Spec of vP, where the Pivot phrase is accessible to nominative/absolutiveCase and A’-extraction (with assumption of an Attract Closest Condition). As illustrated with the CVexample (3), however, both approaches wrongly predict the recipient ‘Aya’ in the CV sentence (1c) to beimpossible to bind into the picture NP reflexive ‘picture of herself ’—as both assume the Pivot-markedpicture NP to be licensed by a high applicative phrase that c-commands the recipient ‘Aya’, as in (3):

(3)

1

Arts User
Text Box
AFLA 26 Victoria Chen (Victoria University of Wellington)
Page 2: CV, wh-cleft · ng cn.gen sarili self niyang 3s.poss inay mother si pivot Riza. Riza. Patient voice (PV) (marginal: ‘The mother of himself  killed Riza .’) Claim

Claim 1: Tagalog ang/si do not mark absolutive/nominative Case. The invariable binding relationsin (1a-c) suggest the absence of voice-conditioned argument structure alternations in Tagalog (contrathe prediction of previous work), indicating that the applicative affix analysis for Tagalog LV/CV affixesis inconsistent with the empirical facts. In addition, it suggests that the Pivot-marked picture NP inthe CV clause (1c), whose reflexive is bound by the recipient ‘Aya’, is not in a position accessible toabsolutive/nominative Case regardless of an EPP on v, given the presence of an intervening recipientDP between v and the Pivot. This indicates that 1) the ‘Pivot’ marker (ang/si) may not be a marker ofsubjecthood, and that 2) promotion to Pivothood is better analyzed as an A’-phenomenon, rather thanan A-phenomenon. (4a-c) lend further support to this proposal. As (4a-b) show, promotion to Pivothoodpatterns with typical A’-operations, in which a picture NP reflexive is always interpreted in its θ-position.(4c) further shows that a Pivot theme (‘Riza’) in a PV clause may bind into an external-argument pictureNP with marginal acceptability, suggesting that promotion to Pivothood shows weak crossover effects,a typical A’-property (cf. the AV counterpart (2b), which reveals that a non-Pivot theme (‘Juan’) cannotbind into an external-argument picture NP reflexive). Together, these examples suggest that promotionto Pivothood shows the hallmarks of an A’-operation.

(4) a. AlingWhich

larawanpicture

nglk

sariliself

niya3s.poss

anglk

i-p<in>akitacv-show<prf>

mo2s.gen

kaypn.acc

Nica?Nica

CV, wh-cleft

‘Which picture of herself<k> did you show Nica<k>?’

b. Angpivot

sariliself

niyang3s.poss

larawan,picture

gustolike.pv

nipn.gen

Alvin.Alvin

topicalization

‘The picture of himself<k>, Alvin<k> likes.’

c. ?P<in>atay<pv.prf>kill

ngcn.gen

sariliself

niyang3s.poss

inaymother

sipivot

Riza.Riza

Patient voice (PV)

(marginal: ‘The mother of himself<k> killed Riza<k>.’)

Claim 2: Tagalog voice affixes are not Case-agreement morphology. A core assumption of the Caseagreement approach to Tagalog voice is that each of the four voice affixes realizes a different type ofCase borne by the Pivot phrase [AV: nom; pv: acc; LV: dat; CV: obl]. According to this analysis, then,the Pivot phrase in the LV clause (1b) is introduced by a low Appl phrase and bears dat Case and thatin the CV clause (1c) is licensed by a high Appl phrase and bears obl Case (Rackowski 2002). Thisanalysis, however, faces two empirical issues. First, given the invariable binding relations in (1a-c)—which suggest a c-commanding relation Agent > Recipient > Theme best analyzed as reflecting a doubeobject construction (DOC)—analyzing the Pivot-marked theme (‘a picture of herself’) in (1c) as a highapplicative phrase contradicts the binding facts. Second, given the DOC structure of both (1b) and (1c),assuming that the Pivot-marked picture NP in the CV clause (1c) bears a special type of Case (i.e. obl)distinct from acc and dat is difficult to maintain, especially since the theme in a DOC occupies a normalacc position (see Rackowski 2002 and Chen 2017 for an acc analysis of Tagalog ditransitive themes).1

Implications. The observations from picture NP reflexives suggest that the LV/CV affixes in Tagalog areneither applicative affixes nor Case agreement morphology. This indicates that Tagalog ‘voice’ is funda-mentally different from the Indo-European-style voice, which marks valency-rearranging operations andsignals argument structure alternation. We entertain several other alternatives for the analysis of Tagalogvoice, in particular a recent claim that Philippine-type voice affixes are the spell-out of different bundlesof abstract Agree relations that probe the topic of a clause (Chen 2017).

Selected references. Aldridge, E. 2004. Ergativity and word order in Austronesian languages. Ph.D. thesis: Cor-nell U. Chen, V. 2017. A reexamination of the Philippine-type voice system and its implications for Austronesianprimary-level subgrouping. Ph.D. thesis: U of Hawai‘i. Rackowski, A. 2002. The Structure of Tagalog: Specificity,Voice, and the Distribution of Arguments. Ph.D. thesis: MIT. Rackowski, A. & N. Richards. 2005. Phase edge andextraction: A Tagalog case study. LI 36(4):565-99.

1Although some Tagalog literature refers to the voice affix i- (e.g. (1c), (4a)) as a ‘PV’ affix, it is in fact a regular reflex of Proto-Austronesian circumstantial voice (CV) affix *Si-. This interpretation is consistent with observations from other Philippine-typelanguages, in which CV is the type of morphology present when a ditransitive theme is Pivot-marked (e.g. Atayal, Seediq, Paiwan,Kavalan, Ilocano, Botolan Sambal, Yami, Muna, Cebuano, Bikol).

2