Ergativity in Indo-Aryan
Ergativity in Indo-Aryan
- Pritty Patel-GroszPritty Patel-GroszUniversity of Oslo
Case and agreement patterns that are present in Old, Middle, and New Indo-Aryan languages have been argued to require the following perspective: since ergative case marking and ergative (object) agreement in these languages are historically tied to having originated from the past perfective morphological marker ta, they can only be fully understood from a perspective that factors in this development. Particular attention is given to the waxing and waning of ergative properties in Late Middle Indo-Aryan and New Indo-Aryan, which give rise to recurring dissociation of case and agreement; specifically, object agreement in the absence of ergative case marking is attested in Kutchi Gujarati and Marwari, whereas ergative case marking without object agreement is present in Nepali. With regard to case, recent insights show that “ergative/accusative” may be regularly semantically/pragmatically conditioned in Indo-Aryan (so-called differential case marking). Pertaining to agreement, a central theoretical question is whether “ergative” object agreement should be analyzed uniformly with subject agreement or, alternatively, as a type of participle agreement—drawing on synchronic parallels between Indo-Aryan and Romance.
- Language Families/Areas/Contact
1. Introducing Case and Agreement in Indo-Aryan
New Indo-Aryan (NIA) languages like Hindi-Urdu and Standard Gujarati are well known to instantiate so-called split-ergativity, a phenomenon where ergative case marking and “ergative” object agreement arise in certain tense/aspect combinations.1 The core pattern is illustrated by (1).2 In (1a), an imperfective habitual sentence, the transitive subject šilaa lacks a case marker and triggers agreement (for feminine singular) on the main verb lakh- ‘write’. In contrast, in the perfective simple past, (1b), the subject carries the ergative marker -e, and verbal agreement on lakh- ‘write’ targets the direct object kaagaL ‘letter’.
Examples such as (1a) are generally subsumed under the notion of accusative alignment, referring to phenomena (such as structural case marking and verbal agreement) that treat transitive subjects (written as A) like intransitive subjects (S), while treating transitive objects (O) differently (e.g., Dixon, 1994), schematized in (2).
In contrast, (1b) can be treated as an instance of ergative alignment, which treats transitive objects (O) like intransitive subjects (S), while treating transitive subjects (A) differently. This is schematically given in (3).
The trigger for ergative case marking (and for the possibility of object agreement) in Indo-Aryan languages is so-called perfective morphology, which is most aptly defined diachronically in line with Butt and Deo (2017), as given in (4). This definition remains neutral as to whether Indo-Aryan “perfective morphology” is semantically perfective or perfect (see Condoravdi & Deo, 2008).
Of course, applying this definition synchronically comes with some practical challenges, since the NIA instantiations of the -ta morphology of Old Indo-Aryan (OIA) are varied, including -y- (as in Kutchi Gujarati kar-y-o ‘did-perf-m.sg’) and -dh- (as in Kutchi Gujarati di-dh-o ‘give-perf-m.sg’). This is of course easily remedied for languages whose history is substantially documented such as Bangla (see Chatterji, 1926); in contrast, it is perhaps significantly more challenging to apply such a diachronic criterion to lesser documented Indo-Aryan languages and dialects. A more operational synchronic approach holds that descendants of the -ta morphology are typically found in tense/aspect combinations such as the simple past, as well as the present perfect, past perfect, and future perfect. Naturally, some of these tense/aspect combinations in a given language may have a different origin; thus, this is at best a necessary condition for unearthing the relevant patterns.
2. Case–Agreement Interactions
Past research on Indo-Aryan has given rise to the insight that case and agreement may have to be dissociated in Indo-Aryan languages even though they often co-occur (see Butt, 2017; Butt & Deo, 2017). This is most clearly instantiated by Nepali, where subject case marking exhibits the same aspectual splits that are present in Standard Gujarati as illustrated by (1)–(3), but verbal agreement uniformly targets the subject, as shown in (5).
In contrast, Hindi-Urdu has generally been assumed to exhibit a closer connection between case and agreement in the sense that agreement systematically tracks the overt realization of subject case (ergative) and object case (optional accusative/differential object marking). In South Asian languages, optional accusative (differential object marking) has a semantic/pragmatic function; its realization is conditioned by factors such as animacy, specificity, and definiteness of the direct object (see de Hoop, 2009; and Malchukov & de Swart, 2009, for discussion).
The central pattern is illustrated by (6). If both subject and object are unmarked (in a non-perfective aspect), subject agreement arises ([6a]); if the subject is case-marked (in the perfective) and the object is unmarked, object agreement arises ([6b]). Finally, if both subject and object are case-marked, only default agreement (masculine singular) is possible ([6c]). In line with Mahajan (2017, and references therein), the differential object marker -ko is glossed dom rather than acc, as it exhibits the semantic properties of differential object marking, yet it does not follow the distribution of structural accusative case; for example, it occurs in the “ergative-absolutive” configuration in Indo-Aryan languages.
In light of the range of different alignment patterns found throughout the Indo-Aryan languages, Butt and Deo (2017) conjecture that Indo-Aryan ergativity cannot be understood from an exclusively synchronic perspective. Section 3 is a brief review of their diachronic discussion.
3. The History of Ergativity in Indo-Aryan
Butt and Deo (2017) consider two possible origins of ergativity in Indo-Aryan that have been proposed in the literature, an origin in a passive construction and an origin in a possessive construction. This article focuses on the former, as given in (7); for recent approaches to syntactic and semantic reanalysis, see Eckardt (2006), Roberts (2007), and Deo (2015). The core idea is that the same surface form, in (7a) and (7b), was, at some diachronic stage of the language, ambiguous between two different structures (as indicated in the glosses). In the course of diachronic development, the original structure ([7a]) was given up in favor of the reanalyzed structure ([7b]), as schematized in (8). The reanalysis from (7a) to (7b) involves an uncontroversial reanalysis of the OIA -ta participle (here illustrated by hataḥ ‘killed’) into a (presumably finite) active verb. Such reanalysis is well attested in the Indo-European languages, including the Romance languages and the Slavic languages (Dombrowski, 2006). (7a) and (7b) also depict a more controversial reanalysis of instrumental case as ergative case.
Butt and Deo (2017) show that a participle-to-active-verb reanalysis must have already been completed in Epic Sanskrit (around 200 bce), using examples like (9), where a resultative analysis in line with (7a) is implausible. Here, the subject, mayā ‘I’ carries instrumental/ergative marking, and the -ta participle visṛṣṭo ‘released’ appears to function as a finite main verb.
While assuming the participle-to-finite-verb analysis, Butt and Deo (2017) challenge the idea that ergative case marking in NIA is derived from the OIA instrumental, drawing on early arguments from Beames (1966/1872–1879). Beames observes that the OIA case system mostly disappeared in the history of the Indo-Aryan languages, so that Middle Indo-Aryan (MIA) had ergative alignment without systematic ergative case marking (see Peterson, 1998). This indicates an early dissociation of “ergative” object agreement and ergative case marking. In order to see this development, compare the expected examples in (10) and (11) to the more revealing examples in (12) and (13). Here in (10), from the Mahāraṣṭrī text Vasudevahiṃḍī, the agent/experiencer is marked with instrumental case, giving rise to object agreement with the theme/stimulus. Butt and Deo (2017) take this to illustrate the MIA ergative system (schematically given in ).
To illustrate the impoverishment of the MIA case system (in contrast to the OIA case system), Butt and Deo present examples from the Paumacariu (a Jaina rendition of the Rāmāyana), as illustrated in (12) (schematically in ). They observe that distinctions such as nominative/accusative and genitive/dative were lost, while the nominative and the (ergative) instrumental case became partially syncretic (namely, in the first and second plural form), written as syncr in (12) and (13). They conclude that agreement became the main realization of ergative alignment in MIA, regardless of morphological case marking.
In Early Hindi, the ergative instrumental marking seems to have undergone complete attrition, becoming syncretic with the nominative in most contexts, as illustrated in (14). Crucially, (14a–b) show that ergative object agreement was preserved in the absence of ergative case marking, indicating the above-mentioned independence of case and agreement in Indo-Aryan.
Butt and Deo show that, in a similar historical period, Old Bengali has a pattern comparable to Early Hindi, as in (15), while still making a phonological distinction between the nonnasalized nominative (cf. kānhē) and the nasalized ergative (cf. kānhẽ).
Moreover, Phillips (2013) documents examples of optional ergative case marking from 16th-century Old Rajasthani, which he takes to be symptomatic for the disappearance of ergative case (or ergative case attrition). Although ergative case marking clearly cannot be considered systematic in (16a–b), these examples exhibit ergative object agreement regardless of whether ergative case is overt or not. Here, (16a) shows object agreement without ergative case marking on the subject (kʊmar ‘prince’). One may suspect that ergative case marking is present in (16a) and not visible for (morpho-)phonological reasons, but (16b) shows that this is not the case, as the example contains a case-marked variant of the same subject DP (kʊmar-i ‘prince’).
At present, Butt and Deo propose that the NIA languages instantiate at least three different systems that have emerged from these MIA stages. Hindi and Nepali have acquired a new ergative case marker that was innovated after the MIA period illustrated in (14). The “strengthened” system of Modern Hindi-Urdu is illustrated by the example in (17b). Butt and Ahmed (2011) argue that the recently innovated ergative clitic -ne of Hindi-Urdu was loaned from neighboring languages, originating in the Old Rajasthani postposition kanhaïN ‘aside, near’.
They show that -ne in Haryani and Kherwada Wagdi has a dative/accusative use as well as an ergative use (see Butt & Ahmed, 2011, p. 562), shown in (18) and (19).3 By contrast, in Gujarati () and Marwari (), -ne has a dative/accusative use only (see also Tessitori, 1913, 1914–1916), while lacking an ergative use.
In sharp contrast to (17)–(19), Bengali and Oriya have lost ergative alignment altogether, in a clear further development of (14)–(15). This loss of ergativity is illustrated for Modern Bengali in (22). Note, in this respect, that ergativity has also been lost in “non-ergative” eastern Hindi dialects, whereas it is robustly preserved, for instance, in Delhi Hindi (Mahajan, 2012, p. 207).
Finally, Marathi and Gujarati have retained a version of the original system from OIA and Archaic MIA. To be specific, the Standard Gujarati ergative marker -e in (1b) is generally traced back to the OIA instrumental, in sharp contrast to the innovated ergative marker -ne in Hindi. At the same time, and somewhat unsurprisingly, the pattern that arose in Early Hindi and Old Bengali (ergative object agreement in the absence of ergative case marking) can still be attested in some present-day Indo-Aryan languages and dialects. Marathi has extended nominative/ergative syncretism to all instances of the 1st and 2nd person, while preserving ergative case marking in the 3rd person. Yet, the agreement split (between subject agreement in the nonperfective and object agreement in the perfective) is robust, as illustrated by (23a–b).
It is difficult to see in which sense (28) and (29) should be classified as ergative alignment, as they instantiate a system where, within the same sentence, agreement on the auxiliary exhibits “accusative” (S/A) alignment, whereas agreement the participle exhibits ergative (S/O) alignment. This seems to further bolster the idea that, in principle, ergative case marking may be dissociated from ergative object agreement in Indo-Aryan languages.4 If verbal agreement were to track morphological case marking on the core arguments (subject and direct object), cases where one agreeing element in the verbal complex targets the subject while the other agreeing element targets the object should not occur (see section 5).
4. Ergative and Accusative Case Versus Differential Case Marking
This section illustrates the mounting evidence that ergative and accusative case marking in Indo-Aryan are partly semantically/pragmatically conditioned and not rigidly derived from structural mechanisms such as dependent case assignment (e.g., Marantz, 1991). As shown in sections 4.1 and 4.2, Indo-Aryan languages (possibly in a sprachbund with Dravidian and Tibeto-Burman languages) exhibit systematic patterns of differential case marking (or rather, in many cases, optional case marking; see McGregor, 2010). This article adopts a broad definition of differential case marking along the lines of Aissen (2003) and Witzlack-Makarevich and Seržant (2018): a minimal pair of two sentences that differ only in the case marking on an argument DP, which has a semantic effect. Before reviewing the range of patterns, it is worth emphasizing that differential case marking is not independent from particular structural configurations. Differential subject marking (DSM)—in particular, optional ergative marking (OEM)—is typically restricted to transitive subjects, which presumably occupy a structural subject position (such as SpecvP), whereas differential object marking (DOM) (or optional accusative marking [OAM]) tends to be restricted to transitive objects, which correspondingly originate in a structural object position (such as the complement position of the VP).5 This means that a syntactic mechanism of dependent case assignment in the spirit of Marantz (1991) may well be the necessary condition for DSM and DOM (at least in some languages), although it is not always a sufficient condition.
4.1 Differential Subject Marking (Subsuming “Optional Ergative”)
Recall the discussion of the attrition of ergative case marking from OIA to MIA. The 3rd-century ce Niya dialect, which has been documented in the Niya documents, exemplifies optionality in the realization of ergative/instrumental case marking at that historical stage (for further discussion, see Butt & Deo, 2017; Jamison, 2000). In some present-day NIA languages, such optionality can still be widely observed, as shown in (32). It has since been well established that languages that allow for a free choice between nominative and ergative case on subjects, the choice ends up being correlated with semantic factors, such as animacy/agentivity, thus giving rise to DSM (e.g., Klaiman, 1987; Mithun, 1991). Similarly, McGregor (2010) proposes that the overt realization of an optional ergative case marker tends to mark a “prominent” subject/Agent, though he leaves prominence as an underspecified notion. Quite generally, our understanding of the semantic/pragmatic conditioning factors for optional ergative case marking in Indo-Aryan languages, as instantiated in (32), is still limited.6
As Butt and Deo (2017) show, differential subject marking in NIA languages such as Bengali and Hindi-Urdu is most clearly visible in modal constructions, where it indicates a difference in modality (that presumably derives from a difference in volitionality). This is illustrated by (33).
Since ergative case is typically reserved for transitive subjects, one could speculate that the case alternation in (33) can be derived from two different underlying structures (a transitive structure in [33a] and an unaccusative structure with an experiencer subject in [33b]). However, this is not trivial, since both sentences contain an unaccusative verb ja ‘go’ in combination with hɛ ‘is’ and literally translate to ‘Nadia is to go to the post office’; therefore, it not self-evident what the source of transitivity in (33a) would be. Moreover, Davison (1999, pp. 185–186) lists a number of transitive verbs in Hindi-Urdu (including bol ‘speak’, la ‘bring’, and laṛ ‘fight’) that never exhibit ergative case on the subject, and proceeds by discussing another class of verbs that exhibits a nominative/ergative alternation. This class is illustrated by samajh ‘understand’, bhul ‘forget’, jan ‘give birth (to)’, phãd ‘leap over’, bak ‘to say nonsense’, har ‘lose, be defeated’, and pahcan ‘recognize’; an illustration is given in (34).
Another instance of a systematic nominative/ergative alternation in Hindi-Urdu concerns intransitive predicates that loosely qualify as “unergative” (Butt, 2017). For this class, Davison (1999, pp. 186–187) lists verbs such as bhõk ‘bark’, jhãk ‘peep, look into/through’, khãs ‘cough’, chĩk ‘sneeze’, muskara ‘smile’, thuk ‘spit’, mut ‘urinate’, hag ‘defecate’, naha ‘bathe’, ro ‘cry’, hãs ‘laugh’, ga ‘sing’, and so ‘sleep’; a classical example is given in (35). As of today, it is unclear whether a systematic structural difference can be posited for (34a) versus (34b), or for (35a) versus (35b). At the very least, such patterns tentatively challenge a view in which Hindi-Urdu has a systematic ergative case pattern in the perfective construction.
Based on such systematic case alternations, which occur throughout the Indo-Aryan languages, Butt (2017) proposes a lexical semantic approach to structural case marking in Indo-Aryan, which treats Indo-Aryan languages as accusative languages with DCM. Moreover, in line with McGregor (2010), Butt links optional ergative marking (as a type of DSM) to semantic properties such as volitionality and agentivity, which she takes to be most clearly illustrated by contrasts such as (36a) versus (36b).
In line with such a DSM approach to ergative case marking in Indo-Aryan languages, it has been argued that optional ergative case in Nepali nonperfective clauses correlates with information structure. In fact, Verbeke (2013, pp. 607–608) takes it to be the “traditional account” that optional ergative marking in Nepali correlates with focus/emphasis (the focus hypothesis). This description fits with examples such as (38), but Verbeke (2013) points out that other corpus examples are not accounted for by the focus hypothesis.
By contrast, Butt and Poudel (2007) suggest that a distinction in terms of temporary (stage-level) versus characteristic (individual-level) properties is more adequate. It is an open question whether this, too, correlates with information structure (e.g., a possible requirement that subjects of individual-level predicates, (39b), are topical; cf. Jäger, 2001). These semantically driven case alternations would be puzzling from a view that treats Indo-Aryan subject case marking as an instance of true ergative alignment, since structural/dependent case is presumably a purely syntactic phenomenon.
4.2 Differential Object Marking (“Optional Accusative”)
Differential object marking (DOM; see de Hoop, 2009; Malchukov & de Swart, 2009) is widespread in the South Asian languages, posing an additional challenge to structural/dependent case approaches to these languages. In South Asian languages, DOM generally amounts to the presence versus. absence of a case marker (e.g., Hindi-Urdu -ko in [40b] and Gujarati -ne in [41b]); that is, it qualifies as optional case marking in the sense of McGregor (2010). The presence of the optional case marker has been shown to correlate with semantic/pragmatic notions such as animacy, definiteness, specificity, and information structure. In Hindi-Urdu, human-denoting objects are always dom-marked; by contrast, dom with non-human-denoting objects correlates with specificity.
Mistry (1997) shows that DOM in Standard Gujarati sometimes just reflects specificity, as in (41); in other cases, he claims that this change in specificity seems to affect the interpretation of the predicate, as in (42).7
It is, as of yet, unclear whether a structural approach to DOM is feasible. Baker (2013) proposes that cross-linguistically DOM may either involve (pseudo-)incorporation of an unmarked object (such as gari ‘car’ in [40a]) into the verb, or object shift of a marked object (such as gari=ko ‘car=dom’ in [40b]) to a position outside the VP (see also Aissen, 2003, on correlations between DOM and object shift).8 Both approaches face a challenge in the apparent flexibility of DOM that emerges from examples such as the following. If unmarked direct objects had to incorporate into the verbal predicate, the existence of (43b–c) would be surprising, where the unmarked xat ‘letter’ has scrambled to a higher position. Moreover, if object shift/scrambling automatically triggered DOM, then we would expect to see obligatory DOM marking (i.e., xat=ko ‘letter=dom’) in (43b), contrary to fact.
Interestingly, notions of DOM also come into play in the realization of ergative object agreement. Patel (2008) observes that Kutchi Gujarati exhibits an alternation between full gender/number agreement (such as masculine singular, -o) and default agreement (neuter singular, -u) in perfective object agreement (though never in nonperfective subject agreement). This is illustrated in (44), where a correlation between default agreement and scrambling of the direct object can be observed in (44b). (Note that superscripted hash marks in the first line of the examples symbolize the dispreferred option rather than true unacceptability.)
The same variation between object agreement and default agreement seems to be possible in Surati Gujarati;9 this is not an isolated fact about Kutchi Gujarati. (See also Suthar, 2006, pp. 185–186, for a discussion of optional agreement on Standard Gujarati adverbs and irrealis markers.)
5. Ergative Agreement Versus Participle Agreement
Having provided an overview of ergativity and in particular case marking in Indo-Aryan, section 5 probes one of the central theoretical questions: what is the status/nature of object agreement? On the basis of a case study on two Western Indo-Aryan languages, Kutchi Gujarati and Marwari, this section establishes a connection between Indo-Aryan and Romance. While such a parallel is not surprising from a diachronic Indo-Europeanist perspective, the connection has only recently been explored in the syntax literature (cf. D’Alessandro, 2011). Although a central focus has been on a unified approach to subject and object agreement (see, e.g., Bhatt, 2005; Bobaljik, 2008), one of the main findings of this case study is that Indo-Aryan object agreement, specifically in Kutchi Gujarati and Marwari, may be a type of participle agreement of the kind found in Romance languages such as French and Italian.
5.1 Case and Agreement as Two Sides of the Same Coin
Many authors attribute subject agreement and object agreement in Indo-Aryan languages to a single mechanism (see, e.g., Bhatt, 2005; Bobaljik, 2008). A slightly simplified rendering of such a view is illustrated in (45) for the example from Mistry (2004, pp. 3–4) from Standard Gujarati. The idea would be that a single probe, say, on the T head, would agree with the closest accessible DP under c-command, where accessibility is determined in terms of case marking; that is, T can agree with a caseless (“nominative”) DP, such as šilaa ‘Sheela’ in (45a) or kaagaL ‘letter’ in (45b), whereas an ergative-marked DP such as šilaa-e ‘Sheela-erg’ in (45b) would be unacceptable.
Such a system allows for some flexibility. For instance, T can agree with DOM-marked objects in Standard Gujarati but not in Hindi. Similarly, T can agree with ergative-marked subjects in Nepali but not in Hindi or Standard Gujarati. Grosz and Patel-Grosz (2014) argue that such a single probe system does not seem to be adequate for languages such as Kutchi Gujarati or Marwari; instead, they argue for a dual probe system, which attributes object agreement and subject agreement to two distinct functional heads in the clausal spine. Section 5.2 revisits the Kutchi Gujarati and Marwari facts and explores a dual-probe system somewhat different from the Grosz and Patel-Grosz (2014) system, which models the observed agreement patterns more in line with the phenomenon of participle agreement in Romance languages.
5.2 An Alternative Scenario: Two Sources of Verbal Agreement
A close connection between case-marking and agreement as in (45) comes apart in Kutchi Gujarati and Marwari, where nested agreement is possible. This is illustrated for Kutchi Gujarati in (46) and for Marwari in (47).
These nested agreement patterns set Kutchi Gujarati and Marwari apart from languages like Standard Gujarati. The ergative subject in Standard Gujarati perfective constructions is truly inaccessible for agreement, which gives rise to default agreement (3rd person) on the present tense auxiliary, as shown in (48).
Note, in this respect, that agreement in languages such as Hindi-Urdu, Gujarati, and Marwari tends to be in gender and number (e.g., joya, dekhī in –). This is due to the origin of present-day verbal morphology in adjectival participles such as the “perfective” -ta participle (see Butt, 2017; Butt & Deo, 2017). As Butt (2017) points out, exceptions include future morphology, as in (46), and the nonpast auxiliaries, as in (47), where person/number agreement has been preserved from OIA. Cases that combine two affixes, which cumulatively express agreement in all three features (person/number + gender/number), can be analyzed as recent innovations where two verbs have fused into one. Butt and Lahiri (2013) derive the Hindi-Urdu future marker -g- in (49) from the light verb ga ‘go’. Similarly, the Kutchi Gujarati progressive -r- in (50) plausibly derives from a light verb rəh ‘stay’.10
Coming back to (46) and (47), it is worth exploring an alternative to treating subject and object agreement on a par. The main predecessor of the view to be outlined in section 5.3 was pioneered, to the best of my knowledge, by Magier (1983a, pp. 321–323), who capitalizes on the observation that the present perfect in Marwari exhibits the same pattern of participle agreement present in the French passé composé. This becomes evident when the Marwari example in (51) is juxtaposed with the French example in (52).11
This parallel suggests that number/gender agreement on Marwari perfective verbs is due to a mechanism of participle agreement, which needs to be distinguished from the “true” verbal agreement (for person/number) found on nonpast auxiliaries and tense markers (see also Friedemann & Siloni, 1997). Importantly, Indo-Aryan object agreement seems to share an origin with the participle agreement found in Romance languages, as illustrated in (53); here, the standard analysis holds that French participle agreement can be traced back to resultative participles in Latin.12 As shown in section 3 (in particular, example ), a similar trajectory can be observed in Indo-Aryan.
Notably, different analyses may be correct for different subsets of the Indo-Aryan languages. It may be true that ergative object agreement in Hindi-Urdu or Standard Gujarati exploits the same mechanism that gives rise to subject agreement. Nevertheless, since participle agreement has barely been actively pursued as an analytical option of agreement in Indo-Aryan, it is worth focusing on this option, at least for languages like Kutchi Gujarati and Marwari.
5.3 A Unified Approach to Indo-Aryan and Romance Participle Agreement?
To begin with, (54a–c) show that the future marker systematically agrees with the subject for person and number; that is, the non-past auxiliary is devoid of an agreement split. The agreement split is only present at the level of the participle, as shown in (54a) versus (54b).
Note also that although past tense configurations typically lack tense auxiliaries, the pattern in (55a,b) is properly contained in the pattern in (54a,b). This makes a uniform analysis desirable, which would assume null past tense auxiliaries (rather than treating the past participle as a finite main verb). Compare in this respect the following quote from Bhatt (2005, p. 772) on Hindi-Urdu: “Finite T0 is, however, not always overtly realized. Simple past tense sentences in Hindi-Urdu do not involve an overt expression of tense.”
Descriptively, the agreement split in Kutchi Gujarati mirrors a split that Belletti (2006) documents for Italian constructions with a reflexive indirect object clitic. The pattern in (56a) (with si ‘to self’) mirrors the Kutchi Gujarati imperfective pattern ([54a]), whereas the pattern in (56b) (with clitic se ‘to self’) mirrors the perfective pattern ([54b]).
A critical reader will notice that participle agreement with objects appears to be less systematic in Romance than in Indo-Aryan, in that it typically requires specific licensing conditions, such as some type of object movement (e.g., clitic movement of li ‘them’ in [56b]). However, archaic and dialectal registers have been argued to exhibit variants without movement (; see also D’Alessandro & Roberts, 2008, p. 479, n.3), which could thus be taken as the direct equivalent of the nested agreement pattern in Kutchi Gujarati and Marwari.
At this point, it is worth exploring what a treatment of Kutchi Gujarati as a participle agreement language would look like. The main idea is that true verbal agreement (as visible on future tense suffixes) always targets the subject; this is presumably an instance of agreement with the T head. As for the split in (54a–b) (and, by analogy, [55a–b]), this is not a split in verbal agreement but rather a split in participle agreement. An outline of the relevant analysis can be modeled after the tentative proposal by Belletti (2006, pp. 510–512) for the subject-/object-agreement split in (56). She proposes that a vP-internal participle-agreement phrase (AgrPstPrtP) is responsible for (object) participle agreement with the object clitic li in (56b). By contrast, a vP-external AgrPstPrtP is responsible for (subject) participle agreement with the subject DP Maria in (56a). Moscati and Rizzi (2014, p. 69) argue that Belletti’s participle-agreement-related AgrPstPrts should be identified with aspectual heads. Based on Travis (1991, 2010) and MacDonald (2008), we could thus propose a vP-internal Asp head (“inner aspect”) and a vP-external OAsp head (“outer aspect”). Accordingly, (58) provides a tentative sketch of the relevant participle agreement relationships in Kutchi Gujarati. The idea would then be that the vP-external OAspP in (58a) agrees with the subject and subsequently passes the agreement information on to the main verb jo-th-i. By contrast, in (58b), the vP-internal AspP agrees with the direct object and subsequently passes the agreement information on to the main verb jo-y-a. As indicated by dotted arrows, each aspectual head agrees with the structurally closest argument that it c-commands.
Although the purpose of this article is to outline the hypothesis space rather than flesh out a specific analysis, it is worth briefly exploring the implications and further predictions of such an approach. For example, Kutchi Gujarati unergatives require subject agreement as opposed to default agreement in the perfective, shown in (59); this indicates that OAsp is active in (58b) as well, but object agreement bleeds subject agreement when both co-occur (see Belletti, 2006, for similar assumptions with regard to Italian).
By contrast, object agreement cannot be possible in (58a), which indicates that the vP-internal Asp is only an “active” agreement probe in the perfective. In fact, although the vP-external OAspP may always be projected, a vP-internal AspP may sometimes lack altogether when it is “inactive” (e.g., in the case of stative predicates; cf. MacDonald, 2008, pp. 27–29).
The proposal that perfective (-y-) participles may involve inner aspect (and thus a vP-internal AspP) makes further predictions that are tentatively confirmed by the following observation. If, as argued by MacDonald (2008), stative predicates lack inner aspect (i.e., AspP) altogether, this gives rise to the prediction that statives cannot occur in the perfective.13 Indeed, this is what one finds. Most stative predicates (javab khabar ha- ‘know the answer’, vajan 80kg ha- ‘weigh 80kg’, ghadero ha- ‘own a donkey’) involve copula constructions in Kutchi Gujarati. As illustrated in (60b), a perfective variant of (60a) is impossible. (On a related note, the grammar of Standard Gujarati [Cardona, 1965, p. 101] explicitly states that the auxiliary [and copula] ha- lacks a perfective form.)
Note, also, that the idea that perfective -y- participles are composed at the level of “inner aspect” captures the observation for Standard Gujarati by Cardona (1965, p. 101): “The perfective designates punctuality of the activity of the verbal root.”
Further, more general, support for unifying Kutchi Gujarati (and Marwari) participle agreement with Romance participle agreement stems from the fact that agreeing adverbs occur in both cases (see Magier, 1983a, p. 325, for Marwari; Grosz & Patel-Grosz, 2014, p. 228, for Kutchi Gujarati; Butt et al., 2016, for Urdu, Sindhi, and Punjabi; D’Alessandro, 2011, for Ripano [Italic, Romance]; as well as Ledgeway, 2011, 2016, and Silvestri, 2016, for Southern Italian). The agreeing adverbs are predominantly aspectual, and presumably base generated in a position between AspP and OAspP (see also Cinque, 1999, p. 106). These patterns can be accounted for by assuming that AspP and OAspP enter an agreement chain that includes all intermediate agreement probes (e.g., the probes that are responsible for agreement on vel- ‘early’ and paach- ‘again’ in ).
Interestingly, Standard Gujarati also exhibits adverb agreement (here: on ekdhaar- ‘continuously’), even though it patterns more like Hindi-Urdu and less like Kutchi Gujarati/Marwari in other respects. If the participle-agreement analysis is on the right track, this may indicate that even Standard Gujarati instantiates (Agr-based) participle agreement on perfective participles (rather than verbal agreement via T).
To conclude, an analysis of Kutchi Gujarati and Marwari object agreement as past participle agreement is feasible, and it would directly account for the nested (Romance-style) agreement patterns present in these languages. This raises the option of treating object agreement as participle agreement in other Indo-Aryan languages as well, even in the absence of nested agreement patterns, for example, in Hindi-Urdu.
Indo-Aryan languages still raise many questions concerning the nature of accusative/ergative alignment. This article recapitulates the insight of Butt and Deo (2017) that Indo-Aryan case/agreement systems can only be understood if we take their diachronic origins into consideration. Moreover, a review of the role that semantics and pragmatics plays in (differential) case marking provides a richer empirical scope, thus deepening the question of how to best model ergative object agreement in Indo-Aryan. One of the core areas of investigations that needs to be addressed in the near future concerns the issue of whether Indo-Aryan object agreement should be subsumed under regular verbal agreement or under the notion of past participle agreement; data from Kutchi Gujarati and Marwari indicate that the latter is more explanatory.
I am very grateful to audiences at LISSIM 10 (Linguistic Summer School in the Indian Mountains) and to two anonymous reviewers whose constructive feedback greatly improved this article.
- Aissen, J. (2003). Differential object marking: Iconicity vs. economy. Natural Language and Linguistic Theory, 21, 435–483.
- Baker, M. (2013). On alignment type and differential object marking [Handout: Workshop on Differential Object Marking]. University of Tromsø.
- Baker, M. (2015). Case. Cambridge: Cambridge University Press.
- Beames, J. (1966). A comparative grammar of the modern Aryan languages of India. Delhi: Munshiram Manoharlal. (Original work published 1872–1879)
- Belletti, A. (2006). (Past) participle agreement. In M. Everaert & H. van Riemsdijk (Eds.), The Blackwell companion to syntax (Vol. III, pp. 493–521). Oxford: Blackwell.
- Bhatt, R. (2005). Long distance agreement in Hindi-Urdu. Natural Language and Linguistic Theory, 23, 757–807.
- Bhatt, R. (2007, November 14). Ergativity in Indo-Aryan languages [Handout: Ergativity Seminar]. MIT Press.
- Binnick, R. I. (2012). The Oxford handbook of tense and aspect. Oxford: Oxford University Press.
- Blanc, L.-G. (1844). Grammatik der Italiänischen Sprache. Halle: C.A. Schwetschke & Sohn.
- Bobaljik, J. (2008). Where’s phi? Agreement as a post-syntactic operation. In D. Harbour, D. Adger, & S. Béjar (Eds.), Phi-theory: Phi features across interfaces and modules (pp. 295–328). Oxford: Oxford University Press.
- Butt, M. (2017). Hindi/Urdu and related languages. In J. Coon, D. Massam, & L. deMena Travis (Eds.), The Oxford handbook of ergativity (pp. 807–831). Oxford: Oxford University Press.
- Butt, M., & Ahmed, T. (2011). The redevelopment of Indo-Aryan case systems from a lexical semantic perspective. Morphology, 21, 545–572.
- Butt, M., & Deo, A. (2017). Developments into and out of ergativity: Indo-Aryan diachrony. In J. Coon, D. Massam, & L. deMena Travis (Eds.), The Oxford handbook of ergativity (pp. 530–552). Oxford: Oxford University Press.
- Butt, M., & King, T. H. (1996). Structural topic and focus without movement. In M. Butt & T. H. King (Eds.), Proceedings of the First LFG Conference. Stanford: CSLI.
- Butt, M., & Lahiri, A. (2013). Diachronic pertinacity of light verbs. Lingua, 135, 7–29.
- Butt, M., & Poudel, T. (2007, April). Distribution of the ergative in Nepali [Presentation]. University of Leipzig.
- Butt, M., Sulger, S., Rahman, M. U., & Ahmed, T. (2016). Adverb agreement in Urdu, Sindhi and Punjabi. In D. Arnold, M. Butt, B. Crysmann, & T. H. King (Eds.), Proceedings of the Joint 2016 Conference on Head-driven Phrase Structure Grammar and Lexical Functional Grammar (pp. 140–160). Stanford: CSLI.
- Bynon, T. (2005). Evidential, raised possessor, and the historical source of the ergative construction in Indo-Iranian. Transactions of the Philological Society, 103(1), 1–72.
- Cardona, G. (1965). A Gujarati reference grammar. Philadelphia: University of Pennsylvania Press.
- Chatterji, S. K. (1926). The origin and development of the Bengali language. 2 vols. Calcutta: Rupa.
- Cinque, G. (1999). Adverbs and functional heads: A cross-linguistic perspective. New York: Oxford University Press.
- Clark, T. W. (1963). Introduction to Nepali: A first-year language course. School of Oriental and African Studies, University of London.
- Comrie, B. (1995). Tense and aspect. In J. Jacobs (Ed.), Syntax. 2. Halbband. Handbücher zur Sprach- und Kommunikationswissenschaft [Handbooks of linguistics and communication science (HSK)] (pp. 1244–1250). Berlin: de Gruyter.
- Condoravdi, C., & Deo, A. (2008). Aspect shifts in Indo-Aryan. In Proceedings of the 18th International Congress of Linguists (CIL 18), Seoul.
- D’Alessandro, R. (2011). Agreement, ergativity, and the parameterization of probes. LUCL Leiden University.
- D’Alessandro, R., & Roberts, I. (2008). Movement and agreement in Italian past participles and defective phases. Linguistic Inquiry, 39(3), 477–491.
- Davison, A. (1999). Ergativity: Functional and formal issues. In M. Darnell, E. Moravcsik, F. Newmeyer, M. Noonan, & K. Wheatley (Eds.), Functionalism and formalism in linguistics, volume I: General papers (pp. 177–208). Amsterdam: John Benjamins.
- de Hoop, H. (2009). Case in optimality theory. In A. L. Malchukov & A. Spencer (Eds.), The Oxford handbook of case (pp. 88–101). Oxford: Oxford University Press.
- Deo, A. (2006). Tense and aspect in Indo-Aryan languages. Variation and diachrony [Unpublished doctoral dissertation]. Stanford University.
- Deo, A. (2015). Diachronic semantics. Annual Review of Linguistics, 1, 179–197.
- Dixon, R. M. W. (1994). Ergativity. Cambridge: Cambridge University Press.
- Dombrowski, A. (2006). Towards a typology of participial development: Evidence from Romance and Slavic. LSO Working Papers in Linguistics, 6, 26–38.
- Eckardt, R. (2006). Meaning change in grammaticalization: An enquiry into semantic reanalysis. Oxford: Oxford University Press.
- Embick, D. (2004). On the structure of resultative participles in English. Linguistic Inquiry, 35(3), 355–392.
- Friedemann, M.-A., & Siloni, T. (1997). Agrobject is not Agrparticiple. The Linguistic Review, 14(1), 69–96.
- Garrett, A. (1990). The origin of np split ergativity. Language, 66(2), 261–296.
- Grierson, G. A. (1969). Linguistic survey of India: Vols. I–XI. Delhi: Motilal Banarsidas. [Reprint of originals published between 1903 and 1928]
- Grosz, P., & Patel-Grosz, P. (2014). Agreement and verb types in Kutchi Gujarati. In P. Chandra & R. Srishti (Eds.), The lexicon-syntax interface. Perspectives from South Asian languages (pp. 217–244). Amsterdam: John Benjamins.
- Hertzenberg, M. J. (2015). Habere + pp and the origin of the periphrastic perfect. Journal of Latin Linguistics, 14(1), 33–63.
- Jäger, G. (2001). Topic-comment structure and the contrast between stage level and individual level predicates. Journal of Semantics, 18(2), 83–126.
- Jamison, S. (2000). Lurching towards ergativity: Expressions of agency in the Niya documents. Bulletin of the School of Oriental and African Studies, University of London, 63(1), 64–80.
- Khokhlova, L. (2001). Ergativity attrition and western new Indo-Aryan languages. In P. Bhaskararao & K. Venkata Subbarao (Eds.), The yearbook of South Asian languages and linguistics 2001. Tokyo symposium on South Asian languages. Contact, convergence, and typology (pp. 159–184). London: SAGE.
- Klaiman, M. H. (1987). Mechanisms of ergativity in South Asia. Lingua, 71(1–4), 61–102.
- Kratzer, A. (2000). Building statives. In Proceedings of the Twenty-Sixth Annual Meeting of the Berkeley Linguistics Society: General Session and Parasession on Aspect (pp. 385–399). Berkeley: BLS.
- Krifka, M. (1989). Nominal reference, temporal constitution and quantification in event semantics. In R. Bartsch, J. van Benthem, & P. van Emde Boas (Eds.), Semantics and contextual expression (pp. 75–116). Berlin: de Gruyter.
- Krifka, M. (1992). Thematic relations as links between nominal reference and temporal constitution. In I. Sag & A. Szabolsci (Eds.), Lexical matters. Stanford, CA: CSLI.
- Ledgeway, A. (2011). Adverb agreement and split intransitivity: Evidence from southern Italy. Archivio Glottologico Italiano, 96(1), 31–66.
- Ledgeway, A. (2016). Parameters in Romance adverb agreement. In M. Hummel & S. Valera (Eds.), Adjective–adverb interfaces in Romance (pp. 47–80). Amsterdam: John Benjamins.
- MacDonald, J. E. (2008). The syntactic nature of inner aspect: A minimalist perspective. Amsterdam: John Benjamins.
- Magier, D. S. (1983a). Topics in the grammar of Marwari [Unpublished doctoral dissertation]. University of California, Berkeley.
- Magier, D. S. (1983b). Components of ergativity in Marwari. In Chicago Linguistic Society (CLS), Papers from the 19th Regional Meeting (pp. 244–255). Chicago: CLS.
- Mahajan, A. (1990). The A/A-bar distinction and movement theory [Unpublished doctoral dissertation]. MIT Press.
- Mahajan, A. (2012). Ergatives, antipassives and the overt light v in Hindi. Lingua, 122(3), 204–214.
- Mahajan, A. (2017). Accusative and ergative in Hindi. In J. Coon, D. Massam, & L. deMena Travis (Eds.), The Oxford handbook of ergativity (pp. 86–108). Oxford: Oxford University Press.
- Malchukov, A., & de Swart, P. (2009). Differential case marking and actancy variations. In A. L. Malchukov & A. Spencer (Eds.), The Oxford handbook of case (pp. 339–355). Oxford: Oxford University Press.
- Marantz, A. (1991). Case and licensing. In Proceedings of the 8th Eastern States Conference on Linguistics (pp. 234–253). Ithaca, NY: CLC.
- McGregor, W. B. (2010). Optional ergative case marking systems in a typological-semiotic perspective. Lingua, 120(7), 1610–1636.
- Menon, M., Simpson, A., & Choudhury, A. (2013). Argument ellipsis and the licensing of covert nominals in Bangla, Hindi and Malayalam. Lingua, 134, 103–128.
- Mistry, P. J. (1997). Objecthood and specificity in Gujarati. In J. Hill, P. J. Mistry, & L. Campbell (Eds.), The life of language: Papers in linguistics in honor of William Bright (pp. 425–442). Berlin: Mouton de Gruyter.
- Mistry, P. J. (2004). Subjecthood of non-nominatives in Gujarati. In P. Bhaskararao & K. Venkata Subbarao (Eds.), Non-nominative subjects (Vol. 2, pp. 1–32). Amsterdam/Philadelphia: John Benjamins.
- Mithun, M. (1991). Active/agentive case marking and its motivations. Language, 67(3), 510–546.
- Mohanan, T. (1995). Wordhood and lexicality: Noun incorporation in Hindi. Natural Language and Linguistic Theory, 13(1), 75–134.
- Moscati, V., & Rizzi, L. (2014). Agreement configurations in language development: A movement-based complexity metric. Lingua, 140, 67–82.
- Öhl, P. (2009). Die Entstehung des periphrastischen Perfekts mit haben und sein im Deutschen—Eine längst beantwortete Frage? Zeitschrift für Sprachwissenschaft, 28, 265–306.
- Patel, P. (2008, May). Exploring case in Kutchi Gujarati [Workshop Handout: A Non-Canonical Perspective on Case]. University of Konstanz.
- Peterson, J. M. (1998). Grammatical relations in Pāli and the emergence of ergativity in Indo-Aryan. München: LINCOM Europa.
- Phillips, M. (2013). Ergative case attrition in central Indo-Aryan. NP-splits and the referential hierarchy. Studies in Language, 37(1), 196–216.
- Posner, R. (1996). The Romance languages. Cambridge: Cambridge University Press.
- Ramat, P. (1982). Ein Beispiel von ‘reanalysis’, typologisch betrachtet. Folia Linguistica, 16, 365–384.
- Ramchand, G. (1997). Aspect and predication. Oxford: Clarendon Press.
- Roberts, I. (2007). Diachronic syntax. Oxford: Oxford University Press.
- Shirani, M. (1987). Urdu ki shakh haryani zaban mein talifat. In M. M. Shirani (Ed.), Maqaalaat-e-hafiz mahmood shirani [Papers of Hafiz Mahmood Shirani]. Lahore: Majilis-e-Taraqqi-e-Adab.
- Silvestri, G. (2016). Adverb agreement in the dialects of the Lausberg area. In M. Hummel & S. Valera (Eds.), Adjective–adverb interfaces in Romance. Amsterdam: John Benjamins.
- Suthar, B. K. (2006). Agreement in Gujarati [Unpublished doctoral dissertation]. University of Pennsylvania.
- Tessitori, L. (1913). On the origin of the dative and genitive postpositions in Gujarati and Marwari. Journal of the Royal Asiatic Society of Great Britain and Ireland, 45, 553–567.
- Tessitori, L. (1914–1916). Notes on the grammar of the Old Western Rajasthani with special reference to Apabhramsa and to Gujarati and Marwari. Indian Antiquary, 43–45.
- Thielmann, P. (1885). Habere mit dem Part. Perf. Pass. Archiv für lateinische Lexikographie und Grammatik, 2, 372–423, 509–549.
- Travis, L. (1991, October). Derived objects, inner aspect, and the structure of vp [Paper presentation]. 22nd Annual Meeting of the North East Linguistics Society (NELS 22). University of Delaware.
- Travis, L. (2010). Inner aspect: The articulation of vp. Dordrecht, the Netherlands: Springer.
- Verbeke, S. (2010). Postpositions and agreement in Rajasthani languages: A comparative analysis of alignment features in Marwari and Harauti. Indogermanische Forschungen, 115, 35–64.
- Verbeke, S. (2013). Differential subject marking in Nepali: The agent marker le in imperfective constructions. Linguistics, 51(3), 585–610.
- Verkuyl, H. (1993). A theory of aspectuality. Cambridge: Cambridge University Press.
- Witzlack-Makarevich, A., & Seržant, I. A. (2018). Differential argument marking: Patterns of variation. In I. A. Seržant & A. Witzlack-Makarevich (Eds.), Diachrony of differential argument marking (pp. 1–40). Berlin: Language Science Press.
1. Although tense/aspect-conditioned split ergativity is the most central for Indo-Aryan languages, many of them exhibit syncretism in their pronominal paradigms. For example, Standard Marathi only exhibits ergative case on 3rd person pronouns and not on 1st or 2nd person pronouns (see Grierson, 1969).
4. See Baker (2015), for rich and detailed discussion on the disassociation between case and agreement systems more generally. Further arguments specifically for South Asian languages have been made by Menon et al. (2013).
5. Not all of the cases that are ruled in as DSM involve transitive subjects or ergative case; for instance, (33) is an example where ergative case alternates with dative case on the subject of ja ‘go’ combined with a modal use of hɛ ‘is’.
6. An anonymous reviewer points out that ergative case in (32b) may be triggered by the light verb nak ‘throw’, which is absent in (32a). Although this may well be the case, the important part of the example is the absence of ergative marking in (32a), since it fulfills the criteria observed in ergative marking in split ergative Indo-Aryan languages: perfective aspect and uncontroversially transitive verb, namely, kap ‘cut’.
7. Examples (39) and (40) are typical of the broader pattern of scalar mapping that arises as a result of explicit case marking and the interaction between the object and the predicate. See Krifka (1989, 1992), Ramchand (1997), and Verkuyl (1993), for discussion.
9. Many thanks to Kinjal Joshi for pointing this out.
10. The Kutchi Gujarati r-progressive is plausibly related to the Hindi-Urdu progressive marker rəha/rəhe/rəhī, which historically derives from rah ‘stay’ (Deo, 2006, p. 176). Another plausible cognate is the Marwari progressive marker riyo (cf. Magier, 1983a, p. 151).
11. Please note that examples (51) and (52) cannot be made more parallel, as French participle agreement depends on cliticization and is not always audible (e.g., the masculine vu [vy] ‘seen’ and feminine vue [vy] ‘seen’ have an identical pronunciation, whereas the masculine pris [pʁi] and feminine prise [pʁiz] are distinct). Many thanks to Philippe Schlenker for discussion of the French example, and the example itself.
12. The origin of this particular example is unclear (though it is sometimes attributed to Cicero), and it may well have been constructed for illustrative purposes. However, the observation that it captures is well established (see Hertzenberg, 2015; Öhl, 2009; Ramat, 1982; and Thielmann, 1885, among others, for a wide range of authentic examples). For comparison, also consider examples (i) in Present-Day English, which is parallel to the Latin example.
- Historical Developments from Middle to Early New Indo-Aryan
- Argument Realization in Syntax
- Morphological Change
- Linguistics in Premodern India
- Tense and Aspect in Morphology
- Grammaticalization in Morphology
- Head/Dependent Marking
- Morphology in Indo-European Languages
- Finite Verb Morphology in the Romance Languages