Covert nominals and argument ellipsis in Punjabi

Abstract

The focus of the study is to find out the potential identity of null arguments in Punjabi. However, the paper provides different patterns of interpretations of null arguments present in Punjabi by keeping in view the propositions of Takahashi (2012, July). The special focus is given on the phenomenon named as Argument Ellipsis (AE), which is used as syntactically mediated process for investigating the relationship of verbal agreement and scrambling with AE in null argument language Punjabi. It was observed that AE in Punjabi licenses the null arguments on direct and indirect objects, subjects and selected PPs. The study finds out by observing and manipulating different configurations of subject-object agreement in Punjabi that there is no role critically played by verbal agreement in order to restrict AE. The research also specifies that scrambling approach for licensing is also not reliable predictor for AE in Punjabi. However, bare nouns approach is more appropriate for the analysis of AE in Punjabi as the bare nouns license the sloppy interpretations, which are marked either indefinite or definite nouns for null arguments in Punjabi. This approach can be tested for further cross-linguistic examination of AE, and hopefully it would be regarded as the only underlying cause of argument ellipsis for future studies.

Keywords:

• Anti-agreement approach
• argument ellipsis
• covert nominal
• null arguments
• scrambling
• syntactic analysis

1. Introduction

Empty nominal has attracted many linguists in the past such as Rizzi (2006), Jaeggli and Safir (1989), and Giusti (1994). They have explored different features of empty category types in different languages across the world especially in the context of nominal ellipsis. Many researchers have already projected the presence of null pronominal pro on the location of argument especially under different restrictive factors (Adams, 1985; Duguine, 2014; Holmberg, 2005). There are different hypotheses of empty nominals in different languages across the globe like the targeted phenomenon is subject to operator-bound variables which involve covert A’- movement (Huang, 1989). However, others believe that empty nominal is a case of VP ellipsis as it is found in Japanese (Otani & Whitman, 1991). Further, a “true empty category” on the argument position (object position) in Chinese is explored (Duguine, 2014; Li, 2014). Moreover, Takahashi (2008), and Şener and Takahashi (2010) have hypothesized the null arguments as null pronominals in Spanish, Turkish and Japanese, whereas other empty categories are subject to argument ellipsis (AE) and they have distinct interpretational features. The focus of the study is to find out the features of empty nominals in Punjabi in terms of interpretation and derivation by keeping in view the propositions of Takahashi (2012, July) i.e. the null arguments are captured in strict vs sloppy interpretations, and they are also bound to certain morpho-syntactic features of a language. According to Takahashi, AE is defined as omission of nominal category which is the result of sloppy interpretations of anaphors/pronouns, whereas the null pronominal pro on the location of argument is a result of strict co-reference possibilities in terms of anaphors/pronouns within preceding (overt) NP antecedents. Japanese licenses sloppy interpretations as it has both null subjects and null objects (see examples (1) and (2)), whereas in the similar context, Spanish null subjects license strict interpretations (3). According to Takahashi, Japanese licenses AE, whereas Spanish is controlled in terms of the presence of pro on subject position as there is no referential null object.

(1) a. John-wa [zibun-no tegami-o] suteta.John-TOP self-gen letter-ACC discarded‘John_i threw out his_i own letters.’ Mary-mo [e] suteta.

b. Mary-too discarded‘Mary did too.’

• (i)Mary also threw out John’s letters. (strict)

• (ii) Mary also threw out her own letters. (sloppy)

(2) c. Taroo-wa [zibun-no kodomo-ga eigo-o hanasu to] omotteiruTaroo-TOP self-gen child-Nom English-Acc speak C thinks‘Taroo_i thinks that his_i child speaks English.’

d. Ken-wa [e furansugo-o hanasu to] omotteiruKen-TOP French-ACC speak C thinks‘Ken thinks that e speaks French.’

1. Ken thinks that his own son speaks French.(sloppy)

2. Ken thinks that Taroo’s son speaks French.(strict)

(3) a. Maria cree que su propuesta sera aceptada.Maria believes that her proposal will be accepted‘Maria believes that her proposal will be accepted.

b. Juan también cree que e sera aceptada.i. Juan also believes that will.be accepted.ii. ‘Juan also believes that it (= Maria’s proposal) will be accepted.’ (strict)

It is also investigated that the null quantificational arguments also license strict and sloppy readings in Japanese (Takahashi, 2012), in which similar set of entities or a set of distinctive features involve such interpretations regarding the position of empty subj/obj (4). Furthermore, Oku (1998) also posited the same observations of null category.

(4) a. Taroo-wa sannin-no sensei-o sonkeishiteiruTaroo-TOP 3-gen teacher-ACC respects‘Taroo respects three teachers.’

b. Hanako-mo e sonkeishiteiruHanako-also respect‘Hanako also respects e.’i. Hanako respects three teachers too. (sloppy)ii. Hanako respects them too. (strict)

c. Sannin-no mahootukai-ga Taroo-ni ai-ni kita.three-GEN wizard-NOM Taroo-DAT see-to came‘Three wizards came to see Taroo.’

d. e Hanako-ni-mo ai-ni kita.e Hanako-DAT also see to came‘e came to see Hanako too.’i. Three wizards came to see Hanako too. (sloppy)ii. They came to see Hanako too. (strict)

Keeping in view the critical factors involved in licensing the AE, Takahashi focused on two significant theories which may lead to AE; one theory is particularly related to scrambling in a language (Oku, 1998) whereas the other theory is relevant to verbal agreement (Saito, 2004).

Bošković (2018) interpretation of scrambling present in different languages was used by Oku (1998); Japanese licenses the base generation of nominal in scrambled position as phi-features are required to get checked from such elements. Though, the probe and goal’s Agree-relation is weak, therefore, under the movement of nominal phrase, they get license at LF. This kind of movement occurs from higher scrambled position to lower scrambled position. However, another suggestion is made by Oku (1998) that the empty argument positions can be filled with the help of LF copying mechanism instead of lowering of scrambled elements’ position to empty arguments at LF. In copying mechanism, the features’ copying of nomimals at null theta position is involved. In other words, instead of the original argument position, a nominal’s referential index with its particular features is supplied from the other source, allowing the sloppy readings and interpretations of argument ellipses. Essentially, the scrambling in the language allows the LF copying mechanism that subsequently results in patterns of AE, whereas the empty nominal languages do not allow scrambling which subsequently result in interdict the AE. Above mentioned examples (1–4) show that Japanese is a scrambling language that allows AE, whereas Spanish does not license AE as it is non-scrambling language.

The second perspective for analysis of argument ellipsis is verbal agreement condition (Saito, 2004; D. Takahashi, 2006), which determines whether argument ellipsis or only pro in null arguments’ positions are licensed for AE in a language. Japanese with no agreement relation between the nominal arguments and the other functional elements generates null arguments until interpreted at LF (S. Takahashi, 2019). He also believes by following Kim (2007) and Saito (2004) that there is not any failure in phi-feature, which has a dependency on agreement examining between v/T and objects/subjects during derivation. In other words, the verbal agreement relation in permitting AE needs the agreement relation between nominal argument and the relevant probe in derivation. The argument is that in such a language, for purpose of interpretation, if a nominal on an argument position at LF is possibly copied then the need of agreement of the heads like T and v will not be satisfied. The reason behind it is that the features which are related to the copied nominals will be removed in course of following agreement relations, which will be getting affected in those clauses from which they have been instigated in and then they will not be available for copying purposes.

In case of Spanish, the copying attempt of nominal-su propuesta (3a) on the location of empty SUB position (3b) will be failure in terms of agreement of copied nominal and T; thus, it shows that only pro can occupy the said position. The languages like Japanese having Null subject/object with no agreement feature have been thus recommended for licensing of AE. On the other hand, the languages like Spanish having agreement are argued for only licensing of pro, which accounts for interpretation of differences of null arguments in these languages (Huang, 1989).

Takahashi compared the two approaches to AE for the pair of languages such as Japanese and Spanish as both the theories predict the same for these two languages. However, there is a need to study more languages in order to check and differentiate AE’s scrambling and agreement-based hypotheses in a more articulated way, and for this reason, S. Takahashi (2019) subsequently examined Turkish language. This language uses scrambling as well as agreement in specific clauses. AE’s agreement-based hypothesis appears to be clearly supported in the results of Takahashi’s study. For empty nominal in Turkish, the availability of only strict interpretations is reported in clauses where agreement happens. However, both strict and sloppy interpretations are permitted in clauses where no agreement is possible. In fact, this is the true expectation in agreement-based method or approach. On the other hand, according to Takahashi, AE’s scrambling based view point predicts a uniformity of strict interpretation and sloppy interpretation in clauses of all types, and so far, no such case is reported.

Even though in the Turkish Language as reported by Şener and Takahashi (2010) and D. Takahashi (2006) a remarkable preliminary support is provided for agreement-based AE’s hypothesis, a question is also raised about hypothesis’ more empirical validation, which could be answered in other languages having same type of properties regarding scrambling and agreement. In order to assess the agreement-based hypothesis, the South Asian languages in many cases are very natural candidates as scrambling is a prominent feature of these languages and variability of agreement’s existence is predictable. In the current study, Punjabi, which is an Indo-Aryan language, is the point of research. In this language, the omission of subjects and objects is licensed in finite and non-finite clauses. Both subject and object agreements are allowed depending on the variable factors such as case marking and tense. The capability of manipulating the agreement in terms of subject and object in different cases and availability of various agreement patterns in Punjabi permit the worthwhile relative examination of correlation between AE and (absence of) agreement.

Following is the outline of the paper: Section 2 articulates comprehensively the investigation and documentation of empty nominal’s distribution and its interpretation in Punjabi, establishing the fact that this language posits AE. Connection between Agreement, scrambling, bare nouns and AE in Punjabi is discussed in section 3. The precise control of the agreement of subject-object in Punjabi illustrates that AE in this language does not have any dependency on agreement is also comprehensively discussed in this section. Finally, it is concluded that agreement and scrambling do not appear as the most important cross-linguistic elements for determining the general presence or absence of AE in different languages. However, bare nouns approach is more appropriate for licensing of AE in Punjabi.

2. The interpretation of empty nominals in Punjabi

2.1. Patterns of sloppy identity

The omission of objects in Punjabi is possible on regular basis, and it posits sloppy interpretations (see the following examples (5)). Interestingly, only sloppy interpretations are mentioned in the examples as the focus is on AE; however, the strict interpretations are also generally available in Punjabi.

(5) a. Amir apni bivi-nu pyar karda ayAmir self’s_k/f wife/acc/ love do/SG.3/pres/m cop.v/pres ‘Amir_k loves his_k/self’s_k wife.’

b. Ali vi pyar karda ayAli also love do/SG.3/pres/m cop.v/pres‘Ali_m also loves (his_m wife).’

On object position, the omission of quantificational phrases (QP) could be possible resulting in the sloppy interpretations, where in first and second sentences, the QP refers to different objects:

(6) a. Amir do kurian-di izzat karda ayAmir two girls/gen respect do/SG.3/pres/m cop.v/pres‘Amir respects two girls.’

b. Ali vi izzat karda ay.Ali also respects do/SG.3/pres/m cop.v/pres‘Ali also respects (two girls).’

Indirect object’s omission similarly allows sloppy interpretations, as illustrated in (7). Ditransitive verb “likhya” involves two objects; direct object (Od) “khat” and indirect object (Oi) “onu”. In the following example, Oi is omitted which licenses AE.

(7) a. Amir-ne onu khat likhya.Amir/erg his/her-Oi letter/Od write/sg.3/pst/m‘Amir_k wrote a letter to him/her_k.’

b. Ali-ne vi khat likhya.Ali/erg also letter/Od write/sg.3/pst/m‘Ali_m also wrote a letter (to him/her_m).’

Comparing with the patterns that are available with direct objects, indirect objects, and selected PPs, the subjects do not permit sloppy interpretations and parallel patterns of ellipsis. An omitted QP subject in Punjabi likewise has a strict interpretation:

(8) a. Punjabi-de do ustad Amina-di izat karde ne. Punjabi/gen two teachers Amina/acc respect do/p.3/pres OP/pres.3‘Two teachers of Punjabi respect Amina.’

b. Hina-di-vi izat karde neHina/f/acc—also rspect do/p.3/pres OP/pres.3‘(They) also respect Hina.’

The embedded clauses’ subjects could be omitted in Punjabi. Also a standard interpretation of null subjects of this type is strict instead of being sloppy:

(9) a. Amina-nu lagda ay odi kuri khana pkayedi.Amina_m/f/dat feel cop.v/pres her_m daughter/dat food cook/m/sg/fut“Amina_m feels her_m/_k daughter will cook the food.”

b. Hina-nu vi lagda ay pkayedi.Hina/f/dat alsofeel cop.v/pres cook/m/sg/fut‘Hina feels (she) will too.’ (Strict only)

2.2. Derivation of sloppy interpretations

When arguments are null phonetically and permit sloppy interpretations, it is usually considered that they are not null pronouns (pro). The fact is that the overt pronouns do not permit sloppy interpretations as per Otani and Whitman (1991) and Şener and Takahashi (2010). Null arguments’ sloppy interpretations are allowed in Punjabi, but in the presence of overt pronoun, sloppy interpretations get blocked and only strict interpretations become possible (see, section 2.1). Therefore, it is safe to conclude that the empty nominals in section 2.1 allow sloppy interpretations are not simply pro/null pronominals.

Sloppy interpretations are possible on object position in English as an outcome of VP ellipsis/VPE:

(10) a. John will invite his girlfriend to the party.b. Bill will [VP_] too. (strict or sloppy) (i)Bill_k will invite his_k girlfriend to the party. (sloppy) (ii) Bill will invite John’s girlfriend to the party. (strict)

The claim that VPE can lead towards sloppy interpretations in a language like English raises the point, whether it is possible that VPE can lead the similar object interpretations and other VP-internal arguments in Punjabi language? The proposition has been made that the verbs might raise out of VP before VPE in different languages, i.e. Finnish Irish, Hebrew, which subsequently result in object omission on the surface level, and it is also called verb stranding VPE (Goldberg & Approach, 2020; Rouveret, 2012). It is important to check if Punjabi has the same features, allowing the empty VP-internal arguments having sloppy interpretations because of the ellipsis of an entire VP, not just singular arguments’ ellipsis. However, in the following sections 2.3 and 2.4, this idea is investigated thoroughly. Nonetheless, the discussion in the following sections conclude that the verb-stranding VPE is a feature of Punjabi but different other patterns indicate that some other mechanism of AE is involved in generating null arguments where there is no need to omit the bigger VP constituent.

2.3. Adjuncts and ellipsis

It is observed that not only null arguments are present in Punjabi, but adjuncts are also licensed to omit in Punjabi; however, still they are interpreted as present in the sentences having parallel preceding sentences, which are illustrated in (11 & 12). Most probably, this omission is due to the deletion of full VP including arguments and adjuncts while following the driving out the verb from VP, i.e. the occurrence of verb-stranding VPE. There are many types of adjuncts present in Punjabi, for example, time, frequency, length, location, means and manners, and they are supposed to be present in the second sentences of all the examples; however, the said adverbials are not specified by any overt pronunciation. All null elements including adjuncts are written in parentheses, which are as follows:

2.4. Time adjuncts

(11) a. Amina raati aayi. Hina vi aayiAmina/f/sg/nom night come/sg.3/f/pst Hina/f/sg/nom also come/sg.3/f/pst‘Amina came at night. Hina also came (at night).’

Above example (11a) includes a transitive verb “aayi” and an adverbial time adjunct “raati” which got omitted and produced sloppy reading (Hina also came (at night)).

The following example shows a transitive verb with time adverbial where the omission of adjunct is possible with or without omission of direct object. See the following example (11b).

(11) b. Hina-ne kal saib khada.Hine/f/sg/erg yesterday apple eat/sg.3/m/pst‘Hina ate an apple yesterday.’

Sloppy readings:Hina-ne___ saib khada. (Hina ate an apple (yesterday))Hina-ne ____khada. (Hina ate (an apple yesterday)Hina-ne kal___khada. (Hina ate yesterday (an apple))

Above mentioned sloppy readings show that object and time adjunct can be omitted from a sentence having transitive verbs. The observations make it clear that adjunct ellipsis is not conditioned with direct object.

2.5. VP adverbs

(12) Amir-ne cheti cheti safai kitti. Hina-ne vi kittiAmir/m/sg.3/nom quickly dusting do/sg/past Hina/f/sg.3/erg also do/sg/past‘Amir did dusting quickly.’ “Hina did too (dusting quickly).”

2.6. Against a VPE analysis of null arguments in Punjabi

The previous examples show that Punjabi has verb-stranding VPE but there are also many other patterns which depict that the omission of nominal arguments is not only the result of verb-stranding VPE.

First, it has been claimed (Goldberg & Approach, 2020; Rouveret, 2012; Zushi, 2003) that VPE only happens in the languages where there is verb-stranding, i.e. Irish, Hebrew, etc., and when the verbs used in the source sentence and the sentence having null arguments are same. However, VPE is not possible where different verbs are present in source sentence and the sentence having null arguments. Interestingly, Punjabi exhibits different verbs in both the sentences, but still sloppy readings are present. See the following examples.

(13) Hina apny munday-nal pyar kardi ay,Hina/f/sg/nom self’s son/m/sg.3/dat love do/f/sg.3/pres cop.v/pres

par sana nafrat kardi ay.but sana/f/sg/nom hate do/f/sg.3/pres cop.v/pres“Hina_k loves her_k son, but Sana_m hates (her_m son).”

Second, a significant asymmetry is present in the form of elided arguments and adjuncts while still interpreting them as present. For the omission of adjuncts and understood as being present is only possible when deletion of other VP-internal components such as direct object occurs, as illustrated in (11). However, in some cases, it is also possible to omit the VP-internal arguments when other VP-internal omission happens as illustrated in (14), where the indirect object gets omitted in the presence of direct object.

(14) a. Amina-ne onu tohfy pejyAmina/f/sg/erg him/her/Oi gifts/Od send/p/pst“Amina sent him gifts.”

b. Amina-ne __ tohfy pejyAmina/f/sg/erg __gifts/Od send/p/pst‘Amina sent gifts (to him).’

When the quantification of the object is possible, similar patterns can emerge. The verb in omitted object sentences can differ from the verb in the preceding sentence. It also allows the “sloppy” readings in which the omitted object is interpreted differently as compared to the preceding overt quantificational object.

(15) a. parchy-tu-baad Hina-ne do dostan-nu call kitti.Exam/m/sg/gen-after Hina/f/sg.3/erg two friends/p/dat call did

b. lekin, sana-ne _ email kittiBut, sana/f/sg/erg _ email did

‘After the exam, Hina called two friends. Sana, however, just emailed (two [possibly different] friends).’

As shown in the “b” examples, elision of adjuncts is possible and regarded as being present if elision of other VP-internal content is being done concurrently. Other VP-internal content (including the verb) is overtly present in the “c” examples, and the covert presence of adjuncts is not understood to be like that. Only if the direct object is also elided then the frequency adjunct ellipsis is possible:

(16) a. Akram-ne mehbooba-da nawa khat teen vari parya.Akram/m/sg/erg girlfriend/f/sg/gen new letter three timeread/m/sg.3/pst‘Akram read the letter by girlfriend thrice.’

b. Amir-ne-vi parya.Amir/m/sg.3/erg -also read/sg/m/pst‘Amir also read (the letter).’

c. Akram-ne vi o khat parya.Akram/m/sg/erg also that letter read/sg/m/pst‘Akram also read that letter.’ (NOT communicated: ‘twice’)

Although (16) depicts the typical pattern of “adjunct ellipsis”, it is worth noting that contrastive focus has an unusual and complex impact that interacts with the prospective adjunct’s ellipsis; however, it does not interact with the arguments’ ellipsis, implying the point that arguments and adjuncts omit differently (Brattico, 2017). Even though adjuncts are usually not elided and interpreted where direct object is also not omitted, this becomes conceivable in Punjabi where the direct object is contrastively focused. Regardless of the overt, focused objects’ presence, ellipsis and the understanding of presence of an adjunct in the second clause/sentence is feasible if object’s identity in the original sentence and in the succeeding sentence differs, as shown below.

The locative adjunct in the existence of an overtly contrasted object:

(17) a. Akram-nu apny kaar vich ek bomb milyaAkram/m/sg.3/dat hishome ina bombfind/m/sg.3/pst‘Akram found a bomb in his home.’

b. Amir-nu ik chitti mili.Amir/m/sg.3/dat a letter/sg.3/f find/f/sg.3/pst‘Amir found a letter (in his office).’

Focus-licensing is not essential for omission of arguments; it is already observed that direct object can be omitted in the presence of overt indirect object and adjuncts without the latter needing to be contrastively focused (14). An analysis is given in the following that could be applied to the examples of successful adjunct ellipsis (17). The argument is that in Punjabi, contrastively focused objects are lifted out of VP for the purpose of focus-licensing, permitting the occurrence of VPE and the deletion of adjuncts that remain in the VP (but not removing the verb, which in Finnish, Hebrew, Irish, and other languages is presumed to move out of the VP, or the object, which is focus-raised out of VP; Brattico, 2017). Contrastively, it is being assumed that during omission of VP internal arguments, the other VP-internal arguments should not be in focused, and hence is not focus-raised out of VP. Consequently, the omission of nominal could be deduced to implement the individual arguments in situ without having any VPE operation; therefore, Punjabi is considered to be having “genuine” AE.

3. Relationship between agreement and AE in Punjabi

Now the potential relationship of agreement to AE can be investigated as it has been established that AE is a significant phenomenon in Punjabi. AE’s agreement-based analysis predicts that AE will only be licensed when the agreement is absent and will be blocked otherwise when the agreement is present. Now the focus is on how agreement-based analysis performs in Punjabi when different patterns of agreement and null arguments are evaluated.

3.1. Tense and case marking in Punjabi

Changes in tenses and presence of case-marking on DPs can be used to control the agreement that exists on Punjabi verbs. In relation to the verbal agreement, a generalization can be made that in Punjabi tenses excluding the simple past tense, verbs have an agreement with nominative subjects and may also have an agreement with objects. Verbal agreement with objects is possible when (a) there is past tense and ergative case—ne is marked to the subject, and overt case-marking of object is not available (having accusative—nu), and (b) there is not any nominative-case marked to the subject (in a number of modal constructions) and there is no overtly case marked object.

When tenses and case-marking are controlled, then it enables a thorough examination of sloppy interpretations’ availability with both the subject and object having verb agreement of subject/ object presence/absence, as it is shown by sections 3.1.1–3.3.3.

3.1.1. Object ellipsis + manipulation of verbal agreement with object

When there a null object of transitive verb is present in Punjabi, sloppy object interpretations are often permitted, especially when there is not any agreement between verb and the object ((18 & 19), where the objects “car” and “bicycle” are feminine). More importantly, when there is an agreement between verb and the object, as illustrated in (20) and (21). The latter patterning plainly contradicts AE’s agreement-based analytical predictions. For more clear illustration see the table 1.

Table

(18)	a.	Akarm apnigadii Akram/m/sg.3 self ’s/f car/f/sg.3 Vaichy-da. sell/sg.3/f/fut ‘Akramk will sell hisk car.’	b.	Aslam-vi Vaichy-da. Aslam/m/sg.3-also sell/sg.3/f/fut “Aslamm will also sell (hism car).”
(19)	a.	Akram apni saikel Akram/m/sg.3 self ’s/f/sg.3 cycle/f theek kareda. good do/m/sg.3/fut “Akramk will repair hisk bicycle.”	b.	Aslam-viitheek kareda. Aslam/m/sg.3-also good do/m/sg.3/fut “Aslamm will also repair (hism bicycle).”
(20)	a.	Akram -neapni Akram /m/sg.3/erg self ’s/f/sg.3 gadiivaichi car/sg.3/f sell/sg.3/f/pst ‘Akramk sold hisk car. PST-F.Sg	b.	Aslam-ne-bhi vechi. Aslam/m/sg.3/erg -also sell/sg.3/f/pst ‘Aslamm also sold (hism car).
(21)	a.	Akram -ne apni saikel Akram /m/sg.3/erg self ’s/f/sg.3 cycle theek kiti. good do/sg.3/f/pst “Akramk repaired hisk bicycle.”	b.	Aslam-ne-vitheek Aslam/m/sg.3/erg-also good kiti. do/sg.3/f/pst “Aslamm also repaired (hism bicycle).”

Table 1. Summary of availability of sloppy interpretations of null subjects and objects from data in section 2.1

Punjabi	Non-QP subject	QP-subject	Non-QP internal argument	QP object
	*	*	U	U

Thus, it is observed that agreement of verb with object has no impact on the sloppy interpretations of objects.

3.1.2. Subject ellipsis + manipulation of verbal agreement with subject

When the embedded finite clauses’ subjects are null and there is an agreement between verb and the subject, then the null subjects’ sloppy interpretations do not be available freely, which are illustrated as follows (22).

(22) a. Amir soch-daay odi kuriAmir/m/sg.3/nom think/sg.3/m/asp cop.v/pres his /f/sg.3 daughter/f/sg.3/nomPunjabi pardi-pai ayPunjabi study/sg.3/f/asp cop.v/pres‘Amir_i thinks his_i/_k daughter is studying Punjabi.’

b. Akram-vi soch-da ay Punjabi pardi-paiAkram/m/sg.3/nom -also think/sg.3/m/asp cop.v/pres Punjabi study/sg.3/f/aspaycop.v/pres‘Akram_m also thinks (his_i/_k/*_m daughter) is studying Punjabi.’

When there is no agreement between verb and subject (the case where the verb has the past tense and the subject has been marked ergative -ne), the patterning is the same:

(23) a. Aslam soch-da ay odi kuri-neAslam/m/sg.3/nom think/sg.3/m/asp cop.v/pres his /f/sg.3 daughter/f/sg.3/ergPunjabi pari ayPunjabi studied/f/sg.3/pst cop.v/pres‘Aslam_i thinks his_i/_k daughter studied Punjabi.”

b. Amirvi soch-da ay Punjabi pariAmir/m/sg.3/nom also think/sg.3/m/asp cop.v/pres Punjabi studied/f/sg.3/pstaycop.v/pres‘Amir_m also thinks (his_i/_k/*_m daughter) studied Punjabi.’

An easy availability of the subject’s sloppy interpretation is not present in (23) just as it is not easy to come by when there is an agreement between verb and the subject. Thus, it is observed that verbal agreement with verb has no impact on the sloppy interpretations of null subjects as the agreement of the verb with object has no impact on sloppy interpretations of the null objects (see table 1).

3.1.3. Null nomianls (dative-marked subject) + no verbal agreement with subject

Further, it is also observed that the subjects marked with dative case and having no agreement with the verb, as per the AE’s agreement-based approach, such patterns are predicted to license the sloppy interpretations once again. However, the observations suggest that there are no sloppy interpretations when the dative case is marked on the subject. See the following examples (24):

(24) a. Akamsoch-da ayAkram/m/sg.3/nom think/sg.3/m/asp cop.v/pres[odi kuri-nu Amir pasanday].his /f/sg.3 daughter/f/sg.3/dat Amir/m/sg.3/nom like cop.v/pres‘Akram_k thinks his daughter_k/_j likes Amir.’

b. Akram-da-pra soch-da ayAkram/m/sg.3/gen -brother think/sg.3/m/asp cop.v/pres[Alipasanday].Ali /m/sg.3/nomlike cop.v/pres‘Akram’s brother_m thinks (his_k/_j//*_m daughter) likes Ali.’

(25) a. Akam soch-da ayAkram/m/sg.3/nom think/sg.3/m/asp cop.v/pres[odi kuri-nu seikal cahiye-da ay].his /f/sg.3 daughter/f/sg.3/dat cycle want/sg.3/m/asp cop.v/pres‘Akram_k thinks his_k/_j daughter needs a bicycle.’

b. Akram-da-pra soch-da ayAkram/m/sg.3/gen -brother think/sg.3/m/asp cop.v/pres[computer cahiye-da ay].computer want/sg.3/m/asp cop.v/pres‘Akram’s brother_m thinks (his_k/_j/*_m daughter) needs a computer.’

Indirect object-arguments marked with dative-nu can be contrasted/compared with the examples where subjects are also marked with ergative-ne. Example (26) is the case where corresponding null arguments’ sloppy interpretations are (easily) possible in latter cases.

(26) a. Akram-ne apni kuri-nu computer dittaAkram/m/sg.3/erg self ’s/f/sg.3 daughter/f/sg.3/dat computer give/v/m/sg/pst‘Akram_k gave his_k daughter a computer.’

b. Amir-ne saikel dittiAmir/m/sg.3/erg bicycle give/v/f/sg/pst‘Amir_m gave (his_m/_k daughter) a bicycle.’

Precisely, different patterns of AE and agreement are investigated in Punjabi, which appears to reflect comprehensive, substantial and empirical evident against the relationship between agreement and sloppy interpretations of null arguments (AE). The sloppy interpretations’ availability appears to be unaffected by agreement’s presence or absence on Punjabi verbs. Whether or not these factors induce agreement on verb, understandings of such types are freely and equally possible/available with objects. Elided subjects’ sloppy interpretations availability is supposed to get unaffected by the presence/absence of verb-subject agreement. Summary of these patterns is presented in Table 2, which shows the availability of null arguments’ sloppy interpretations in either case where there is a potential agreement between the verb and the argument or where there is not. The table shows that in presence of a null argument, the agreement’s presence or absence has no variable effect on the sloppy readings’ availability and thus argument ellipsis becomes possible.

Table 2. Summary of availability of sloppy interpretations of null subjects and objects matched against agreement, from data in sections 2.1 and 3.1–3.3

Punjabi	Subjects		Objects
	Verb agrees with subject	Verb does not agree with subject	Verb agrees with object	Verb does not agree with object
	Sloppy not ok	Sloppy not ok	Sloppy ok	Sloppy ok

In general, an examination of AE patterns in Punjabi language (when the agreement patterns could be amended) specifies that anti-agreement-based and scrambling approaches for licensing AE are disproved. The presence of agreement in South Asian languages evidently does not look like a major element limiting AE and sloppy interpretations’ availability; however, AE/sloppy interpretations licensing normally functions in a fashion, which is independent of agreement. Similar to anti-agreemen approach, previous researches (Sakamoto, 2016; Sato & Karimi, 2016; Sugisaki, 2012; S. Takahashi, 2019) have investigated that argument ellipsis also occurs in scrambling languages such as Hindi, Malayalam, Turkish, Bangla and Japanese, and Punjabi is also a scrambling language having AE. However, Li (2014) and Huang (1989) have also confirmed the presence of AE in Chinese, which is not a scrambling language. Therefore, it can be established that the scrambling is also not the only reason for argument ellipsis in all languages across the world like anti-agreement approach.

Other than scrambling and anti-agreement, another approach is very appropriate for the analysis of AE in Punjabi, i.e. bare nouns approach (Hoji, 1998), which is precisely discussed in the following section.

4. Bare nouns-approach

Hoji (1998) suggested that bare nouns are responsible for sloppy interpretations in Japanese, and these bare nouns are marked either indefinite or definite nouns (Izumi, 2011). See the following examples:

27. a. John-ga zibun-o suisen-shitaJohn-NOM self-ACC recommended‘John recommended himself.’

b. Bill-mo _ suisen-shitaBill-also recommended‘Bill also recommended ec.’ ec interpreted as a + definite N = personec identified as either (a) John = strict (b) Bill = sloppy

The above example shows that definite interpretation is available for null argument. The (b) example is interpreted as “the person”- a definite identity; “the person” could be read as strict interpretation such as “John” and for sloppy interpretation, “the person” would be considered as “Bill”. So far, all the languages who exhibit AE also show the features of bare nouns, which can be interpreted as either definite or indefinite. Further, those languages that have null subjects without argument ellipsis do not permit the definite and indefinite interpretations for bare nouns. Significantly, Punjabi also allows definite and indefinite interpretations of bare nouns for AE. The examples of Punjabi mentioned in previous sections can be interpreted in terms of definite and indefinite interpretations for bare nouns, for example, 18 (a/b) and 28(a/b), respectively. 18 (a) is Akram_k will sell his_k car’ whereas 18 (b) “Aslam_m will also sell (his_m car)” can be considered as definite interpretation of bare noun such as “his car” or “Akram’s car”.

28. a. Amir-ne apny daftar-wich ik computer rakhya.Amir/erg self ’s office/loc a computer place/sg.3/pst/f‘Amir_k placed a computer on his_k desk.’

b. Ali-ne vi ik computer rakhya.Ali/erg also a computer place/sg.3/pst/f‘Ali_m also placed__ in his_m office (a computer).’

28(a) is “Amir_k placed a computer on his_k desk” whereas 7 (b) “Ali_m also placed- in his office (a computer)” can be considered as indefinite interpretation of bare noun such as “a computer”.

Therefore, it can be assumed that bare nouns approach is more compatible for all the languages bearing AE as compared to scrambling and anti-agreement approaches.

5. Conclusion

This paper focuses on determining the potential identity of null arguments in Punjabi. It also explored the phenomenon of AE, which is used as syntactically mediated process for the identification of null arguments by providing sloppy interpretations. The process of omission of nominals is same as VP ellipsis; however, the full VP-level constituent is not deleted. A special attention was given to find out different linguistic factors, which license the argument ellipsis in Punjabi. It was argued during the course of investigation that Punjabi is a language, in which AE is responsible to produce empty null arguments on different positions such as direct objects, indirect objects, selected PPs and subjects.

An analysis of null subjects in Punjabi concerning pre and post verbal placement of complement clauses provides a method for further exploration in this language, especially the effective AE’s essential licensing scenario in which the anaphors instead of pronouns in antecedents NPs are copied on the ellipsis position at LF. In addition, it was determined that anti-agreement methodology to license AE is not confirmed as a universal proposition, Punjabi posits many agreement patterns, which can occur with null subjects/objects. Cross-linguistically, it is well known that the use of a substitute scrambling approach in order to permit AE is also not a reliable predictor, given that languages like Chinese and Vietnamese demonstrate AE patterns but do not scramble.

Other than scrambling and anti-agreement, another approach is very appropriate for the analysis of AE in Punjabi, i.e. bare nouns approach (Hoji, 1998). He suggested that bare nouns are responsible for sloppy interpretations in Japanese, and these bare nouns are marked either indefinite or definite nouns; however, this study also found this approach more suitable for Punjabi argument ellipsis. This approach will be more beneficial for further cross-linguistic examination of AE and would be regarded dynamic for future studies.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This article has not received any funding. Please change it accordingly.