Observations on Tagalog Genitive Inversion

Tagalog is a strongly head-initial language: arguments without special discourse status typically follow their lexical heads. However, genitive-marked pronominal arguments display a word order alternation where instead of following their lexical head, they may precede it. This alternation, which I refer to here as Genitive Inversion, has received comparatively little attention in the research on Tagalog, even though it is relatively commonplace. This paper offers a detailed description of the behavior of Genitive Inversion, showing what kinds of arguments it can apply to and what environments it can apply in. I show that this process raises questions about the basic properties of Tagalog and discuss directions for potential analyses and avenues for further research into this topic.

1. INTRODUCTION.¹

This paper discusses a phenomenon found in Tagalog that I refer to as Genitive Inversion. Under this phenomenon, certain pronominal arguments may precede the lexical head of their phrase rather than follow it. This leads to alternations like the one illustrated in (1), where we see that the third-person singular agent can be rendered as postverbal niya or preverbal kanyang and the first-person singular possessor as postnominal ko or prenominal aking.

(1)

 

As is characteristic of Philippine languages, Tagalog is strongly head-initial. Lexical heads of phrases overwhelmingly precede their dependents (such as agents and possessors), with two major exceptions, examples like (1b) notwithstanding. Certain classes of modifiers either precede their head or show flexible word order, and clausal arguments with special discourse status (such as topic and focus) prominently appear preceding their lexical predicate.³ Genitive Inversion thus presents an interesting instance of word order variation, as the element preceding its head is intuitively an argument (rather than a modifier), and the effect of this variation appears to be largely stylistic rather than semantic or discourse-related. Genitive Inversion also has quite a restricted distribution, being sensitive to a few properties of its target—specifically nominal type and grammatical function—that interact with other properties of Tagalog. Thus, careful investigation of this phenomenon has the potential to deepen our understanding of the general mechanisms underlying word order and argument encoding in Tagalog and related languages.

Prior research on Tagalog has not given much attention to this phenomenon. This cannot be because the construction is rare in any sense. In fact, examples exhibiting Genitive Inversion are frequently encountered in naturalistic speech and text. They are also not infrequently reported in works on Tagalog in the context of other issues (often pronominal binding), such as (2) from Kroeger (1991), reproduced with glosses modified for consistency.

(2)

 

The primary goal of this paper is therefore to offer a careful description of the distribution and structure of the Genitive Inversion construction. Secondarily, I critically discuss prior literature on this construction, highlighting notable points and issues that arise. Ultimately, more work is needed to provide a formal analysis of this phenomenon, but the hope is that this paper can serve as a starting point for further research on this issue not only in Tagalog but also in related languages where similar phenomena may be found.

2. BACKGROUND

2.1. BASIC PROPERTIES

Let us begin with a brief overview of some properties of Genitive Inversion. A handful of differences distinguish this construction from its default counterpart, as schematized in (3) and exemplified in (4) and (5). To facilitate discussion, I use the term "default" to describe examples like (4a) and (5a) without Genitive Inversion, and the term "inverted" for examples like (4b) and (5b) that display Genitive Inversion. Comparing the default examples to their corresponding inverted examples, we see three differences. The genitive pronoun that normally follows its head noun or verb (italicized) instead (i) appears in the oblique form, (ii) precedes the head, and (iii) is followed by the linker morpheme =ng. Examples (4c,d) and (5c,d) also show that the oblique pronoun cannot appear in the default position and the genitive pronoun cannot appear in the inverted position, regardless of the presence or absence of the linker.

(3) Genitive Inversion Schema

[Head pro_gen … ] ↔ [pro_obl =ng Head … ]

(4)

 

(5)

 

Speakers do not report any obvious change in meaning between default constructions and their corresponding inverted constructions. However, many perceive the latter examples to be of a more formal or poetic register, especially in verbal environments like (5). The more formal register may cause some speakers to disprefer this construction. However, naturally occurring examples do exist, and some are included in this paper.

Two facts support the view that the inverted examples are related to the default examples in some way. First, the genitive pronoun in the default examples and the oblique pronoun in the inverted examples share the same grammatical function. Both pronouns in (4) are interpreted as possessors, while those in (5) are interpreted as agents. Second, the oblique (inverted) pronoun and the corresponding genitive (default) pronoun cannot co-occur, as (6) illustrates.

(6)

 

2.2. COMPARISON TO SIMILAR CONSTRUCTIONS

It is worth distinguishing Genitive Inversion from other processes that result in the apparent fronting of pronouns, as such word order alternations differ in other ways from the pattern sketched in (3). One of these processes is second-position clitic placement. Nominative and genitive pronouns in Tagalog belong to a class of second-position clitics, which encliticize to the first element within a certain domain (Schachter and Otanes 1972:§3.29; Kroeger 1991:§5.4.1; Kaufman 2010; Erlewine and Levin 2018). In the basic case, pronouns immediately follow their verbal head (7a). However, in the presence of preverbal clitic hosts like negation in (7b), the pronouns instead precede the verbal head.

(7)

 

Despite the pronoun's preverbal position in (7b), this is distinct from Genitive Inversion because (i) the pronoun does not alternate in form, (ii) no linker appears, and (iii) a preceding element (i.e., hindi) is required to serve as a clitic host. In fact, we will see in section 4 that an inverted pronoun has a different position in environments like (7b).

Another relevant process is focus fronting, specifically of oblique phrases. As (8) shows, this process involves the fronting of an oblique-marked argument to a preverbal position, resulting in a focused interpretation (see, e.g., Richards 1991; Kroeger 1991; Aldridge 2002; Mercado 2004; Hsieh 2020, for more details). We can also distinguish this process from Genitive Inversion. Most clearly, the fronted constituent in focus fronting lacks a following linker morpheme and is marked with the oblique marker sa. Section 3 will show that the two processes in fact target two distinct types of arguments and therefore do not overlap in their distribution.

(8)

 

2.3. PREVIOUS RESEARCH

An analysis of Tagalog Genitive Inversion must, at minimum, account for the differences highlighted in (3) between two structures that express the same meaning. Again, these differences relate to (i) word order, (ii) pronominal form, and (iii) the presence of the linker morpheme. There is limited prior research on this construction, and much of it has focused mainly on the phenomenon as observed in nominal environments (see Ramos 1971:§10.1; Schachter and Otanes 1972:§3.20). Nevertheless, this existing work has taken a number of divergent approaches to accounting for the main points highlighted above. Here, I summarize these existing analyses, highlight some points of interest, and point out issues that they encounter.

2.3.1. Genitive Inversion as relativization

One view of the Genitive Inversion construction is as a relative clause(-like) modifier. This is proposed by Schachter and Otanes (1972:§3.20). Discussing the phenomenon in nominal contexts, they propose that inversion derives from a type of possessive clause with an oblique phrase predicate, shown in (9).⁴ The idea is thus that Genitive Inversion is the result of forming relative clauses from such sentences, deriving (10) from (9).

(9)

 

(10)

 

Under this view, the default and inverted structures have different sources, so the alternation in (3) should not be understood as such. Instead, the form of the pronoun in the inversion structure is tied to the form found in the predicate construction, while the word order and presence of the linker are expected with relative clauses.

The relativization view leaves some details unexplained, however. Most notably, we expect the set of possible possessive predicates and the set of possible Genitive Inversion constructions to be identical, but this is not what we find. Section 3.1 shows that only pronouns may participate in Genitive Inversion, but (11) shows that nonpronominal possessive predicates can serve as (postnominal) relative clauses.⁶

(11)

 

In addition, (12) shows that certain types of possession, such as inalienable possession, are degraded when expressed as predicates but are perfectly natural with Genitive Inversion.⁷

(12)

 

We also have no explanation for the pattern Schachter and Otanes (1972) observe, whereby sa is ungrammatical when the pronoun is prenominal; compare (10) with (13). They also claim that sa is obligatory when the pronoun is postnominal (also problematic under the relative clause view), but counterexamples to this claim are attested in certain common expressions like (14) as well as naturalistic data like (15).⁸

(13)

 

(14)

 

(15)

 

Finally, the analysis does not readily extend to the phenomenon in the verbal domain, as no parallel to the possessive predicate construction exists. Schachter and Otanes (1972:§5.25(1)) do discuss Genitive Inversion in the verbal domain separately and point out a similarity to the nominal domain, but they do not propose an analysis for the verbal version and simply give a brief description.

2.3.2. Genitive as attributive modification

More discussion on Genitive Inversion in the verbal domain comes from Naylor (1980), who notes the alternation between the two clausal examples in (16). She draws a parallel to the alternation in nominal phrases like (4) above and proposes that the same process gives rise to the alternation in both domains.

(16)

 

To explain this parallelism, Naylor (1980:42) adopts a view of the Tagalog genitive as "a marker of attributive relation" in both nominal and verbal environments. She does not make it explicit what "attributive" entails, though her discussion suggests that it has both syntactic and semantic import. Despite this, the idea is intuitively attractive, as it allows us to understand Genitive Inversion in the context of a similar word order alternation found with manner adverbs.

A large class of manner adverbs in Tagalog are morphologically indistinguishable from adjectives and may appear either pre- or postverbally. We see in (17a) that the postverbal mapayapa 'peaceful' is introduced by a marker nang. Prescriptively, nang is distinct from the homophonous genitive marker ng, but there is evidence that this distinction is purely orthographic, as other Philippine languages (e.g., Cebuano) also have similarly homophonous particles (Kaufman 2006). With this in mind, (17b) is noteworthy. Here, the preverbal adverb is not marked by nang but instead is followed by the linker =ng. This pattern is similar to Genitive Inversion, in that we find a word order alternation where the post-head version shows genitive morphology while the prehead version involves the linker morpheme instead.

(17)

 

It is also notable that this pattern contrasts with the behavior of other modifiers with variable word order, such as relative clauses and adjectives, where the linker consistently appears between the modifier and its head. The behavior shared by both Genitive Inversion and manner adverbs is thus rather distinctive, and it is tempting to formulate a unified analysis where the pre- and post-head word orders are derivationally related (contra Schachter and Otanes 1972). However, to do so requires spelling out how or why adverbs, possessors, and agents form a natural class, what makes such a class distinct from other types of arguments or modifiers, and how these result in the particular morphological signature of this word order alternation.

2.3.3. Genitive Inversion through head movement

Another work discussing Genitive Inversion in the verbal domain comes from Culwell-Kanarek (2005), who also adopts the view that the default and inverted structures are derivationally related. Along these lines, he sketches an analysis of verbal Genitive Inversion within a generative framework. First, he proposes that the inverted word order is derived via an operation of head-lowering of Asp⁰/T⁰ to v⁰, resulting in the agent in Spec,vP preceding the fully inflected verb on the surface. This contrasts with head-raising to Asp⁰/T⁰ or higher, which is typically assumed to derive verb-initial word order in Tagalog.

Second, and perhaps more interestingly, he proposes an account for the change in the form of the pronoun, which is not explicitly addressed by Naylor (1980). The idea is that the different pronominal forms observed are not the result of changes in formal features (e.g., case, as framed in this paper), but instead in morphophonology. To illustrate, table 1 gives the pronominal paradigm in Tagalog. Under Culwell-Kanarek's proposal, the "gen" and "obl" series are treated as the clitic and free variants (respectively) of a single pronominal form, in contrast with the classical view, which treats them as formally distinct (cf. Schachter and Otanes 1972:§3.3).

This approach ties together the morphological alternation found in Genitive Inversion with that exhibited by the 2sg.nom pronoun. As (18) shows, the freestanding form ikaw is required when prosodically prominent (e.g., when focused), otherwise the clitic form =ka is used.¹⁰ This analysis thus provides a way of accounting for the morphological alternation while also treating the default and inverted forms as derivationally related, contrasting with the relativization approach discussed in section 2.3.1.

Table 1.

TAGALOG PRONOMINAL PARADIGM.

(18)

 

However, as with Naylor's (1980) approach, some details remain to be worked out. In particular, the reanalysis of the pronominal system has potentially far-reaching implications for general issues such as case assignment and noun phrase structure in Tagalog. Such issues deserve separate, careful consideration. Why the linker morpheme appears in the inverted structure is also not explained. Finally, there is evidence that head-lowering is not necessarily the correct approach to take, as I argue in section 4 that the inverted pronoun occupies a syntactically high position, above Spec,vP.

We have thus seen that previous research on Genitive Inversion has proposed a diverse range of approaches to analyze the phenomenon, each with its own advantages and disadvantages in terms of accounting for the three salient properties outlined at the beginning of this section. With this in mind, let us turn to a discussion of some more properties of Genitive Inversion, aimed to provide a more complete picture of this phenomenon for future research.

3. DISTRIBUTION

This section is concerned with the distribution of Genitive Inversion. I show that this process occurs in nominal and verbal environments, but valid targets of inversion are restricted by properties such as nominal type. Specifically, we will see that Genitive Inversion only applies to agents and possessors and that it is also largely restricted to pronouns in the modern language.

3.1. NOMINAL TYPE

For the most part, only pronominal arguments may participate in Genitive Inversion, although there are minor differences between nominal and verbal environments. Let us begin by considering its behavior in nominal environments, as in (19), which shows various types of genitive-marked possessors: a pronoun, a demonstrative, a proper name, and a common noun.

(19)

 

Among these, only the pronominal possessor is fully acceptable with Genitive Inversion, as (20) shows. Inversion is marginal with proper names and ungrammatical with demonstratives (which have a pronominal use) as well as common nouns.¹¹

(20)

 

The pattern we find in verbal environments is more restricted. Parallel to (19), (21) shows various types of genitive-marked agents in a Patient Voice transitive clause. Compare this with (22) showing that only pronouns may participate in inversion in verbal environments. Proper names are robustly ineligible; compare kay Vilma in (22b) and (20b).

(21)

 

(22)

 

It is worth noting that the distribution of Genitive Inversion in Tagalog may have been broader in the past, as examples of nonpronominal inversion can be found in older texts, as in (23). The first two examples come from Tagalog poetic literature from the Spanish colonial era (1565–1898), while the third comes from an early work in Tagalog linguistics.¹² This diachronic change suggests that we might expect variation in terms of what can and cannot undergo inversion among related languages that exhibit similar processes.

(23)

 

Besides personal pronouns, the wh-pronouns nino 'who.gen' and kanino 'who.obl' also alternate under Genitive Inversion, though here we find an asymmetry. Nino follows both nouns and verbs (typically forming echo questions), as (24) shows. Kanino, on the other hand, is only well-formed in nominal contexts, as in (25a), which is a typical possessor question in Tagalog.¹⁵ This contrasts with the verbal context in (25b), where kanino is degraded, potentially because it only has a possessive reading.

(24)

 

(25)

 

This different behavior seen with kanino is one of the few exceptions to the otherwise robust parallelism in the morphosyntactic expression of possessors and agents in the language, and it is not immediately clear why such a difference exists. That is, why should the oblique wh-expression kanino only allow a possessive reading when the agentive reading is available in other similar expressions, namely oblique personal pronouns as well as the genitive wh-expression nino? These differ minimally from kanino in different ways: the personal pronouns are not wh-expressions, while nino appears in a different position and case form. Thus, ascribing the behavior of kanino to one of these properties would likely not fully explain the difference.

Whatever the answer to this question might be, it has implications for the long-standing issue of syntactic category in Tagalog, as the strong agent–possessor parallelism is often taken as evidence in favor of collapsing the noun–verb distinction in the language (see, e.g., Kaufman 2009). If the behavior of kanino cannot be ascribed to independent factors, then the case for a noun–verb distinction may be strengthened.¹⁶ In this case, explaining the agent–possessor parallel in Tagalog likely lies in the body of research that observes similar parallels cross-linguistically, even in languages with a clear-cut noun–verb distinction (Chomsky 1970; Abney 1987; Szabolcsi 1994, among many others).

3.2. GRAMMATICAL FUNCTION AND CASE MARKING

Nominal type is not the only property that Genitive Inversion is sensitive to. Here, I argue that Genitive Inversion is also restricted by the grammatical function of its target, though this effect is hard to tease apart from case marking.

Only external arguments, such as possessors and (external) agents, may undergo Genitive Inversion. Ungrammatical examples attempting inversion of a causee (i.e., an internal agent) and an internal argument are given in (26), where the pronouns' noninverted positions and forms are indicated with angle brackets. Examples with both angle brackets and a star are meant to convey that the inverted structure is ungrammatical while the default (noninverted) structure is well-formed.

(26)

 

Looking at these examples more closely, however, we might ask whether it is in fact case marking that is to blame for (26) being ill-formed.¹⁷ Whereas these examples show oblique-marked default forms, the grammatical ones we have seen so far have genitive-marked default forms. Indeed, inversion of agents is ill-formed just in case the agent is nominative in its default form (regardless of the inverted form), as (27) shows.

(27)

 

If Genitive Inversion were restricted by case marking rather than grammatical function—as the contrast between (22a) and (27) suggests—then we should find examples of inversion involving genitive-marked internal arguments (i.e., neither agents nor possessors). Such examples appear to be unattested.

On the one hand, this is partly due to the case marking patterns of Tagalog. Genitive case does mark internal arguments (cf. ng diyaryo in (27)), but pronouns trigger differential object marking that results in oblique marking (cf. sa inyo in (26b); see Ramos 1974; Latrouite 2012; Sabbagh 2016; Collins 2019, for further discussion). Given that there are no genitive-marked internal argument pronouns in Tagalog and that Genitive Inversion is restricted to pronouns (as discussed in section 3.1), we independently do not expect to find Genitive Inversion of internal arguments in the modern language. We therefore cannot test this hypothesis synchronically.

On the other hand, earlier stages of Tagalog may provide some insight since the historical examples in (23) suggest Genitive Inversion of common nouns was possible (though the examples are of definite nouns, which may also require differential object marking). Even with this broader distribution, however, the Genitive Inversion of internal arguments appears to be unattested.

The evidence is thus slightly in favor of a grammatical-function-based restriction on Genitive Inversion, as this would allow us to explain why internal argument inversion is absent throughout the historical data. The main issue with this approach would be to account for the contrast between nominative and genitive agents (e.g., (27) vs. (22a)). One possibility is to take a structural approach to this contrast, as it is commonly assumed that the nominative argument in Tagalog occupies a dedicated syntactic position or is otherwise syntactically distinct from the corresponding non-nominative argument (e.g., Schachter 1976, 1996; Guilfoyle, Hung, and Travis 1992; Kaufman 2009; Hsieh 2023). Here again, comparing the behavior of similar processes in other Philippine (or, more broadly, Austronesian) languages would be fruitful.

3.3. TYPES OF ENVIRONMENTS

As we saw in section 1, Genitive Inversion can generally apply to arguments of both nouns and verbs. Here we consider in more detail the types of environments where Genitive Inversion can occur, paying particular attention to the verbal domain.

First, Genitive Inversion in the verbal domain is general across different argument structure and voice configurations. Aside from the transitive Patient Voice clause examples we have seen so far, the naturalistic examples in (28)–(30) show this alternation in ditransitive clauses (both Conveyance and Locative/Benefactive Voice) and a causative. I have reconstructed the default form and position of the inverted pronoun, indicating it (as before) in angle brackets.

(28)

 

(29)

 

(30)

 

Turning to more reduced clauses, we find variable behavior with respect to the availability of Genitive Inversion. We first consider infinitival or aspectless clauses, as exemplified below. We see in (31a,b) that imperatives and certain clausal adjuncts allow Genitive Inversion. We can compare this to modals like dapat 'should', which also allow it and appear to exhibit some variability in where the pronoun may appear. As the naturalistic examples in (32) show, the inverted pronoun may immediately precede either the lexical verb or the modal, in contrast to (31b,c), showing that the pronoun may only precede the verb and not the particle nang.

(31)

 

(32)

 

Interestingly, the modal-pronoun order of (32a) is less common than the pronoun-modal order of (32b), based on the number of search results on Google. However, almost all valid examples of the latter involved relativization of some kind, while non-relativized examples were impressionistically more common with the former. It is not immediately clear why the data looks this way, but finer details of reduced clause structure likely play a role and may also explain the contrast between (31b,c) and (32). Genitive Inversion is thus a potential diagnostic for differences in clause size and may shed light on the nature of finiteness in the language. At the same time, a deeper understanding of clausal embedding in Tagalog is potentially illuminating for the structure of Genitive Inversion (see also section 4).

Genitive Inversion is also possible in gerundive constructions, as in (33). This is perhaps unsurprising given that such constructions intuitively exhibit a mix of nominal and verbal properties. Interestingly, Genitive Inversion of 'who' is also possible with gerunds and is in fact preferred over the corresponding default construction (cf. (24) and (25) and fn. 16). On the other hand, (34) shows that Recent Perfective clauses disallow Genitive Inversion.

(33)

 

(34)

 

Outside of the gerunds considered above, other types of derived nominals also allow Genitive Inversion. Specifically, possessors of abstract nouns derived from adjectival roots participate in this alternation, as (35) shows.

(35)

 

In contrast, other adjectival constructions disallow Genitive Inversion entirely. For regular predicative adjectives, this may seem unsurprising because, as (36) shows, these typically mark their subjects nominative (recall section 3.2). However, a number of intensified forms do assign genitive case to their subjects, and (37) shows that even with these forms, Genitive Inversion is impossible.²⁶

(36)

 

(37)

 

3.4. SUMMARY

To summarize this section, we have seen that Genitive Inversion generally may only apply to pronominal arguments that are genitive-marked, though this characterization obscures some details. We saw that Genitive Inversion is more flexible in nominal environments, where it may target proper names marginally as well as the wh-pronoun 'who', in addition to the personal pronouns that may undergo inversion in verbal environments. We also saw that the class of genitive-marked pronouns in Tagalog happens to coincide with (non-nominative) external arguments, so it is not clear whether Genitive Inversion is sensitive to case marking itself or to argument structural properties, though historical behavior may suggest the latter. I then showed that Genitive Inversion is found in both nominal and verbal (but not adjectival) environments, with some restrictions on the range of hosting environments.

4. STRUCTURAL POSITION

Having discussed the distribution of Genitive Inversion, let us now consider the question of what structural position the inverted pronoun occupies. Here, I assume that word order in the domain preceding the head noun or verb of a phrase generally indicates structural height.²⁷ We will see that the inverted pronoun appears in an intermediate position within the extended projection or left periphery of its containing clause or DP (in the spirit of Rizzi 1997, and subsequent work), as it precedes certain functional particles but follows other kinds of material.

4.1. WORD ORDER AND STRUCTURAL POSITION

One indication of the inverted pronoun's height is that it precedes certain inflectional particles, namely the plural marker mga and the negative particle hindi. These are illustrated in (38) and (39), with naturally occurring examples in (40). Note that the difference in the position of the clitic pronoun ko between (38) and (39) is due to a difference in clitic-hosting ability between mga and hindi (see also section 2.2).

(38)

 

(39)

 

(40)

 

It should be noted that previous work has claimed that examples like (39a), where Genitive Inversion precedes negation, are ill-formed. Schachter and Otanes (1972:381) assert that inversion is impossible in negated sentences, regardless of word order. Separately, Culwell-Kanarek (2005:54) remarks that the pre-negation position of the inverted pronoun is "not acceptable, or at least not as acceptable," but does not report on the acceptability of the post-negation position, like that in (39b).

These discrepancies in the data may be due to the overall markedness of Genitive Inversion in the verbal domain. Speakers I have consulted prefer the pre-negation order over the post-negation order, though some remark that the co-occurrence of Genitive Inversion and negation is unusual to some degree. The naturalistic data in (40) provide evidence that the pre-negation order is at least possible, even if somehow marked.³⁰

In contrast to their height relative to hindi and mga, we can also establish that inverted pronouns occur below other material in the clause or DP periphery. In nominal environments, inverted pronouns follow case markers, as we have seen in many examples above with ang. Genitive ng and oblique sa show the same behavior, as shown in (41). Similarly, prenominal demonstratives (which generally replace ang and ng) also precede the inverted pronoun, as (42) shows, though such examples appear to be less common.

(41)

 

(42)

 

In the verbal domain, the inverted pronoun follows complementizers such as the linker na introducing the relative clause in (43a) and kung 'if' in (43b). Various kinds of topics—specifically those that involve ay-inversion and comma intonation—also precede the inverted pronoun, as in (44).

(43)

 

(44)

 

The inverted pronoun's comparatively lower position in the clausal or DP left-periphery is also reinforced by the fact that it may host second-position clitics. Cliticization in Tagalog has been used as a diagnostic for the structural position of left-peripheral material in a clause (see Kroeger 1991; Aldridge 2002; Mercado 2004; Hsieh 2020, among others). For example, the placement of sila in (45) is taken to show that the adjunct bukas 'tomorrow' lies within the same clitic-placement domain as a pronoun when it is focused (45a) but not when it is topicalized (45b).

(45)

 

Examples involving Genitive Inversion have previously been reported by Schachter and Otanes (1972) and Schachter (1973), but naturalistic examples can also be found. These are provided in (46) and (47), showing clitic (nominative) pronouns and nonpronominal clitics, respectively. The relevant second-position clitic is italicized, while the inverted pronoun and the associated linker appear in boldface. As we see in these examples, inverted pronouns pattern with focused adjuncts (45a) rather than topicalized ones (45b).³⁶

(46)

 

(47)

 

The evidence we have seen so far suggests that the position of the inverted pronoun is determined by syntactic factors rather than, say, morphophonological ones, as would be the case for second-position clitichood. In particular, we have seen that inverted pronouns are structurally higher than inflectional particles but lower than material from the far left-periphery. Specific details remain unanswered, however. What specific position does the inverted pronoun occupy, and what process underlies the alternation between default and inverted pronouns? Below, I discuss two interesting puzzles related to the position of the inverted pronoun that could potentially bear on these questions.

4.2. TWO PUZZLES

The first puzzle relates to relative clauses. Among the relative clauses that relativize nominal arguments, different word orders are possible (see Aldridge 2004, 2017; Law 2016). Some of these word orders seem to restrict the possible positions for Genitive Inversion.

Head-initial relative clauses are the least restrictive, allowing inversion of the possessor of the head noun and the agent of the relative clause, as (48) shows. That is, inversion is independently possible in both the nominal and verbal domains, as we might expect from what we have seen so far. In fact, both instances of inversion may occur at the same time, as in (49), which also shows that the grammatical roles of the pronouns are indeed different (cf. (6)).

(48)

 

(49)

 

In headless and head-final relative clauses, however, only the nominal position appears to be possible. In (50) and (51), we can detect this restriction as a requirement that the inverted pronoun precede the plural marker mga (if present). Despite this syntactic restriction, the inverted pronoun can be interpreted as either a possessor or an agent.

(50)

 

(51)

 

Note that the inverted pronoun also cannot immediately precede the head noun in a head-final relative. Similar behavior is found with prenominal adjectives, like in (52), where the inverted pronoun cannot immediately precede the head noun either.

(52)

 

The word order restriction relative to mga is unexpected if we view all relative clauses as involving a clause embedded within a larger nominal structure. Under such a view, we would expect the immediately preverbal position to be available in headless and head-final relatives, reflecting the possibility of Genitive Inversion within the clause. The difference in behavior thus suggests that there are nontrivial structural differences between the different types of relatives, along the lines of previous research (in particular, Aldridge 2017).

Alternatively, we might also take the above data as evidence against a formal noun–verb distinction in Tagalog (e.g., Kaufman 2009). This kind of approach maintains that Tagalog does not have a distinct process of relativization since apparently verbal elements such as binili in (50) and (51) are in fact nominal. Thus, no extra functional structure is needed to compose with ang or mga, and we expect only one possible position (i.e., preceding mga) for the inverted pronoun.

Genitive Inversion thus has implications regarding the structural differences (or lack thereof) between these relativization constructions. Further detailed work exploring these implications may in turn give us a better picture of the structural position that the inverted pronoun occupies, which is poorly understood at present.

The second puzzle relates to the focus constructions of oblique-marked phrases and adjuncts. Specifically, Genitive Inversion is incompatible with such constructions. We see in (53) that while default pronouns are compatible with such focused elements (here sa gubat 'in the jungle'), inverted ones are not, regardless of the relative word order.

(53)

 

We can easily show that this incompatibility is not tied to the semantics of focus, as Genitive Inversion is possible with focus constructions that target DPs, as (54) shows.

(54)

 

In Tagalog, this kind of focus construction is structurally distinct from adjunct focus (Aldridge 2002; Mercado 2004; Hsieh 2020), strongly suggesting that the incompatibility of adjunct focus with Genitive Inversion is syntactic in nature. For example, it may be the case that the focused adjunct and the inverted pronoun compete for the same syntactic position or that the formation of one construction creates a configuration that blocks the other in some way. Pursuing the former hypothesis would require finding independent evidence for the shared syntactic position (e.g., other shared properties), which seems unlikely. The latter hypothesis, on the other hand, would require a more detailed investigation into the Tagalog clausal left periphery.

5. DISCUSSION AND CONCLUSION

As a phenomenon in and of itself, Genitive Inversion is interesting in the context of Tagalog, as it is an instance of word-order variation outside of the post-head field that does not have an obvious truth-conditional or information-structural effect. The discussion in this paper has thus endeavored to show what this process is. What types of arguments may it apply to, where may it apply, and what is its behavior in various contexts? However, it remains largely unclear why Genitive Inversion behaves the way it does. That is, what underlying processes account for the distribution of the construction? In some cases, the path to answering this question is blocked by confounds that require more data to tease apart, such as the issue of whether inversion is sensitive to grammatical role or case.

Thinking more broadly, we have also seen that Genitive Inversion is a potentially useful tool for uncovering details of Tagalog syntax. Specifically, we saw that Genitive Inversion was sensitive to differences in clause size or finiteness (section 3.3) and to differences in relative clause word order (section 4.2). We also saw a parallel between Genitive Inversion and manner adverbs observed at least as early as Naylor (1980) that suggests a perhaps unexpected connection between argumenthood and modification. Finally, the near-parallel behavior of this phenomenon between nominal and verbal domains has important implications for the debate on syntactic category in Tagalog.

A better understanding of Genitive Inversion therefore has the potential for deepening our understanding of various areas of Tagalog syntax, some of which have been the subject of long-standing debate. Reaching this goal likely requires not only further research into the open issues posed in this paper for Tagalog but also a better understanding of similar phenomena within related languages, as this may help to address questions (e.g., grammatical role and case) that are confounded in Tagalog. Of course, it is also generally necessary to consider the wide range of attested variations in order to formulate a sufficiently general analysis.