(formal) Approaches to South Asian Languages 16

I investigate the internal structure of plural expressions in Ganjami Odia, an Eastern Indo-Aryan language that productively combines a classifier system with multiple overt plural markers. This configuration raises a central puzzle for theories of the nominal domain: whether plurality is realized in a unified projection (PluralP) or distributed across distinct syntactic positions (Chierchia 1998; Borer 2005). I address two related questions: (i) whether Odia projects a dedicated PluralP, and (ii) why the absence of -maane yields obligatory singular interpretations with animate nouns, while bare inanimate nouns remain number-neutral.

I argue against a unified PluralP and propose instead a split architecture in which plurality is distributed across the nominal spine (Wiltschko 2021). I analyze -maane as a functional head in #⁰ encoding count plurality, and -gurika as a DP-adjoined modifier encoding collective reference (Moltmann 2016). This structural asymmetry is supported by diagnostics including complementary distribution with numeral–classifier constructions (Dékány 2011), agreement asymmetries, number neutrality (Wiltschko 2008), and scope under coordination (Butler 2021).

Crucially, I develop a feature-licensing account of Odia plurality. I propose that animate nPs introduce an interpretable i[ANIM] feature (Kramer 2015) which triggers the projection of #P, where [+Num:pl] is introduced. Following Kalin’s (2019) “derivational time-bomb” model, I argue that the feature bundle {i[ANIM], [+Num:pl]} must be licensed via DP-internal Agree with Agr⁰ (Fong 2023). The morphological exponent -maane emerges as the spell-out of this licensed configuration; its absence yields default singular interpretations. In contrast, -gurika lacks such licensing requirements and patterns as a lexicalized, definite collective adjunct.

This analysis supports a non-uniform view of plurality and situates cross-linguistic variation within feature selection under the Borer–Chomsky Conjecture.

Akankshya Mishra <amishra1132000@gmail.com>

Indian Institute of Technology Delhi

Plurality, Split-NumP, Classifier Systems, Agreement, Animacy, Collective vs. Count Plurality

Colloquial Sinhala Conditional Structures

While a core sense of conditional meaning clearly validates the cross-linguistic construct of conditional structure, attempts to categorize conditionals and related structures, both within and across languages (e.g. Comrie 1986, Athanasiou & Dirven 1995, Declerck & Reed 2001, Xrakovskij 2005, etc.) indicate considerable variation in not only structural realization but also situated meaning and discourse-pragmatic function. Such research is not yet well distributed across languages, and to achieve a comprehensive understanding of the structure and role of conditionals in language requires analysis from a much wider variety of languages. This research contributes to that discussion by describing the two principal conditional structures of Colloquial Sinhala (CS) and examining their relationship to topic and focus structures. Sinhala is a diglossic language of which CS, our focus, is less described and analyzed than the Literary variety. We show that the morphological conditional is more limited in usage than the particle structure, notably with regard to factuality status. Classified as New Indo-Aryan, it exhibits some features unusual in that group, of which many are related to influence from Tamil and other Dravidian languages. One such feature is the dual nature of the conditional system, which includes a conditional structure based on a particle that attaches to a clause, much like the English if structure, but also a morphological conditional formed by means of a verbal suffix, similar to that in Tamil. Neither the structural distribution nor the discourse functions of these structures has yet been described in detail. In this research, we address these gaps, using a combination of the limited corpus data available and acceptability judgments. Concerning the two structures, we show that both are common. The conditional particle, nam, can attach to finite past or nonpast clauses, as well as certain nonfinite clauses. It shares a shape with the topic particle but can be distinguished. A marked difference is that the topic marker can attach to sub-clausal constituents, while the conditional marker must take scope over a clause, even if that clause contains a focus-marked constituent. The morphological conditional is effected by means of an ending on the head verb of the antecedent. This suffix attaches to the past tense stem of a verb, in place of a tense suffix. The clause thus has no tense. Morphological conditionals are unable to express counterfactual meaning, strengthening the case for connection between past tense and factuality phenomena. Morphological conditionals can be followed by nam, suggesting the possibility that some conditional clauses are double-marked. However, we show that this nam is the topic marker rather than the conditional particle. It cannot occur when the function of the morphological conditional is to indicate discourse topic, as this function obviates the possibility of marking sentence topic. A further discourse restriction we note is that the morphological conditional does not express anankastic meaning. Thus, as has been shown for other languages, the nonfinite conditional structure is more restricted in usage than the correlative conditional, which accords with its more marked status cross-linguistically. However, the specific constraints are different from those noted for, e.g., Bangla.

References Athanasiou, Angeliki, and René Dirven. 1995. Typology of if-clauses. Cognitive linguistics in the redwoods: The expansion of a new paradigm in linguistics Vol. 6, ed. by Eugene Casad, 609-654. Berlin: DeGruyter. Comrie, Bernard. 1986. Conditionals: A typology. On conditionals, ed. by Elizabeth Close Traugott, Alice ter Meulen, Judy Snitzer Reilly, and Charles A. Ferguson, 77-99. Cambridge: Cambridge University Press. Declerck, Renaat, and Susan Reed. 2001. Conditionals: A comprehensive empirical analysis. Berlin: de Gruyter Mouton. Xrakovskij, Victor. 2005. Typology of conditional constructions. Munich: LINCOM Europa.

Kapugollawe Anandakiththi and Theresa McGarryTheresa McGarry <MCGARRY@mail.etsu.edu>

University of Kelaniya and East Tennessee State University

Sinhala, conditionals, topic, factuality

In this paper, I argue that Bangla (Bengali; Eastern Indo-Aryan) lacks a true instance of Copy Raising (henceforth, CR; Rogers 1971, 1972) even though it exhibits all the properties of a CR except one, i.e., the presence of a related (Johni seems like [hei knows everything]) or unrelated (Nora smells like [John was cooking biryani]) embedded subject inside the finite embedded clause. I suggest this ‘gap’ inside the embedded clause is due to A'-movement (topicalization) of the embedded nominal to the front of the matrix clause with the presence of a dropped pro in the matrix subject position. This ties up neatly with another claim that P(erceptual)-Source is not a ө-role but a semantic one, resolving the apparent violation of Theta Criterion (Chomsky 1981) on the moved nominal.

In lieu of this discussion, I show that the moved NP in the concerned Bangla data does give a true P-Source reading. I also give tests from the language which includes an overt dependency test to see whether or not there is a dropped pro, and further to tease out the difference in movement to A-position (hyper-raising) and the structures in concern, and a control constructions diagnostic to account for the lack of an overt or dropped subject inside the embedded clause. I further propose that rather than a dropped pro dependency analysis, A-movement (raising/hyper-raising), base-generation (control), it is a case of A'-movement to the sentence initial position where the topicalized phrase is an aboutness topic (Frascarelli & Hinterhölzl 2007; Bianchi & Frascarelli 2010; via Miyagawa 2017). I follow Georgiou (2023) who suggests a topicalized phrase moves due to EPP feature on the functional Top head, and there is feature checking between the functional Top head and the topicalized NP via an Agree mechanism (Chomsky 2000, 2001) where the Top head is the probe and the topicalized phrase is the goal.

I conclude by showing that Bangla hence lacks a true instance of CR construction even though the PRV in the matrix clause gives a P-Source reading to the moved nominal. This opens up further discussion on how to tease apart subjects and topics across languages and what makes A-movement distinct from A'-movement at large.

Mouma Banik

University of Delhi

Copy Raising, Perceptual Source, A'-movement, Theta Criterion

The Central Magadhan Prakrit languages such as Bhojpuri, Maithili, and Magahi have distinct historical roots in the eastern branch of Middle Indo-Aryan (Bhattacharya 2016). Here, we argue that agreement with conjoined noun phrases in these languages is determined by a process of sequential feature probing, which accesses person, honorificity, and gender in a fixed order. This stepwise probing mechanism results in a typologically rare agreement hierarchy: Person (1st ≻ 2nd ≻ 3rd) ≻ Honorificity (H ≻ NH) ≻ Gender (Mas ≻ Fem).Data: The following data are from Bhojpuri, collected in the Bihta district of India. As shown in (1), when the conjoined subject contains a first-person and a second-person DP, verbal agreement systematically targets the first person. Similarly, in combinations involving first-and third-person DPs (2), agreement continues to target the first person. In combinations involving second- and third-person conjuncts (3), agreement consistently aligns with the second person. When both conjuncts are third person (4) and (5), the verb agrees with the honorific DP, irrespective of whether the honorific DP is masculine (4) or feminine (5). Finally, when both DPs are non-honorific (6), agreement targets the masculine conjunct. ’Proposal: We propose that the conjunct agreement hierarchy can be explained by a standard probe–goal Agree mechanism, which operates through a sequential feeding–bleeding relationship across features (Preminger 2014, Keine 2020). Specifically, person features [+Author; +Participant] are probed first, followed by honorificity [+Hon], and then gen- der [+Mas]. Successful agreement with person features bleeds agreement based on honorificity. Conversely, failure to agree on person features feeds the probe to target honorificity. In the same way, successful agreement on honorificity bleeds gender agreement, while its failure feeds agreement based on gender. Assumptions: First, we assume that T is the agreement head that hosts the probing features (Bhatt 2005, Bhatt and Walkow 2013). Second, we assume a conjunct DP structure in which the features of each conjunct are represented as distinct sets within the DP. This means that when Agree applies, the probe has equal access to the features of both conjuncts but agrees with the one that maximally satisfies its features (Arregi and Nevins 2008, Béjar and Rezac 2009) Analysis: In the derivation of (1) and (2) the probe first searches for person features [+Author; +Participant]. This requirement is maximally satisfied by the first person DP. As a result, agreement targets the first person rather than the second or third. Because the person feature has been successfully valued, subsequent probing for honorificity and gender does not take place. In the derivation of (3), where the conjunct DPs consist of a second person and a third person, the probe initially seeks to value its person features. The second person DP satisfies at least one of the person feature [+Participant] of the probe, thereby allowing successful agreement. As a result, the agreement is realized with the second person DP. The successful valuation of the person feature preempts further probing for honorificity and gender. In the derivation of (4) and (5) , where both conjunct DPs are third person, the probe cannot value its person feature, as both DPs do not have matching person features of the probe. This failure to establish person agreement feeds the next step: probing for honorificity. Between the two conjuncts, one is honorific and the other non-honorific. The honorific DP provides the matching value for [+Hon], satisfying the probe. As a result, agreement aligns with the honorific conjunct. Since honorificity agreement is successful, gender probing does not proceed. In the derivation of (6), both conjuncts are third person and non-honorific. The probe’s search for a person value again fails, prompting it to probe for honorificity. The honorific feature of the probe [+Hon] also could not find a match. As a result, the probe moves to the final feature in the hierarchy:[+Mas] , to establish agreement. When one conjunct is masculine and the other feminine, the probe values its feature from the masculine conjunct.Typology: We proposed an analysis of sequential probing in which the probe accesses features in a fixed order: person ≻ honorificity ≻ gender. We have intentionally refrained from arguing that this is the only available order in these languages. Instead, we suggest that the ordering of feature probing is subject to parametric variation within this language group. This means that while Bhojpuri exhibits the person-first hierarchy, other languages in the group may prioritize features differently. For example, Maithili appears to follow the reverse pattern, where honorificity takes precedence over person in agreement (7), such that 3h agreement is preferred over 2nh. (Complete data in the pdf of abstract)

Sonal Sinha

Indian Institute of Technology, Jodhpur

Conjunct Agreement

This study presents a competition-based analysis in the tense–aspect domain in Bangla, focusing on the present perfect and the simple past (past perfective). Drawing on empirical data and pilot experimental results, the study demonstrates a systematic division of labor between the two forms. Another important observation concerns their behavior in the negative paradigm, where negation of the simple past gives rise to a quirky implicature. The goal is to try to analyze an evidentiality puzzle concerning the two forms and the quirky implicature from the perspective of the nature of the perfective operator and the competition between the two types of perfectivity.

Sumedha Gupta

University of Göttingen

Present Perfect, Past, Perfective, Maximality, Alternatives, Implicature

Modeling Syntactic Prominence: Aspect as a Structural Attractor in Haryanvi

Does Tense always anchor the clause? This poster challenges the universal TP-centric assumption of generative syntax using novel data from Haryanvi. Under the Syntactic Operation Competition (SOC) framework, we demonstrate that Haryanvi is an Aspect-prominent language where Asp-head acts as the primary structural attractor. Through evidence from morphosyntactic finiteness, split-ergativity, negation control, and vector (light verb) selection, we show that Aspect, not Tense, hosts the uninterpretable features driving Case and phi-agreement. This argues for a parameterized, rather than universally fixed, distribution of probes across functional categories.

Madan Mohan <mmohangupta277@gmail.com>

Indian Institute of Technology, Kanpur

Syntactic Prominence, Parameters, TAM, Haryanvi, Aspect, Split-ergativity

Therese Liam Tom

Indian Institute of Technology Delhi

This study examines Object Gaps (OGs) in Punjabi ditransitive constructions, which challenge existing accounts. We argue that prevailing analysis based on Argument ellipsis is insufficient to explain the observed patterns. Instead, we propose a derivation of OGs through Verb-Stranding Ellipsis (VSE) combined with focus movement. We further demonstrate that information structure plays a crucial role in the formation of Object Gaps with ditransitives in Punjabi.

Piyush Sharmna <Sharmapiyush442@gmail.com>

University of Delhi, Department of Linguistics

Object gaps, Argument ellipsis, Verb-stranding ellipsis, Information structure.

The research investigates the interaction between complex predicates, negation, and ellipsis, grounded in the intriguing controversy between the existence and non-existence of verb-stranding VP ellipsis. It adopts a phase-based approach showing verb-stranding TP ellipsis in Bangla. Since verb-stranding VP ellipsis violates the constraint on head movement proposed by Landau (2018). Complex predicates also raise unresolved puzzles about the size and type of the constituent that should be elided.

Sudipta Saha

IIT Delhi, Jadavpur University

Verb-stranding VP/TP ellipsis, Phase-based anlysis, complex predicates, negation

In this paper, I analyze polar questions in Marathi in the context of their better known Indo-Aryan counterparts to show that despite their surface complexity, Marathi polar questions are strictly a clause-edge phenomenon as apparent clause-medial instances of polar question particles do not lead to polar questions. In doing so, I argue that a radically simplified version of existing accounts of Indo-Aryan polar questions would be sufficient to account for Marathi polar questions.

[Full abstract: https://drive.google.com/file/d/1L0fPCSCdZmNJrLz70IePyLq7EPJDgD3d/view?usp=sharing ]

Aaditya Kulkarni

All India Institute of Speech & Hearing

polar questions, polar question particles, hybrid interrogatives, pseudo interrogatives, Marathi, Indo-Aryan

Nominal predication in Bangla Introduction. Languages are known to exhibit restrictions with respect to the form of a nominal phrase in the predicate position. For instance, Adger & Ramchand 2005 show that Scottish Gaelic cannot have a noun phrase with a definite determiner in the predicate position (*is Calum an tidsear “Calum is the teacher”) while English can (“Calum is the teacher”). In French, a predicate nominal with an indefinite article is shown to express ‘classifying predication’ while a bare noun is responsible for ‘attributive predication’ (Beyssade & Dobrovie-Sorin 2005). English simply does not allow predication with a bare noun (*“Calum is teacher”). In this paper we will look at how nominal predication takes place in Bangla, an eastern Indo-Aryan language. Bangla does not have articles and can employ bare singular nominals as well as nominal+classifier DPs depending on the context (Dayal 2012, Biswas 2012 etc). It will be shown that Bangla allows the two types of nominal predication argued for in French but here both structures can use bare nominal predicates and the classifying predication is actually instantiated by a special copula. This work further indicates that there is a strong connection between the default semantic type of nominals in a language and the type of predication structure that it allows. Predication is generally characterised in terms of a property that is a function taking the subject as its argument: P(x), attributing said property ‘P’ to the subject ‘x’ (Den Dikken 2006:18). This is what Beyssade & Dobrovie-Sorin have called attributive predication, which manifests with bare nominal predicates in French as seen in (1). They also identify a type of predication dubbed ‘classifying predication’ in the language, which establishes the membership of the subject to the set of individuals who are dancers, involving the use of an indefinite article with the nominal as seen in (2). It is this indefinite article that helps in the set formation by projecting a NumP over the bare nominal. Jean est danseur 2. Jean est un danseur Jean is dancer Jean is a dancer ‘Jean is a dancer.’ ‘Jean is a dancer.’ [French] Question. It will be argued that this classification explains the following structures in Bangla involving a zero-copula structure (3) and a quirky copula structure (4). Garden-variety predication structures with adjectives in Bangla are obligatorily zero-copula constructions [John lOmba “John (is) tall”]. However, nominal predication can have two possible instantiations as seen in (3) and (4); how can these two available structures be accounted for? 3. Bruto kukur 4. Bruto holo kukur Bruto dog Bruto cop dog Bruto is a dog. Bruto is a dog. [Bangla] Analysis. This paper makes the case that the nature of bare nominals in Bangla has a role to play in the existence of the quirky copula structure. The consensus in the literature is that Bangla being a classifier language, bare nouns denote kind terms (Saha 2025:520). This idea is developed on top of Dayal’s (2012) kind characterisation of Bangla bare nominals to claim that in Bangla bare nominals have the ‘singular kind’ status – which means that they “behave as a compact whole” (Saha 2025:525). Thus, the usual type shift operators such as pred available in plural kind languages, like Hindi, for instantiating members of the kind are not available here. When it comes to nominal predication structures, the ramification is that two different strategies for predication become available. One, the typical structure where the property of the bare nominal is ascribed to the subject through the mediation of a functional head like Pred – this is a zero-copula structure in Bangla and the PredP small clause containing the nominal would have the type < e, t >. The quirky copula sentence in (4) on the other hand is like the French ‘classifying predication’ type structure – establishing the subject as a member of the set expressed by the post-copular kind term of the type < s, e >. We claim that the quirky copula instantiates a ‘belong to’ relationship argued to exist independently in singular kind languages (such as, Turkish) to connect an object with the group it is a part of (Sağ 2022:764). Since bare nominals in French denote objects, an indefinite article is required to yield a set-denoting term for classifying predication, through the projection of NumP. In the case of Bangla, the inherent ‘singular kind’ nature of bare nominals makes them available directly for the set denotation required for classifying predication, albeit with the help of the quirky copula. Empirical evidence supporting the two types of predication in Bangla comes from modification structures – while a kind-level modifier like ‘labrador’ is accepted in both, the object-level modifier ‘tall’ is not suitable for the quirky copula structure (6). In other words, ‘hunting dog’ is an established sub-kind while ‘tall dog’ is not, making the attempted class membership of Bruto to the set of ‘tall dogs’ infelicitous in (6). 5. Bruto shikari \ lOmba kukur 6. Bruto holo shikari \ *lOmba kukur Bruto hunting tall dog Bruto cop hunting tall dog Bruto is a hunting dog \ tall dog. Bruto is a hunting dog.Unavailable:Bruto is a tall dog.

Another piece of evidence comes from reciprocal nominal predicates, where the holo structure becomes infelicitous (8). This is expected if the bare nominal predicates in these structures are singular kind terms since they do not allow a reciprocal relation between the members of the group (9). 7. East Bengal ar Mohonbagan protiddondi dOl East Bengal and Mohonbagan rival team East Bengal and Mohonbagan are rival teams. 8. *East Bengal ar Mohonbagan holo protiddondi dOl East Bengal and Mohonbagan cop rival team East Bengal and Mohonbagan are rival teams. 9. *kukur eke-opor-er sathe maramari kore dog each-other-gen with fight do Dogs fight with each other. [adapted from Saha 2025:522] Conclusion. There is empirical evidence and independent theoretical premise to argue for two types of nominal predication structures in Bangla. Thus, this work i) shows that there is a close relationship between the semantic type of nominals in a language and the type(s) of nominal predication available in it, ii) provides further empirical evidence in favour of the ‘singular kind’ account for bare nominals in Bangla. Selected references: Beyssade, Claire & Dobrovie-Sorin, Carmen. (2005). A Syntax-based Analysis of Predication. SALT 15; Biswas, Priyanka. (2012). Reanalyzing Definiteness in Bangla. Berkeley Linguistics Society 38; Dayal, Veneeta. (2012). Bangla classifiers: Mediating between kinds and objects. Italian Journal of Linguistics 24(2):195-226; Sağ, Yağmur. 2022. Bare singulars and singularity in Turkish. Linguistics and Philosophy 45(4): 741-793. Saha, Ankana. (2025). Reference to Kinds: The Perspective from Bangla. Proceedings of the 41st WCCFL (519-529).

Srabasti Dey

Autonomous University of Barcelona

copulas, nominal predicates

Many South Asian languages, including Hindi-Urdu, are pro-drop which allows them to drop any argument (Butt, 2001). We might not expect Hindi-Urdu to exhibit expletives, but it would appear that there are at least two which are associated with clausal complements: yeh and aesa. While yeh is relatively well-studied (Dayal, 1996 and 2017; Kidwai, 2013), aesa remains underexplored as an expletive. This paper aims to examine aesa as an expletive/clausal associate in the following constructions: (1) (aesa) lag-ta hai ki baarish ho-gi (EXPL) seem.IMPF.M.SG be.PRS.3SG that rain be.FUT.F.SG It seems that it will rain. (2) Amna-ne (aesa) soch-a ki baarish ho-gi Amna-ERG (EXPL) think.PST that rain be.FUT.F.SG Amna thought that it will rain. Although aesa also functions as an adjective or adverb in Hindi‑Urdu, its behavior in expletive constructions is syntactically distinct. Adjectival and adverbial aesa shows gender and number agreement (e.g. aese, aesi) and is referential, linking to contextual or discourse‑salient entities. In contrast, expletive aesa is invariant, non‑referential, and semantically vacuous, consistently surfacing with singular masculine morphology. I argue that aesa in the constructions (1) and (2) behaves as an expletive with a clausal associate. It is non-referential, semantically vacuous, and invariant in form (Booth & Groothuis, 2024). Moreover, aesa behaves similarly to the clausal expletive yeh. Aesa, in these constructions, not only blocks leftward scrambling out of the extraposed complement clause but also long-distance wh-dependency and matrix scope of wh-questions. The analysis raises questions about the structural position of aesa: whether it originates in Spec vP or as a head of CP that undergoes extraposition. Finally, this work motivates future inquiry into why a pro drop language like Hindi Urdu employs multiple expletives and how the optionality of aesa should be modeled.

Almas Ashraf

University of South Carolina

aesa, expletive, Urdu, Hindi

Biswanath Dash

Indian Institute of Technology Delhi

Discourse particles are grammatical elements that contribute to the expressive content of an utterance, and not its core propositional content (Zimmerman 2011). The current project investigates the formal and functional properties of the discourse particle ko, which is found in Hakha Lai, a Chin language in the South Central sub-branch. Preliminary research on this discourse particle shows that it is often used to strengthen the assertive force of the propositional content encoded in the sentence it accompanies. Put another way, ko is used in a sentence to express that the speaker firmly believes the proposition to be true. Properties investigated in this study include the syntactic distribution as well as its semantic and pragmatic contributions to the interpretation of the sentences with which ko occurs. The investigation makes use of corpus data from the Linguistically Underserved Communities and Health (LUCAH) project, which includes interviews conducted with members of the Indianapolis Chin community revolving around health and health information access particularly during the COVID-19 pandemic (Berkson et al. 2023). The corpus contains naturalistic speech from the interviews, which were conducted in Hakha Lai. Examples of ko from dictionary sentences are shown in (1-3) below:

1. Pathian nawl a-si [ko] = “It is God’s will.” (chin-dictionary.com)

2. Zanriah ei ah cun ka-ra-manh [ko] lai = “I will/shall be on time for dinner.” (chin-dictionary.com)

3. Zanlei cu inn=ah a-um [ko] lai = He will be home all evening. (chin-dictionary.com)

In examples (1-3), ko is used in the sentence to express that the speaker strongly believes what is being said. For example, the use of ko in (2) expresses that the speaker is strongly committed to what they are promising to the addressee, namely that they shall be on time for dinner. This study, using data from the LUCAH corpus and the Hakha Lai translation of the Bible, presents an analysis of the Hakha Lai discourse particle ko wherein it takes scope over clauses as well as illocutionary operators such as politeness markers to express speaker certainty. This analysis thus makes comparisons to similar attitudinal discourse particles such as Mandarin Chinese discourse particles ba (Xu 2003) and ma (Luo & Huang 2022). Additional questions addressed in this study include the diachronic origins of ko, its syntactic position within the clausal spine, and polysemous functions such as the encoding of modal force. Investigations of corpus data containing both naturalistic and constructed speech in under-researched languages like Hakha Lai are a key contribution to our understanding of the range of cross-linguistic typological variation found in discourse particles.

James Wamsley

University of Rochester

Tibeto-Burman, Chin, Pragmatics, Discourse Particles