Flicking Switches: Ayeri and the Austronesian Alignment

  • There is a relatively more recent (1991) Ph.D. thesis by Paul Kroeger on Tagalog alignment than the article by Schachter from 1976 (!) which much of the article below is based on if memory serves. Kroeger (1991) is available from his staff page at GIAL. Another article, Kroeger (1993; linked at the bottom), summarizes tests for subjecthood from the thesis briefly. I should probably go and reevaluate my findings below accordingly.
  • I seriously need to reevaluate these things. Just from having read Kroeger (1991) by now, it looks like Austronesian alignment may in fact not be the best framework to analyze Ayeri in, even though it gave the original impetus. I’m leaving this article here for the sake of documentation, but please take it with a grain of salt.
  • I am currently in the process of publishing revised results regarding Ayeri’s syntactic alignment; the twelve-part series starts here.

In this article, a number of features of the Austronesian alignment will be discussed in hope to gain a better understanding about the difference between what David J. Peterson called the ‘conlang trigger system’ – of which Ayeri uses a variety – and the naturally occurring ‘trigger system’ of South-East Asia, namely the ‘Philippine’ or ‘Austronesian’ alignment which served as an inspiration for part of Ayeri’s grammar, albeit with some misunderstandings. In order to examine and test how Ayeri works with regards to the terms topic, focus, and subject and in how far it differs from Austronesian languages such as Tagalog, papers by Charles N. Li and Sandra A. Thompson, Paul R. Kroeger, and Paul Schachter were taken into consideration.

As the focus of this article is mainly on how Ayeri relates to the Austronesian alignment system, the language’s detailed way of assigning case to the different semantic roles as well as the details of its handling of morphologic case marking with regards to verb transitivity and the classic constituent-order typology will only be touched on superficially.

0. Introduction

I have been wondering about how to analyze Ayeri’s syntactic alignment for a while now. In another blog post, I have already mentioned that it was originally inspired by languages such as Tagalog, an Austronesian language of the Philippines: that is, by the infamous Austronesian alignment. This alignment system is a feature that has been present in Ayeri right from the beginning, however, I did not really know what I was doing when I started working on Ayeri in late 2003.

Ayeri has occasionally been mentioned as a go-to for a ‘trigger’ conlang in the past,1 however, as David J. Peterson pointed out regarding the explanations about the Austronesian alignment that were floating around on the Conlang Mailing List in earlier years:

As I said […], just because the conlang trigger system doesn’t exist in the real world doesn’t mean it’s a bad – or even necessarily unrealistic – system. It simply means that it’s unattested. (Peterson, 19 Mar. 2008)

My understanding is that the Conlang Trigger Language is very similar to the first analysis of, for example, Tagalog, but that work since has shown that those analyses were “wrong” (replace that with the academic word for “wrong”). (Peterson, 23 Mar. 2008)

Since I am somewhat concerned with naturalism, it has long been on my mind to figure out how similar Ayeri actually is to the original system as found in languages on the Philippines. Consequently, I have so far been cautious to call Ayeri a proper ‘trigger language’ myself.

In addition to the uncertainties about the classification and the naturalism of Ayeri’s syntactic alignment, different people have pointed out to me in the past that my habit of glossing those preverbal particles typical of Ayeri as focus and/or topic marking respectively were wrong and there was confusion as to what the language actually marks. Unfortunately, however, “‘Topic’ means many things to many people” (Kroeger 2) – and I suppose, so does ‘focus’. Thus, what I will be trying to come to terms with in this blog article is probably still contestable, depending on whose theories and terminology you subscribe to.

The rest of this essay-length blog article is written in the style of a linguistic paper, as the “abstract” at the beginning already suggests. You may only want to read on if you are not bored by a wall of text in academic prose (that I labored on for about two weeks) and not scared of syntax.

1. Questions of Interpretation

In fact, I was reminded to have a look at the issue of Ayeri’s syntactic alignment by Conlangery Podcast 53, which dealt with topicalization – and what Tagalog does to case- and definiteness marking is sometimes described in terms of topicalization as well (cf. Schachter/Otanes 60 and Schachter 494). The theoretical grounding of the episode seemed to be mostly based on the 1976 paper by Charles N. Li and Sandra A. Thompson, called “Subject and Topic: A New Typology of Language,” which was cited in the episode’s annotations as well.

In this paper, Li and Thompson are concerned with establishing the topic-comment structure as a basic sentence structure underived from the subject-predicate structure, which they criticize was long assumed to be the single basic way sentences could be built (cf. Li/Thompson 460). The typology the authors suggest consists of a four-fold distinction between subject-prominent, topic-prominent, both subject- and topic-prominent, and neither subject- nor topic-prominent languages (cf. 459).

Although topicalization is an important keyword in the description of Tagalog and other languages of the Philippines (cf. Schachter 494), the language is grouped as ‘neither’ in their study (cf. Li/Thompson 460) with reference to Schachter and Otanes, who “stated that the Tagalog basic sentence structure should not be described in terms of the notion subject” (460–61). Li and Thompson conclude that in order for a language to be neither subject-prominent nor topic-prominent, its “topic becomes more closely integrated into [the] case frame of [the] verb” (485). However, Tagalog itself is not subjected to Li and Thompson’s analysis, as the focus of their survey is on languages that exhibit a strong preference for the topic-comment structure. But as we will see, the notion of topic is indeed tied to the case frame of a verb in Tagalog as well as in Ayeri.

Contrary to the quotation from Li and Thompson above, at least in that part of their 1972 Tagalog Reference Grammar describing the basic sentence structures of the language, Schachter and Otanes persistently assume the language to be essentially topic-prominent in Li and Thompson’s terms in so far as they define the basic structure to have a predicate followed by a topic which “expresses the focus of attention in the sentence” (Schachter/Otanes 60). Their definition of ‘predicate’ as “what is said about the topic” (61) seems to correspond to Li and Thompson’s use of ‘comment’ in this regard.

In his 1976 article on “The Subject in Philippine Languages,” Schachter, however, does not quite agree with the definition of topic mentioned above anymore and gives a counterexample where the “focus of attention” and the topic do not coincide (Schachter 496):

  1. Speaker A:

    Nasaan
    where
    si Maria?
    T-Maria

    “Where is Maria?”

    Speaker B:

    Hinuhugasan
    GT-is-washing
    niya
    A-she
    ang mga pinggan.
    T-dishes

    “She’s washing the dishes.”

In this example, the emphasis is put on “Maria,” however, the corresponding pronoun niya in the answer is not marked on the verb, but instead, ang mga pinggan ‘the dishes’ are. This sentence also shows the way Tagalog marks constituents as ‘topics’ (for a definition see below): the noun phrase is marked with the particle ang, and the verb at the head of the clause corresponds with the according topic marker (in this case goal-topic), indicating that the marked constituent is definite. Consequently, Schachter writes on the definition of the term ‘topic’:

According to Li and Thompson […], the term topic, as generally used by non-Philippinists, designates a sentence constituent that has the following semantic properties: 1) the topic is always “definite”; 2) the topic functions as the “center of attention” established by the discourse context. In the usage of Philippinists, however, it is only the first of these properties that is consistently associated with the constituent that is identified as the topic. (496)

This is to say, the term ‘topic’ for the purpose of the present analysis is limited to its referential function – information assumed to be established in discourse by the speaker, thus definite. Definiteness, then, is tied to highly referential information, which in turn depends on discourse context that may also be established, for example, by the speaker pointing at something in their vicinity (cf. Li/Thompson 461). This highly referential information, however, does not need to be a center of attention in order to constitute a topic as shown in example 1 above.

With the term ‘definiteness’ in mind, one may wonder what happens when two sentence constituents whose referentiality is presupposed appear in a sentence. Firstly, there may only be a single topic per basic sentence, and every case role may only occur once per basic sentence as well (cf. Schachter 494). Secondly, “it is not always clear why one of these noun phrases, rather than another, is chosen as topic. […] [I]t seems that there is often a good deal of leeway with regard to the choice of topic” (497). However, Payne suggests that topic-worthiness is related to agent-worthiness, based on “a survey of languages that rely, at least partially, on pragmatics to distinguish A from P in some subset of their transitive clauses” (Payne 150). He illustrates this with the following chart, but cautiously reminds the reader that this chart just represents tendencies, the argument being that the probabilities of both being an actor and a topic are tied to animacy (150):

1 > 2 > 3 > 1 > 2 > 3 > proper names > humans > non-human animates > inanimates
  agreement > pronouns
definite  >  indefinite

If definiteness by means of high topicality/referentiality is what is specially marked in the Austronesian alignment, the term ‘focus’ may probably be disposed of for the sake of this essay, since “Focus ‘expresses CONTRAST […]; it designates something that is NOT presupposed (relative to some context)’” (Kroeger 3). However, even though Kroeger refutes the term ‘focus’ for his analysis of Tagalog syntax, he still includes it in a parenthetical comment as a synonym for “the system of voice marking […] in Tagalog and other Philippine languages” (1), so the term seems to have some currency.

Another important term that should be defined is the notion of an ‘actor’ as used by the papers consulted for this essay. Schachter notes in reference to Fillmore that “the label actor should not be taken as equivalent to agent, at least if agent is associated with some such role as ‘the typically animate perceived instigator of the action’” (Schachter 497). He goes on to show that an animate actor NP is not necessarily the ‘instigator’ of an action and that the actor does not need to be animate either (cf. 498). Quoting Benton, he defines it as “the entity to which the action of the verb is attributed” (498) – which does not need to be an agent in the sense of thematic relations (cf. Van Valin 31).

In the course of this essay, the terms ‘actor’, ‘topic’, and ‘actor-topic’ are used to refer to various elements of the sentence. The following illustration will explain which term refers to which part of a sentence (cf. Schachter 494):

Angkond-yaØKanseygo-ley
ATeat-3SG.MT-Kanapple-P
actor-topic
marker
 topic
marker
actor 
  actor-topic 

“Kan eats an apple.”

Different than in Tagalog, the topic marker of NPs is a null-morpheme, i.e. it has no realization at the surface of a statement, but constitutes an underlying morpheme. An actor that is not topic-marked will be referred to as a ‘non-topic actor’.

2. Testing Hypotheses

Since we have now established a definition of the terms ‘topic’ and ‘actor’ and have also seen an example of how Tagalog marks definiteness and cases, let us now proceed with an analysis of how Ayeri handles a number of different syntactic issues that are related to the notions of actor, topic, and subject. Kroeger’s and Schachter’s respective papers will be of value here since they elaborate on a number of different properties that subjects are supposed to exhibit universally, and especially Schachter makes a painstaking effort to carefully test the different relevant syntactic elements (actor, topic, and actor-topic) for these properties. In the course of the following analysis I will compare their findings about Tagalog and other Philippine languages to Ayeri and see if they are in any way applicable in order to determine similarities and differences.

2.1. Basic sentence structure

The first basic sentence pattern in Ayeri is the declarative sentence, which consists at least of a predicate and an actor (but cf. 2.7. below):

  1. Tahanya
    write-3SG.M
    ayonang.
    man-A

    “The man writes.”

Besides that, there are also equative sentences which – at least at the surface – only consist of two NPs in juxtaposition:

  1. Ayonang
    man-A
    karomayās.
    doctor-P

    “The man is a doctor.”
     
    Ayonang
    man-A
    alingo.
    intelligent

    “The man is intelligent.”

There are also imperative sentences, which may consist of as little as just a verb marked for the imperative mood – a covert second person actor is implied:

  1. Silvu!
    see-IMP

    “Look!”
     
    Ilu
    give-IMP
    duramas
    tea-P
    yayam!
    3SG.M.DAT

    “Give him the/some tea!”

However, the “double subject” type sentence that Li and Thompson mention to be typical of topic-prominent languages (cf. Li/Thompson 468) does not normally appear at least in formal language:

  1. *​Ada-mehirang,
    that=tree-A,
    baloyjas
    leaf-PL.P
    kāryo.
    big

    “That tree, leaves are big.”

Furthermore, sentences with dummy subjects do occur, contrary to their predictions about topic-prominent languages, and in favor of subject-prominence (cf. 467):

  1. Seyarareng.
    rain-3SG.INAN

    “It is raining.”

A striking difference to Tagalog is the lack of topic-marking on both the verb and the marked noun phrase in intransitive sentences. Compare example 2 above to the following example from Schachter (499):

  1. Magtatrabaho
    AT-will-work
    ang lalaki.
    T-man

    “The man will work.”

Another difference is in how Ayeri handles equative sentences. By default, the constituent a statement is made about will be stated first and then the property NP follows; there is no overt copula morpheme. Inversion resulting in Tagalog’s preferred order is, however, possible for contrast:

  1. Karomayās
    doctor-P
    ayonang.
    man-A

    “The man is a doctor.”
     
    Alingo
    Intelligent
    ayonang.
    man-A

    “The man is intelligent.”

Interestingly, the predication in equative sentences seems to be interpreted in the way of a transitive sentence, although it is lacking an overt predicate. The assumption that the predicate might in fact be a covert element at the head of the clause seems problematic as well, though, if, for example, an emphatic affirmative or negative adverb is involved:

  1. Tahanya
    write-3SG.M
    māy
    AFF
    ayonang.
    man-A

    “The man does write.”
     
    Ayonang
    man-A
    māy
    AFF
    karomayās.
    doctor-P

    “The man is a doctor.”
     
    Ayonang
    man-A
    māy
    AFF
    alingo.
    intelligent

    “The man is intelligent.”

If there were an underlying predicate in the form of a copula, one might expect the affirmative māy to appear at the head of the clause, following the covert copula, and staying in the VP, however, this does not seem to be the case.

Furthermore, Ayeri does not allow topic marking in imperative sentences (cf. example 5 above), contrary to Tagalog, which does (506):

  1. Ibigay
    GT-give
    mo
    A-you
    sa kaniya
    D-him
    ang kape.
    T-coffee

    “Give him the coffee.”

Another Philippine language, Waray, may drop the actor from an imperative statement as it is implied anyway, like Ayeri does, but an actor-topic can nonetheless be marked on the verb, even if it is referring to a covert constituent (506):

  1. Paglutu
    AT-cook
    hit panihapun.
    G-supper

    “Cook supper.”

Yet another difference to Tagalog related to basic sentences with regards to constituent order is that according to Schachter, there seems to be no preferred order of NPs in the predication (cf. 495). Ayeri, on the other hand, displays a customary preference in constituent order at least for verb—actor—patient—dative, in spite of morphologically marking all cases, thus theoretically not strictly relying on word order alone to signify syntactic relations. This order changes according to syntactic weight, though, if relative clauses are involved (cf. 2.6. below). Schachter, however, also notes that another Philippine language, Pangasinan, has a fixed order of verb—actor—goal (cf. Schachter 507), so at least this feature is not completely unprecedented in the language family of interest.

2.2. Verb agreement

Another morphosyntactic feature that distinguishes Ayeri from Tagalog is that of verb agreement. Certainly, there is concord in Tagalog’s verbs, but as could be observed from the previous examples, this is basically limited to the case of the topic NP. In Ayeri, however, there usually is mandatory number and person agreement with the actor for finite verbs, regardless of the actor being the topic or not:

  1. Yam
    DT
    naraye
    speak-3SG.F
    ang Maha
    A-Maha​​F​
    Ajān nay Pulan.
    T-Ajān​​M​​ and Pulan​​M​

    “Maha speaks to Ajān and Pulan.”

    Tangyan
    hear-3PL.M
    ang Ajān
    [A-Ajān​​M​
    nay
    and
    Pulan
    Pulan​​M​​]​PL

    “Ajān and Pulan hear/listen.”

If the actor consists of only a personal pronoun, the person agreement on the verb does double duty as person marking and a cliticized pronoun:

  1. Ang
    AT
    tangyan
    hear=3PL.M.T
    yam Maha.
    D-Maha

    “They listen to Maha.”
     
    *​Ang
    AT
    tangyan
    hear-3PL.M
    yan
    3PL.T
    yam Maha.
    D-Maha

     
    Tangtang.
    hear=3PL.M.A

    “They hear/listen.”
     
    *​Tangyan
    hear-3PL.M
    tang.
    3PL.M.A

Note, however, that this feature is not unprecedented in Philippine languages again, although it is marginal. Thus, Kroeger quotes examples from Tagalog with optional plural agreement with the topic (cf. Kroeger 6), and Schachter quotes examples from Kapampangan, which has mandatory actor agreement at least under some circumstances, although he notes that this is unusual among Philippine languages (cf. Schachter 501–02).

2.3. Reflexivization

Another test Schachter uses as evidence against the topic – and for the actor – being the subject in Tagalog is that of reflexivizability. According to him, one notable feature of the subject is the ability to control reflexivization, but not be reflexivized itself (cf. Schachter 503, 505). It would also be interesting to see how Ayeri fares in this regard.

Of course, Ayeri allows the actor-topic to be an antecedent for reflexivization:

  1. Ang
    AT
    silvye
    see-3SG.F
    envan
    woman-T
    sitang-yes
    self=3SG.F.P
    puluyya.
    mirror-LOC

    “The woman sees herself in a mirror.”

However, it also allows the actor to be reflexivized upon even if it is not the topic, and moreover, it is even possible for the reflexive NP to constitute the topic:

  1. Sa
    PT
    silvye
    see-3SG.F
    envanang
    A-woman
    sitang-ye
    self=3SG.F.T
    puluyya.
    mirror-LOC

    “A woman sees herself in the mirror.”

Besides, even a non-actor can control reflexivization (cf. Kroeger 9), which Schachter does not consider:

  1. Sa
    PT
    ningya
    tell-3SG.M
    ang Nasan
    A-Nasan
    kalam
    truth-T
    yam Cānvay
    D-Cānvay
    sitang-yena.
    self=3SG.F.GEN

    “Nasan tells Cānvay the truth about herself.”

Since Ayeri marks gender on third-person pronouns, the sentence presented in example 16 is not ambiguous as for whose truth is revealed, unlike its Tagalog counterpart (cf. Kroeger 9). But otherwise, examples 14–16 work the same way in Tagalog as well. And just like in Tagalog, the latter two examples in 17 are ungrammatical (cf. Schachter 505):

  1. Sa
    PT
    niltang
    think-3PL.M.A
    sitang-yan.
    self=3PL.M.T

    “They think of themselves.”
     
    *​Sa
    PT
    nilyan
    think-3PL.M.T
    sitang-tang
    self=3PL.M.A

     
    *​Ang
    AT
    niltas
    think-3PL.M.P
    sitang-yan.
    self=3PL.M.T

The latter two examples in 17 show that reflexive NPs cannot be actors, not even if they are topic-marked. And conversely, reflexivization can only be controlled by actors, but actors cannot be reflexive themselves.

2.4. Equi-NP deletion

Another test applied by both Schachter and Kroeger is that of deleting coreferential complement subjects, or equi-NP deletion (cf. Schachter 504–05, Kroeger 10–11). According to theory, it is always an instance of the subject that is deleted in sentences like the following (Schachter 504):

  1. I want to leave.
    = I want [ISUB leave].

Provided that the topic is the subject in Tagalog, Schachter then goes to check whether topics become targets of deletion as they should. However, he finds that in all tested configurations, it is always the actor that is deleted instead (cf. 504–05). Something to the same effect, albeit with slightly different syntax, also happens in Ayeri:

  1. Le
    PT
    ihaya
    borrow-3SG.M
    ang Mican
    A-Mican
    pangis
    money-T
    butayana.
    lender-GEN

    “Mican borrowed the money from a money lender.”
     
    Le
    PT
    perisaya
    hesitate-3SG.M
    ang Mican
    A-Mican
    ihayam
    borrow-PTCP
    pangis
    money-T
    butayana.
    lender-GEN

    “Mican hesitated to borrow the money from a money lender.”
     
  2. Na
    GENT
    ihaya
    borrow-3SG.M
    ang Mican
    A-Mican
    pangisley
    money-P
    butaya.
    lender-T

    “Mican borrowed money from the money lender.”
     
    Na
    GENT
    perisaya
    hesitate-3SG.M
    ang Mican
    A-Mican
    ihayam
    borrow-PTCP
    pangisley
    money-P
    butaya.
    lender-T

    “Mican hesitated to borrow money from the money lender.”
     
  3. Ang
    AT
    ihaya
    borrow-3SG.M
    Mican
    T-Mican
    pangisley
    money-P
    butayana.
    lender-GEN

    “Mican borrowed money from a money lender.”
     
    Ang
    AT
    perisaya
    hesitate-3SG.M
    Mican
    T-Mican
    ihayam
    borrow-PTCP
    pangisley
    money-P
    butayana.
    lender-GEN

    “Mican hesitated to borrow money from a money lender.”

The difference to Tagalog here is that the controlled predicate (iha- ‘borrow’) does not bear any agreement marking, but is instead marked as a participle. The topic of the complement clause is instead marked on the controlling verb (perisa- ‘hesitate’). Like in Tagalog, the agent NP is dropped, so that “ang Mican” is not repeated in the complement, and there is no person marking on the complement phrase’s predicate either (cf. Schachter 504).

However, while Tagalog can form sentences like (Kroeger 10):

  1. Nagpilit
    PERF.AV-insist.on
    si
    NOM
    Maria=ng
    Maria=COMP
    bigy-an
    give-DV
    ng
    GEN
    pera
    money
    ni
    GEN
    Ben.
    Ben

    “Maria insisted on being given money by Ben.”

Ayeri does not allow for such complexity due to the lack of agreement marking on the controlled verb and phrases the statement instead as:

  1. Sapriye
    insist-3SG.F
    ang Maria,
    A-Maria,
    yam
    DT
    mya
    shall
    ilya
    give-3SG.M
    ye
    3SG.F.T
    ang Ben
    A-Ben
    trendaley.
    money-P

    “Maria insisted that she be given money by Ben.”

Note that in example 22 above, Kroeger glosses the topic as ‘NOM’, the goal as ‘GEN’, and the topic-marking on the verb as ‘V’ for ‘voice’. There is one instance in Schachter where the particle ni is glossed as an actor (cf. Schachter 497), but in the example above it must logically be “Maria” who is the dropped actor that the “DV” in the complement phrase refers to.

2.5. Conjunction reduction and pro-drop

Another topic related to deletion is that of dropping constituents in phrases joined by ‘and’. Kroeger only spends one short paragraph on this topic, however, he incldues a number of examples to demonstrate that only the topic can be deleted in coordinated clauses (Kroeger 8):

  1. Huhugasan
    wash-DV
    ko
    1SG.GEN
    at
    and
    pupunasan
    dry-DV
    mo
    2SG.GEN
    ang
    NOM
    mga
    PL
    pinggan.
    plate

    “I will wash and you dry the dishes.”

In this example, the first instance of ang mga pinggan ‘the dishes’ has been deleted, just like in the English translation. However, Ayeri is rather averse to this kind of dropping, no matter whether the dropped phrase is the topic or not:

  1. *​Le
    PT
    kecang
    wash-1SG.A
    nay
    and
    tirisavāng
    dry-2.A
    vekamdekey.
    dishes-T

Rather, what would happen in Ayeri is another kind of dropping, namely that of coreferential unstressed third-person pronouns of transitive verbs as implied by context:

  1. Le
    PT
    kecang
    wash-1SG.A
    vekamdekey
    dishes-T
    nay
    and
    tirisavāng.
    dry-2.A

    “I wash the dishes and you dry (them).”

This also works with relative clauses:

  1. samanas
    two-NMLZ-P
    mitanyena
    palace-PL-GEN
    si
    REL
    midaytong
    surround-3PL.N.A

    “of a second palace which surrounds (it)”

However, since there is mandatory actor agreement, the actor cannot be dropped completely, but there is always some trace remaining by means of verb agreement, since verbs cannot be uninflected in Ayeri:

  1. *​Le
    PT
    kecang
    wash-1SG.A
    vekamdekey
    dishes-T
    nay
    and
    tirisa.
    dry

     
    Le
    PT
    keca
    wash-3SG.M
    ang Mican
    A-Mican
    vekamdekey
    dishes-T
    nay
    and
    tirisaya.
    dry-3SG.M

    “Mican washes the dishes and dries (them).”

Also compare Ayeri to the following example from Kroeger (8):

  1. Tagalog:

    Hinalikan
    PERF-kiss-DV
    ni
    GEN
    David
    David
    si
    NOM
    Linda
    Linda
    atsaka
    and.then
    umalis.
    PERF.AV-leave

    “Linda was kissed by David and then (he) left.”

    Ayeri:

    Sa
    PT
    vengaya
    kiss-3SG.M
    ang David
    A-David
    Linda
    T-Linda
    nay
    and
    saraya
    leave-3SG.M
    epang.
    after

    “Linda was kissed by David and then (he) left.”

The Ayeri sentence in example 29 above is not ambiguous as to who leaves due to gender agreement. If Linda left, the verb would need to be marked as feminine, however, since Linda is not the coreferential actor, there also needs to be case marking to disambiguate:

  1. nay
    and
    sarayeng
    leave-3SG.F.A
    epang.
    after

    “… and then she left.”

Furthermore, there is topic polarity to consider in the dropping of coreferential NPs:

  1. Ang
    AT
    ketay
    wash-1S.T
    vekamley dekey
    dishes-P
    nay
    and
    tirisāy
    dry-1S
    raley.
    3PL.P

    “I wash dishes and dry them.”
     
    ​?​​Ang
    AT
    ketay
    wash-1S.T
    vekamley dekey
    dishes-P
    nay
    and
    tirisāy.
    dry-1S

    “I wash dishes and (I) dry.”

The example above shows that the actor-topic would be the subject to drying rather than the dishes if the third-person pronoun were dropped. Consequently, in order to keep the topic continuous across the conjoined phrases in example 29 and 30 respectively, the conjoined phrase must be:

  1. nay
    and
    saraye
    leave-3SG.F.T
    epang.
    after

    “… and then she was left.”

2.6. Relativization

An important constraint in Tagalog is that only topics can be relativized, while the topic is deleted in the subordinate clause, since it is understood as as the coreferent head of both the matrix- and the relative clause (cf. Schachter 500, Kroeger 6). The relative clause itself is introduced with an enclitic linking particle, -ng (adapted from Schachter 500):

  1. Bumasa
    AT-read
    ang lalaki
    T-man
    ng diyaryo.
    G-newspaper

    “The man read a newspaper.”
     
    Matalino
    intelligent
    ang lalaki=ng
    T-man=COMP
    bumasa
    AT-read
    ng diyaryo.
    G-newspaper

    “The man who read a newspaper is intelligent.”

In Ayeri, on the other hand, any constituent can be relativized, regardless of it being the topic. First, consider the following simple statements:

  1. Ayonang
    man-A
    alingo.
    intelligent

    “The man is intelligent.”
     
    Ang
    AT
    layaya
    read-3SG.M
    ayon
    man-T
    budangas.
    message-P

    “The man reads a message.”
     
    Budangisa
    message-CAU
    yāng
    3SG.M.A
    mino.
    happy

    “The message makes him happy.”

Combining the first two examples of 34 has the following result:

  1. Ang
    AT
    layaya
    read-3SG.M
    budangas
    message-P
    ayon
    man-T
    si
    REL
    alingo.
    intelligent

    “The man who is intelligent reads a message.”

In this case, the actor-topic is the constituent that has been relativized. As mentioned before, the word order shifts according to syntactic weight, so that the part of the sentence that carries most information comes last.

Incorporating the the third sentence in example 34 above into the second sentence has in turn the following result:

  1. Ang
    AT
    layaya
    read-3SG.M
    ayon
    man-T
    budangas
    message-P
    sisā
    REL-P-CAU
    yāng
    3SG.M.A
    mino.
    happy

    “The man reads a message that makes him happy.”

Here, the relative clause refers to a non-topic NP. The relative pronoun, si, is even inflected for the role of the relative clause’s head in the relative clause, that is, a causative.

2.7. Relational Annihilation Law and Passivization

In order to test whether the actor-topic might in fact be the subject in Tagalog, Schachter tests whether the Relational Annihilation Law holds. According to this law, “[i]f an NPi assumes a grammatical relation j previously borne by NPj, then NPj ceases to bear any grammatical relation; it becomes a chômeur (French for ‘unemployed person’)” (Schachter 512). This is the case e.g. for passives in English, where the subject of an active sentence is demoted to an adverbial while the previous object takes its place. In terms of relational grammar, the previous subject ceases to be a ‘term’ (i.e. a subject, direct object, or indirect object) on demotion, thereby losing all its associated functional properties and becoming a chômeur (cf. 512). It turns out, however, that the Relational Annihilation Law is not applicable to Tagalog at least in that actors can control reflexivization and equi-NP deletion, regardless of being a topic or not, so that a non-topic actor does not become a chômeur (cf. 513). The same has been shown to be true for Ayeri (cf. 2.3. and 2.4. above).

Since passives were already mentioned with regards to the Relational Annihilation Law, I want to add that Kroeger consistently renders sentences with goal/direction-topic as passives in English translation (Kroeger 8):

  1. Hinalikan
    PERF-kiss-DV
    ni
    GEN
    David
    David
    si
    NOM
    Linda
    Linda
    at.saka
    and.then
    umalis.
    PERF.AV-leave

    “Linda was kissed by David and then left.”

Taking topicality into account, this makes sense in English. However, apart from different marking of topic, nothing special seems to happen here. Ayeri can do the same, as we have seen in example 29 above:

  1. Sa
    PT
    vengaya
    kiss-3SG.M
    ang David
    A-David
    Linda
    T-Linda
    nay
    and
    saraya
    leave-3SG.M
    epang.
    after

    “Linda was kissed by David and then (he) left.”

Furthermore, Schachter and Otanes show the following example (adapted from Schachter/Otanes 73):

  1. Kinain
    GT-eat
    ang pagkain.
    T-food

    “The food was eaten.”

Unfortunately, they do not add interlinear glossing here, but they explain that “In the case of most goal-focus verbs” – such as kain ‘eat’ (cf. “Tagalog Dictionary”) – “the actor complement may be omitted, if the sentence expresses an action performed by unknown persons, or if the performer of the action is irrelevant” (Schachter/Otanes 73). They also inform that this construction is usually translated as a passive in English.

Ayeri can do something similar in this case, although again, it differs due to person-agreement and lack of topic-marking in intransitive sentences:

  1. Kondara
    eat-3SG.INAN
    kondanley.
    food-P.INAN

    “(The) food was eaten.”

Similar to the goal-topic marking on the verb in example 39 above, the verb in example 40 agrees in person, number, and gender with the patient NP that is the only overt constituent in this clause. Verb agreement with non-actors is thus also possible, but restricted to quasi-intransitive sentences with covert actors.

3. Subjecthood in Ayeri

The question of subjecthood has come up a number of times in the course of the above survey, and it is the central topic in Schachter’s paper that has served as an important theoretical backing.

While Li and Thompson categorize Tagalog as neither topic-prominent nor subject-prominent on the grounds of a quotation from Schachter and Otanes’ Tagalog Reference Grammar which suggests that the term ‘subject’ is unsuitable for Tagalog (cf. Li/Thompson 460–61), Kroeger assumes that the NP marked by ang, i.e. the actor-topic, is the best candidate for subjecthood (cf. Kroeger 1). Schachter, however, arrives at a much more differentiated result at the end of his paper. According to his theory, the notion ‘subject’ in fact is a bundle of three sub-categories: reference-related properties, role-related properties, and either reference- or role-related properties. While the subject constitutes a single unit in many languages, Tagalog and other Philippine languages split these properties and associate them with the topic, the actor, and verb agreement (cf. Schachter 513–15).

In comparison, however, the topic in Ayeri does not have all the subject-like properties that it has in Tagalog and other related languages from Schachter’s survey. While topics in Ayeri signify definiteness, can be relativized, play a role in coreferential NP dropping with regards to conjunction reduction, control of relativization is not limited to topics, and neither is coreferential NP dropping limited to them in conjunction reduction. Additionally, sentences without topic-marking can occur, namely intransitive sentences; floating quantifiers (cf. 501), on the other hand, do not occur at all.

With regards to subject-like properties of actors, Ayeri seems to put a stronger emphasis on this category than Tagalog does. Thus, actors do not need to be topics in order to control relativization, there is obligatory verb agreement with actors, and existential sentences are formed like ordinary declarative sentences by means of the verb yoma- ‘exist’ and thus do not lack actors (cf. 502). Similarities between Philippine languages and Ayeri in this case are the ability of actors to control reflexivization while not being reflexivizable themselves, being the only target of equi-NP deletion, deletion in imperatives, and an almost fixed constituent order – albeit only customary, but not theoretically necessary – that resembles VSO.

An argument against the actor-topic being the subject in both Tagalog and Ayeri is that in demotion, actors do not become chômeurs. However, Ayeri’s actor-topic verbs do not show a special prominence or special abilities with regards to morphologic categories (cf. 510).

4. Conclusion

In this essay the question was posed, in how far Ayeri resembles the group of languages that inspired a major part of its basic syntactic structure, namely the syntactic alignment system typically employed by Philippine languages. Other, related grammatical similarities than those primarily concerned with syntactic alignment have been discussed shortly for good measure.

Ayeri proved to likely not be completely categorizable in terms of the standard syntactic typology that assumes subjects to exist as single units in that the notion ‘subject’ does not completely conincide with either the topic and the actor, but both show similarities and differences to the standard. However, the actor would be the more likely candidate for subjecthood in comparison to Tagalog, which was commonly used as an example in the scientific literature consulted.

Certainly, my bias as a speaker of an Indo-European language with a “well-defined subject” (Schachter 515) shows, as well as my initial cluelessness of what the Austronesian alignment entails besides ‘triggeriness’, but I hope to have illustrated in this essay that Ayeri is more than just Standard Average European in disguise. And although Tagalog served as a source of inspiration, I think by now that copying its mechanics exactly is not necessarily desirable, but researching them was fascinating. And as Van Valin writes, “there is considerable variation among Philippine languages; it should not be assumed that other Philippine languages parallel Tagalog in every detail” (Van Valin 70), so in order to be reminiscent of Philippine languages, it is not even necessary to fulfill every criterion, as criteria and their fulfillment vary anyway.

Abbreviations

1, 1st person
2, 2nd person
3, 3rd person
A, actor
AFF, affirmative
AT, actor-topic
B, beneficiary
CAU, causative
COMP, complementizer
D, direction
DV, direction voice
F, feminine
GEN, genitive
GENT, genitive-topic
GT, goal-topic
IMP, imperative
INAN, inanimate
LOC, locative
M, masculine
N, neuter
NOM, nominative
P, patient
PL, plural
PT, patient-topic
S, singular
SUB, subject
T, topic

  1. Cf. this Wikibooks article based on discussions about “trigger languages” on Conlang-L way back. Take that wiki page with a modicum of salt and continue reading this blog post instead for a better informed account of how it works in real life, and also in my conlang. — 2013-08-02

One thought on “Flicking Switches: Ayeri and the Austronesian Alignment

  1. This was really great, and I enjoyed reading it thoroughly. I learned some things about Tagalog, about “trigger systems”, and about Ayeri. I love a good, meaty, linguistic discussion, and one with a conlang aspect even more.

    [Thanks, Jesse! —CB]

Comments are closed.