Null Expletives in Malay

Malay is not known to have expletives due to their absence in the grammar, but this does not necessarily entail that expletives do not exist altogether. Gaps are observed in constructions with positions in which one would expect an expletive in a strictly non-pro-drop language such as English, hence making it seem as though Malay has a phonetically null variety of expletive. This phenomenon is common in attested pro-drop languages; however, data from Malay have not been properly documented and a principled account of the hypothesised existence of null expletives in Malay is lacking in the literature. This paper attempts to confirm whether Malay does have expletives by examining certain null-subject constructions through Chomsky’s (1982) Extended Projection Principle and comparing Malay with English and Irish. All cited examples in Malay are from the online archive of Utusan newspapers and those in Irish are from McCloskey (1991) and McCloskey (1996). All other examples are from consultants who are native speakers. It is found that Malay obeys the EPP, based on the obligatory movement of a subject into TP, which consequently makes it necessary for the use of a null expletive when no other constituent is able to raise to satisfy the EPP. The merging of a null expletive therefore creates an expletive-argument chain that causes the definiteness restriction in the existential construction, which can be overridden by substituting the expletive with clitic -nya on the verb.


INTRODUCTION
Expletives constitute some of the most mysterious words there are in human languages. According to the Minimalist Program by Chomsky (1995), languages obey principles of economy and grammatical constructions should have as few constituents as is grammatically possible. Regardless, there seem to be certain phenomena that apparently do not observe such principles. For instance, pleonastic subjects, or more commonly known as expletives, occur in a great many of the world's languages despite their being meaningless place markers for subjects in a clause. Indo-European languages such as English and French are taken as typical examples of languages with expletives subjects. The French examples below show that the pronominal subject of the clause is non-referential, such that it has no antecedent. In other words, the pronouns do not refer to anything and are meaningless.
(French) 3.SG COP clear COMP-he PRF slept 'It is clear that he has slept.' b. Il y a un oiseau. there LOC has a bird 'There is a bird.' It seems as though the existence of expletives goes against such principles of economy as they do not contribute to the meaning of an utterance but exist nonetheless. This being the case, languages can be said to be better off without them as they are redundant and meaningless elements of a clause. Nevertheless, there must be an underlying linguistic principle which governs the possibility of there existing such semantically vacuous words.
In spite of the apparent meaninglessness of expletives, they are obligatory in languages such as English and French for reasons that concern agreement and the obligatory presence of the subject in a clause. According to Chomsky (1998), expletives serve the purpose of satisfying the Extended Projection Principle (EPP). They do not affect full interpretation at LF, which means that their existence is purely functional and concerns the syntax of an expletive construction to the exclusion of the semantics.
Some other languages appear to be more logical in the sense that expletives do not occur anywhere. In Malay and Mandarin Chinese, expletives are not employed by the syntax, even though the canonical word order of these languages is similar to that of languages that do have expletives, like English and French - SVO. 1 (2) a. Jelas bahawa dia telah tidur.  (2) and (3) illustrate that there is a gap preceding the main predicate of the clause where subjects are usually realised. The absence of overt subjects in the examples above begs the following question: does the absence of any overt linguistic material in such a gap position entail that there is nothing occupying said position?
The aim of this study is to investigate the syntax and semantics of a small class of constructions in Malay that lack overt subjects to substantiate the overlooked claim that expletives exist in Malay, although they lack phonological form. In order to do that, a comparison is done between English (a language known to have expletives), Irish (a language argued to lack expletives), and Malay (a language claimed to have expletives) with regard to the EPP and the definiteness restriction in existential constructions. Novel data on clitic doubling in Malay is also presented to further consolidate the argument that null expletives are responsible for the definiteness restriction observed in existentials in Malay.

OVERVIEW OF EXPLETIVES TESTING THE PRESENCE OF NULL ELEMENTS
Many languages are attested to have null elements, both heads and phrases, such as pro, PRO, zero copulas, and more. The existence of most of these postulated zero elements is supported by various semantic and syntactic evidence. For example, there have been various studies on control and the presence of PRO in non-finite control clauses since the early days of generative grammar (see Chomsky, 1981;Martin, 2001;Radford, 2016;Sigurđsson, 1991)  The examples above demonstrate that condition A of binding theory is not violated despite there not being any overt antecedent binding the reflexives within the same clause. It should not be possible for the reflexives to occur in the embedded clauses if there were no antecedents to bind them. This predicament has led to the postulation of PRO as subjects of nonfinite clauses. In such control clauses, PRO is analysed to occupy the subject position preceding the predicate and acts as the antecedent that binds the reflexive.
In languages which allow pro-drop, such as Italian, the subject of a clause may be dropped. Thanks to the rich morphology of Italian, the subject is recoverable through conjugations on the verb, which agree with the dropped subject, marking φ-features such as person and number on the verb itself. TABLE 1 shows the different conjugations on the verb despite there being no overt subject with which the verb could visibly agree. If it were taken that there was no subject at all, the verbal conjugations would not have been able to surface, as no agreement relation could have been established. A semantic diagnostic test that takes advantage of binding relations and a syntactic diagnostic test using agreement markers could be used to detect the presence of some nominal in the covert syntax. Unfortunately, binding only works with referential nominals, so it cannot be used to detect a DP that might not have any semantic content. To make matters worse, it is not possible to use the other diagnostic test using agreement markers if the language being studied is morphologically impoverished.
Malay is known to be an analytic language. What this means is that the language makes minimal use of any inflectional morphology to indicate relations between constituents. Therefore, it is not possible to determine the presence of a phonologically null subject using agreement markers, as in Italian. This is further exacerbated by the fact that expletives are nonreferential. So, it is not totally straightforward how one can substantiate the claim that there might be an invisible, semantically vacuous nominal in a language that does not have agreement markers of any sort.
Constructions like those in (2) present a challenge as it is difficult to test the presence of linguistic material in the subject position when binding and agreement tests cannot be used. If it truly is empty, the constructions would make up an exceptional class of construction in the language as no other kind of construction patterns with them. Instead of accepting this apparent exceptionality and further complicating the grammar of Malay, it is hypothesised that the language has a silent variety of expletive which would account for the presence of gaps in environments where an overt subject is expected. The postulation of this null expletive also makes it possible to explain the restriction on definite nominals in existential constructions in the language, as will be explained in following sections.

COMPARISON BETWEEN ENGLISH AND MALAY EXPLETIVE CONSTRUCTIONS
As shown in examples (1a-b), English and French are languages that employ overt expletives to satisfy a syntactic requirement known as the EPP, first introduced by Chomsky (1982), which makes it obligatory that all clauses have a subject, whether it be overt or covert, There are attested to be 2 kinds of expletives in English: it and there. The former is used with clausal associates, whereas the latter is used with nominal associates 2 , and both types have their own unique morphosyntactic properties. For example, since there occurs with nominal associates, there tends to be a requirement of some sort for case in there constructions. Previous analyses of existentials have claimed that the expletive receives nominative case and transmits it to the associate through the chain formed between the two constituents (see K. Safir, 1982). In a language which makes case visible outside the pronominal paradigm, such as German, nominative case marking is seen on the nominal associate of the expletive. The same is not said of it constructions as the clausal associate does not require case, as is commonly assumed.
(German) there was a.NOM bird 'There was a bird.' Further, agreement is established between the copula and the nominal associate, not the expletive, as the expletive is invariant and plural marking on the copula is seen when the nominal associate is in the plural. In constructions with it, the verb agrees with the expletive as coordinated clausal associates do not trigger plural agreement on the copula.
a. There {*is/are} birds on that branch. b. It {seems/*seem} that he slipped and that he died. On the other hand, in Malay, the subject position of constructions in which an expletive is expected is not occupied by any sort of linguistic material. There seems to be a very consistent absence of the subject, even in nominal expletive constructions such as existential and unaccusative constructions, unlike Italian in which even though an equivalent of it is absent, the equivalent of there (ci)often analysed as a locative cliticis present in the language. Based on the observation that the subject in Malay is absent in the same environments in English, a preliminary (but rather shallow) assessment would be that Malay is the opposite of English with respect to expletives.  Salleh (1995) states "Malay does not allow referential pronouns to be dropped but requires that all kinds of pleonastic subjects be dropped" (p. 30), and argues that there are no expletives in Malay that correspond to it and there in English. This claim is made based on the eISSN: 2550-2131 ISSN: 1675-8021 229 observation that the subject is missing in constructions akin to those in TABLE 3; however, further evidence to support or refute the postulation of expletives is not provided. In concord with the objective of this paper, Abd. Wahab and Nomoto (2017) argue against Salleh (1995) and claim that there exists a null variety of expletive in Malay. Other than that, Selvanathan (2018), postulates null expletives in null-subject constructions of the tough kind, but does not provide further elaboration. It is apparent that, although there are different scholars who have said something about the (non-)existence of expletives in Malay, any argumentation on the topic has not been materialised. Claims have been made based on gaps in null-subject constructions in Malay, but no actual evidence has been brought forth to argue whether expletives occupy those gaps.
The null hypothesis is that Malay does not have any expletives, be they overt or null. To claim that the constructions in TABLE 3 do not have expletives or any form of linguistic element in subject position would necessarily require a rigorous explanation as to how it is possible for an SVO language not to have subjects in a very narrow set of constructions. This is obviously a counterintuitive move as it introduces a very specific exception to the fact that all other sentences and constructions in Malay do have overt subjects, notwithstanding cases of ellipsis and other forms of deletion. If the null hypothesis is correct, this counterexample to a widely accepted generalisation must be addressed and an explanatory account of how such constructions could avoid the principle that all clauses must have a subject, needs to be formulated. That is, of course, not the position taken by this paper.
To reiterate, what this study seeks to achieve is a confirmation of there being a null variety of expletive in Malay, which would account for the gaps in the subject position of the constructions in TABLE 3.

EXPLETIVES AND THE EXTENDED PROJECTION PRINCIPLE
The presence of an EPP feature on T 0 in a language is what determines whether it has expletives. This is because expletives are generated to satisfy the EPP feature when no other nominal can, by causing the non-referential element, to surface in subject position. For example, in a raising construction with a CP complement in a language such as English, the T head probes for a nominal with which to agree but raising predicates do not assign any external θ-roles. According to Chomsky (2008), it is not possible for a head to probe into CPs as they constitute phases, and probing would cause a violation of the phase impenetrability condition. Effectively, the absence of any sort of nominal should cause the derivation to crash due to the EPP feature not having been satisfied; however, the derivation continues with the use of an expletive. Since the raising predicate in (8) takes a clause as its complement, expletive it is generated. There is no other means but to use an expletive to ensure that a subject is realised, and the derivation converges.
As for expletive there, consider unaccusative verbs. The EPP is intricately interwoven in the analysis of unaccusative verbs by Burzio (1986). This generalisation correlates the assignment of external θ-roles with the assignment of accusative case, such that if a verb cannot assign an external θ-role, it cannot assign accusative case. Consequently, passive verbs are also regarded as unaccusative verbs as they cannot assign an external θ-role. The agent is not an argument of the verb but is realised as an adjunct by-phrase, and the theme which originates as the object of the verb is promoted to subject status because of the unaccusative quality of eISSN: 2550-2131 ISSN: 1675-8021 230 passive verbs. The actual object is raised to the surface subject position to satisfy the EPP and to obtain case, which cannot be assigned by the verb.
Burzio's generalisation predicts that the single theme argument of unaccusative verbs is forced by the EPP to raise from object position to subject position. In cases where the option of the argument remaining in-situ is open, an expletive is used to satisfy the EPP and prevent the derivation from crashing.

SUBJECTS IN A VSO LANGUAGE
According to Haider (2013), "OV languages do not implement obligatory functional subject positions and therefore they do not need obligatory expletives if the obligatory functional subject position cannot be lexicalised" (p. 222). He also points out that such a position is a genuine SVO requirement. This entails that the subject of SOV and VSO languages is not required to move to TP as the specifier of TP is not lexicalised. If this is true, examining a non-SVO language should provide a relevant point of comparison for Malay, an SVO language.
McCloskey (1996) argues that Irish, a VSO language, lacks the EPP, which obviates the need for expletives. Based on his observations of a construction that he calls salient unaccusatives, he argues that it is not possible to postulate expletives in Irish, thus dispelling the idea that they exist in the language.
(Irish) decreased on his strength 'His strength decreased.' The salient unaccusative construction is characterised by a verb that selects a PP but lacks a subject, be it referential or expletive. The only DP in the construction, having been satisfied of its case needs by the preposition, is no longer active for probing. If it is assumed that there is a probe that seeks to check an EPP feature, the derivation of the construction would be jeopardised, as there is apparently no DP that can satisfy that probe. As seen in (9), there is neither a compatible DP nor an expletive that can rescue the construction from the scenario, but the construction is grammatical nevertheless.
He takes the 2 possible analyses of the linking of expletives and their associates by Chomsky (1986) and explains that the Irish data cannot be compatible. On the expletiveargument chain analysis, because the so called salient unaccusative verb takes a PP as its complement, it is not possible for the PP to enter an expletive-argument chain, given that a chain needs to form between two constituents of the same syntactic category. As for the analysis with LF movement of the associate to SpecTP, the PP complement cannot adjoin the expletive as, again, the two constituents are not of the same syntactic category. The conclusion is that, if it is assumed that Irish does not have expletives to begin with as a direct consequence of the language not obeying the EPP, many of the problematic data regarding salient unaccusatives, perfective passives and passives of intransitives can be explained.
One then needs to ask whether Malay belongs in the special class of languages that do not have expletives owing to the absence of the EPP feature. Although Malay is an SVO language like English, it is necessary to determine whether the language has an EPP feature, just to be sure of the possibility of it also having expletives.

METHOD
A comparative method is utilised to see whether Malay patterns with English (a language that is known to have expletives) or Irish (a language that is argued to lack expletives) with respect to the EPP by Chomsky (1982). By observing the movement of subjects and any restrictions on definite nominals in existentials in Malay, it can be determined whether the language behaves according to the effects of the EPP and expletives that are possibly present in a clause. Therefore, the postulation of null expletives in the language can be motivated, if it is observed that Malay obeys the EPP.
Formally, the movement of a nominal to TP and the use of expletives in a language are attributed to and caused by the EPP. Inversely, if the T head of a language does not have an EPP feature, there should be no requirement that a nominal occupy TP, hence no use of expletives. In other words, the absence of the EPP feature on T 0 precludes the following: movement of a nominal to TP; the generation of expletives. Consider the following argument: If a language obeys the EPP, it requires a subject in TP Premise 2: Language X obeys the EPP Conclusion: Language X requires a subject in TP If any of the premises is false, it is then possible to conclude that the language does not have expletives, unless the language has some other mechanism to satisfy the EPP without violating other conditions. Assuming that premise 1 is true, as according to Chomsky (1982), then premise 2 needs to be examined. There does not seem to be any other condition on the use of expletives in the literature, so the logic presented in the argument above should hold true. Through this line of reasoning, in order to determine whether there are expletives in a language, one could ask whether the language obeys the EPP.
The argument above can be applied to English and the conclusion is known to be valid as there is overt movement of the subject from its base-generated position within vP. In the event that a referential subject cannot occupy SpecTP to satisfy the EPP, an expletive is merged and occupies the position. Therefore, the next step is to see whether Malay behaves in the same way by examining subject movement, if any, in the language.
The authentic linguistic examples in Malay that have been cited in the following section come from the online archive of one of Malaysia's most popular Malay-language newspapers, Utusan Malaysia. By a simple Google search of the relevant predicates, such as berlaku (occur) or wujud (exist), using the Google search operator [" "], an exact match can be found and selected. In addition to that, use of the operator [site:] is required in order to limit the search to the Utusan website. For example, the following string is used to search for the word berlaku in the Utusan website: site:www.utusan.com.my "berlaku". As for the Irish data, the examples are taken and cited from McCloskey (1991) and McCloskey (1996)

ANALYSIS OF EXPLETIVES IN MALAY SUBJECTS AND TP
It is possible to take the position of the subject, relative to other constituents such as the verb and auxiliaries, to determine whether a language obeys the EPP. A language without an EPP feature is expected to have its subject remain in its base-generated position or between that position and the highest inflectional position (i.e. TP). There should be no motivation for the subject to move to TP if the language has no EPP feature. This is evident in OV languages in which the subject remains VP-internal, in accordance with the claim by Haider (2013), or raises to a position between the verb and the highest inflectional position (e.g. AgrP) to check for agreement, as is suggested by data in Irish by McCloskey (1996). In OV languages, the subject does not raise to TP because such languages are believed to lack an EPP feature.
Raising of a nominal from a subordinate clause to a matrix clause is attested to be an instance of A-movement, i.e. the need to obtain case and to check the EPP feature in the matrix clause. For example, a reporting verb such as 'claim', in the passive voice, is able to move the subject of a non-finite subordinate clause to its matrix subject position by virtue of the nonfinite clause lacking nominative case. Such a condition leaves the subordinate subject caseless and active for probing by matrix T 0 , which seeks to dispense its nominative case feature and to get its EPP feature checked. Subsequently, the agreement relation between the 2 constituents triggers movement of the subordinate subject to the TP of the matrix clause.
In the case that the complement of the passive reporting verb is a finite clause, the subordinate subject is inactive as it has been assigned nominative case by subordinate T 0 . An expletive occupies SpecTP as, without it, the derivation of the clause would crash because of the unchecked EPP and case features. English, which readily obeys the EPP, exhibits this behaviour, as does Malay. The only difference is that expletives in Malay are null.
3.SG PASS-claim PRF sleep throughout robbery DIST 'He is claimed to have slept throughout the robbery.' b. Ø di-dakwa dia telah tidur sepanjang perompakan itu. Pro PASS-claim 3.SG PRF sleep throughout robbery DIST 'It is claimed that he had slept throughout the robbery.' According to McCloskey (1991) verbs in Irish undergo V-T movement but not T-C movement in normal circumstances. This should yield word order in which the subject of a clause follows the verb. On the other hand, in clauses with auxiliary verbs, it intervenes the verb as well as any possible auxiliaries, which are attested to occupy the inflectional layer above the VP. The examples below by McCloskey (1991) show just that: If a language obeys the EPP, the subject should be driven to move to the highest inflectional position, TP. Therefore, it should be possible to hypothesise that languages whose subjects raise past verbs and auxiliaries to TP do possess an EPP feature. This is seen to happen in English, and it so happens that Malay is the same in this respect. Examples (12a-b) show that the subject precedes the verb and auxiliaries, which means that the subject has moved past intermediate inflectional and verbal elements to TP.
3.PL NEG might PROS can ACT-pursue progress 'They might not be able to pursue progress.' ("Melayu ditindas jika abai revolusi maklumat," 2001) b. Kawasan tersebut mungkin sudah tidak lagi sesuai. Area mentioned might PRF NEG anymore suitable 'The area mentioned might not be suitable anymore.' (Yatimin, 2014) Certain unaccusative verbs make it obligatory for the argument to raise to subject position. Die is such a verb in English, and in Malay it is mati. 4 The argument cannot remain in-situ and the use of an expletive is not permitted. Assuming that there is no other requirement on the argument to raise except the EPP, there should be no reason for the argument to raise in a language that does not obey the EPP because nominative case can be assigned remotely via agree. Because English is known to obey the EPP, the argument moves to SpecTP. In Irish (13a), the argument moves to AgrSP to check for a strong agreement feature, but not to SpecTP.
Unsurprisingly, Malay patterns with English such that the argument moves to SpecTP, as in (13b). Because subject-verb agreement is non-existent in Malay, there is no AgrP in the language and there is no reason for the argument to raise above the VP other than the EPP. (Malay) Mary PRF die 'Mary has died.' Another relevant unaccusative verb in Malay is gagal (to fail). In Irish, the argument of this unaccusative verb may come inside a PP, whose head obviously can assign accusative case. Since PPs do not agree, they should therefore not be forced to move to AgrSP in Irish. Again, since Malay obeys the EPP, the DP obligatorily moves to SpecTP. (Malay) 3.PL PROG fail 'They are failing.'

EXPLETIVE-ARGUMENT CHAINS
The presence or absence of the EPP feature in a language has repercussions on other aspects of the language such as the presence of expletives and expletive-argument chains. Linguists such as K. J. Safir (1985) have observed that existential constructions and other constructions with expletive there as the subject exhibit an incompatibility with definite DPs. This is claimed to be a property of expletive-argument chains, such that the argument in a chain created by an expletive and its associate must be indefinite. 'There is (the) bird.' By reversing this causal relation, it is possible to determine whether a language has an EPP feature and expletives. In other words, if a language exhibits a definiteness restriction on arguments of unaccusative verbs, which have not been promoted to subject, there can be said to be an expletive-argument chain, which can further be traced back to the EPP.
Consider examples (17a-b) and (18a-b), which illustrate that Malay patterns with English in that definite DPs cannot occur in existential constructions. Based on this observation, it can be said that an expletive-argument chain is present in existentials in Malay, meaning that the language does have underlying expletives.  That Malay shows signs of the definiteness restriction in the existential construction is not surprising, given that it has been determined that the language obeys the EPP. The definiteness restriction is a sign that a null expletive occupies the subject position of the clause and heads the chain whose tail is the offending definite DP.

OVERRIDING THE DEFINITENESS RESTRICTION WITH CLITIC DOUBLING
An interesting observation is that the 3 rd person clitic -nya can optionally cliticise onto existential verbs. The definiteness restriction can be overridden by the cliticisation of -nya, which seems to be an instance of clitic doubling in an environment whose definite argument would otherwise render the sentence ungrammatical. (19) a. …bahawa terdapat=nya perkara itu untuk meŋ-tarik minat pembaca.
…COMP EXIST=3 matter DIST for ACT-pull interest reader '…that the matter exists to capture the interest of the reader.' ("Utusan-AlHijrah kongsi maklumat dengan pendekatan sinergi," 2016) b. Ramai orang tahu akan berlaku=nya pembohongan itu. many person know PROS occur=3 lie DIST 'Many people know that the lie will occur.' (Mohmad Shaid, 2007) Example (19b) shows that movement of the definite DP has not occurred as it remains within the VP with the aspectual marker preceding both the DP and the verb. This can be further supported by the following example showing that movement of only the topmost auxiliary has occurred, leaving behind the lower auxiliary and the verb. The inverted auxiliary is affixed by interrogative -kah, meaning that it has moved to CP whilst everything else remains in the TP.  Moreover, the argument of an existential construction is argued not to raise to TP as there is a difference in interpretation between "moved" and unmoved DPs as noted by Den Dikken (1995) in examples (23a-b). If it were true that the DP can replace the expletive by movement, be it at PF or LF, example (23b) should rightfully be grammatical. To the contrary, movement renders the example ungrammatical. (23b) is not a paraphrase of (23a), which is an unanticipated result, as replacing an expletive with a referential DP should not alter the interpretation of a sentence. Example (23b) could arguably be interpreted as a locative construction, but it definitely does not mean the same thing as example (23a). This difference in construction is dealt with in the next section.
Since existentials do not have the option to move the DP to the TP to satisfy the EPP, an expletive should be merged; however, a problem arises if the DP is definite, which would lead to the formation of an illegitimate expletive-argument chain, as definite DPs cannot enter the chain. To ensure that the derivation converges, a clitic copy of the definite DP is generated and forced to raise to TP by the EPP feature. Therefore, it appears that, in lieu of merging an expletive to satisfy the EPP, existential clauses have the alternative of clitic doubling, as do other nominal expletive constructions, as in examples (19a-b). In other words, the clitic can be said to replace the null expletive, which avoids the formation of an expletive-argument chain. As a consequence, the absence of such an expletive-argument chain makes it possible for a definite DP to occur as the argument of nominal expletive constructions. Otherwise, if a chain were formed, the construction would prevent the cliticisation of -nya, and this would cause the derivation to crash, as the offending definite DP could not enter a chain with the expletive due to the definiteness restriction. It is commonly assumed in Minimalism that the clitic in clitic doubling is a D head that undergoes head movement from the DP to its host. For example, Nevins (2011) argues that the clitic moves from what is called a Big-DP structurea DP to which the clitic adjoinsto its landing site. By adopting the movement approach, it can be analysed that the clitic undergoes head movement to the verb host before the complex head formed is subsequently triggered by the EPP feature on T 0 to move to it, picking up any auxiliaries available along its course. 5 Finally, the clitic is assigned nominative case by T 0 , if needed at all. 5 Although the verb does not usually move to T in Malay, it is possible in polar questions for the verb to move to C via T, especially in spoken Malay. The following examples show that the verb has undergone head movement to CP, forming a complex V-T-C head with the auxiliary and the interrogative particle: How case is assigned is of no particular importance in this study. Whether the associate obtains case via transmission from the expletive or by other means does not affect the main thesis of the study, which is that Malay obeys the EPP and expletives exist in the Malay language.

TP T [EPP] Akan berlaku=nya
To summarise, Malay patterns with English with regard to the movement of the subject to TP and the definiteness restriction in existentials. Both languages are the exact opposite of Irish, which has been argued extensively to lack an EPP feature as well as expletives. The findings unveiled from the comparison of these languages should be construed as evidence that Malay obeys the EPP and has expletives, albeit null.

DISCUSSION EXISTENTIALS VS. LOCATIVES
Instances in which a strong quantifier such as semua (all) occurs in what appears to be an existential clause with the verb ada can be found, and it seems, at first blush, as though such examples constitute a counter-example to the analysis presented above. Consider the example below: (24) Ada siapa lagi di kenduri tadi?
Ada semua kawan kita. EXIST who else LOC feast just.now EXIST all friend 1.PL.INCL 'Who else was at the feast just now?' 'All of our friends were (there).' The quantifier should render the answer to the question ungrammatical because an expletive cannot form a chain with a DP with a universal quantifier such as 'all'. However, it is clear that the example has a locative reading, instead of an existential reading. It can be said that the locative adverbial is omitted in the answer, provided that it is given in the question.
The difference between example (24) and the analysis beforehand is in the type of construction. It must be acknowledged that existentials are different from the type of construction in (24), which happens to be a locative construction. Although existentials and locatives are often difficult to tell apart formally, there are certain features that are not shared by the 2 constructions. For one, certain adverbials may only modify locative clauses. The example below, does not have an existential reading, as existence is not contingent on invitation. (25) Tanpa sebarang undangan, ada semua musuh kita. without any invitation EXIST all enemy 1.PL.INCL 'Without any invitation, all of our enemies were (there).' Obviously, the use of a locative PP is obligatory in a locative clause, but optional in an existential clause, unless the information pertaining to the location is given, as in example (24). Such a feature entails the different selectional properties of the 2 constructions. This is shown (iii) a. Boleh jadi-kah keramat ulama lebih hebat…? can become-Q sacred ulama more great 'Can the sacred power of the ulama be greater?' (Astora Jabat, 2003) b In spite of the grammaticality of the example, whose constituent order is assumed to be derived via focus movement, it does not obtain a locative reading. An existential reading persists, especially because the prepositional phrase is not a locative. Examples (27a-b) demonstrate that one construction is not derived from the other via movement.
Lastly, it is no surprise that strong quantifiers and various other definite DPs may occur in existential clauses, albeit quite restrictedly, as this is also attested in other languages, such as English. For example, the existential cleft construction features such definite DPs: (28) a. Let me recall who was at the feast. Well, there was Mary that came.
b. I woke up from a nightmare. There was this huge monster that was chasing me.

PRO-DROP
The lack of any phonological realisation of expletives in Malay does not seem to be a property of case or the mechanism of its assignment. This is because other nominal elements are able to occur overtly in environments in which the null expletives are found. The Exceptional Case Marking environment in the examples below show that the expletive remains null even after having obtained accusative case. It can be assumed that the expletive is not given any phonological content due to Prodrop. Various languages are known to exhibit pro-drop. For example, referential nominals can be dropped in Italian, largely due to the rich morphological distinctions that can be found on the verb in Italian. The subject of a clause in Italian is inferable despite the absence of the subject, due to the overt morphological marking of the φ-features of the subject on the verb. Because of this, subjects can be dropped freely in Italian. Needless to say, non-referential subjects should also be able to delete.

Existential
Ci deve essere un uccello. there must be a bird 'There must be a bird.' ✓ English does not permit pro-drop because, unlike Italian, it is morphologically impoverished. Although verbs need to agree with subjects, they do not mark for many of the φ-features of their subjects. The correlation between pro-drop and richness of overt morphological marking seems to hold as, besides these 2 languages, many other languages also exhibit (non-)pro-drop properties according to their morphology, as argued by Speas (1994).
Nonetheless, there exists a certain class of languages which does not appear to conform to this generalisationradical pro-drop languages. Many languages that are morphologically impoverished in terms of φ-feature agreement, such as Japanese and Mandarin Chinese, allow pro-drop for some reason. Although Salleh (1989) claims that Malay is a non-pro-drop language like English, the fact is that subjects can be omitted despite there being no overt morphological agreement between subject and verb, as illustrated in example (30). This makes Malay similar to Japanese and Chinese, classifying it as a radical pro-drop language. Since Malay is a radical pro-drop language, referential DPs can be omitted. The obvious conclusion is that non-referential DPs should also be allowed to drop. Therefore, this property of the language is responsible for the apparent invisibility of the expletives in Malay.

FURTHER RESEARCH
The case of clitic-doubling to avoid the formation of illegitimate expletive-argument chains outlined in this paper is merely a sketch and can benefit from more in-depth examination. As suggested by a reviewer, it might prove fruitful to look at this phenomenon as a case of nominalisation. Indeed, nominalisation of whole VPs is possible via cliticisation of -nya. For instance, consider the following examples in which a clausal constituent occupies nominal positions, e.g. subject, object and complement of preposition. The whole constituent headed by the verb cliticised by -nya, including its own argument, is the argument of a matrix verb. On the other hand, the clitic-doubling examples cannot be construed to be parallel to the nominalised verb examples above, due to the lack of a matrix predicate that would take the nominalised verbal constituent as its argument. Example (19a) is a clear example of an embedded clause headed by a complementiser, meaning that it is not nominalised, or at least, not straightforwardly.
In addition to that, sensitivity of -nya (if any) towards factors such as specificity, animacy, and information structure is also worth investigating; however, due to the constraint of space, this shall be left to future research.

CONCLUSION
This study has attempted to substantiate the claim that there exists a null variety of expletive in the Malay language. There is a dire lack in literature on expletives in Malay, and studies that do assume that the language has expletives tend to postulate null expletives without motivating their existence.
The comparison of Malay with English and Irish has yielded the finding that Malay patterns with English in 2 important respects: movement of the subject to TP; the definiteness restriction in existentials. Judging by the position of the subject in relation to the verb and possible auxiliaries, Malay can be concluded to obey the EPP. This is because the subject is observed to move to a position preceding the verb and inflectional heads that are known to occupy the inflectional layer, e.g. TP, AspP, etc. Since Malay does not show agreement phenomena with respect to the subject, such movement of the subject can only be analysed to be triggered to raise to SpecTP by the EPP, i.e. the requirement that every clause has a subject. Moreover, the observation that existentials in Malay exhibit sensitivity towards definite DPs further confirms that there exists an expletive-argument chain in such a construction. Without an expletive in the construction heading the chain with the definite DP as its tail, there is no reason for there to be such a restriction on definite DPs, as seen in Irish, which is argued to lack expletives.
Therefore, the hypothesis that Malay has expletives can be accepted as the 2 diagnostic tests show positive signs of the language obeying the EPP and having expletives. Malay patterns with likewise languages such as English and French to the exclusion of Irish, which has been argued to lack an EPP feature and, consequently, expletives as well.
To conclude, this study has provided evidence for and a principled account of the existence of null expletives in Malay. The findings in this study are valuable to linguistic analysis in Malay syntax in general, as null expletives are no longer a postulate in the grammar of the language.

ACKNOWLEDGEMENT
I would like to thank the Faculty of Languages and Linguistics, University of Malaya, for funding my attendance at the Linguistic Society of Korea Conference, 2018, where this paper was presented. I am also indebted to the attendees of my presentation and the reviewers of the paper for the insightful comments.