Update on the Grammar Writing Process III

I’ve recently done a lot of proofreading of basically anything besides the introduction chapter of the new Ayeri Grammar. I did this to weed out errors I’ve previously overlooked and also to make sure that what I’d written earlier in the morphology chapter was consistent with the rather extensive work I did in order to come to terms with why certain pre- and suffixes should be clitics. This detour took quite a while—from January to April—but it was probably worth it, since it clarified some questions I had. My quest for clarity on clitics versus affixes in Ayeri culminated in a lengthy blog article, a version of which, revised in parts, can be found in the new grammar as section 3.2.5.

Starting to document Ayeri’s syntax is the logical next step now after I tried to describe its phonology and morphology as well as I could. So, what I’m up to now is trying to describe the morphosyntactic structure of the various syntactic constituents: noun phrases, verb phrases, adjective phrases etc. Since there is very little agreement morphology in Ayeri, this should be rather straightforward for the most part, except for verb phrases (I recently discovered that Dalrymple (2001) contains a whole lot more examples than Bresnan (2016), so that might come in handy). Currently, however, I’m still only concerned with NPs and DPs. What’s still making me self-conscious about all this is that I still have never really studied syntax formally, as I pointed out earlier. So, if you take a look at the grammar and see something implausible, please let me know!

When I tried to figure out clitics in Ayeri earlier, I also came up with a lot of examples of coordination, and one thing I wondered is if the following is actually reasonable.

An attempt to describe formally the distribution of the progressive clitic over two coordinated verbs

What you can see here is an attempt to apply LFG to an example sentence which contains a coordinated constituent: manga sahaya rangya ‘is coming home’ is coordinated with nedraya ‘sits (down)’. The question now is, how to formally describe that manga as the (enclitic) progressive marker is to be understood as distributing over both verbs, sahaya ‘comes’ and nedraya ‘sits’? I actually looked up a few articles (Belyaev et al. 2015; Kaplan and Maxwell 1988; Maxwell and Manning 1996; Peterson 2004) and at least took a casual glance at them, but nowhere did I see any discussion of how to indicate when certain markers in the verb phrase distribute to multiple conjuncts. Instead, I could only find discussions of how to indicate the distribution of the subject to conjuncts. The distribution of the subject is also indicated in the argument-value matrix on the right in the illustration above, namely, in that the first verb’s SUBJ(ect) is connected by a line to the second verb’s empty SUBJ slot.

The question I now have is whether connecting items this way is possible also for other features, like ASP(ect). From what little I know, I don’t see why it shouldn’t be plausible to analogize here, but I might as well be wrong. If you know, please let me know as well. What is slightly frustrating is that a lot of times, you can only easily find information on English.

Also, I’ve been working on writing this grammar for almost a whole year now. Wow.

  • Belyaev, Oleg, et al. “Number Mismatches in Coordination: An LFG Analysis.” Proceedings of the LFG ’15 Conference, Tokyo, Japan, 18–20 Jul. 2015. Ed. by Miriam Butt and Tracy Holloway King. Stanford, CA: CSLI Publications, 2015. 26–46. Web. 25 May 2017. ‹https://web.stanford.edu/group/cslipublications/cslipublications/LFG/20/papers/lfg15belyaevetal.pdf›.
  • Bresnan, Joan et al. Lexical-Functional Syntax. 2nd ed. Chichester: Wiley Blackwell, 2016. Print. Blackwell Textbooks in Linguistics 16.
  • Dalrymple, Mary. Lexical Functional Grammar. San Diego, CA: Academic Press, 2001. Print. Syntax and Semantics 34.
  • Kaplan, Ronald M., and John T. Maxwell, III. Constituent Coordination in Lexical-Functional Grammar. Palo Alto, CA: Xerox PARC, 1988. Web. 18 Apr. 2017. ‹http://www2.parc.com/isl/groups/nltt/xle/coord.ps›.
  • Maxwell, John T., III, and Christopher D. Manning. “A Theory of Non-constituent Coordination Based on Finite-State Rules.” Proceedings of the LFG ’96 Conference, Rank Xerox, Grenoble, France, 26–28 Aug. 1996. Ed. by Miriam Butt and Tracy Holloway King. Stanford, CA: CSLI Publications, 1997. Web. 18 Apr. 2017. ‹https://web.stanford.edu/group/cslipublications/cslipublications/LFG/1/lfg96maxwellmanning.pdf›.
  • Peterson, Peter G. “Coordination: Consequences of a Lexial-Functional Account.” Natural Language & Linguistic Theory 22.3 (2004): 643–679. Web. 18 Apr. 2017. ‹http://www.jstor.org/stable/4048099›.

Some Further Thoughts on Agreement in Ayeri

Ayeri makes use of grammatical gender to at least some degree. As a category, gender is by far not as pervasive as in Romance languages, though, since in Ayeri it only matters in picking the correct case suffix for agents and patients (animate vs. inanimate) as well as in 3rd person pronouns and verb agreement (masculine/feminine/neuter animate vs. inanimate). The fact that gender plays a role in verb agreement opens up possibilities for adding some quirks. As pointed out in a previous blog article,

Ayeri usually exhibits verbs as agreeing with agents and occasionally patients, not topics as such. This may be a little counterintuitive since the relation between topics and subjects is close, but is possibly due to the fact that the unmarked word order is VAP. This means that agent NPs usually follow the verb, and it strikes me as not too unnatural to have an agreement relation between the verb and the closest NP also when non-conjoined NPs are involved (Corbett [2006:] 180). This conveniently explains why verbs can agree with patients as well if the agent NP is absent.

This brings up the idea that Ayeri should likely exhibit agreement with the closest conjunct when a word is forced to agree with coordinated NPs of different genders, instead of finding some way to resolve conflicting gender features. Yet, however, I have often been following the rule that a masculine entity and a feminine entity, regardless of number, resolve to masculine as the default gender, though sometimes I have also used the neuter as a third category to escape to (Wechsler 2009 actually quotes Icelandic as following this route):

M + F = M (N attested)
F + M = M (N attested)

I was originally wondering if I should get rid of this system and instead use nearest-conjunct agreement throughout, but—why not have both? In this little blog article I basically want to sketch out an idea I had in the hope that it is well within the confines of what natural languages do. Note that the tables given in the following refer strictly to verb agreement, where the verb precedes any NPs it agrees with. Thus, for the outcome of nearest-conjunct agreement, it is assumed that the agreement target precedes the agreement controller, so that the verb should in most cases agree with the first conjunct.

Now, if NPs referring to people or other entities to which masculine and feminine apply as grammatical categories have their diverging gender features resolved to masculine as the default, what about combining either masculine or feminine with neuter? This is shown in the following table:

M + N = M
N + M = N
F + N = M (F possible)
N + F = N

From this table we can gather that in general, there is nearest-conjunct agreement for combinations of masculine/feminine and neuter, though feminine and neuter equally resolve to masculine as the default if the feminine conjunct is closer to the agreement target. Note that for agreement with simplex NPs, there is no default gender, so masculine gender will trigger masculine agreement, feminine gender will trigger feminine agreement, etc. As indicated in the table above, true nearest-conjunct agreement with the feminine conjunct is possible as well, however, basically ignoring any further complications. It takes not a lot of imagination to assume that the regularization towards nearest-conjunct agreement would be a point of divergence between the formal and the colloquial language and also very likely a fertile ground for prescriptivist bickering.

So far, we have only looked at the combinations within the animate tier. The following table lists the possible permutations for combinations of animate and inanimate NPs:

M + INAN = M
INAN + M = INAN
F + INAN = M (F possible)
INAN + F = INAN
N + INAN = N
INAN + N = INAN

For consistency, the same rules as above operate here: masculine and feminine mixed with inanimate show nearest-conjunct agreement. If, however, a feminine conjunct comes first, agreement will default to masculine, though again, nearest-conjunct agreement to feminine is possible.

Wechsler (2009: 571–73) furthermore discusses Corbett (1991)’s observation that there may be differences in how languages go about gender resolution with regards to semantic and syntactic resolution—i.e. resolution of grammatical gender according to biological gender vs. formal gender—and finds that animacy plays a role in that. This is especially relevant for cases where grammatical and semantic gender diverge, as in hybrid nouns like German Mädchen ‘girl’, which semantically refers to a female person but which is formally of neuter gender for grammatical purposes, since the diminutive suffix -chen always derives neuter nouns.

While Ayeri assigns masculine and feminine semantically (with neuter for the remainder of animate entities which are neither male nor female), there are occasional idiosyncrasies with nouns very obviously referring to non-living things being assigned animate neuter gender, such as nanga ‘house’. It might be interesting to develop some further ideas for likely outcomes in that regard even though canonically, nearest-conjunct agreement should operate in those cases and conlangs, by their nature, probably produce a lot less variation than actual natural languages do.

  • Corbett, Greville G. Agreement. Cambridge: Cambridge UP, 2006. Print. Cambridge Textbooks in Linguistics.
  • ———. Gender. Cambridge: Cambridge UP, 1991. Print. Cambridge Textbooks in Linguistics.
  • Wechsler, Steven. “‘Elsewhere’ in Gender Resolution.” The Nature of the Word: Studies in Honor of Paul Kiparsky. Ed. by Kristin Hanson and Sharon Inkelas. Cambridge, MA: MIT Press, 2009. Print. 567–586. Current Studies in Linguistics.
  • I can’t seem to keep my terminology straight: I corrected *nearest-conjunct resolution to nearest-conjunct agreement in a few places.

Markov-Chain Generator for Ayeri Words

Since I’m sometimes a little lazy to come up with new words, I wrote myself a little Python script which pulls a certain subset of words from the dictionary database I’m using and applies a Markov chain algorithm to it in order to generate new similar words. The script is sophisticated enough to filter out duplicates and some other undesirable outcomes. You can adapt the code shared below to your needs if you wish to.


Clitics in Ayeri: Thoughts and Notes

For the past months, work on writing my grammar has been stalling since I’ve been trying to figure out which of the functional morphemes that are maybe a little untypical as inflections might in fact be clitics. After having read Spencer and Luís (2012) on the topic nearly cover-to-cover, I’ve become a little disenchanted about the whole effort, since clitics appear to be very elusive things which can’t be easily defined in a formal way (126). Based on my reading, some characteristic traits appear to be:1

  • Clitics behave in part like function words and in part like affixes, but in any case they are not free morphemes (38, 42).
  • Clitics tend to be phonologically weak items (39).
  • Clitics prominently—and importantly—tend to attach ‘promiscuously’ to surrounding words. That is, unlike inflection, they are not limited to connect to a certain part of speech or to align with their host in semantics (40, 108–109).
  • Clitics tend to appear in a second position, whether that is after a word or an intonational or syntactic phrase (41)
  • Clitics tend to be templatic and to cluster, especially if they encode inflection-like information (41, 47–48).
  • Clitics have none of the freedom of ordering found in true words and phrases (43).
  • Positions of ‘special’ clitics tend to not be available to free words (44).
  • Clitics tend to be functional morphemes, and to realize a single morphosyntactic property (67, 179).
  • There are no paradigmatic gaps (108–109).
  • There tends to be no morphophonemic alteration like vowel harmony, stress shift or sandhi between a clitic and its host (108–109).
  • There tends to be no idiosyncratic change in meaning when a clitic and a clitic host come together, unlike there may be with inflection (108, 110).
  • Similar to affixes, clitics and their host tend to be treated as a syntactic unit, that is, there is lexical integrity, so you can’t put word material in between a clitic and its host (108, 110).
  • Clitics usually get joined to a host word after inflection (108, 110)
  • Affixes tend to go on every word in a conjunct (narrow scope), while clitics have a tendency to treat a conjunct as a unit to attach to (wide scope; 139, 196 ff.).

However, Spencer and Luís (2012) point out many counterexamples in order to drive their point home that the border between clitics and affixes is often fuzzy. It comes as no surprise that according to their assessment, there’s a lot of miscategorization in individual grammars as a result (107). Another consequence of definitional fuzziness is that since not all of the traits described above are always present, making a checklist and summing up the tally is only of limited value. The traits enumerated above are sufficient, but not necessary, conditions.

This blog article has become rather long and technical in retrospect, but I needed to write it all down in order to get a clear head about the status of various particles and affixes in Ayeri which behave not quite like function words or inflection, and which I’ve thus long suspected to belong to the interstitial category of clitics. The first part of this article will detail all the particles and affixes which precede the lexical heads they modify, the second one will elaborate on suffixes. This article is also likely to find its way into the 2016/17 edition of the Ayeri Grammar, which I’m still working on, as a subchapter to the chapter on morphological typology.

1. Preposed particles and prefixes

Now, I think what should be rather unproblematic with regards to analysis as clitics in Ayeri are the preverbal elements, that is, the topic marker, one or several modal particles, the progressive marker, and also the emphatic affirmative and negative discourse particles. All of these particles essentially have functional rather than lexical content, and are usually unstressed. They come in a cluster with a fixed order, and they appear in a position where no ordinary word material could go, since Ayeri is strictly verb-initial. In conjuncts it’s also unnecessary to mark every verb with one or several of them:

    1. Ang
      ang
      AT
      kece
      ket-ye
      wash-3SG.F
      nay
      nay
      and
      dayungisaye
      dayungisa-ye
      dress-3SG.F
      māva
      māva-Ø
      mother-TOP
      yanjas
      yan-ye-as
      boy-PL-P
      yena.
      yena
      3SG.F.GEN

      ‘The mother washes and dresses her boys.’
    2. Manga
      manga
      PROG
      sahaya
      saha-ya
      come-3SG.M
      rangya
      rang-ya
      home-LOC
      nay
      nay
      and
      nedraya
      nedra-ya
      sit-3SGM
      ang
      ang
      A
      Tikim.
      Tikim
      Tikim

      ‘Tikim is coming home and sitting down.’
    3. Ang
      ang
      AT
      mya
      mya
      be.supposed.to
      ming
      ming
      can
      sidegongya
      sideg-ong-ya.Ø
      repair-IRR-3SG.M.TOP
      nay
      nay
      and
      la-lataya
      la~lata-ya.Ø
      ITER~sell-3SG.M.TOP
      ajamyeley.
      ajam-ye-ley
      toy-PL-P.INAN

      ‘He should be able to repair and resell the toys.’

In (1a), therefore, the agent-topic marker ang only occurs before kece ‘(she) washes’, and the conjoined verb dayungisaye ‘(she) dresses’ is also within its scope. Repeating the marker as well before the latter verb could either be considered ungrammatical because there is only one topic there—māva ‘mother’—or the sentence could be interpreted as having two conjoined clauses with different subjects: ‘She1 washes and mother2 dresses her boys.’ The latter outcome has māva as the topic only of dayungisaye, while kece‘s topic is the person marking on the verb—a pro-drop subject, essentially.

In (1b), then, the progressive marker manga equally has scope over both verb conjuncts, sahaya ‘(he) comes’ and nedraya ‘(he) sits’ in what I presume is a case of extended/distributed exponence, since the verb conjuncts can be represented by the incomplete f-structure matrix (cf. Bresnan et al. 2016; Butt and King 2015) shown in (2). Manga is treated there as being part of things the verb inflects for, that is, progressive aspect. The topic marker ang is not a semantic property of the verb, but announces the case and—for agents and patients—the animacy value of the topicalized noun phrases (NPs), so the f-structure in (2) lists this information under the TOP relation.

  1. Part of the f-structure of saha- ‘come’ in (1b):

    PRED ‘come ⟨(↑ SUBJ), (↑ OBLloc)⟩’
    ASP PROG
    TOP
    CASE A
    ANIM +

Modal particles, exemplified in (1c), are probably slightly less typical as clitics since it seems feasible for them to be stressed for contrast. What is not possible, however, is to front either mya or ming, and the verb itself also can’t precede the particles, which is demonstrated in (3). It’s also not possible to coordinate any of the elements in the preverbal particle cluster with nay ‘and’, as we will see in (4).

    1. * mya ang ming sidegongya
    2. * ming ang mya sidegongya
    3. * sidegongya ang mya ming
    1. * ang nay mya ming sidegongya
    2. * ang mya nay ming sidegongya
    3. * ang mya ming nay sidegongya

It needs to be pointed out that unlike verbs, modal particles in Ayeri resist inflection, so in (1c) the irrealis suffix -ong is realized on the verb sidegongya ‘(he) would repair’ instead of on one or both of the modal particles as * mingong and * myong, respectively. The combination of mya ‘be supposed to’ with an irrealis-marked verb together indicates that the speaker thinks the action denoted by the verb should be carried out. On the other hand, the marking on the verb may also be interpreted as being valid for the whole verb complex, and just generally adds the feature [MOOD IRR] to the verb’s feature list in analogy to [ASP PROG] in (2). With regards to negation, it’s just the same: only the verb can be negated, but not the modal particle. Possibly, it would be useful in this case to abstract the modal particles as a feature [MODALITY] as listed by ParGram for purposes of functional representation. At least superficially, it looks as though Ayeri acts differently from English here in that the verb is possibly not a complement of the modal. I suppose that this is mostly apparent from the fact that in Ayeri, the verb inflects, not the modal particle. Furthermore, modal particles in Ayeri can’t be modified by adverbs like regular verbs can:

    1. Ming
      ming
      can
      tigalye
      tigal-ye
      swim-3SG.F
      ban
      ban
      well
      nilay
      nilay
      probably
      ang
      ang
      A
      Diya.
      Diya
      Diya

      ‘Diya can probably swim well.’
    2. * 
      Ming
      ming
      can
      nilay
      nilay
      probably
      tigalye
      tigal-ye
      swim-3SG.F
      ban
      ban
      well
      ang
      ang
      A
      Diya.
      Diya
      Diya

In order to not confuse things even more, it seems advisable to decree that combinations of topic particle and modal particle, as well as modal particle and verb, likewise not be interrupted by parenthetical material like naratang ‘they say’, so that:

    1. ​Naratang,​
      nara-tang
      say-3PL.M.A
      ang
      ang
      AT
      ming
      ming
      can
      tigalye
      tigal-ye
      swim-3SG.F
      ban
      ban
      well
      Diya
      Diya
      Diya
      kodanya.
      kodan-ya
      lake-LOC

      ‘They say Diya can swim well in a lake.’
    2. * Ang, naratang, ming tigalye ban Diya kodanya.
    3. * Ang ming, naratang, tigalye ban Diya kodanya.
    4. ? Ang ming tigalye, naratang, ban Diya kodanya.
    5. Ang ming tigalye ban, naratang, Diya kodanya.
    6. Ang ming tigalye ban Diya, naratang, kodanya.
    7. Ang ming tigalye ban Diya kodanya, naratang.

Of all the other parts of speech, only nouns also have preposed modifiers. This is the case with proper nouns, where the name will be preceded by a case marker instead of receiving a case-marking suffix. This case marker is phonologically weak in that it is no longer than other affixes, and unstressed, with the exception of the causative case marker , which bears at least secondary stress since it contains a long vowel. We already saw case particles preceding names in (1b) and (5) above: ang Tikim and ang Diya; ang marks the proper-noun NPs as agents in both cases. The case marker is missing when the NP is topicalized, as exemplified by (6), where the agent NP appears as just Diya, not ang Diya. While case suffixes have narrow scope as in (7a) and thus need to be repeated on every NP in a conjunct, preposed case markers may be used with wide scope if both conjuncts are proper nouns as in (7c). Narrow scope with proper nouns may add an individuating connotation in (7d).

    1. Toryon
      tor-yon
      sleep-3PL.N
      veneyang
      veney-ang
      dog-A
      nay
      nay
      and
      badanang.
      badan-ang
      father-A

      ‘The dog and father are (both) sleeping.’
    2. * 
      Toryon
      tor-yon
      sleep-3PL.N
      veney
      veney_
      dog_
      nay
      nay
      and
      badanang.
      badan-ang
      father-A
    3. Sa
      sa
      PT
      sobisayan
      sobisa-yan
      study-3PL.M
      ang
      ang
      A
      Niva
      Niva
      Niva
      nay
      nay
      and
      Mico
      Mico
      Mico
      narānye.
      narān-ye-Ø
      language-PL-TOP

      ‘Languages is what Niva and Mico study.’
    4. Sa
      sa
      PT
      sobisayan
      sobisa-yan
      study-3PL.M
      ang
      ang
      A
      Niva
      Niva
      Niva
      nay
      nay
      and
      ang
      ang
      A
      Mico
      Mico
      Mico
      narānye.
      narān-ye-Ø
      language-PL-TOP

      ‘Languages is what Niva and Mico (each) study.’

Taking the above characteristics into account, one may argue that the preposed case markers are clitics. They also often enough follow other phonetic material, so it should be possible to analyze case markers as indiscriminately attaching to the word on the left while syntactically modifying the word to the right:

  1. ? Sa sobisayan=ang Niva nay Mico narānye.

There is no audible break or change in intonation between sobisayan and ang, and there is also no morphophonemic modification occurring between them either: if ang were a suffix, stress would shift from /sobiˈsajan/ to /sobiˌsajaˈnang/. According to Klavans (1985), a clitic’s phonological leaning to the preceding word while modifying the following one is a process which is attested in natural languages, and she quotes examples from Kwak’wala, a Wakashan language of British Columbia (Hammarström et al. 2017), for instance:

  1. Adapted from Klavans (1985: 106):
      [Kwak’wala]
    1. nəp’idi-da
      throw-DEIC
      gənanəm
      child
      [x̣a
      OBJ
      gukʷ]​​​N​
      house
      [sa
      OBL
      t’isəm]​​​N​
      rock

      ‘The child hit the house with a rock by throwing.’ (Levine 1980)

Since clitics should be treated as syntactically coherent with their hosts, it shouldn’t be possible, then, to interrupt sobisayan and ang with parenthetical word material in the same way it wasn’t possible in (6). This assumption proves false, however, since Ayeri does not take an issue with placing things other than a case marker after the verb cluster and before the first NP, while the interruption of ang and Niva, on the other hand, is indeed ungrammatical:

    1. Sa
      sa
      PT
      sobisayan,
      sobisa-yan
      study-3PL.M
      naratang,
      nara-tang
      say-3PL.M.A
      ang
      ang
      A
      Niva
      Niva
      Niva
      nay
      nay
      and
      Mico
      Mico
      Mico
      narānye.
      narān-ye-Ø
      language-PL-TOP

      ‘Languages is what Niva and Mico, they say, study.’
    2. * 
      Sa
      sa
      PT
      sobisayan
      sobisa-yan
      study-3PL.M
      ang,
      ang
      A
      naratang,
      nara-tang
      say-3PL.M.A
      Niva
      Niva
      Niva
      nay
      nay
      and
      Mico
      Mico
      Mico
      narānye.
      narān-ye-Ø
      language-PL-TOP

It is furthermore possible for the case marker to begin an utterance, namely, in equative sentences like the one in (11a). In this case, there is no way for the case marker to lean on a host to its left, but only to its right. In these cases as well, it isn’t possible for parenthetical material to be placed between the case marker and its target of modification, as in (11b).

    1. Ang
      ang
      A
      Misan
      Misan
      Misan
      lajāyas
      lajāy-as
      student-P
      puti.
      puti
      zealous

      ‘Misan is a zealous student.’
    2. * 
      Ang,
      ang
      A
      paronyang,
      paron-yang
      believe-1SG.A
      Misan
      Misan
      Misan
      lajāyas
      lajāy-as
      student-P
      puti.
      puti
      zealous

Since the case marker and its modification target cohere that closely, we have to assume that here as well, the case marker is proclitic rather than enclitic. Unlike “typical” clitics, however, its attachment is not strictly speaking ‘promiscuous’, unlike in our first hypothesis, but always with a proper noun very specifically. This property puts it very close to an affix—just like the suffixed case markers. More typical of function words, though, there is no morphophonemic interaction between the flexive and the word it inflects, for example, there is no /saːdʒaːn/ from sa (P) + Ajān. This overlap in form between affix and function word is typical of clitics, according to the traits excerpted from Spencer and Luís (2012) above. Moreover, , as mentioned above, represents an exception to other case particles in that it takes secondary stress due to being a long syllable, which cannot be unstressed. Ayeri does not allow words to end in secondary-stressed syllables, but only allows secondary-stressed syllables to precede a stressed one, since word stress spreads backwards from the right edge of phonological words. If we don’t want to create an exception for alone, thus, it is more likely for the preposed case markers to lean to the right rather than to the left also on phonological grounds, like in (12).

  1. Sa sobisayan ang=Niva nay Mico narānye.

From this discussion of prenominal particles, let us return to verbs again for a moment. Besides the preverbal particles discussed above, there is also what is spelled as a prefix on the verb which appears to be a little odd as such in that it can have wide scope over conjoined verbs. This is the prefix da- often meaning ‘so, thus’, displayed in (13).

  1. Ang
    ang
    AT
    da-pinyaya
    da-pinya-ya
    so-ask-3SG.M
    nay
    nay
    and
    hisaya
    hisa-ya
    beg-3SG.M
     
    Ø
    TOP
    Yan
    Yan
    Yan
    sa
    sa
    P
    Pila.
    Pila
    Pila

    ‘Yan asks and begs Pila to (do so).’

Da-, where it is not used for presentative purposes,2 is a functional morpheme in that it basically acts as an anaphora for a complementizer phrase (CP) the speaker chooses to drop. Thus, it does not mark any of the intrinsic morphological categories of the verb (tense, aspect, mood, modality, finiteness), just like the topic marker refers to a syntactic relation the verb subcategorizes but none of its inflectional categories. As an anaphora, da- cannot stand alone, though it is possible to use a full demonstrative form danya ‘such one’ in its place:

  1. Ang
    ang
    AT
    pinyaya
    pinya-ya
    ask-3SG.M
    nay
    nay
    and
    hisaya
    hisa-ya
    beg-3SG.M
     
    Ø
    TOP
    Yan
    Yan
    Yan
    sa
    sa
    P
    Pila
    Pila
    Pila
    danyaley.
    danya-ley
    such.one-P.INAN

    ‘Yan asks and begs Pila such.’

Unlike the preverbal particles, da- can be associated with a full form, though it still displays special syntax in that unlike English -n’t or ‘ll, for instance, it does not occur in the same place as the full form. Note also how da- is appended to the right of tense prefixes, which express a property of the verb, as shown in (15).

    1. Ang
      ang
      AT
      da-məpinyaya
      da-mə-pinya-ya.Ø
      so-PST-ask-3SG.M.TOP
      sa
      sa
      P
      Pila.
      Pila
      Pila

      ‘He asked Pila to.’
    2. Ang
      ang
      AT
      da-məpinyaya
      da-mə-pinya-ya.Ø
      so-PST-ask-3SG.M
      nay
      nay
      and
      məhisaya
      mə-hisa-ya
      PST-beg-3SG.M
       
      Ø
      TOP
      Yan
      Yan
      Yan
      sa
      sa
      P
      Pila.
      Pila
      Pila

      ‘Yan asked and begged Pila to.’

The verb form in (15) becomes ungrammatical with the order of its prefixes reversed, so it is not acceptable to say: * məda-pinyaya, although note that pre- and suffixes proper also have a fixed order in Ayeri, so this alone is probably not enough evidence to claim that da- is not possibly a prefix. Furthermore, while the tense prefixes undergo crasis, this is not the case with da-:

    1. Da-amangreng.
      da-amang-reng
      thus-happen-3SG.INAN.A

      ‘It happens thus.’
    2. * Dāmangreng.
    1. Māmangreng.
      mə-amang-reng
      PST-happen-3SG.INAN.A

      ‘It happened.’
    2. * Məamangreng.

Besides the characteristic of not seeking out certain parts of speech, the da- prefix at least satisfies the criteria of being a phonologically reduced form of an otherwise free functional morpheme, and it occurs in a place where normal syntax would not put its corresponding full form. It has wide scope over conjuncts, is attached outside of inflection for proper categories of the verb, and doesn’t interact with its host with regards to morphophonemics. Besides these more typical traits of clitics, there is also no way to place words between da- and the verb stem:

  1. *
    Da,
    da
    thus
    naratang,
    nara-tang
    say-3PL.M.A
    amangreng.
    amang-reng
    happen-3SG.INAN.A

    Intended: ‘It happens, they say, thus.’

The prefix sitang- ‘self’ behaves in the same way as da-, since it also abbreviates a reflexive NP, for instance, sitang-yes ‘herself’ where ‘herself’ as a patient is coreferential with the agent of the clause. Reflexivity, however, is a category a verb in Ayeri could be said to inflect for that way, though on the other hand, Ayeri also does not have any verbs which are obligatorily reflexive to indicate anticausativity like in Romance languages:

    1. Adruara
      adru-ara
      break-3SG.INAN
      biratayreng.
      biratay-reng
      pot-A.INAN

      ‘The pot broke.’3
    2. * 
      Sitang-adruara
      sitang-adru-ara
      self-break-3SG.INAN
      biratayreng.
      biratay-reng
      pot-A.INAN

      Intended: ‘The pot broke.’ (an unspecified force broke it)
    1. [French]
    2. Le
      le
      the
      pot
      pot
      pot
      sest
      se=est
      self=be.3SG.PRES
      cassé.
      cassé
      broken

      ‘The pot broke.’ (an unspecified force broke it)
    3. Le
      le
      the
      pot
      pot
      pot
      est
      est
      be.3SG.PRES
      cassé.
      cassé
      broken

      ‘The pot is broken.’

Ayeri has a tendency to reuse prefixes with different parts of speech, and thus da- is also used with nouns, forming part of the series of deictic prefixes, da- ‘such (a)’, eda- ‘this’, ada- ‘that’. The prefix in all these cases represents a grammatical function, is unstressed, and may have wide scope over conjoined NPs, unless an individuating interpretation is intended, as in (21b). These traits are typical of clitics, as we have seen, though (22) shows that unlike with verbs, the deictic prefixes do undergo crasis here, which is a trait more typically associated with affixes.

    1. Sinyāng
      sinya-ang
      who-A
      eda-ledanas
      eda-ledan-as
      this-friend-P
      nay
      nay
      and
      viretāyās
      viretāya-as
      supporter-P
      tondayena-hen?
      tonday-ena-hen
      art-GEN-all

      ‘Who is this friend and supporter of all arts?’
    2. Sinyāng
      sinya-ang
      who-A
      eda-ledanas
      eda-ledan-as
      this-friend-P
      nay
      nay
      and
      eda-viretāyās
      eda-viretāya-as
      this-supporter-P
      tondayena-hen?
      tonday-ena-hen
      art-GEN-all

      ‘Who is/are this friend and this supporter of all arts?’
  1. Sa
    sa
    PT
    ming
    ming
    can
    nelnang
    nel-nang
    help-1PL.A
    edāyon.
    eda-ayon-Ø
    this-man-TOP

    ‘This man, we can help him.’

The deictic prefixes also cannot be used with all types of NPs, only with those headed by generic and proper nouns; the picky nature of the deictic prefixes also makes them more typical of affixes than of clitics. The preverbal particles, on the other hand, also only occur with verbs, and I’ve nonetheless argued for them being clitics above.

As mentioned initially, Spencer and Luís (2012) give numerous counterexamples to the catalog of traits typically associated with clitics. One of this counterexample is what they term ‘suspended affixation’, which occurs in Turkish, for instance, where the plural suffix -lEr and subsequent suffixes can be left out in coordination (23a), as well as case markers (23b), and adverbials with case-like functions (23c):

    1. [Turkish]

      bütün
      all
      kitap(…)
      book
      ve
      and
      defter-ler-imiz
      notebook-PL-1PL.POSS

      ‘all our books and notebooks’ (199)
    2. Vapur
      boat
      hem
      and
      Napoli(…)
      Naples
      hem
      and
      Venedik’-e
      Venice-LOC
      uğruyormuş
      stops.EVIDENTIAL

      ‘Apparently the boat stops at both Naples and Venice’ (ibid.)
    3. öğretmen-ler(…)
      teacher-PL
      ve
      and
      öğrenci-ler-le
      student-PL-WITH

      ‘with (the) students and (the) teachers’ (ibid.)

They note that, in “the nominal domain especially, wide scope inflection is widespread in the languages of Eurasia, becoming more prominent from west to east”, and that wide scope affixation “can be found with inflectional and derivational morphology in a number of languages, and it is often a symptom of recent and not quite complete morphologization” (200). They report Wälchli (2005) to find that this is especially the case with ‘natural coordination’, that is, the combination of items very frequently occurring in pairs like knife and fork or mother and father, as opposed to cases of occasional coordination (Spencer and Luís 2012: 200). Whether this is also true for Ayeri would require a separate survey.4

Given the evidence from Turkish, the categorization of deictic prefixes as either affixes or clitics is unclear, especially since the diagnostic of scope is devalued by the Turkish examples. On the other hand, suffixes on nouns don’t behave this way, as demonstrated in (24)—they rather behave like typical affixes in that they mandatorily occur on each conjunct. The question is, thus, whether an exception should be made for prefixes on nouns. We may as well assume that they are clitics.

    1. sobayajang
      sobaya-ye-ang
      teacher-PL-A
      nay
      nay
      and
      lajāyjang
      lajāy-ye-ang
      student-PL-A

      ‘(the) teachers and (the) students’
    2. * 
      sobayaye
      sobaya-ye
      teacher-PL
      nay
      nay
      and
      lajāyjang
      lajāy-ye-ang
      student-PL-A
    3. * 
      sobaya
      sobaya
      teacher
      nay
      nay
      and
      lajāyjang
      lajāy-ye-ang
      student-PL-A

From a functional point of view, the exact nature of the deictic prefixes shouldn’t matter either way—ParGram also cites a [DEIXIS] feature with PROXIMAL and DISTAL as its values, which fits eda- ‘this’ and ada- ‘that’ just fine. At present I’m at a loss about the feature representation of ‘such (a)’, however, since it is clearly deictic, but neither proximal nor distal. In this case it should be possible to just use [DEIXIS this/that/such] as well, I suppose, hence:

    1. edāyon
      eda-ayon
      this-man

      ‘this man’
    2. PRED ‘man’
      DEIXIS this

As described above, proper nouns are case marked by probably clitic case markers in front of the noun. In fact, these markers are somewhere at the left periphery of the NP, so the deictic prefixes stand in between the case marker and the proper noun itself, which is unproblematic for lexical integrity, since the deictic prefixes are not free morphemes. And even if they were part of inflection, the case markers, as clitics, would be on the outside—the order deictic prefix – case marker – noun is ungrammatical. An example is given in (26).

    1. Ang
      ang
      AT
      koronay
      koron-ay.Ø
      know-1SG.TOP
      sa
      sa
      P
      eda-​Kagan.
      eda-​Kagan
      this-Kagan

      ‘I know this Kagan.’
    2. * 
      Ang
      ang
      AT
      koronay
      koron-ay.Ø
      know-1SG.TOP
      eda-​Kaganas.
      eda-​Kagan-as
      this-Kagan-P
    3. * 
      Ang
      ang
      AT
      koronay
      koron-ay.Ø
      know-1SG.TOP
      eda-sa
      eda-sa
      this-P
      Kagan.
      Kagan
      Kagan

The question now is, what happens to coordinated proper nouns? Since the suffixed case markers on common nouns have the distributional properties of affixes, they occur on every conjunct, the deictic prefix, however, only occurs on the first unless an individuating reading is intended, as shown in (20). For proper nouns it ought to be possible for both a case marker and a deictic prefix to have scope over coordinated proper nouns (27a). Yet, however, my gut tells me that this is slightly odd-sounding (German/English bias?), so I would prefer the strategy in (27b), which avoids the problem altogether by making the names an adjunct to the demonstrative edanya ‘this/these one(s)’.5 The example in (27c) is unproblematic and here as well indicates that the two people are referred to individually and not as a group.

    1. ? 
      Ang
      ang
      AT
      koronay
      koron-ay.Ø
      know-1SG.TOP
      sa
      sa
      P
      eda-​Kagan
      eda-​Kagan
      this-​Kagan
      nay
      nay
      and
      Ijān.
      Ijān
      Ijān

      ‘I know these Kagan and Ijān.’
    2. Ang
      ang
      AT
      koronay
      koron-ay.Ø
      know-1SG.TOP
      edanyās,
      edanya-as
      this.one-P
      Kagan
      Kagan
      Kagan
      nay
      nay
      and
      Ijān.
      Ijān
      Ijān

      ‘I know these, Kagan and Ijān.’
    3. Ang
      ang
      AT
      koronay
      koron-ay.Ø
      know-1SG.TOP
      sa
      sa
      P
      eda-​Kagan
      eda-​Kagan
      this-​Kagan
      nay
      nay
      and
      eda-​Ijān.
      eda-​Ijān
      this-​Ijān

      ‘I know this Kagan and this Ijān.’

Of the deictic prefixes, da- is not only available to verbs and nouns, but also to adjectives. Like with verbs, it is short for danya ‘such one’ in this case (28a). The resulting meaning is ‘the adj. one’; da- essentially acts as a nominalizer, at least to the extent the compound of da- + adjective inherits the distributional properties of danya as a demonstrative pronoun. Thus, it can be case- and topic marked (28bc), and also be modified by another adjective (28c). On the other hand, it can’t be reduplicated for diminution, and also also can’t be pluralized. Since adjectives follow their heads, the original order of demonstrative – adjective remains intact.

    1. Le
      le
      PT.INAN
      noyang
      no-yang
      want-1SG.A
      danyaley
      danya-Ø
      such.one-TOP
      tuvo.
      tuvo
      red

      ‘The red one I want.’
    2. Ang
      ang
      AT
      noay
      no-ay.Ø
      want-1SG.TOP
      da-tuvoley.
      da-tuvo-ley
      one-red-P.INAN

      ‘I want the red one.’
    3. Le
      le
      PT.INAN
      noyang
      no-yang
      want-1SG.A
      da-tuvo
      da-tuvo-Ø
      one-red-TOP
      kivo.
      kivo
      small

      ‘The little red one I want.’

The prefix, again, coheres tightly in that no additional material can be inserted. Like with nouns above, inflecting each form in a group of coordinated adjectives results in an individuating reading (29a). It should be possible for the prefix to take wide scope (29b), though it seems better to me to instead rephrase the coordinated adjective as a relative clause (28c), for instance, besides using the full form danya + adjectives. Since case markers go on every conjunct, (29d) is not grammatical.

    1. Ang
      ang
      AT
      noay
      no-ay.Ø
      want-1SG.TOP
      da-tuvoley
      da-tuvo-ley
      one-red-P.INAN
      nay
      nay
      and
      da-lenoley.
      da-leno-ley
      one-blue-P.INAN

      ‘I want the blue one and the red one.’
    2. ? 
      Ang
      ang
      AT
      noay
      no-ay.Ø
      want-1SG.TOP
      da-tuvoley
      da-tuvo-ley
      one-red-P.INAN
      nay
      nay
      and
      lenoley.
      leno-ley
      blue-P.INAN

      ‘I want the red and blue one.’
    3. Ang
      ang
      AT
      noay
      no-ay.Ø
      want-1SG.TOP
      adaley
      ada-ley
      that-P.INAN
      si
      si
      REL
      tuvo
      tuvo
      red
      nay
      nay
      and
      leno.
      leno
      blue

      ‘I want that which is red and blue.’
    4. * 
      Ang
      ang
      AT
      noay
      no-ay.Ø
      want-1SG.TOP
      da-tuvo
      da-tuvo
      one-red
      nay
      nay
      and
      lenoley.
      leno-ley
      blue-P.INAN

Possessive pronouns like ‘my’, vana ‘your’, etc. behave the same way when derived to free-standing anaphora (da-nā ‘mine’, da-vana ‘yours’, etc.) from their usual role as modifiers, except they can’t themselves be modified by adjectives in the way da-tuvo ‘the red one’ is in (28c). Taking all of the examples above into account, da- with adjectives and possessive pronouns seems to be most like a simple clitic according to Zwicky’s (1977) definition, compared to the other contexts it can appear in:

Cases where a free morpheme, when unaccented, may be phonologically subordinated to a neighboring word. Cliticization of this sort is usually associated with stylistic conditions, as in the casual speech cliticization of object pronouns in English; there are both formal full pronouns and casual reduced pronouns. (Zwicky 1977: 5)

Typical of a simple clitic as well, the distribution of da- is restricted by grammatical context, as pointed out regarding example (27b). Unlike in English, which Zwicky (1977) gives examples of, the condition in Ayeri is likely not phonological in this case. The nature of the condition, however, is not predetermined in Spencer and Luís (2012), when they write that

we may therefore need to define simple clitics along the lines of Halpern (1998), namely, as clitics that may be positioned in a subset of the positions within which the full forms are found, rather than as clitics that have the same distribution as their full-form counterparts as in Zwicky (1977). Under this broader definition, we capture the fact that simple clitics differ from special clitics in that they can appear in some of the positions that are occupied by their corresponding full forms, while special clitics never can. (Spencer and Luís 2012: 44)

Besides deictic prefixes, nouns may also receive a prefix expressing likeness, ku-. This prefix is also applicable to adjectives, and is maybe more adverbial in terms of semantics than purely functional morphemes like da-. In contrast to da-, ku- has no full-form equivalent. Some examples of it leaning on nouns are given in (30). Like the deictic prefixes, ku- appears in a position which is privileged to dependent functional morphemes in that any modifiers which appear as free words or phrases (adjectives, relative clauses, nominal adjuncts) follow nouns. Slightly untypical of a clitic, it is not fully ‘promiscuous’ regarding its phonological host in that it requires a nominal, adjectival or phrasal host.

    1. Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ku-depangas.
      ku-depang-as
      like-fool-P

      ‘Amān acts like a fool.’
    2. Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ku-depangas
      ku-depang-as
      like-fool-P
      nay
      nay
      and
      karayās.
      karaya-as
      coward-P

      ‘Amān acts like a fool and a coward.’
    3. Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ku-depangas
      ku-depang-as
      like-fool-P
      nay
      nay
      and
      ku-karayās.
      ku-karaya-as
      like-coward-P

      ‘Amān acts like a fool and like a coward.’
    4. Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ku-ada-depangas.
      ku-ada-depang-as
      like-that-fool-P

      ‘Amān acts like that fool.’
    5. Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ku-ada-depangas
      ku-ada-depang-as
      like-that-fool-P
      nay
      nay
      and
      ada-karayās.
      ada-karayās
      that-coward-P

      ‘Amān acts like that fool and that coward.’
    6. * 
      Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ada-ku-depangas.
      ada-ku-depang-as
      that-like-fool-P

Generally, ku- fulfills the function of the preposition like in English in (30). However, if it were a preposition in Ayeri, it should trigger the locative on its dependent. In the examples above, however, the NP ku- modifies takes the patient case, like predicative NPs are otherwise wont to do. While prepositions like kong ‘inside’ in (31) are free morphemes in Ayeri, ku- is bound, which becomes apparent by introducing a parenthetical remark again:

    1. Ang
      ang
      AT
      yomāy,
      yoma-ay.Ø
      be-1SG.TOP
      surpareng,
      surpa-reng
      seem-3SG.A.INAN
      kong
      kong
      inside
      sayanya.
      sayan-ya
      cave-LOC

      ‘I am, it seems, inside a cave.’
    2. Ang
      ang
      AT
      yomāy
      yoma-ay.Ø
      be-1SG.TOP
      kong,
      kong
      inside
      suprareng,
      surpa-reng
      seem-3SG.A.INAN
      sayanya.
      sayan-ya
      cave-LOC

      ‘I am inside, it seems, a cave.’
    1. Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān,
      Amān
      Amān
      surpareng,
      surpa-reng
      seem-3SG.A.INAN
      ku-depangas.
      ku-depang-as
      like-fool-P

      ‘Amān acts, it seems, like a fool.’
    2. * 
      Ang
      ang
      AT
      misya
      mis-ya
      act-3SG.M
       
      Ø
      TOP
      Amān
      Amān
      Amān
      ku,
      ku
      like
      surpareng,
      surpa-reng
      seem-3SG.A.INAN
      depangas.
      depang-as
      fool-P

      ‘Amān acts like a, it seems, fool.’

Examples (30ab) show that similar to the deictic prefixes, ku- precedes its target of modification and can have wide scope with coordinated NPs. As (30c) shows, narrow scope is possible as well, and in this case, again, each conjunct is to be interpreted separately instead of ku- modifying both conjuncts collectively. If combined with ada- as a deictic prefix, for instance, ku- even precedes that (30d), and changing the order of the prefixes is not possible, as is shown in (30f). As (30e) shows, ku- may also have scope over two individuating noun phrase conjuncts. Ku- is also applicable to pronouns, so (33) is possible, for example.

    1. Ang
      ang
      AT
      silvye
      silv-ye
      look-3SG.F
       
      Ø
      TOP
      Pada
      Pada
      Pada
      ku-yes.
      ku-yes
      like-3SG.F.P

      ‘Pada looks like her.’
    2. Sa
      sa
      PT
      silvye
      silv-ye
      look-3SG.F
      ang
      ang
      A
      Pada
      Pada
      Pada
      ku-ye.
      ku-ye
      like-3SG.F.TOP

      ‘Like her Pada looks.’

With proper nouns, we get the same distributional properties as with generic nouns, except that ku- appears (rather idiosyncratically) as a suffix at the right edge of an NP—or at the right edge of the first NP conjunct—if the NP is preceded by a case marker, as shown in (34). Admittedly, I hadn’t considered the behavior of ku- with names before and felt adventurous in introducing this little twist. I hope it’s viable.

    1. Ang
      ang
      AT
      lentava
      lenta-va
      sound-2.TOP
      sa
      sa
      P
      Tagāti
      Tagāti
      Tagāti
      diyan-ku.
      diyan-ku
      worthy-like

      ‘You sound like Mr. Tagāti.’
    2. Ang
      ang
      AT
      lentava
      lenta-va
      sound-2.TOP
      sa
      sa
      P
      Tagāti
      Tagāti
      Tagāti
      diyan-ku
      diyan-ku
      worthy-like
      nay
      nay
      and
      diranas
      diran-as
      uncle-P
      yana.
      yana
      3SG.M.GEN

      ‘You sound like Mr. Tagāti and his uncle.’
    3. Sa
      sa
      PT
      lentavāng
      lenta-vāng
      sound-2.A
      ku-​Tagāti
      ku-​Tagāti
      like-​Tagāti
      diyan.
      diyan
      worthy

      ‘Like Mr. Tagāti you sound.’

With adjectives, however, there are no idiosyncrasies to this degree; ku- appears only as a prefix here, as with generic nouns:

    1. Surpya
      surp-ya
      seem-3SG.M
      ku-suta
      ku-suta
      like-busy
      ang
      ang
      A
      Maran.
      Maran
      Maran

      ‘Maran seems like he’s busy.’
    2. Surpya
      surp-ya
      seem-3SG.M
      ku-suta
      ku-suta
      like-busy
      nay
      nay
      and
      baras
      baras
      gruff
      ang
      ang
      A
      Maran.
      Maran
      Maran

      ‘Maran seems like he’s busy and gruff.’

As (35b) shows, ku- again can have wide scope over conjuncts. What further distinguishes ku- from a prefix here is that it doesn’t undergo crasis if the adjective begins with an /u/, hence ku-ubo /kuˈubo/ ‘like bitter’, not * kūbo /ˈkuːbo/.6 Again, the position ku- appears in is special in that whatever modifies adjectives usually trails after them.

Another case I hadn’t considered before is that ku- should be able to subordinate infinite CPs. Since ku- leans on a whole phrase in (35), which affixes can’t do, its status as a clitic should be unmistakable in this context.

  1. Silvyeng
    silv-yeng
    look-3SG.F.A
    ku-tahayam
    ku-taha-yam
    like-have-PTCP
    misungas.
    misung-as
    secret-P

    ‘She looks as though having a secret.’

2. Suffixes

I hope that I could shed some light on the prefixes and particles occuring before lexical heads so far, however, Ayeri also has a number of morphemes trailing lexical heads as suffixes which do not seem quite like typical inflection. These are, for one, person suffixes on the verb, which I already tried to come to terms with in a previous blog article. In this article, I had assumed a priori that the case- and topic-marked suffixes on the verb are clitics. However, what I didn’t do is to test whether they actually fulfill formal criteria of clitics. Especially tricky in this regard is maybe that “a pronominal affix or incorporated pronominal is effectively a clitic masquerading as an affix. Therefore, if there are pronominal affixes then they should behave exactly like clitics with respect to crucial aspects of morphosyntax” (Spencer and Luís 2012: 144). Spencer and Luís (2012) then proceed to give examples from Breton and Irish where the person marking on the verb is in complementary distribution with full NPs:

    1. [Breton]
    2. Bremañ
      now
      e
      PRT
      lennont
      read.PRES.3PL
      al
      the
      levrioù
      books

      ‘Now they are reading the books’ (145, from Borsley et al. 2007)
    3. Bremañ
      now
      e
      PRT
      lenn
      read.PRES.3SG
      ar
      the
      vugale
      children
      al
      the
      levrioù
      books

      ‘Now the children are reading the books’ (ibid.)
    4. * 
      Bremañ
      now
      e
      PRT
      lennont
      read.PRES.3PL
      ar
      the
      vugale
      children
      al
      the
      levrioù
      books

      (ibid.)
    1. [Irish]
    2. Chuirfinn
      put.COND.1SG
      (*mé)
      (​I​)
      isteach
      in
      ar
      on
      an
      the
      phost
      job
      sin
      that

      ‘I would apply for that job’ (145, from McCloskey and Hale 1984)
    3. Chuirfeadh
      put.COND.3SG
      sibh
      you
      isteach
      in
      ar
      on
      an
      the
      phost
      job
      sin
      that

      ‘You would apply for that job’ (ibid.)
    4. Chuirfeadh
      put.COND.3SG
      Eoghan
      Owen
      isteach
      in
      ar
      on
      an
      the
      phost
      job
      sin
      that

      ‘Owen would apply for that job’ (ibid.)

What we can see in (37) is that, according to Spencer and Luís (2012), the verb shows no number marking, defaulting to the singular form, in non-negative clauses if the subject of the verb is overt as either a full noun or a pronoun: plural marking on the verb and a full subject can’t coincide in this case, which is why (37c) is marked ungrammatical. In (38a) we can see that there is no need for an explicit first-person pronoun, since that function is already expressed by person marking on the verb—person inflection on the verb seems to be in complementary distribution with full subject pronouns at least for some parts of the paradigm. In (38b) we have an overt 2nd-person subject pronoun, but in this case, the verb does not agree with it and instead defaults to the 3rd-person form, a clear case of which is given in (38c).

While in Ayeri, there is no defaulting to a certain person in the presence of an overt subject NP as such, there is still the effect of complementary distribution between a pronominal suffix in the absence of an overt subject NP and a functionally impoverished as well as phonologically reduced form in its presence:

    1. Suta
      suta
      busy
      ang
      ang
      A
      Niyas.
      Niyas
      Niyas

      ‘Niyas is busy.’
    2. Yāng
      yāng
      3SG.M.A
      suta.
      suta
      busy

      ‘He is busy.’
    1. Lampya
      lamp-ya
      walk-3SG.M
      ang
      ang
      A
      Niyas.
      Niyas
      Niyas

      ‘Niyas is walking.’
    2. Lampyāng.
      lamp-yāng
      walk-3SG.M.A

      ‘He is walking’
    1. * 
      Lampyāng
      lamp-yang
      walk-3SG.M.A
      ang
      ang
      A
      Niyas.
      Niyas
      Niyas
    2. * 
      Lampya
      lamp-ya
      walk-3SG.M
      yāng.
      yāng
      3SG.M.A

Example (39b) shows the free form of the third singular masculine agent pronoun, yāng. This is in complementary distribution with a full NP, which in (39a) is ang Niyas. In (40a) we can see that the verb agrees with the subject NP in person, gender and number in that it exhibits the suffix -ya. If the overt subject NP is missing, the verb is marked with the same form as the free pronoun, -yāng, which feeds the verb as a syntactic argument. That is, the person suffix itself fills the SUBJ relation of the verb’s subcategorization frame. (42) lists the constituent parts of lampyāng and their associated grammatical features.

  1. lamp- Vstem (↑ PRED) = ‘walk ⟨(↑ SUBJ)⟩’
    -yāng Vinfl (↑ SUBJ PRED) = ‘pro’
    (↑ SUBJ PERS) = 3
    (↑ SUBJ NUM) = SG
    (↑ SUBJ GEND) = M
    (↑ SUBJ ANIM) = +
    (↑ SUBJ CASE) = A
  2. PRED ‘walk ⟨(↑ SUBJ)⟩’
    SUBJ
    PRED ‘pro’
    PERS 3
    NUM SG
    GEND M
    ANIM +
    CASE A

Since (43) shows that functionally, the inflection takes the role of the subject relation, (41a) is ungrammatical in that the pronominal suffix -yāng on the verb is redundant in the presence of a full NP which expresses the same features except that the subject NP’s [PRED] value is ‘Niyas’, not ‘pro’. Example (44) shows the annotations for lampya as agreeing with an overt NP. Here, the suffix does not have a [PRED] feature—it is not available for predication. The agreement suffix -ya defines that the subject NP has a certain set of person features. The NP which controls verb agreement (in canonical cases the agent NP) needs to match up with these features in order to establish an agreement relationship. I suppose that by constraining (=c) the subject’s predicator not to be a pro-form in (44) it should also be possible to rule out cases like in (41b), where person agreement is triggered by a pronominal NP, which is ungrammatical, since Ayeri basically supplants person agreement with a pronominal suffix in those cases. If -yāng were a simple inflectional affix, one of the two examples in (41) should be grammatical.

  1. -ya₁ Vinfl (↑ SUBJ PRED) =c ¬ ’pro’
    (↑ SUBJ PERS) = 3
    (↑ SUBJ NUM) = SG
    (↑ SUBJ GEND) = M
    (↑ SUBJ ANIM) = +

The behavior of the pronominal person marking on the verb is thus rather complex, and decidedly unlike inflection in that what looks like an affix on the verb is also an argument of it, like a pronoun, as displayed in (42). Another layer of complexity is added by the fact that such an incorporated pronoun is also eligible for topicalization. As we have seen above, topic marking on nouns is realized by suppressing the realization of the overt case marker, whether it is a proclitic or a suffix. The topic-marked forms of pronouns are also underspecified for case, and they happen to be the same as those of the person-agreement suffixes as exemplified by -ya in (40a). Thus, a topic-marked pronominal suffix on the verb will look exactly like ordinary agreement with a full NP, except that there is no full NP to agree with—hence the subscript numbers in (44) and (45) to distinguish between both kinds of -ya.7

  1. -ya₂ Vinfl (↑ SUBJ PRED) = ‘pro’
    (↑ SUBJ PERS) = 3
    (↑ SUBJ NUM) = SG
    (↑ SUBJ GEND) = M
    (↑ SUBJ ANIM) = +
    (↑ SUBJ CASE) = Ø ⇒ (↑ TOP) = ↓

Comparing the feature list in (45) with that in (42) and (44), we see that (45) is basically the same as (42), except that either the [CASE] feature is not set, or that the suffix is underspecified for case. In absence of an NP to agree with, it follows from this definitional lack that the person marking on the verb itself is to be identified as constituting the topic, and the correspondent of the preverbal topic marker. In the following case, the preverbal topic marker defines that the topic is an animate agent:

  1. ang Cl (↑ TOP CASE) = A
    (↑ TOP ANIM) = +

Instances of other case-unmarked nouns can be ruled out as being also part of the topic relation on the grounds of cohesion: if the topic is defined as an agent and it can’t be assumed from context that the case-unmarked noun in question is also part of the agent NP, discard it as a candidate. Besides, every core θ-role (agent, patient, recipient) can only be assigned once, so if the role specified by the topic marker is already assigned, another NP in the same clause can’t also be assigned it. This gets more difficult with non-core roles, though I assume that oblique arguments are less likely to be topicalized.

What has led me to confusion about the status of the pronominal suffixes in the past is essentially that “a pronominal affix or incorporated pronominal is effectively a clitic masquerading as an affix” (Spencer and Luís 2012: 144). While the pronominal suffixes in Ayeri behave in a special way regarding syntax, they lack wide scope, which is typical of affixes (apart from the examples from Turkish quoted earlier). Unlike Breton or Irish, Ayeri’s pronominal affixes do not default to some form, and verbs cannot be unmarked either, that is, verbs always have to be inflected in some way, mostly for phonotactic reasons. Thus, in coordination, every conjunct has to be inflected for person features:

    1. Nedrayāng
      nedra-yāng
      sit-3SG.M.A
      nay
      nay
      and
      layayāng.
      laya-yāng
      read-3SG.M.A

      ‘He is sitting and reading.’
    2. * 
      Nedrayāng
      nedra-yāng
      sit-3SG.M.A
      nay
      nay
      and
      laya.
      laya
      read
    3. * 
      Nedra
      nedra
      sit
      nay
      nay
      and
      layayāng.
      laya-yāng
      read-3SG.M.A

In the case of nedra- and laya- in (47), leaving off the person marking would theoretically work, since * nedra and * laya satisfy phonotactic constraints. However, Ayeri also has a great number of verb stems which end in a consonant cluster, such as anl- ‘bring’ or tapy- ‘set’, which don’t form valid words as bare stems. What would be possible instead is that one conjunct might carry the full pronominal suffix as a “strong” form and the other one might only partially co-index the required features by using the less specific corresponding agreement marker as a “weak” form. Differential marking of this kind, though, is simply not established.

After briefly delving into the realm of syntax, let’s return to morphology for the second group of suffixes which could use some clarification. While Ayeri has quantifiers which are independent words, there are also a number of very common ‘little’ quantifiers and degree adverbs which are customarily spelled as suffixes, for instance, -ikan ‘much, many; very’, -kay ‘few; a little’, -nama ‘almost’, -nyama ‘even’. All of these are adverbial in meaning and while they are comparatively light in their semantics compared to regular content words, I can’t say that they strike me as functional morphemes in particular. In the course of writing this article, an article which has come up in my searches a number of times is Bittner (1995) on quantification in West Greenlandic. According to her terminology, -ikan in Ayeri would be a D-quantifier, which “forms a constituent with, a projection of N” (Bittner 1995: 59), in contrast to A-quantifiers, which form “a constituent with some projection of V” (59). That is, A-quantifiers are adverbs like almost (-nama in Ayeri), mostly, or never, while D-quantifiers are words like most, some, or every. Ayeri makes no distinction between A- and D-quantifiers with regards to their being treated as suffixes, so one can find suffixed quantifiers in both groups. While my main interest here is in the morphosyntax of these quantifiers, Bittner’s is in their semantics, though the article gives evidence notwithstanding of a language in which suffixed quantifiers are attested, for example:

  1. [West Greenlandic]
    1. qaatuur-tuaanna-ngajap-p-a-a
      break-always-almost-IND-TR-3SG.3SG

      ‘he almost always breaks it’ (adapted from Bittner 1995: 60)
    2. qaqutigu-rujussuaq
      rarely-very

      ‘very rarely’ (63)
    1. ang
      ang
      AT
      adruya
      adru-ya.Ø
      break-3SG.M.TOP
      tadayen-ngas
      tadayen-ngas
      always-almost
      adaley
      ada-ley
      that-P.INAN

      ‘he almost always breaks it’
    2. kora-ikan
      rarely-very

      ‘very rarely’

As we can see in (48a), West Greenlandic incorporates the quantifier suffixes into the verb, while Ayeri—not a polysynthetic language—proceeds more freely in (49a) in that tadayen ‘always, every time’ is an adverb and as such a free morpheme which is, however, in turn modified by a suffixed quantifier. Since orthography may be treacherous, let’s first try to establish whether -ngas and -ikan and their like are free morphemes or not. As discussed initially regarding the preverbal particles, it is possible to reorder free morphemes, while clitics, as bound morphemes, can’t move around. Adverbs and adjectives are, if they optionally add additional information to a lexical head, adjuncts, and according to Carnie (2013) it is possible for adjuncts to switch places within the same syntactic domain. Adjuncts can also be coordinated with other adjuncts in the same syntactic domain. Furthermore, it is possible to replace X’ nodes with pro-forms, like one in English.

    1. kipisānye-ikan
      kipisān-ye-ikan
      painting-PL-many
      bino
      bino
      colorful
      kāryo
      kāryo
      big

      ‘many big colorful paintings’
    2. kipisānye-ikan kāryo bino
      ‘many big colorful paintings’
    3. ! kipisānye bino-ikan kāryo
      ‘very colorful big paintings’
    4. ! kipisānye bino kāryo-ikan
      ‘very big colorful paintings’
    1. kipisānye-ikan
      kipisān-ye-ikan
      painting-PL-many
      bino
      bino
      colorful
      nay
      nay
      and
      kāryo
      kāryo
      big

      ‘many big and colorful paintings’
    2. * kipisānye-ikan nay bino kāryo
      ‘many and colorful big paintings’
    3. ! kipisānye bino-ikan nay kāryo
      ‘big and very colorful paintings’

As (50cd) shows, moving -ikan ‘many, much, very’ into different positions results not necessarily in ungrammatical expressions, but in ones with meanings different from what was intended, since -ikan‘s scope changes from the noun to the adjective it is appended to. On the other hand, comparing (50a) and (b), it is possible for kāryo ‘big’ and bino ‘colorful’ to switch places with no adversary effects. Example (51b) demonstrates that placing a coordinating conjunction between -ikan and bino ‘colorful’ doesn’t work. The coordination in (51c), on the other hand, is not a problem—not because it is possible to coordinate -ikan and kāryo, but because bino-ikan ‘very colorful’ is considered one syntactic unit which is coordinated with kāryo. Thus, in (50b), we have actually been trying to coordinate kipisānye-ikan ‘many paintings’ with bino ‘colorful’, which does not work, since it is not possible to coordinate a lexical head with an adjunct supposed to modify it, since they are of different syntactic categories. In this regard it is worth mentioning that Ayeri’s quantifier suffixes are rather not complements either, since they are not required in order to satisfy their head’s argument structure.

One might argue that in (50) and (51) we tried to compare apples to oranges in that -ikan and bino are of different categories, since -ikan and bino don’t appear to operate on the same levels. So instead, let’s look at possibilities of word order change and coordination between different quantifiers to ensure that we actually stay on the same level. With this there is the problem, however, that it seems strange to modify the same lexical head with multiple different quantifiers, so this test does not really seem feasible to produce grammatical results. Also, with regards to coordination of quantifiers, it is maybe more natural to oppose them with soyang ‘or’ than to coordinate them; the grammatical structure of two categorially identical elements connected by a grammatical conjunction (even if the meaning is disjunctive) remains the same in either case.

    1. * 
      keynam-ikan-kay
      keynam-ikan-kay
      people-many-few

      ‘few many people’
    2. ? 
      keynam-ikan
      keynam-ikan
      people-many
      soyang
      soyang
      or
      -kay
      -kay
      few

      ‘few or many people’

In example (52a) we see that it is indeed not possible to combine multiple quantifiers to jointly modify a head in the way it is possible for multiple adjectives to modify the same head as in (50a), for instance. The example of quantifier disjunction in (52b) is also odd unless we permit a reading where keynam ‘people’ has been suppressed in the second disjunct to avoid repetition, although in the corresponding case of (53b) below, da-kay would be preferable.

    1. ? keynam[-ikan soyang kay]
      ‘[few or many] people’
    2. [keynami-ikan] soyang [_i-kay]
      ‘[few _i] or [many peoplei]’

Both tests, moving -ikan into other positions and coordination, have failed so far, and we have evidence that -ikan forms a syntactic unit with its head, which points to it being a bound morpheme similar to an affix in spite of its adverb-like meaning. As with free words, it is also possible to replace a quantifier’s head with a pro-form, as mentioned above in the comment on (53b), and shown in more detail in (54). With quantifier suffixes there seems to be an overlap between word-like and affix-like properties, which is typical of clitics.

    1. Ang
      ang
      AT
      vacyan
      vac-yan
      like-3PL.M
      keynam-ikan
      keynam-Ø-ikan
      people-TOP-many
      seygoley.
      seygo-ley.
      apple-P.INAN

      ‘Many people like apples.’
    2. Ang
      ang
      AT
      vacyan
      vac-yan
      like-3PL.M
      danya-ikan
      danya-Ø-ikan
      such.one-TOP-many
      seygoley.
      seygo-ley.
      apple-P.INAN

      ‘Many of them like apples.’
    3. Ang
      ang
      AT
      vacyan
      vac-yan
      like-3PL.M
      da-ikan
      da-ikan-Ø
      one-many-TOP
      seygoley.
      seygo-ley.
      apple-P.INAN

      ‘Many (of them) like apples.’

Somewhat untypical of affixes, it seems to be possible to modify suffixed quantifier with adverbs like ekeng ‘too’ and kagan ‘far too’, as (55) shows. This suggests that at least in this context, -ikan may actually be the lexical head of an adverbial phrase, which is at odds with its status as a bound morpheme.

  1. Ang
    ang
    AT
    vacyan
    vac-yan
    like-3PL.M
    keynam-ikan
    keynam-Ø-ikan
    people-TOP-many
    kagan
    kagan
    far.too
    disuley.
    disu-ley.
    disu-P.INAN

    ‘Far too many people like bananas.’

I previously tried to insert parenthetical word material in between morphemes, and this test may be especially interesting in face of (55), since here it is not entirely clear whether keynam-ikan kagan ‘too many people’ forms a single unit. Since signs point to the status of quantifier suffixes as clitics, chances are good that it does, in fact, constitute a clitic cluster similar to the preverbal one. Example (56), therefore, lists examples which try to split up the expression at every relevant point. According to this test, it looks indeed as though keynam-ikan kagan forms a syntactic unit, in that -ikan kagan ‘too many’ cannot be split up internally and also cannot be divided from -ikan‘s head, keynam ‘people’.

    1. Ang
      ang
      AT
      vacyan,
      vac-yan
      like-3PL.M
      ​narayang​​,
      nara-yang
      say-1SG.A
      keynam-ikan
      keynam-Ø-ikan
      people-TOP-many
      kagan
      kagan
      far.too
      disuley.
      disu-ley
      disu-P.INAN

      ‘Far too many people, I say, like bananas.’
    2. * Ang vacyan keynam, narayang, ikan kagan disuley.
    3. * Ang vacyan keynam-ikan, narayang, kagan disuley.
    4. Ang vacyan keynam-ikan kagan, narayang, disuley.

Another interesting distributional property of suffixed quantifiers in Ayeri is that in spite of their being suffixed, to verbs for instance, they can form arguments of the verb, similar to pronominal suffixes. Thus, with verbs like kond- ‘eat’, -ma ‘enough’ appears suffixed to the verb instead of as a predicative adverb. Incidentally, the examples in (56) also show that a quantifier attaches after pronominal suffixes, which we have already established as being clitics. An inflectional affix would not normally appear in post-clitic position, which is further evidence to the hypothesis that quantifier suffixes in Ayeri are clitics.

    1. Kondanang-ma.
      kond-nang-ma
      eat-1PL.A-enough

      ‘We ate enough.’
    2. Ang
      ang
      AT
      tangay-ikan
      tang-ay.Ø-ikan
      hear-1SG.TOP-much
      vana.
      vana
      2.GEN

      ‘I’ve heard much about you.’

Since Ayeri possesses a zero copula, equative phrases which treat quantifier suffixes as predicative adverbs pose a difficulty in that quantifier suffixes cannot stand alone like predicatives normally would. Thus, in a similar fashion to -ma‘s behavior in (57a), the predicative -ma in (58b) cliticizes to the only available word: the subject, adareng ‘that’.

    1. Adareng
      ada-reng
      that-A.INAN
      edaya.
      edaya
      here

      ‘It is here.’
    2. Adareng-ma.
      ada-reng-ma
      that-A.INAN-enough

      ‘That/It is enough.’

If quantifier suffixes are clitics, they should also have wide scope over conjuncts. Here as well, quantifier suffixes behave typically of clitics, though, in that they can have scope over a conjunct as a whole, although not totally unambiguously so.

    1. Ang
      ang
      AT
      tahisayan
      tahisa-yan.Ø
      own-3PL.M.TOP
      koyās
      koya-as
      book-P
      nay
      nay
      and
      kihasley-ikan.
      kihas-ley-ikan
      map-P.INAN-many

      ‘They own many books and maps.’
    2. Yeng
      yeng
      3SG.F.A
      alingo
      alingo
      clever
      nay
      nay
      and
      para-ven.
      para-ven
      quick-pretty

      ‘She’s pretty clever and quick.’

Thus, in (59a), while koyās nay kihasley-ikan is translated as ‘many books and maps’ (nouns do not mark plural if modified by a quantifier which indicates plurality), another possible reading is ‘a book and many maps’. Ways to force the latter reading explicitly are, for one, to use koyās men ‘one/a single book’, or alternatively, to reduplicate the coordinator nay ‘and’ to naynay ‘and also’. Context should be sufficient to indicate the correct reading of (59a) under normal circustances, however. The same applies to (59b), where the non-distributive reading can be made explicit by using naynay instead of simple nay. In both (59a) and (b), if the first conjunct is modified by an adjective, the distribution of the quantifier over both conjuncts is also blocked. Thus, in (60a), there is ‘a big book and many maps’, and in (60b) ‘she’ is ‘surprisingly clever and pretty quick’.

    1. Ang
      ang
      AT
      tahisayan
      tahisa-yan.Ø
      own-3PL.M.TOP
      koyās
      koya-as
      book-P
      kāryo
      kāryo
      big
      nay
      nay
      and
      kihasley-ikan.
      kihas-ley-ikan
      map-P.INAN-many

      ‘They own a big book and many maps.’
      Not: ‘They own many big books and maps.’
    2. Yeng
      yeng
      3SG.F.A
      alingo
      alingo
      clever
      patu
      patu
      surprisingly
      nay
      nay
      and
      para-ven.
      para-ven
      quick-pretty

      ‘She’s surprisingly clever and pretty quick.’
      Not: ‘She is surprisingly pretty clever and quick.’

The false interpretations in (60) can be correctly achieved by ordinarily placing the adjective after the compounds so that the adjective itself has scope over both conjuncts. This is demonstrated in (61) and (62). Again, an unambiguous and individuating interpretation can be achieved by placing the quantifier suffix on each conjunct.

    1. Ang
      ang
      AT
      tahisayan
      tahisa-yan.Ø
      own-3PL.M.TOP
      koyajas
      koya-ye-as
      book-PL-P
      nay
      nay
      and
      kihasyeley
      kihas-ye-ley
      map-PL-P.INAN
      kāryo.
      kāryo
      big

      ‘They own big books and maps.’
    2. Ang
      ang
      AT
      tahisayan
      tahisa-yan.Ø
      own-3PL.M.TOP
      koyās
      koya-as
      book-PL-P
      nay
      nay
      and
      kihasley-ikan
      kihas-ley-ikan
      map-P.INAN-many
      kāryo.
      kāryo
      big

      ‘They own many big books and maps.’
    1. Yeng
      yeng
      3SG.F.A
      alingo
      alingo
      clever
      nay
      nay
      and
      para
      para
      quick
      patu.
      patu
      surprisingly

      ‘She’s surprisingly clever and quick.’
    2. Yeng
      yeng
      3SG.F.A
      alingo
      alingo
      clever
      nay
      nay
      and
      para-ven
      para-ven
      quick-pretty
      patu.
      patu
      surprisingly

      ‘She’s surprisingly pretty clever and quick.’

3. Conclusive remarks

I have applied various tests, mostly of morphosyntactic nature, to particles occurring in front of verbs, nouns, and adjectives, as well as to suffixes indicating person features on verbs, and to quantifier suffixes on a variety of parts of speech in order test whether these particles, prefixes, and suffixes behave like clitics in spite of appearing like function words, adverbs, prefixes, or suffixes. By testing morphosyntactic properties of the various morphemes evidence was presented to argue that (1) the preverbal particles form a clitic cluster together with the verb; (2) da- and sitang- on verbs are likely clitics; (3) da- on nouns, adjectives and possessive pronouns are clitics; (4) the preposed case markers of names are likely clitics; (5) deictic prefixes on the noun are likely clitics; (6) pronominal suffixes on verbs are clitics for their special syntax and in spite of the suffixes’ narrow scope, which is at least in part due to phonotactic requirements; and (7), suffixed quantifiers are clitics in spite of their adverbial meaning.

  • Bittner, Maria. “Quantification in Eskimo: A Challenge for Compositional Semantics.” Quantification in Natural Languages. Ed. by Emmon Bach et al. Dordrecht: Springer, 1995. 59–80. Print. Studies in Linguistics and Philosophy 54. DOI: 10.1007/978-94-017-2817-1_4.
  • Borsley, Robert D., Maggie Tallerman, and David Willis. The Syntax of Welsh. Cambridge: Cambridge UP, 2007. Print.
  • Bresnan, Joan et al. Lexical-Functional Syntax. 2nd ed. Chichester: Wiley Blackwell, 2016. Print. Blackwell Textbooks in Linguistics 16.
  • Butt, Miriam and Tracy Holloway King. “Lexical-Functional Grammar.” Syntax—Theory and Analysis: An International Handbook. Vol. 2. Ed. by Tibor Kiss and Artemis Alexiadou. Berlin: De Gruyter Mouton, 2015. 839–74. Handbücher zur Sprach- und Kommunikationswissenschaft 42. Web. 10 Apr. 2015. DOI: 10.1515/9783110363708-002.
  • Carnie, Andrew. Syntax: A Generative Introduction. 3rd ed. Chichester: Wiley Blackwell, 2013. Print. Introducing Linguistics 4.
  • Halpern, Aaron. “Clitics.” The Handbook of Morphology. Ed. by Andrew Spencer and Arnold M. Zwicky. Oxford: Blackwell, 1998. 101–22. Print.
  • Hammarström, Harald et al. “Language: Kwak’wala.” Glottolog. Version 3.0. Max Planck Institute for the Science of Human History, n.d. Web. 10 Apr. 2017. ‹http://glottolog.org/resource/languoid/id/kwak1269›.
  • Klavans, Judith L. “The Independence of Syntax and Phonology in Cliticization.” Language 61.1 (1985): 95–120. Web. 20 Jul. 2016. ‹http://www.jstor.org/stable/413422›.
  • Levine, Robert D. “On the Lexical Origin of the Kwakwala passive.” International Journal of Linguistics 46.4 (1980): 240–59. Web. 10 Apr. 2017. ‹http://www.jstor.org/stable/1264707›.
  • McCloskey, James and Ken Hale. “On the Syntax of Person-Number Inflection in Modern Irish.” Natural Language and Linguistic Theory 1.4 (1984). 487–534. Print. DOI: 10.1007/BF00417057.
  • [Smedt,] Koenraad [de]. “Feature Table.” ParGram/ParSem: An International Collaboration on LFG-Based Grammar and Semantics Development. 2017. U of Bergen, 1 Feb. 2011. Web. 8 Apr. 2017. ‹https://pargram.b.uib.no›.
  • Spencer, Andrew and Ana R. Luís. Clitics: An Introduction. Cambridge: Cambridge UP, 2012. Print. Cambridge Textbooks in Linguistics.
  • Wälchli, Bernhard. Co-compounds and Natural Coordination. Oxford: Oxford UP, 2005. Print.
  • Zwicky, Arnold M. On Clitics. Bloomington: Indiana U Linguistics Club, 1977. Web. 22 Jul. 2016. ‹https://web.stanford.edu/~zwicky/on_clitics.pdf›.
  • ——— and Geoffrey K. Pullum. “Cliticization vs. Inflection: English N’T.” Language 59.3 (1983): 502–13. Web. 21 Jul. 2016. ‹http://www.jstor.org/stable/413900›.
  1. This list also subsumes the criteria proposed in Zwicky and Pullum (1983), though the page numbers given here refer to Spencer and Luís (2012).
  2. Although this use is probably related to the anaphoric use.
  3. Actually, I’m slightly tempted to make the S a patient here to reflect that the pot does not cause itself to break but rather suffers breaking.
  4. Or, since it’s a fictional language and I’ve never thought about this before, making up supplemental rules.
  5. Honestly, it’s these cases where you wish you could just ask a speaker of your fictional language for their judgement instead of relying on your own intuition, which will most certainly be tainted by interference from your native language. While I am the one to make up the rules for Ayeri, I try to be wary of carrying all too familiar patterns over into my creation accidentally.
  6. The phonology chapter of the new grammar has so far read in a footnote that this would be permissible, though given ku-‘s behavior elsewhere it doesn’t make sense to keep this rule, in retrospect.
  7. Assuming that the person suffix on the verb always co-indexes the topic and that one therefore doesn’t need to distinguish a person-agreement suffix from a homophonous topicalized pronominal suffix is a premature conclusion. In fact, the agreement suffix always agrees with the subject of the verb, which is most often the agent NP but may, in absence of an agent NP, also be the patient NP. The topic may consist of any NP, also oblique ones, which the verb then does not agree in person with.

Last month I asked whether I should get a top level domain for this site here, since benung.nfshost.com has long seemed a bit unwieldy to me. Most people of the dozen who gave feedback spoke out of in favor of it. And there I went, after six years and some consideration, and registered ayeri.de. I hope that this is a lot more succinct now.

In spite of the German country code, contents here will continue to be mainly in English, since the conlanging community as I know it is international. Old links will continue to work, as I set the server to rewrite them; benung.nfshost.com also remains the canonical domain for DNS purposes (hat tip to kaleissin).

Just a quick status update on the Grammar: I’m currently trying to come up with example sentences in order to figure out the status of various elements regarding their classification as an affix or a clitic. Not everything I’ve been calling ‘clitic’ so far is one, most probably, since my understanding had been faulty. Especially the declined person markers (-yang, -vāng, -yāng etc.) seem to behave more like affixes, in fact.

Should I stay or should I go?

I was recently wondering again whether I should move my Ayeri stuff to my own top level domain from its current residence at nfshost.com. I haven’t planned anything concrete, I’m just curious for your opinion. There’s a poll embedded in the following tweet.

The thing is, Ayeri’s been living at its current address since July 2008, and thus, things from this site are linked in quite a few other places. In this regard, I feel somewhat uneasy breaking continuity. On the other hand, people usually come in from a handful of sources only, besides search engines, and I can always put a redirect up here.

Update on the Grammar Writing Process II

A problem I have recently come to see with conlanging is that while a whole number of people may research a natural language at any time, each researcher contributing to scholarly discourse from their area of expertise, your typical conlanger is working on their fictional language all by themselves. I’m no exception with regards to this. This also means, however, that only you are acquainted with your conlang, which also means that while fleshing it out, you have to be a kind of jack-of-all-trades if you want to do it well. On the other hand, a single person does not have talent for or interest in all areas of a field to the same degree, nor can you know everything about a field as variegated as linguistics. In addition to this, acquiring some deeper knowledge and experience just in a part of a field takes time.

While writing my new Ayeri grammar, describing phonology at least roughly, and morphology with a little more attention to detail seemed fair enough.1 Describing a language, however, doesn’t end at elaborating on how to form words. Syntax is just as important, as it describes how to form larger units of meaning, which is certainly no trivial issue either. Since Ayeri’s structure departs from English in some basic ways, it definitely warrants more serious attention.

Most conlangers I know seem to be mainly interested in morphology, and may even go so far as meeting formal syntax theories with suspicion. Moreover, I have never had a proper introduction to syntax myself either, for instance, in class at university. However, since Ayeri is rather different from German or English, I have long had an itch to figure its syntax out in a more structured way, in order to find out and describe in standard terms what I have been doing so far without giving it too much of a second thought. Since I’ve been trying to keep up a certain level of seriousness in the grammar, simply stating that Ayeri is VSO and heads mostly go first, and treating everything within 5 pages won’t do. Dealing with such a complex topic this superficially does not seem satisfying to my own curiosity and ambition. I am hoping that finding out more about Ayeri’s syntax would uncover more remaining blank spots, the filling of which would allow me to add yet more depth.

A colleague of mine had suggested to get acquainted with Lexical-Functional Grammar, actually with regards to my day job as a grad assistant. Describing Ayeri in this framework, however, might be interesting as well, since LFG was developed with flexibility in mind so that configurational, non-configurational, and mixed languages can all be dealt with in a straightforward manner. With its VSO constituent order, Ayeri may fall somewhere in the middle of this spectrum, though this needs further analysis, which I can’t provide just yet. I have been trying to work through Bresnan et al. (2016), but I realized that trying to study these things on my own is no adequate replacement for correction by teachers, since it’s too easy to accidentally gloss over important details by reading a textbook without discussing its contents. Furthermore, this book presupposes familiarity with common structuralist paradigms, such as Generative Grammar (Carnie 2002/2013 seems to be a popular introduction), Government and Binding, and X-bar theory, which it seems reasonable to acquaint myself with before I continue.

Yet, I am impatient to keep on writing, since I really don’t want to let the grammar drift off into negligence again this time. I had written some 20 pages on syntax earlier this month, however I realized that much of what I had written is probably wrong, since, for example, I disregarded lexical integrity as a fundamental principle with regards to what I assume to be clitics, simply for the reason of not being aware of this principle for the lack of formal training in a very formal discipline. For the time being, I have deleted what I wrote about the phrase structures of DPs/NPs and AdjP/AdvPs from the PDF in the main development branch on Github (‘master’) to not spread misinformation. Once I know more and have reevaluated some things, development on this part will go on in the ‘trunk’ branch, which I will merge back into ‘master’ once I am confident enough that my analyses are at least not completely off.

Thus, for the time being, the grammar will have to pause at morphology, and hopefully not for another 5 years. Alternatively, I may need to find a way to adequatly describe how to form clauses and sentences without getting too deeply into theories, at least provisionally, if that is possible.

  • Bresnan, Joan et al. Lexical-Functional Syntax. 2nd ed. Chichester: Wiley-Blackwell, 2016. Print. Blackwell Textbooks in Linguistics 16.
  • Carnie, Andrew. Syntax. A Generative Introduction. 3rd ed. Chichester: Wiley-Blackwell, 2013. Introducing Linguistics 4.
  • Spencer, Andrew and Ana R. Luís. Clitics. An Introduction. Cambridge: Cambridge UP, 2012. Print. Cambridge Textbooks in Linguistics.
  • Zwicky, Arnold M. On Clitics. 1977. Arnold M. Zwicky. 21 Apr. 2015. Stanford U. 21 Apr. 2015. Web. 22 Jul. 2016. ‹https://web.stanford.edu/~zwicky/on_clitics.pdf›.
  1. I will still have to rewrite some things with regards to cliticization, though. For instance, I am not quite sure whether manga with verbs is inflection or rather a special clitic; the term ‘bound word’ from Zwicky (1977) I used in the grammar hasn’t stood the test of time. I’m currently reading up on more recent research and positions on clitics in Spencer & Luís (2012), so corrections to the morphology chapter will follow eventually.

Subordinating Verbs: A Small Blast from the Past

I was recently thinking about this with regards to writing my New and Improved (tee-em) grammar of Ayeri and my previous post on subordinating verbs. I saw subordinating verbs as posing the problem of putting too much stuff in the constituent that holds the verb. As a solution, I described moving the complement of the main verb into a finite complement clause if it’s more than intransitive. However, when I did some analysis of verbs yesterday to maybe shed some light on the alternation between -isa and -isu in deverbal adjectives, I came across the following example sentence in the entry for pinya ‘ask’, entered October 24, 2008:

  1. Sa
    Sa
    PT
    pinyayāng
    pinya=yāng
    ask=3SG.M.A
    ye
    ye
    3SG.F.TOP
    rimayam
    rima-yam
    close-PTCP
    silvenoley.
    silveno-ley
    window-P.INAN

    ‘Her he asks to close the window.’

Material from 2008 is not quite fresh anymore, but going through my example texts, I also found the following sentence fragment in the 2010/11 Conlang Holiday Card Exchange (interlinear glossing updated to current standards):

  1. nārya
    nārya
    but
    le
    le
    PT.INAN
    tavisayang
    tavisa=yang
    receive=1S.A
    takan
    takan-Ø
    chance-TOP
    incam
    int-yam
    buy-PTCP
    dagangyeley
    dangang-ye-ley
    card-PL-P.INAN

    ‘but I got the chance to buy cards’

In both cases, the subordinating verb is transitive: (1) ‘he asks her’, (2) ‘I got the chance’; pinya- ‘ask’ in (1) is a raising an object-control verb (the logical subject of the subordinate verb is the object of the verb in the matrix clause), while int- ‘buy’ in (2) should simply be an infinite clausal complement. However, in both cases we do neither get the complement awkwardly placed in the middle, nor are the sentences rephrased so as to result in a finite complement clause or a nominalized complement to avoid the infinite verb form:

    1. ??
      Sa
      Sa
      PT
      pinyayāng
      pinya=yāng
      ask=3SG.M.A
      rimayam
      rima-yam
      close-PTCP
      silvenoley
      silveno-ley
      window-P.INAN
      ye.
      ye
      3SG.F.TOP

      ‘Her he asks to close the window.’
    2. Pinyayāng,
      pinya=yāng,
      ask=3SG.M.A,
      ang
      ang
      AT
      rimaye
      rima=ye.Ø
      close=3SG.F.TOP
      silvenoley.
      silveno-ley
      window-P.INAN

      ‘He asks that she closes the window.’

    1. nārya
      nārya
      but
      le
      le
      PT.INAN
      tavisayang
      tavisa=yang
      receive=1S.A
      takan
      takan-Ø
      chance-TOP
      intanena
      intan-ena
      purchase-GEN
      dagangyena
      dangang-ye-na
      card-PL-GEN

      ‘but I got the chance of a purchase of cards’
    2. nārya
      nārya
      but
      le
      le
      PT.INAN
      tavisayang
      tavisa=yang
      receive=1S.A
      takan,
      takan-Ø,
      chance-TOP,
      ang
      ang
      AT
      incay
      int=ay.Ø
      buy=1SG.TOP
      dagangyeley
      dangang-ye-ley
      card-PL-P.INAN

      ‘but I got the chance that I buy cards’

Both constructions, (1) and (2) are not widely attested in my materials, and the new grammar doc as it currently is does not rule out cases like (2), insofar I only need to make up my mind about constructions like in (1): continue allowing them as a variant, declare them ungrammatical, or simply ignore them? In the first case I might be required to keep a VP or a functional equivalent of it, after all, since there would be a post-subject position associated with verbs, then. In any case, raising and control should be interesting topics to come to terms with in my conlang.

Unclogging the Sink: Avoiding Center-Embedding with Subordinating Verbs

While working on the chapter on verb morphology in the revised Ayeri grammar I’ve been working on all the way since July, I had to stop the other day and consider how to deal with participles. With regards to Ayeri, these are the infinite subordinated verbs which appear together with verbs that may take another verb – or even a whole phrase headed by such an infinite verb – as a complement:

    1. Cunyo
      Cun-yo
      begin-3SG.N
      makayam
      maka-yam
      shine-PTCP
      perinang.
      perin-ang
      sun-A

      ‘The sun began to shine.’
    2. Manangyeng
      Manang=yeng
      avoid=3SG.F.A
      pengalyam
      pengal-yam
      meet-PTCP
      badanas
      badan-as
      father-P
      saha
      saha
      in.law
      yena.
      yena
      3SG.F.GEN

      ‘She avoids to meet her father-in-law.’

Example (1a) shows two intransitive verbs being combined, cun- ‘begin’ (subordinating) and maka- ‘shine’ (complement); example (1b) illustrates an intransitive verb which takes a transitive verb as a complement, manang- ‘avoid’ (subordinating) and pengal- ‘meet’ (complement). We can also observe word order differences in that the agent NP in (1a) follows the initial verb complex while the agent NP is an enclitic pronoun directly following its governing verb in (1b), which is not surprising, since agent pronouns take verbs as clitic hosts. That is, the agent pronoun -yeng ‘she’ (3SG.F.A) bundles together with its governing verb manang- ‘avoid’ rather than cliticizing to the whole verb phrase,1 i.e. appending -yeng ‘she’ (3SG.F.A) to pengalyam ‘meeting’. As far as describing the morphology of participles goes, we are basically done here, and this is also where I stopped thinking previously.

However, if we actually continue from here and take a morphosyntactic point of view, an interesting question arises: what happens to the constituents’ linearization if we (a) use full NPs instead of cliticized agent pronouns, and (b) not only combine an intransitive verb with a transitive one, but try all possible combinations–intransitive, transitive, ditransitive, and any of these with adverbial adjuncts? I think that for the sake of sketching out my thoughts here, it won’t be necessessary to give examples of all 18 possible combinations, since the more complex cases should all be treated alike anyway.

Assuming that she in (1b) is Maha, who just doesn’t get along with her husband’s father, where does the agent NP go? As noted above, the position of the agent clitic is somewhat special, so I take it that (1a) should be the basic word order. Thus, for (1b), we get:

  1. Manangye
    Manang-ye
    avoid-3SG.F
    pengalyam
    pengal-yam
    meet-PTCP
    badanas
    badan-as
    father-P
    saha
    saha
    in.law
    yena
    yena
    3SG.F.GEN
    ang
    ang=
    A=
    Maha.
    Maha
    Maha

    ‘Maha avoids to meet her father-in-law.’

Besides the fact that the agent winds up at the very end of the clause instead of at its otherwise preferred position after the verb, this still seems to be comprehensible. What happens, however, if we use a transitive subordinating verb? The dictionary, for instance, lists pinya- ‘ask (s.o. to do sth.)’ as a candidate. With regards to the normal constituent order of Ayeri, we can assume that the patient NP will follow the agent one:

  1. Ang
    Ang
    AT
    pinyaya
    pinya-ya
    ask-3SG.M
    sahayam
    saha-yam
    go-PTCP
     
    Ø=
    TOP=
    Yan
    Yan
    Yan
    sa
    sa=
    P=
    Pila.
    Pila
    Pila

    ‘Yan asks Pila to go.’

This also still looks very harmless with regards to parsability. However, things become more complicated if we increase the complexity of the embedded phrase by making the subordinate verb transitive (4a) or even ditransitive (4b):

    1. Ang
      Ang
      AT
      pinyaya
      pinya-ya
      ask-3SG.M
      konjam
      kond-yam
      eat-PTCP
      inunas
      inun-as
      fish-P
       
      Ø=
      TOP=
      Yan
      Yan
      Yan
      sa
      sa=
      P=
      Pila.
      Pila
      Pila

      ‘Yan asks Pila to eat the fish.’
    2. Ang
      Ang
      AT
      pinyaya
      pinya-ya
      ask-3SG.M
      ilyam
      il-yam
      give-PTCP
      koyaley
      koya-ley
      book-P.INAN
      ledanyam
      ledan-yam
      friend-DAT
      yana
      yana
      3SG.M.GEN
       
      Ø=
      TOP=
      Yan
      Yan
      Yan
      sa
      sa=
      P=
      Pila.
      Pila
      Pila

      ‘Yan asks Pila to give the book to his friend.’

The distance between the subordinating verb and its arguments grows by the increasing number of constituents in the embedded phrase, which in turn becomes increasingly “deep” in terms of underlying syntactic structure or “heavy” with regards to syntactic weight. The parser in both the speaker’s and the listener’s brain thus has to keep track of more and more relations in parallel. In terms of information flow, this does not strike me as beneficial or intuitive either to a speaker constructing the phrase, or to a listener having to decode the utterance. For the same reason, there is already a rule that relative clauses constitute “heavy” elements which pull their referent NP all the way to the back of the clause to keep the upper right field free of clutter. While (4a) is a little awkward since there are two patients to keep track of at the same time and at the time the first patient NP occurs we don’t know yet whether -ya is just for agreement or a cliticized agent-topic pronoun, example (4b) seems even more impenetrable with its three referents in the embedded clause and two in the matrix clause, placed at the end. Putting less important information before important information also goes against information-flow intuition.2 Restructuring seems advisable here, thus, and there are two possibilities:

    1. Ang
      Ang
      AT
      pinyaya
      pinya-ya
      ask-3SG.M
      _​​i​
       
       
       
      Ø=
      TOP=
      Yan
      Yan
      Yan
      sa
      sa=
      P=
      Pila
      Pila
      Pila
      [ilyam
      il-yam
      give-PTCP
      koyaley
      koya-ley
      book-P.INAN
      ledanyam
      ledan-yam
      friend-DAT
      yana]​​i​​.
      yana
      3SG.M.GEN

      ‘Yan asks Pila to give the book to his friend.’
    2. Ang
      Ang
      AT
      (da​​i​-)pinyaya
      (da=)pinya-ya
      (so=)ask-3SG.M
       
      Ø=
      TOP=
      Yan
      Yan
      Yan
      sa
      sa=
      P=
      Pila
      Pila
      Pila
      [, ang
      ang
      AT
      ilye
      il=ye.Ø
      give=3SG.F.TOP
      koyaley
      koya-ley
      book-P.INAN
      ledanyam
      ledan-yam
      friend-DAT
      yana]​​i​​.
      yana
      3SG.M.GEN

      ‘Yan asks Pila to give the book to his friend.’

In the past I might have preferred (5a) as a solution, but since the analogy to relative clauses suggests itself and complement clauses are useful, I find that I tend towards the sentence in (5b) currently. To indicate that the sentence structure has been remodeled and information will be following, it might also be useful to add the da- particle to the verb: a more literal translation of (5b) could be ‘Yan asks Pila such that she give the book to his friend.’ Either way, however, the normal VSO word order is restored this way, important information is present early on, and the heavy constituent is banned to the back. Information should now be able to flow easily again.

  1. If Ayeri has such a thing, which is kind of a problem for me right now. I need to read up on VSO-language syntax before I can make conclusive claims.
  2. Hat-tip to Oliver Schallert (blog article in German).