Hostname: page-component-74656f9884-j7cmm Total loading time: 0 Render date: 2026-11-05T16:13:13.607Z Has data issue: false hasContentIssue false
  • English
  • Français

L-syntax and phono-symbolism: on the status of ideophones in complex predicates *

Published online by Cambridge University Press:  09 February 2017

Ludovico Franco
Ludovico Franco*
Affiliation:
Universidade Nova de Lisboa
*
franco.ludovico@gmail.com
Rights & Permissions [Opens in a new window]

Abstract

In this article, the author demonstrates that verbal compound constructions involving an ideophone and a light verb represent a widespread syntactic device in the world's languages. The author provides evidence that phono-symbolic morphemes cannot be treated as ‘bare’ direct objects in such constructions. Ideophones appearing in the light verb-adjacent position form a semantic unit with the verbal predicate, despite the fact that in some languages they can be syntacticized as (bare) nouns and appear in argumental position. Specifically, ideophones in complex predicates are part of the verbal domain with which they ‘blend’ (yielding a single predicate) through the mechanism of conflation, along the lines of Hale and Keyser (1993, 2002), and building on Ramchand (2008).

Résumé

Dans cet article, l'auteur montre que les constructions verbales composées impliquant un idéophone et un verbe léger constituent un dispositif syntaxique répandu dans les langues du monde. L'auteur argumente que dans de telles constructions, les morphèmes phono-symboliques ne peuvent pas être traités comme des objets directs « nus ». Les idéophones qui apparaissent dans la position adjacente à un verbe léger forment une unité sémantique avec le prédicat verbal, bien que dans certaines langues ils puissent être syntactisés comme des noms « nus » et apparaître en position argumentale. Plus précisément, les idéophones dans les prédicats complexes font partie du domaine verbal avec lequel ils se combinent, grâce au mécanisme de la conflation, pour produire un prédicat unique, selon la proposition de Hale et Keyser (1993, 2002) et en s'appuyant sur Ramchand (2008).

Keywords

ideophonescomplex predicatesinternal argumentslight verbsL-syntaxIdéophonesprédicats complexesarguments internesverbes légersL-syntaxe

1. Introduction

Ideophones generally represent a significant part of the lexicon of human languages (see Alpher Reference Alpher, Hinton, Nichols and Ohala1994, Childs Reference Childs, Hinton, Nichols and Ohala1994, Dingemanse Reference Dingemanse2012, among many others, for a general overview). They are uninflected words (‘bare lexemes' according to Creissels Reference Creissels, Erhard Voeltz and Kilian-Hatz2001), often of an onomatopoeic character (or at least with a phonologically deviant organization), which typically encode pivotal events (natural sounds, movements, lights, beats, etc.).Footnote 1

Ideophones have often been characterized in the literature as an independent syntactic category (e.g., Evans and Levinson Reference Evans and Levinson2009). It is true that they usually exhibit distinctive phonology (such as vowel changes to signal switches in size or intensity, and/or unusual phonemes or tonal patterns) and distinctive morphology (such as special morphemic reduplication patterns), and occur sometimes in highly constrained morpho-syntactic constructions, but I believe that it is better to describe them as overlapping with (a set of) the lexical categories of a given language. Indeed, as Childs (Reference Childs, Hinton, Nichols and Ohala1994: 178) writes, “In reality, the features of ideophones are not unique, or even qualitatively different from those possessed by other word categories. No feature is unique to ideophones, but they do function somewhat apart from the matrix language, possibly due to their greater reliance on universals and lesser reliance on language-specific conventions.”

The goal of this work is to analyse the morpho-syntactic characteristics of a specific syntactic environment in which ideophones appear, namely the construction in which they are introduced by a light verb with meanings such as ‘do’, ‘say’, ‘have’, or ‘be’. Descriptively, at least three main syntactic contexts in which ideophones usually appear have been distinguished (Childs Reference Childs, Hinton, Nichols and Ohala1994; Creissels Reference Creissels1997, Reference Creissels, Erhard Voeltz and Kilian-Hatz2001). First, they can occur in isolation (as a kind of interjection, or functioning as an independent clause, (e.g., Kilian-Hatz Reference Kilian-Hatz, Erhard Voeltz and Kilian-Hatz2001)), as shown in the Italian example in (1).

  1. (1) glu glu,  la  barca  affondò.

    idph  the  boat  sink.pst.3sg

    ‘Gobble gobble, the boat sank.’

Second, they can occur as adverbials, in combination with verbs that express the same or a related meaning, as in the case of Japanese so-called ‘mimetics’ (cf. Shibatani Reference Shibatani1990; see also Nuckolls Reference Nuckolls1996 on Pastaza Quechua). Consider the Italian example in (2).

  1. (2) la  vespa  volava  a  zig zag.

    the  wasp  fly.ipfv.3sg  at  idph

    ‘The wasp was flying in a zigzag pattern.’

Finally, as already pointed out, ideophones can appear attached to a light verb often meaning ‘say’, ‘do’, ‘have’, ‘be’, etc. Such constructions are the precise focus of my investigation. Consider an example of this structure from Italian in (3).

  1. (3) il  mio  cuore  fa  bum bum.

    the  my  heart  do.prs.3sg  idph

    ‘My heart beats.’

Two aims will be pursued here. The first is to provide a detailed comparative illustration of the phenomenon of ideophonic complex predicates from a cross-linguistic perspective. The second is to provide evidence that ideophones appearing in the light verb-adjacent position form a semantic unit with the verbal predicate, despite the fact that in some languages they can appear in argumental position. Complex predicates are standardly defined as “predicates which are multi-headed; they are composed of more than one grammatical element (either morphemes or words), each of which contributes part of the information ordinarily associated with a head” (Alsina et al. Reference Alsina, Bresnan and Sells1997: 1). In this article, I argue that ideophones in complex predicates are part of the verbal domain with which they ‘blend’ (yielding a single predicate) through a mechanism of (hidden) conflation, along the lines of Hale and Keyser (Reference Hale, Keyser, Hale and Keyser1993, Reference Hale and Keyser2002; see also Mateu Reference Mateu2002), and building on Ramchand (Reference Ramchand2008). I will also suggest that the light-verb construction with ideophones is likely to originate from a genitive-like noun phrase involving the ideophone as the ‘possessum’.

2. Cross-linguistic evidence for a widespread light verb + ideophone pattern

In this section I show, with a broad range of cross-linguistic examples, that verbal compound structures involving an ideophone and a light verb represent a widespread syntactic device in the world's languages. The observation that ideophones are quite commonly found together with ‘neutral’ verbs is far from new (e.g., Childs Reference Childs, Hinton, Nichols and Ohala1994: 187, Alpher Reference Alpher, Erhard Voeltz and Kilian-Hatz2001: 10–11, Schultze-Berndt Reference Schultze-Berndt, Erhard Voeltz and Kilian-Hatz2001: 360–367, Güldemann Reference Güldemann2008: 280, Amha Reference Amha, Amberber, Baker and Harvey2010, Dingemanse Reference Dingemanse2012, among others), but, to my knowledge, no attempt to provide a comparative illustration of the phenomenon has been available until now.

In such constructions, the basic universal pattern is the following: the lexical meaning is determined by the (uninflected) ideophone, while the grammatical content (e.g., tense, aspect, mood (the TAM markers)) is carried by the light verb. This section also provides some empirical evidence that ideophones are part of the verbal predicate rather than internal arguments of the (light) verb. This point, namely the evidence that ideophones cannot be treated as (‘bare’) direct objects, could be useful to support the formal characterization of ideophonic complex predicates to be sketched in section 3.1. I begin with a prototypical African example. As shown in Creissels (Reference Creissels1997, Reference Creissels, Erhard Voeltz and Kilian-Hatz2001: 80–83) ideophones in Setswana (Bantu) normally combine with the light verb re ‘say’. Consider the example in (4) where re combines with the ideophone phatla.Footnote 2

  1. (4) Dikokwana  ts-a  re  phatla  fa  di  bona  segodi.

    10.chicken  sc10-cons  say  idph  when  sc10  see  7.hawk

    ‘The chickens suddenly scattered when they saw the hawk.’

    Setswana (Creissels Reference Creissels, Erhard Voeltz and Kilian-Hatz2001: 81)

Creissels (Reference Creissels, Erhard Voeltz and Kilian-Hatz2001: 79) argues that “Setswana ideophones are basically predicative lexemes that constitute the lexical element of compound predicates in which the function of auxiliary is fulfilled by the verb re.” Interestingly, re is unusual among Setswana verbs (cf. Creissels Reference Creissels, Erhard Voeltz and Kilian-Hatz2001: 83) in that it cannot take a noun phrase as its complement, but only clauses and ideophones. This fact militates against taking the ideophone to be an internal argument of the light verb, at least in Setswana.

Afro-Asiatic languages also display large inventories of ideophones entering a complex predicate construction. Amberber (Reference Amberber1996) has shown that in Amharic the light predicate al ‘to say’ is commonly used to create complex predicates with a full set of different lexical bases, including ideophones, in a configuration that he assumes is parallel to quotative constructions. In their description of Somali ideophones, Dhoorre and Tosco (Reference Dhoorre and Tosco1998: 129) assume that “ideophones are limited to use with the verb yiri ‘to say’ when intransitive […] and the verb sii ‘to give’ when transitive; if the use of the ideophone with the verb yiri ‘to say’ is transitive, its use with the verb sii ‘to give’ involves a causative meaning.”

More details, from the case of Wolaitta (Omotic), are given below. As shown in Amha (Reference Amha, Amberber, Baker and Harvey2010: 278–279), Wolaitta ideophones obligatorily occur with the verb g- ‘say’ when they enter intransitive predicates, and with ʔoott- ‘do’ when they enter transitive predicates. These two verbs act as dummy verbs uniquely devoted to hosting TAM morphology, as illustrated in (5). Note that in Wolaitta, according to Amha, the ideophone and the light verbs g- and ʔ oott- cannot be separated by any other constituents. This rigid ordering may be considered as evidence that ideophones form a unit with the verbal predicate.

  1. (5)

    1. a. kúnd-étt-íyo  t’úlʔu-g-aási

      fall-nmz-f.acc  idph-say-1sg.f

      ‘I failed.’ [when talking about a test on the Wolaitta language]

      Wolaitta (Amha Reference Amha, Amberber, Baker and Harvey2010: 262)

    2. b. guútta  moóre  ʔašo  cácácácáca  ʔoott-ádá  tamá-n

      little  fat  meat.acc  idph  do-ss.a.cnv  fire-loc

      t'iit't’-aásu

      roast-3sg.fs.g.pf

      ‘She roasted a little fat meat on the fire.’

      Wolaitta (Amha Reference Amha, Amberber, Baker and Harvey2010: 278–279)

Continuing the crosslinguistic overview of verbal complexes including an ideophone, it becomes apparent that these constructions are quite common in the languages spoken in Asia. I begin with the languages spoken in Siberia. In many Mongolic languages the verb for ‘say’ can occur as a light verb in conjunction with ideophones (cf. Matić and Pakendorf Reference Matić and Pakendorf2013: 387). Below we provide an example from Buryat in which the verb ge ‘say’ acts as a light verb, combining with the non-inflected lexeme teb-teb (ideophone for chew) and encoding TAM features.

  1. (6) Tiixeden  baabgajn  teb-teb  ge-xyn  duula-ad.

    then  bear  idph.chew  say-fut.ptc.gen  hear-pf.cvb

    ‘Then the bear heard (the fox) chewing.’

    Buryat (Matić and Pakendorf Reference Matić and Pakendorf2013: 387)

In Nivkh (isolate), a language spoken in Eastern Siberia and on Sakhalin Island, complex predicate structures including ideophones display the same basic pattern and are usually accompanied by the verb ha- (‘be so’), as shown in (7), where the light predicate is used together with the expressive item ŋəa ŋəa (doze).

  1. (7) Q'o-jnə-r  ŋəa ŋəa  ha-d.

    sleep-des/inch-conv.nar.3sg  doze doze  be.so.ind

    ‘Feeling sleepy he is dozing.’  Nivkh (Nedjalkov and Otaina Reference Nedjalkov and Otaina2013: 90)

In the Tungusic languages spoken in Eastern Siberia and Manchuria we observe an analogous system. Consider the case of Evenki (Nedjalkov Reference Nedjalkov1997: 304) in which ideophones form complex predicates with the light verb o:- ‘to make/do’. Examples include kachuso:- ‘make a noise’ from the ideophone kachus ‘hush’; taso:- ‘crack’ (of ice) from tas ‘cracking/crackling’; d'usero:- ‘flash’ from d'user ‘flashing’, etc.Footnote 3

Ideophonic complex predicates seem to be a syntactic object widely available in other language families of the Asian continent. Such constructions are attested in Tibeto-Burman languages, as reported for example in Matisoff (Reference Matisoff, Hinton, Nichols and Ohala1994: 120), who described Lahu's complex predicates in which an ideophone occurs directly before the dummy verb qáy (commonly translated as ‘go’), as in qáw-qáw ideo qáyverb ve ‘go bow-bow’. In recent work on Japhug, a Tibeto-Burman language spoken in the Barkam County of Sichuan, China, Jacques (Reference Jacques2013) shows that in this language ideophones can be used with up to four distinct light verbs: (a) the semantically empty stative verb pa, from proto-Rgyalrong ‘to do’; (b) the light verb ti ‘say’; (c) the manner verb stu ‘do like…’; (d) its reflexive form ʑɣɤ-stu ‘act like…’.

The verb pa is a stative intransitive verb. It is exclusively attested as a light verb and requires the presence of an ideophone. It is used with ideophones for shape, colour, or spatial disposition. Consider the example in (8).

  1. (8) ɯ-phoŋbu  nɯ  rcanɯ  ʁɲɟliʁɲɟli  ʑo  nɯ-pa.

    3sg.poss-body  top  top  idph  emph  testim-light.verb

    ‘Its body, it is enormous.’  Japhug (Jacques Reference Jacques2013: 271)

The verb ti ‘say’ is used intransitively as a light verb with ideophones expressing sound and sensations, as illustrated in (9).

  1. (9) ɯ-thoʁ  nɯ  tɕu  zɟɯɣ  ʑo  ti  ɲɯ-ŋu.

    3sg.poss-ground  top  loc  idph  emph  fact.say  testim-be

    ‘[The stone] made a loud noise (as it fell) on the ground.’

    Japhug (Jacques Reference Jacques2013: 272)

The transitive manner deixis verb stu ‘do like…’ always indicates a volitive (causative) action, unlike the light verbs pa and ti. The verb can either have no semantic patient, as in (10), or a definite patient, like ɯ-ku ‘its head’ in (11). On the contrary, its reflexive form ʑɣɤ-stu ‘act like…’, which also describes a volitional activity, shows only an intransitive use.

  1. (10) ɯ-sŋɯro  lu-lɤt  tɕe  ɬɯɣnɤlɯɣ  ɬɯɣnɤlɯɣ

    3sg.poss-breath  ipfv.upstream-throw  lnk  idph  idph

    tu-ste  ɲɯ-ŋu.

    ipfv-do.like[III]  testim-be

    ‘When it breathes, [one can see its body] expanding and contracting with each breath.’

    Japhug (Jacques Reference Jacques2013: 273)

  2. (11) ɯ-ku  ra  pjɯ-nɯ-χtɕi  tɕe  ɯ-ku  ra  rloʁnɤrloʁ

    3sg.poss-head  pl  ipfv-auto-wash  lnk  3sg.poss-head  pl  idph

    tu-ste  ɲɯ-ŋu

    ipfv-do.like[III]  testim-be

    ‘When it cleans its head, it moves it with rhythm’  Japhug (Jacques Reference Jacques2013: 273)

In Lao (Tai-Kadai) ideophones, when not employed as manner adverbs, can occur in a light-verb construction with hêt1 ‘do/make’ as a host verb (Enfield Reference Enfield2007: 301). Normally, the verb juu1 ‘be.at, continuous’ is also present in the construction, in final position, as shown in (12).

  1. (12) mia2  laaw2  hêt1  tòk1-pòk1  juu1.

    wife  3sg.f  do  idph  cont

    ‘His wife is being squat, comical.’ (tòk1-pòk1)   Lao (Enfield Reference Enfield2007: 301)

Dravidian languages are also able to ‘syntacticize’ ideophones in light-verb constructions, specifically together with an inflected form of the verb for ‘say’ (Emeneau Reference Emeneau and Dil1980, Abbi Reference Abbi1992, Krishnamurti Reference Krishnamurti2003). For instance, examples from Telugu include complex predicates formed with the verb –an (‘to say’), such as curukku-m-an ‘a body part burns with a heated object suddenly’, thalukku-m-an ‘flash like lightning’, etc. An analogous construction is at work in the cognate language Kota, where ideophones are again accompanied by the verb -in ‘say’, as in danak in- ‘become limp with fatigue’ or kulak in- ‘body feels cool’ (Krishnamurti Reference Krishnamurti2003: 485).

At the crossroads of Asia and Oceania, Papuan and Austronesian languages offer interesting insights into the syntax of verbal predicates including a phono-symbolic morpheme. Mian (Ok / Trans New Guinea / Papuan) has two ‘light’ verbs, ge/ga~gena ‘do’ and ke ‘make’, which combine with a noun (e.g., tekein ‘knowledge’) or an ideophone (e.g., kalkal, ‘sizzle’, bokbok, ‘boil’) to form a complex predicate (Fedden Reference Fedden2007: 306). Two examples of these complex predicate constructions are given in (13). Interestingly, ge/ga~gena ‘do’ only occurs in light-verb constructions, whereas ke ‘make’ can act as an independent verb and license object arguments. Note that ge/ga~gena merges almost exclusively with ideophones, yielding unergative structures. This pattern can be taken as further empirical evidence of the verbal-predicate–internal status of ideophones, because these elements are very unlikely to be licensed as bare direct objects in the Mian pattern.

  1. (13)

    1. a. naka.l  tekein  ke-b-i=be.

      man.m dem.m.sg  knowledge  make-ipfv-1sg.sbj=decl

      ‘I know this man.’

    2. b. kalkal  ge-bi-n-e=a…

      idph  do.ipfv-aux.ipfv-seq-3sg.n1.sbj=med

      ‘It (some meat) was sizzling and then…’  Mian (Fedden Reference Fedden2011: 154)

In the cognate language Oksapmin, the lexical morphemes that combine with the light verbs li- ‘say’ and pl- ‘tell’ show a certain degree of sound symbolism and “appear to have consistent sound-meaning correlations” (Loughnane Reference Loughnane2009: 313).Footnote 4 Consider for instance the ideophonic item xoj ‘make noise as when one engages in traditional singing and dancing’, as shown in example (14). The ideophonic nature of this item is confirmed by the fact that the syllable rhyme /oj/ is not attested elsewhere in Oksapmin.

  1. (14) mə=ma  sjap  ma  ixile  kom  san  mə-de=x

    dem.prx=rel  cassowary  rel  3p.poss  back  body  dem.prx-across=3sm

    pla-t-pel=xən  xoj  li-n-gop=li.

    pull-pfv-if.pl=sbrd  idph  say-pfv-vis.fp.sg=rep

    ‘When the cassowaries pulled with their backs, they made singing noises.’

    Oksapmin (Loughnane Reference Loughnane2009: 314)

Interestingly, in Oksapmin the verb pl- is morphologically the causative counterpart of li-, although the meaning of pl- is not strictly the causative of li-, but rather its ‘transitive’ form. In such cases the transitive object encodes the addressee/hearer, as shown in (15), where the phono-symbolic morpheme goŋ ‘whistle’ is present.

  1. (15) goŋ goŋ  pli-l  tap  oxə  pli-n-gwel

    idph  tell-ipfv.per.todp  pig  3sm  come-pfv-vis.yestp

    ‘I whistled to him and then (I saw that) the pig came’

    Oksapmin (Loughnane Reference Loughnane2009: 315)

In Austronesian languages ideophones have not been the object of detailed studies, the only exceptions I know of being Klamer (Reference Klamer2002) and Bradshaw (Reference Bradshaw2006). Bradshaw identifies a set of phono-symbolic items in Numbami and Jabêm, two Austronesian languages spoken in Papua New Guinea. In both languages ideophones employed as manner adverbs/movement imitatives appear with specific grammatical suffixes.

  1. (16) Ai-sarja  i-yotomu  pakapdka-adala.

    tree-branch  it-severed  idph-ly

    ‘The tree branch snapped with a crack.’  Numbami (Bradshaw Reference Bradshaw2006: 56)

  2. (17) Tuàmbiŋ  gêlôb  ôŋôŋ-geŋ  gêja.

    hombill  it-flew  idph-ly  it-went

    ‘The hornbill flies with a flapping noise.’  Jabêm (Bradshaw Reference Bradshaw2006: 58)

Hinton, Nichols, and Ohala (Reference Hinton, Nichols and Ohala1994: 3) have shown that a subgroup of sound-symbolic words is precisely that of ‘movement imitatives’, namely sound-symbolic expressions for the characterisation of movement. For instance, Japanese has a large set of ‘mimetics’ movement imitatives (Hirose Reference Hirose1981, Hamano Reference Hamano1998, among others) as shown in (18).

  1. (18)

    1. a. Dosa-dosa(-to) aruku ‘to walk with a loud noise’

    2. b. Doka-doka(-to) aruku ‘to walk noisily and violently’

    3. c. Kotu-kotu(-to) aruku ‘to walk with hard-soled shoes’

    4. d. Saku-saku(-to) aruku ‘to walk in soft snow’

Note that the examples in (18) are optionally grammatically marked with the case morpheme/ complementizer –to, patterning in this respect with the Austronesian examples reported in (16) and (17). Another language that, like Japanese, shows a large sound symbolic vocabulary of movement imitatives (more than 800, according to Ibarretxe-Antuñano Reference Ibarretxe-Antuñano2004) is Basque. Movement imitatives in Basque are the result of a derivational process by means of the suffix -tu when used in adverbial contexts and, as expected, the result of compounding with the light verb egin ‘make’ in verbal predicate context. Note that in Indo-European languages also, adverbs signalling bodily movements/movement imitatives are signalled by the means of special derivational suffixes, for instance in Italian by means of the suffix -one/i (e.g., carpone/i ‘on all fours’, tentone/i ‘gropingly’), in French by means of the circumfix à …-on(s) (e.g., à reculons, ‘backward’, à califourchon, ‘astride’, and in German by means of the suffix -lings (e.g., näslings ‘face down’, bäuchlings, ‘upon one's belly’, etc.). See Franco (Reference Franco2015) for a recent morphosyntactic analysis of such adverbial items in Italian. See also Torrence (Reference Torrence2013) for a similar analysis of manner-adverbial ideophones in Wolof.

Returning to Austronesian languages, Klamer (Reference Klamer1998, Reference Klamer2002) shows that Kambera (Malayo-Polynesian/Austronesian) also displays complex predicates with idephones. In Kambera, ideophonic roots are an open class, and they make up approximately 10 per cent of the language lexicon. Klamer (Reference Klamer2002: 262) specifically writes that “[s]yntactically, the ideophonic roots are exceptional because they can surface only in the position of a quote in a special ‘quotative’ construction […]. Morphologically, they are special because they are the only root forms that, in order to be used as verbs, must be derived by circumfixation rather than by prefixation or suffixation, as ‘normal’ roots are.”

The circumfix employed to derive verbal predicates from ideophones is ka-…-k, as shown in (19). Without such a morphosyntactic tool, the ideophones can appear only as ‘bare roots’ in constructions with the verb wà (‘say/report’), as illustrated in (20b).

  1. (19) jila ‘flash’ (sight)  >  ka-jila-k ‘to gleam; to flash (as lightning)’

  2. (20)

    1. a. Na-ka-jila-k  na  uma.

      3sg.nom-idph  art  house

      ‘The house gleams/shines.’

    2. b. Jila  wa-na  na  uma.

      idph  report-3sg.gen  art  house

      ‘The house gleamed.’      Kambera (Klamer Reference Klamer1998: 245–247)

Another Austronesian language that employs the light verb-ideophone construction is Muna (Malayo-Polynesian/Austronesian). In Muna there are many ideophonic roots that form the basis of a verbal derivation with the morpheme ko-, yielding roughly the meaning ‘to make the sound of’ (cf. Van den Berg Reference van den Berg1989: 194). Consider the examples in (21). Notably, verbal derivations that include the light verbal prefix ko ‘have’ usually have the meaning ‘have, possess’, as illustrated in (22).

  1. (21)

    1. a. no-ko-bhondu ‘to make the sound of a falling coconut’

    2. b. no-ko-pisi ‘to make the sound of a thin, flat surface hitting something’

    3. c. no-ko-rangku ‘to make the sound of falling rice, maize or stones’

      Muna (Van den Berg Reference van den Berg1989: 194)

  2. (22) no-ko-bhake-mo  ghai  aitu.

    3sr-have-fruit-pf  coconut  that

    ‘That coconut tree has borne fruit.’  Muna (Van den Berg Reference van den Berg1989: 90)

Unlike languages of the Austronesian family, Australian languages have been widely studied with respect to the syntactic behaviour of their ideophone inventories. Compound verbs are common in Australian languages (Heath Reference Heath and Dixon1976; Schultze-Berndt Reference Schultze-Berndt2000, Reference Schultze-Berndt, Erhard Voeltz and Kilian-Hatz2001; McGregor Reference McGregor2002; Bowern Reference Bowern2008; among many others) where they are described as an open class of non-inflecting lexical roots – of which ideophones are a subset – that, together with a reduced set of light (auxiliary) verbs hosting TAM morphology, form complex predicates. An example from Jawoyn (Gunwinyguan) is given in (23).

  1. (23) ga-wutjwutj-mar/mang.

    3sg-idph:bubble-lv.prs

    ‘It bubbles, it boils.’  Jawoyn (Merlan Reference Merlan, Simpson, Nash, Laughren, Austin and Alpher2001: 371)

Amerindian languages also seem to make heavy use of ideophone–verb compounds, even if there are not yet any systematic comparative studies on the topic. Amazonian languages make extensive use of light verbs to accommodate ideophones in their syntax. In Macushi (Carib), ideophones can occur with an inflected form of the verb taa ‘say’, as shown in (24).

  1. (24)

    1. a. tuna-ya  tiko tiko  taa  tanne,  tiaron  witi-'pi

      water-erg  idph:bubble  say  while,  another  go-past

      ‘While the water was bubbling, another one went.’

    2. b. sisiu  ta-i-ya.

      idph:lightning  say-3-erg

      ‘It is lightning.’    Macushi (Abbott Reference Abbott, Derbyshire and Pullum1991: 149)

A similar pattern is attested elsewhere in Amazonia and in the Andes. For instance, in Hixkaryana (Carib), ideophones can occur with the verb ka- ‘to do’ (Derbyshire Reference Derbyshire1977: 179). Similar constructions have been described in other Amazonian (Panoan) languages such as Kashibo-Kakataibo and Matses, which show a class of ideophones that are embedded in a complex predicate with the light verbs ke ‘say’ and ka ‘say/do’, respectively (Fleck Reference Fleck2003: 197; Zariquiey Biondi Reference Zariquiey Biondi2011: 244, 376; see also Mihas Reference Mihas2012). Recently Reiter (Reference Reiter2013) has also shown that in Awetí (Carib), an ideophone may be the uninflected part of a complex predicate with an inflected light verb based on the root ‘e ‘to say’. Similarly, in the non-cognate language Alto Perené (Arawak) spoken in Eastern Peru, ideophones merge as complement of the speech verb opoimatatzi or with the light verb kant ‘to be’, ‘to say’, ‘to do’ (Mihas Reference Mihas2012: 305).

Complex predicates including phono-symbolic expressions have also been described in Mosetén, a Mosetenan language spoken in the foothills of the Bolivian Andes (Sakel Reference Sakel2004, Reference Sakel, Miestamo and Wälchli2007). Most actions and events commonly expressed by lexical verbs in other languages are complex predicates in Mosetén. Thus, Sakel (Reference Sakel, Miestamo and Wälchli2007: 329), in her description of Mosetén complex predicates, draws an explicit parallel with ideophones in Australian languages, in which compound verb constructions are abundant. Complex predicates always consist of a so-called ‘verbness marker’, such as for example -yi-, meaning ‘do/be/say’ (Sakel Reference Sakel, Miestamo and Wälchli2007: 316).Footnote 5

The verbness marker determines the syntactic environment in which the (syntactically inert) lexical base appears, as shown in the examples in (25), with the phono-symbolic item wai’ ‘boil’, that appears as a bare stem.Footnote 6 The verbness markers specify different aspectual interpretations as well as different possible syntactic derivations. In particular, the morphemes -yi- (25a) and -tyi- (25b) give rise to predicates that differ in the amount of control of the subject over the action in terms of volition and causation; the morpheme -ki- has a stative interpretation (25c), and finally the light verb -jo- in (25d) signals an accidental action.

  1. (25)

    1. a. Yäe  wai’-ye-’  öjñï’.

      1sg.m  boil-do/be-1m.sg>3f.sg  water.f

      ‘I boil the water.’

    2. b. Mö’  öjñï’  wai’-tye-’.

      3f.sg  water.f  boil-put-1m.sg>3f.sg

      ‘I put the water to boil.’

    3. c. Mö’  öjñï’  wai’-ki-’.

      3f.sg  water.f  boil-be-3f.sg

      ‘The water boils.’

    4. d. Mö’  wej  wai’-jo-’  arosh-khan.

      3f.sg  seed.f  boil-become-3f.sg  rice.loan.f-in

      ‘The seed was (accidentally) boiled in the rice.’

Ideophones in North American languages also appear in the same syntactic construction. For instance, Mithun (Reference Mithun1982) showed that in Iroquoian languages such as Mohawk, Oneida, Cayuga and Seneca, they invariably appear in compound predicates directly accompanying a verb like ‘say’ or ‘do’.

The languages of the Caucasus provide further striking evidence for the universality of the observed tendency for ideophones to combine with light verbal items when they are used as verbal predicates. In Ingush (Nichols Reference Nichols2011: 122) phono-symbolic elements appear in light-verb constructions, usually with the light verbs joax ‘say/do’, oal ‘say’, or d.uoll ‘insert’. Interestingly, the light verb–ideophone construction in such languages usually appears as an adverbial clause, as illustrated in (26). In Ingush the light verb is required in order to syntacticize the ideophone as a manner adverb, while elsewhere in this syntactic configuration ideophones may surface as bare lexical items, accompanied by an adposition or a case marker, or bearing a derivational affix (see Nichols, Part 1 and fn. 4).

  1. (26) zou zou  eanna  oarqanjg  diezhar

    idph.ring  say.cvant  plate  d.fall.wp

    ‘The plate fell with a ringing sound’  Ingush (Nichols Reference Nichols2011: 122)

The case of the cognate language Udi is also interesting. In Udi most verbal predicatives are composed, in all TAM categories, of a light verb and a lexical base. According to a recent analysis put forth by Harris (Reference Harris2008), Udi light verbs act as classifiers, classifying the verb type. Thus, there is a specific light verb/classifier for inchoatives, -bak; a specific item for other unaccusatives, -eγ-; a classifier for unergatives, which are invariantly marked with -p-; a classifier for transitive verbs of inherently directed motion, -č-; one for transitive change-of-state verbs, -b-; and finally a classifier for the other transitives, marked with the -d- morpheme. Udi ideophones are used exclusively with the light verb -p- for unergatives, which originally meant ‘say’. Examples include verbs expressing non-language noise as in (27a) or verbs expressing sounds that animals make, as in (27b).

  1. (27)

    1. a. giʕzgiʕz-p- ‘laugh’

      gügü-p-  ‘thunder’

      xrp-xrp-p-  ‘crackle, rustle’

    2. b. boʕγoʕ-p-  ‘low, moo (of cattle, livestock)’

      q'iʕlaʕnc'i-p-  ‘bray of a donkey or buffalo’

      γaʕγaʕ-p-  ‘snap, snarl’  Udi (Harris Reference Harris2008: 223)

The construction under scrutiny here is also attested in Uralic languages. Consider, as an example, the structure in (28) below from Erzá Mordvin, a language spoken in the Middle Volga region of Russia, described in Wälchli (Reference Wälchli2005). Here, ideophones merge with the verb meaning ‘do’.

  1. (28) śeźeź  pŕa  marto  sandal'anzo  čikor-lakor  t'ejst’

    tear.ptc.pass  head  with  sandal.pl.poss.3sg  idph  do.pst.3pl

    ‘[The woman walked on tiptoe, from which] her sandals with torn tips squeaked.’

    Erzá Mordvin (Wälchli Reference Wälchli2005: 164–165)

To conclude this brief cross-linguistic overview of ideophone–light verb compounds, I now turn to some Indo-European languages. This will be the main focus of section 3, where a formal characterization of such constructions is proposed. Ideophones have generally been underdescribed in Indo-European, despite their widely attested use. Indeed, as Dingemanse (Reference Dingemanse2012: 657) points out, “in Classical Sanskrit imitative words were followed by the quotative verb –iti according to Pāṇini's 4th century BCE grammar (Pāṇini Reference Pāṇini1962: 196), and this same construction was used to mark reported speech and gestures.”

Phono-symbolic expressions entering the light verb pattern, for instance, are very common in Romance languages. Consider the case of Italian in (29) (see also example (3) above) and the Catalan example in (30).Footnote 7 Here, both languages use the dummy verb fa ‘does/makes’ to introduce the ideophone in an unergative form. Ideophones normally cannot form transitive predicates in such languages, as shown with an Italian example in (31).

  1. (29) Il  treno  fa  ciuf ciuf

    the  train  do.prs.3sg  idph:chug

    ‘The train chugs along’

  2. (30) el  ventijol  fa  xiu xiu

    the  breeze  do.prs.3sg  idph:whisper

    ‘the breeze whispers’

  3. (31)

    1. a. *Il  treno  fa  ciuf ciuf  i  binari

      the  train  do.prs.3sg  idph:chug  the  tracks

    2. b. Il  treno  fa  ciuf ciuf  sui  binari

      the  train  do.prs.3sg  idph:chug  on.the.pl  tracks

      ‘The train chugs on the tracks’

Ideophones in Indo-Iranian languages present a very similar pattern. Consider the case of Persian, in which most ideophones are reduplicated monosyllabic words (Mahootian Reference Mahootian1997: 343). In complex predicates, ideophones form invariantly intransitive predicates accompanied by some tensed version of kærdæn ‘to do’ or zædæn ‘to hit’.Footnote 8

  1. (32)

    1. a. gorbe  xor xor  mi-kone

      cat  idph:purr  dur-do.3sg.prs

      ‘the cat purrs’

    2. b. in  pesær  negh negh  mi-zane

      this  boy  idph:lament  dur-hit.3sg.prs

      ‘this boy cries’

The case of Persian is quite intriguing because, like many other Iranian languages, it relies heavily on light-verb constructions. In fact, it has been argued that simple verbs in these languages form a small class or even a closed class. Consequently, most light-verb constructions do not have lexical verb counterparts (Folli et al. Reference Folli, Harley and Karimi2005, Megerdoomian Reference Megerdoomian2012, among others).Footnote 9

Megerdoomian (Reference Megerdoomian2012: 179–180) writes that “the status of bare lexical elements in light verbal predicates has been a controversial topic for languages with a productive complex predicate formation process. The bare nouns […] display mixed properties, sometimes behaving as the internal argument and at other times forming a semantic unit with the verbal predicate.” In section 3.1, based on an analysis of the syntactic behaviour of Persian ideophones, which relies in turn on a set of diagnostics developed in Megerdoomian (Reference Megerdoomian2012), we will see that the syntactic configuration in which phono-symbolic expressions are introduced is that of a ‘bare’ predicative root and not that of an internal argument.

Nevertheless, ideophones can also appear in argumental position, as internal or external arguments, as shown in (33) for Italian, Catalan and Persian:

  1. (33)

    1. a. Non  sopporto  il  tic tac  dell' orologio

      not  stand.prs.1sg  the  idph  of.the  clock

      ‘I cannot stand the ticking of the clock’

    2. b. M’  agrada  el  xiu xiu  del  vent

      cl.dat.1sg  like.prs.3sg  the  idph  of.the  wind

      ‘I like the whisper of the wind’

    3. c. vaq vaq-e  sag  asab-am-o  khoord  mi-kone

      ideo-lnk  dog  nerve-my-dom  break  dur-do.prs.3sg

      ‘The bark of the dog is getting on my nerves’

The morphosyntactic properties of such kinds of ‘argumental’ ideophones are discussed in section 3.2.

The main results of this crosslinguistic overview can be summarized as follows. The verb most used in ideophone–verb compounds is the verb for ‘say’ but there are many other light items. such as ‘do’, ‘be’, and ‘have’, that can merge with phono-symbolic morphemes. Our investigation casts some doubts on the fact that ‘say’ verbs in this context can always be semantically interpreted as quotative verbs. Consider again the case of Udi, described by Harris (Reference Harris2008). As mentioned above, here the -p originally derived from ‘say’ acts as a classifier and introduces all the intransitive predicates (including ideophonic predicates), as shown below in (34). No quotative meaning is detectable in such constructions.

  1. (34)

    1. a. gölöš-p- ‘dance’

    2. b. axšum-p- ‘laugh’

    3. c. mərə-p-  ‘mew’ (idph)

Similarly in Persian, another language where complex predicates are far more common than lexical verbs, the light verb kærdæn ‘do, make’ is usually combined with nominals to form intransitives, as shown in (35). The same verb, as we have already seen in (32), is usually employed with ideophones. Again, there is no trace of a quotative semantics (Potts Reference Potts, Barker and Jacobson2007).

  1. (35) gerye  kærdæn (crying do)  ‘cry’

    šena  kærdæn (swim do)  ‘swim’

    vez-vez  kærdæn (idph: buzz do)  ‘buzz’

In addition, Plank (Reference Plank2005: 473), following the insight of Benveniste (Reference Benveniste1966), argues against a mere ‘quotative’ interpretation of ‘delocutive’ ideophone–verb compounds, claiming that “the syntax of expressive constructions is not quite like that of regular speech reporting” and that in “‘say’ constructions with expressives […] the action denoted is not really one of speaking but of doing, and a semantic development from ‘saying’ via ‘saying and/or doing’ to ‘doing’ seems more plausible than the reverse.”Footnote 10

Furthermore, ideophonic complex predicates crosslinguistically seem to obey a very basic implicational hierarchy [unergative > transitive]. This hierarchy captures the observation that, while in all the languages we have encountered, ideophone–verb compounds give rise to intransitive predicates, only in a subset of them is a transitive counterpart available, normally encoded with a different light verb.Footnote 11 This main finding is recapitulated in Table 1.

Table 1.

Ideophonic Complex Predicates

As we have seen, there are also languages (e.g., Mosetén or Japhug) that allow ideophones in a full set of verbal configurations, but the relevant fact is that, invariably, ideophonic predicates seem to typically describe internally caused eventualities, construed as arising from inherent properties of their single agent argument. In this respect, they pattern with the verbs of sound/light emission described in Levin and Rappaport Hovav (Reference Levin and Hovav1995, Reference Levin and Hovav2005). Interestingly, many objects ‘transitively’ licensed by the ideophonic predicates seem actually to be addressees/hearers/beneficiaries, as shown in (15) for the Oksapmin language, and in (11) for Japhug. We will give a comprehensive interpretation of these aspects of the syntax of ideophone–light verb formation in the next section.

3. The l-syntax of ideophones

At the end of the previous section, it was claimed that ideophones, crosslinguistically speaking, generally give rise to unergative structures when merged with a light verb. It is not very common to find languages in which such structures are able to license direct objects.

Since the early 1990s, many inquiries into the structure of verbal predicates have embraced a view of the verbal layer whereby syntactic derivations are built compositionally by combining the various components of the verb (Jackendoff Reference Jackendoff1990, Hale and Keyser Reference Hale, Keyser, Hale and Keyser1993, Levin and Rappaport Hovav Reference Levin and Hovav1995, Harley Reference Harley1996, Travis Reference Travis, Tenny and Pustejovsky2000, Mateu Reference Mateu2002, Ramchand Reference Ramchand2008, among many others). In their seminal work on argument structure, Hale and Keyser (Reference Hale, Keyser, Hale and Keyser1993, Reference Hale and Keyser2002) argue for a derivation of argument structure that is strictly dependent on the syntactic organization. Within the lexical module, which Hale and Keyser label l-syntax, lexical items are decomposed into atomic primitives that are combined by the syntactic tools of complementation and adjunction, and are subject to criteria of grammatical well-formedness (e.g., the Empty Category Principle of Chomsky Reference Chomsky1981 and the Head Movement Constraint of Travis Reference Travis1984). As for the configuration in which ideophones appear more frequently, Hale and Keyser (Reference Hale, Keyser, Hale and Keyser1993, Reference Hale and Keyser2002) basically argue that unergative verbs (laugh, dance, etc.) involve a hidden transitive structure where an (empty) light verb selects for (merges with) a bare noun (or a √Root; see Marantz Reference Marantz1997, Harley Reference Harley, Erteschik-Shir and Rapoport2005, Kayne Reference Kayne2009, among others). Consequently, the l-syntax of a unergative verb like laugh would be as illustrated in (36). ‘Unergative’ ideophones would represent cases in which this covert transitive structure is de facto spelled out (just as in those languages – e.g., Basque, Persian, Mosetén, etc. – in which complex predicates represent a usual way of encoding verbal meanings), as in (37).

  1. (36)

  2. (37)

In what follows I argue that ideophones can have a structure like (37), roughly parallel to bare noun roots such as laugh in (36), but that the structures in such derivations cannot behave like transitive constructions; that is, ideophones and bare noun roots do not behave like direct objects in transitive structures.

3.1. Ideophones are not internal arguments

To show that ideophones cannot occupy an argumental position, I provide an analysis of Persian light verbs combining with phono-symbolic expressions. As pointed out in section 2, light verbs are an important feature of many Iranian languages. Indeed, in Persian, complex predicates frequently express what in English would be expressed by a lexical verb, as in the unergative examples in (35).

With respect to the structure of the Persian VP, some authors (e.g., Ghomeshi and Massam Reference Ghomeshi and Massam1994, Vahedi-Langrudi Reference Vahedi-Langrudi1996, Samvelian Reference Samvelian2001) have assumed that bare nouns in a light-verb construction are not syntactically different from bare direct objects of transitive verbs; that is, the two sentences in (38) below would have the same underlying structure. The predicative root and the bare object would occupy the same position in the syntactic configuration and should therefore receive identical treatment. According to such analyses, ideophones – which can be employed as nouns in PersianFootnote 12 – should be parallel to direct object NPs. Note that both sentences in (38) contain the verb xordæn (eat), which in Persian may be used either as a heavy or as a light verb.Footnote 13

  1. (38)

    1. a. Mæn  hævij  xord-æm.  ‘bare’ DP object

      I  carrot  ate-1sg

      ‘I ate carrot(s).’

    2. b. mærdom  færib  xord-æn.  ‘bare’ √ element

      People  deceit  ate-3pl

      ‘(The) people were deceived.’     (Megerdoomian Reference Megerdoomian2012: 180)

The two types of preverbal items, bare object DPs and nominal root elements, actually display many similar properties. Among other similarities, they are both unmarked for number, they appear immediately before the verb, and they receive the main VP stress. However, Megerdoomian (Reference Megerdoomian2012) (see also Folli et al. Reference Folli, Harley and Karimi2005) has presented clear evidence, based on a set of diagnostics, that the two elements are distinct. In what follows, I show that ideophones pattern with nominal root elements in many of the tests employed by Megerdoomian. Their behaviour in fact shows that ideophones form a semantic unit with the light verbal predicate. Consequently, their syntax must be different from that of true bare DP objects.

First, evidence for a distinct distribution and for different syntactic shapes of the preverbal root item and the true bare DP object is provided by interrogatives. Bare DP objects can be questioned, as shown in (39b).

  1. (39)

    1. a. nima  ketab  mi-xun-e.

      Nima  book  dur-read-3sg

      ‘Nima is reading a book’

    2. b. Q:  nima  či  mi-xun-e?

      Nima  what  dur-read-3sg

      ‘What is Nima reading?’

      A:  ketab.

      ‘Book.’     (Megerdoomian Reference Megerdoomian2012: 190–191)

In contrast, Persian nominal roots in complex predicates cannot be questioned, as shown below again for færib xordæn, ‘to be deceived’.

  1. (40) Q:  mærdom  či  xord-æn?

    people  what  ate-3pl

    ‘What did people eat?’

    A:  *færib.

    ‘Deceit.’

Persian ideophones pattern with nominal roots in complex predicates and cannot be questioned, as shown in (41). The only way to obtain a grammatical question/answer pair is to provide an answer that includes the whole complex predicate, clearly interpreted as a conceptual unit.Footnote 14

  1. (41)

    1. a. Sag  vaq vaq  mi-kone

      dog  idph  dur-do.prs.3sg

      ‘the dog barks’

    2. b. Q:    Sag  či  mi-kone?

      Dog  what  dur-do.prs.3sg

      ‘What is the dog doing?’

      A:  *?  vaq vaq /  vaq vaq mi-kone

      idph   idph dur-do.prs.3sg

      ‘Barking.’

Similar question formation data with ideophonic predicates can be found in Italian. Indeed, questioned ‘predicative’ ideophones cannot surface as the wh-pronouns usually designating direct objects, (che) cosa ‘what’ and chi ‘who’. In these cases, questions can be formed only with the wh-particle normally employed in Italian with PP adverbials, come ‘how’, as shown in the examples in (42).Footnote 15 The use of ideophones in such contexts is interesting because the Italian light verb fare ‘to do/make’ can licence a direct object DP elsewhere, as shown in (43). Thus, in Italian as in Persian there is a clear contrast between ideophonic roots and true internal arguments.

  1. (42)

    1. a. Q:  Cosa/*come mangia Gianni?

      ‘What/*how does Gianni eat?’

      A:  Una banana

      ‘A banana’

    2. b. Q:  Come/*cosa mangia Gianni?

      ‘How/*what does Gianni eat?’

      A:  Con le mani/voracemente

      ‘with hands/ greedily’

    3. c. Il cane fa bau bau - il treno fa ciuf ciuf

      ‘the dog idph:barks - the train idph:chugs’

    4. d. Q:  Come /*cosa fa il cane? - Come/*cosa fa il treno?

      ‘How/*What does the dog do? - How/*What does the train do?’

      A:  ‘Bau bau / Ciuf ciuf’

  2. (43) Q:  Cosa fa la gallina?

    ‘What does the hen do?’

    A:  ‘L'uovo’

    ‘The egg’

Returning to Persian, a second test developed by Megerdoomian to disentangle nominal roots and internal arguments bears on ‘number neutrality’, a common property of bare nouns. For instance, in (44), “the professors were eating an unspecified number of pomegranates: they could have been eating many pomegranates or sharing the same one” (Megerdoomian Reference Megerdoomian2012: 191).

  1. (44) ostad-a  ænar  mi-xord-æn.

    professor-pl  pomegranate  dur-ate-3pl

    ‘The professors were eating a pomegranate/pomegranates’

By contrast, in complex predicates, (nominal) roots do not entail an interpretation involving on a ±singular entity and are instead conceived as predicate-internal. Indeed, bare DPs can also appear with numerals, taking on a numeral reading, while bare ‘predicative’ VP roots cannot show up with numerals. This is shown in (45) and (46), adapted from Megerdoomian Reference Megerdoomian2012: 193. With respect to number neutrality, ideophones again pattern with bare roots, as shown in (47).

  1. (45) ostad  se-ta  ænar  xord

    professor  three-clf  pomegranate  ate.3sg

    ‘The professor ate three pomegranates’

  2. (46) * mærdom  se-ta  færib  xord-æn.

    People  three-clf  deceit  ate-3pl

    Lit. ‘(The) people ate three deceits’

  3. (47) *? Sag  se-ta  vaq vaq  mi-kone

    dog  three-clf  idph  dur-do.prs.3sg

    Lit. ‘(The) dog is doing three vaq vaq’

Finally, another crucial test distinguishing DP objects from bare roots/ideophones concerns the availability of the definite/specific dom suffix –ra. In standard Persian, the –ra morpheme is optionally attached to definite/specific direct objects (Ghomeshi Reference Ghomeshi2003, Cagri Reference Cagri2007; see also Manzini and Franco Reference Manzini and Franco2016). The sentences in (48) show a bare object (48a) and its definite/specific counterpart (48b) The object in (48b) can also bear plural morphology and be accompanied by a demonstrative.

  1. (48)

    1. a. ostad  ænar  xord  bare direct object

      professor  pomegranate  ate.3sg

      ‘The professor ate a/some pomegranate(s)’

    2. b. ostad  in  ænar-a-ro  xord  specific direct object

      professor  this  pomegranate-pl-dom  ate.3sg

      ‘The professor ate these pomegranates’

By contrast, as illustrated in (49) (Megerdoomian Reference Megerdoomian2012: 194), no definite/specific counterpart is available for the nominal root in light-verb constructions, suggesting again that the nominal root is part of the verbal predicate rather than the internal argument of the verb.Footnote 16 As expected, ideophones behave exactly like bare roots with regard to the availability of the specific/definite –ra morpheme, as shown in (50).Footnote 17

  1. (49)

    1. a. pesær-æk  dobare  færib  xord

      boy-dim  again  deceit  ate

      ‘The little boy was deceived again’

    2. b. ?* pesær-æk  dobare  færib-ro  xord

      boy-dim  again  deceit-dom  ate

      ‘The little boy fell for the deception again’

  2. (50)

    1. a. *Sag  vaq vaq-ro  mi-kone

      dog  idph-dom  dur-do.prs.3sg

      Lit. ‘(The) dog is doing the vaq vaq’

    2. b. *gorbe  xor xor-o  mi-kone

      cat  idph:purr-dom  dur-do.3sg.prs

      ‘the cat purrs’

In sum, I have shown, using diagnostics developed by Megerdoomian (Reference Megerdoomian2012), that ideophones do not have the properties of direct objects. I now turn to the syntactic representation of their behaviour, relying mainly on Ramchand (Reference Ramchand2008). Both the cross-linguistic overview in section 2 and the data illustrated in this section for Persian show that ideophonic predicates are prototypically unergative. In addition, like other unergatives, they are typically atelic (Harley Reference Harley, Erteschik-Shir and Rapoport2005). Examples are given in (51).

  1. (51)

    1. a. Il treno ha fatto ciuf ciuf *in un'ora/per un'ora.  Italian

      ‘The train chugged *in an hour/for an hour.’

    2. b. il ragazzo ha tossito *in un'ora/per un'ora.

      ‘The boy coughed *in an hour/for an hour.’

    3. c. Sag *dær ye saæt/saætha vaq vaq kærd.  Persian

      ‘The dog barked *in an hour/for hours.’

    4. d. pesær *dær ye saæt/saætha sorfe kærd.

      ‘The boy coughed *in an hour/for an hour.’

    5. e. pesær  dær ye saæt/??saætha in ænar-a-ro xord.

      ‘The boy ate these pomegranates in an hour/??for an hour.’

Using an event-based decomposition of the VP along the lines of Ramchand (Reference Ramchand2008) (see also Svenonius Reference Svenonius2008, Lundquist Reference Lundquist2008) suppose that ideophones, as archetypal unergative predicates, are endowed with [init, proc] category features. The argument of such a predicate initiates or gives rise to an event, and thus undergoes the event. Ramchand (Reference Ramchand2008) basically argues that the VP can be split into three projections (init, proc, res), corresponding to subevental predications. Unergative/ideophonic complex predicates involve only two (init, proc) of the three subevents shaped in Ramchand, with res not relevant to the unergative domain. Given the basic event decomposition, I assume that in ideophonic predicates the undergoer features are underassociated, following Ramchand's terminology.Footnote 18 What this means is that the [proc] feature is essentially suspended, or deactivated. The syntactic derivation of ideophonic predicates can thus be thought of as involving only the higher level of the VP phase; that is, [init].

The ideophonic root is inserted in the complement position of a verb of initiation, often spelled out as a light verb. According to this analysis, the rough representation of a Persian complex predicate involving an ideophone is as shown in (52)

  1. (52)

In a structure like (52), nothing prevents the morphological incorporation or conflationFootnote 19 of an ideophonic root into a verbal slot, roughly as in the English example to laugh in (36), where the complement root is incorporated into a null init verb or verbal affix. Indeed, there are languages where ideophones can alternate between the two possibilities. For instance, in Hindi (Kachru Reference Kachru2006: 127) verbs may be derived from ideophonic roots by means of the verbal suffix –na, which is also employed with other parts of speech, as in (53), or the ideophonic root may merge with the light verb kərna ‘to do’,Footnote 20 to derive an ideophonic predicate, as in (54).

  1. (53) bəɽbəɽ-ana ‘to grumble’; hinhin-ana ‘to neigh’ [film-ana ‘to film’]

  2. (54) cε̃cε̃ kərna ‘to chirp’ - bəkbək kərna ‘to jabber’ [saf kərna ‘to clean’]

An analogous pattern can be seen in English, as shown in (55) with the ideophone ‘bubble’, which is syntacticized as a verbal predicate without the aid of a light verb.

  1. (55) The soapy water bubbled down the drain.

Actually, the strong crosslinguistic tendency of ideophones to appear with light verbs may be interpreted as a consequence of their peculiar phonological and morphological properties,Footnote 21 which do not allow them to easily undergo derivational processes such zero conversion, affixal derivation, etc. Phono-symbolic expressions are sui generis items in natural language lexica. To this end, it is interesting to note that verbal ideophones frequently pattern with loanwords.Footnote 22 The fact that loanwords are generally accompanied by a light verb when they enter the lexicon of a given language has been related to the observation that the borrowing of verbs as verbs is crosslinguistically marked (Moravcsik Reference Moravcsik1975, Muysken Reference Muysken2000, Schultze-Berndt Reference Schultze-Berndt, Erhard Voeltz and Kilian-Hatz2001, Wichman and Wohlgemuth Reference Wichmann, Wohlgemuth, Stolz, Bakker and Palomo2008, among others). Consider the data in (56) from Persian, in which loanwords usually enter verbal predicates by the means of the same light verbs recruited for use with ideophones.

  1. (56)

    1. a. danlod  kærdæn  b.  fæks kærdæn

      download  do    fax do

      ‘to download’    ‘to fax’

    2. c. imeyl  zædæn  d.  klik  kærdæn

      email  hit    click  do

      ‘to email’    ‘to click (on a mouse)’

Returning to the representation given in (52), I propose that the structures produced in these derivations must be distinguished from transitive constructions (contra Hale and Keyser Reference Hale and Keyser2002; see also Gallego Reference Gallego2012, Bosque and Gallego Reference Bosque and Gallego2014). In addition to the empirical evidence given above, based on Megerdoomian's (Reference Megerdoomian2012) diagnostics, a further reason to argue against a transitive interpretation of these structures is that many different lexical categories can be incorporated into, or act as complements of a possibly null light verb. For instance, in Basque, unergative predicates may be formed by the light verb egin (‘do’) plus an ideophone (irrist ideo egin ‘slip’), an NP (disdira NP egin ‘glitter’), an AdvP (hagin-ka AdvP egin ‘bite’) or a PP (hega-n PP egin ‘fly’), but crucially not with a case-marked DP (Ibarretxe-Antuñano Reference Ibarretxe-Antuñano2004, Berro Reference Berro2012). If we follow the standard minimalist assumption that case is an uninterpretable T feature on D (Pesetsky and Torrego Reference Pesetsky, Torrego and Kenstowicz2001, Reference Pesetsky, Torrego, Gueron and Lecarme2004; Chomsky Reference Chomsky and Kenstowicz2001); in other words, a form of agreement involving D, the lexical items occupying the root position in a structure like (52), which cannot be DPs, must be caseless and thus non-argumental.

I have argued that ideophonic predicates are prototypically unergative, but as seen in the crosslinguistic overview there are also some languages in which ideophones seem to form transitive predicates. I tentatively propose that such cases are actually causativized unergatives and again, broadly following Ramchand (Reference Ramchand2008: 174), I claim that these structures include a causative projection above the Vinit postulated in (52), as illustrated in (57) (see also Folli and Harley Reference Folli and Harley2007, Horvath and Siloni Reference Horvath and Siloni2011).

  1. (57)

The relevant point is that the structures in (57) have two external arguments in specifier position, a causative agent in DP1 and a lower agent (initiator or causee) in DP2 (see also Pylkkänen Reference Pylkkänen2008 for a similar interpretation of causativized unergatives in Finnish). The lower, or internal agent of the construction is introduced as the internal argument of VPcause + {VPinit; √root}. This structure does not seem to be universally available; this would explain why only a subset of the languages introduced in section 2 displays an unergative/transitive alternation with respect to ideophones.Footnote 23

Evidence that this approach is on the right track comes from the behaviour of unergatives in some Indo-Aryan languages (Bhatt and Embick Reference Bhatt and Embick2004, Ramchand Reference Ramchand2008). It was shown in (53) that Hindi ideophonic predicates may be formed by suffixation. Many other unergative predicates are formed with the same suffix, as shown in (58).

  1. (58) hɑ̃s-na  ‘to laugh’

    naach-na  ‘to dance’

    ur-na  ‘to fly’

Many Hindi unergative verbs can appear with a transitive/causative interpretation when the suffix -aa is added, as shown in (59) and (60).

  1. (59) hɑ̃s-aa-na  ‘to make laugh’

    naach-aa-na  ‘to cause to dance’

    ur-aa-na  ‘to fly (tr.)’

  2. (60)

    1. a. patang   ur  rahii  hai  unergative

      kite  fly  prog.f  be-prs.sg

      ‘The kite is flying.’

    2. b. anjali  patang  ur.aa  rahii  hai

      Anjali  kite  fly  prog.f  be-prs.sg

      ‘Anjali is flying a kite’       Hindi (Ramchand Reference Ramchand2008: 174)

According to Richa (Reference Richa2008: 43, 178), at least a subset of the ideophones (Levin and Rappaport Hovav's Reference Levin and Hovav1995 ‘verbs of external sound/light emission’) of the type illustrated in (53) can participate in the alternation as shown in (61).

  1. (61) kəɽək-na  ‘boil noisily’  >  kəɽək-aa-na  ‘to make boil noisily’

    cəmək-na  ‘to shine’  >  cəmək-aa-na  ‘to make shine’

These data from Hindi validate the view of ‘transitive’ ideophonic predicates as causativized unergatives, as illustrated in (62), where the morphemes –na and –aa are hosted respectively in Vinit and Vcause.Footnote 24

  1. (62) [ VPcause aa [ VPinit na ]]

When morphological transitivizing suffixes are not available, possibly due to a parameterization of the Cause projection along the lines of Pylkkänen (Reference Pylkkänen2008), languages like Italian or French must use analytical or periphrastic causatives (Kayne Reference Kayne1975, Rizzi Reference Rizzi and Keyser1978, Guasti Reference Guasti1996, Cinque Reference Cinque and Belletti2004, Folli and Harley Reference Folli and Harley2007, among many others) to spell out ideophonic predicates, as shown for Italian in (63). In such cases, according to a standard minimalist view, a functional head Appl checks the (dative) DP initiators, il treno ‘the train’ or il cane ‘the dog’ (Cuervo Reference Cuervo2003, Pylkkänen Reference Pylkkänen2008).Footnote 25

  1. (63)

    1. a. Il macchinista ha fatto fare ciuf ciuf al treno.

      ‘The driver made the train do idph.’

    2. b. L'allarme ha fatto fare bau bau al cane.

      ‘The alarm made the dog do idph.’

An alternative configuration in which the ideophone predicate may in principle occur is that of ditransitive structures (causative, for many authors; e.g., Harley Reference Harley2002, Manzini and Franco Reference Manzini and Franco2016). Indirect evidence comes from Italian examples like (64) and (65), where the verb dare ‘give’ is used instead of dire ‘say’, as might be expected for certain routine social acts (Plank Reference Plank2005). The verb fare ‘do’ may also be recruited in this context, as in (65).

  1. (64)

    1. a. Ho  dato  la  buonanotte  ai  bambini.

      have.prs.1sg  give.pst-ptcp  the  goodnight  to.the  children

    2. b. * Ho  detto  la  buonanotte  ai  bambini.

      have.prs.1sg  say.pst.ptcp  the  goodnight  to.the  children

      ‘I said goodnight to the children’ (lit. I give the goodnight to the children)

  2. (65)

    1. a. Gianni  mi  ha  dato/fatto  gli  auguri

      G.  cl.dat.1sg  have.prs.3sg  give/make.pst-ptcp  the  wishes

      di  Natale.

      of  Christmas

    2. b. *Gianni  mi  ha  detto  gli  auguri  di  Natale

      G.  cl.dat.1sg  have.prs.3sg  say.pst-ptcp  the  wishes  of  Christmas

      ‘Gianni wished me a Merry Christmas’ (Lit. ‘give/made me’)

Interestingly, some verbs of sound emission (not necessarily ideophonic, but semantically linked to ideophones in many languages) may follow an analogous pattern in Italian, employing the light verb fare ‘do’, as in (66) or dare ‘give’ as in (67).

  1. (66)

    1. a. Ho  fatto  un  fischio  a  Gianni.

      have.prs.1sg  made  a  whistle  to  Gianni.

    2. b. ?? Ho  fischiato  a  Gianni.

      have.prs.1sg  whistled  to  G.

      ‘I whistled to Gianni’ (Lit. ‘I made a whistle to G.’)

  2. (67) Ho  dato  un  colpo  di  telefono  a  Gianni.

    have.prs.1sg  give.pst-ptcp  a  bang  of  telephone  to  Gianni

    ‘I phoned Gianni’ (Lit. ‘I give a bang of telephone to Gianni.’)

Following Manzini and Franco (see also Manzini and Savoia Reference Manzini and Savoia2011a,Reference Manzini and Savoiab), I assume that the primitive content of the preposition a ‘to’ in (63)–(65) is ‘inclusion’ (part-whole). I notate inclusion with ⊆, though the inclusion relation must be interpreted not mathematically but loosely as zonal inclusion, in the sense of Belvin and Den Dikken (Reference Belvin and den Dikken1997). Prepositions like English to/of or Italian a/di that have predicative inclusion content are notated as P(⊆), as in the structure of (64), given in (67). P(⊆) takes as its internal argument the DP i bambini ‘the children’ and as its external argument the sister to its projection, that is, the DP la buonanotte ‘the goodnight’.

  1. (67)

While this issue will not be further explored here, since the languages for which I have primary data (Romance and Iranian languages) do not exhibit relevant examples, an interpretation of transitive ideophone-based predicates based on ‘give’ might explain the crosslinguistically attested pattern in which the syntax of verbal ideophones encodes an addressee or beneficiary, as for instance in the Papuan language Oksapmin – see (15) – or in Somali, as described by Dhoorre and Tosco (Reference Dhoorre and Tosco1998). In any case, the idea just introduced of a primitive (part-whole, possessive) content preposition such as to or of will be relevant for a brief discussion of the syntax of ideophones in noun phrases, a largely unexplored topic to which I now turn.

3.2 Ideophones and noun phrases

In this section, I propose that the widespread availability of the complex predicate construction with ideophones might be explained if it is derived from a genitive-like noun phrase with the ideophone as the ‘possessum’. Indeed, in many different languages ideophones can appear as nouns in a seemingly possessive construction. Some examples were shown in (33), repeated in (68).

  1. (68)

    1. a. il  tic tac  dell'orologio  Italian

      the  idph  of.the clock

      ‘the ticking of the clock’

    2. b. el  xiu xiu  del  vent  Catalan

      the  idph  of.the  wind

      ‘the whisper of the wind’

    3. c. vaq vaq-e  sag  Persian

      ideo-lnk  dog

      ‘The bark of the dog’

In Italian and Catalan, the ideophone noun is accompanied by a genitive phrase introduced by the preposition for ‘of’ (di/de). In the Persian example, the ideophone bears the ezafe morpheme, which introduces genitive DPs as well as adjectival modifiers (Ghomeshi Reference Ghomeshi1997, Samvelian Reference Samvelian2007, Kahnemuyipour Reference Kahnemuyipour2014, among others).

Similar construction types have been observed in other languages. Consider (69) from Finnish, and (70) from the Amazonian language Macushi.

  1. (69) Puu-n  kanttura

    tree-gen  idph.nom

    ‘a tree bent by wind, weight of snow, etc.’

    Finnish (Armoskaite and Koskinen Reference Armoskaite and Koskinen2014)

  2. (70)

    1. a. sarai sarai  u-saraisara-ri

      ‘combing’  1sg-comb-possn

      ‘my comb’

    2. b. kiri kiri  u-kirikiri-ri

      ‘filing action’  1sg-file-possn

      ‘my file’  Macushi (Abbott Reference Abbott, Derbyshire and Pullum1991: 150)

To my knowledge, the only syntactic analysis so far proposed for such structures is that of Armoskaite and Koskinen (Reference Armoskaite and Koskinen2014) for Finnish. They point out that in constructions such as (69), the merging of a nominative ideophone kanttura with a genitive noun, Puu-n ‘tree-gen’ triggers an evaluative (pejorative, in the example) interpretation. They show that the [NGen NidphNom] complex behaves as a morphosyntactic unit in Finnish, and it is possible to modify the entire string, but not each individual noun, and when the compound inflects for number, only one suffix is allowed.

Their evaluative analysis for Finnish seems not to work for Romance, or for Iranian languages. Consider the Italian example in (71).

  1. (71) il ciuf-ciuf leggero delle vecchie macchine a vapore

    ‘the light idph of the old steam machines’

Here the string [NIDPHNom NGen ] has no evaluative meaning, and both the ideophone and the genitive DP can take independent modifiers. The example in (71) therefore cannot be interpreted as a single morphosyntactic unit. Note that Italian allows prepositional compounds forming a N-prep-N unit, such as topo di biblioteca ‘bookworm’, lit. ‘mouse of library’; colpo di fulmine ‘love at first sight’, lit. ‘bolt of lightning’, (Delfitto and Melloni Reference Delfitto and Melloni2009, Franco et al. Reference Franco, Zampieri and Meneghello2013). Nevertheless, Armoskaite and Koskinen (Reference Armoskaite and Koskinen2014) argue correctly that – in our terms – syntactically, the sound-symbolic [NGen NIDPHNom] string in Finnish is a kind of possessive construction.

Freeze (Reference Freeze1992), Kayne (Reference Kayne1993) den Dikken (Reference den Dikken1995), Larson and Choi (Reference Larson and Cho2003), among others, propose that possessive sentences with ‘have’ are built on an underlying small clause predication relation between a possessor and possessee (see also Barwise and Cooper Reference Barwise and Cooper1981, Partee and Borschev Reference Partee, Borschev, Lang, Maienborn and Fabricius-Hansen2003, among others, for relevant semantic characterizations of genitive/part-whole relations). Following this characterization, which can be traced back to Benveniste (Reference Benveniste1966) and Fillmore (Reference Fillmore, Bach and Harms1968), Freeze (Reference Freeze1992) proposes that possessive have constructions such as ‘John has a bike’ are composed underlyingly from the copula be and a part-whole-like/locative prepositional phrase as in ‘the bike of John’. A possible derivation for (68a) is given in (72), roughly based on Freeze (Reference Freeze1992). For our current purposes, I will assume that it is basically correct, but see Levinson (Reference Levinson2011) for Icelandic data that may weaken Freeze's account. I assume that the ‘inclusion’ (part-whole) relation ⊆, introduced above for the verbal domain, plays a role in nominal contexts as well. See Manzini and Savoia (Reference Manzini and Savoia2011b), for morphological evidence based on the widely attested genitive/dative syncretism that is present in many genetically diverse languages.

  1. (72)

A sentential derivation of a light verb-ideophone complex predicate such as l'orologio fa tic tac ‘the clock goes tick tock’ based on the configuration in (72) would have the structure in (73).

  1. (73)

In (73) the part-whole preposition incorporates into verbal element be. The possessive light verb have spells out the be + of (to) complex. The ideophonic morpheme, following Ramchand (Reference Ramchand2008), would then underassociate its N feature, which allows it to merge with a D item, and appear as a bare (non-argumental) root in the structure.

While there are some languages, such as Muna, illustrated in (21) and (22), in which complex ideophonic predicates are introduced by a have-like verb, the vast majority of the languages of the world use a do/make or a say light verb.Footnote 26 However, the ‘possessive interpretation’ of even sentential ideophones finds support in the fact that there is a clear semantic link between (inalienable) possession and verbs of internal sound emission such as whistle, hum, squeak, click, etc. As shown in the cross-linguistic overview, these verbs are often realized by the means of a light verb + ideophone compound. There is a physical constraint: the need to possess certain characteristics in order to produce the sound rendered by a given verbal predicate. According to Folli and Harley (Reference Folli and Harley2008: 192; see also Levin and Rappaport Hovav Reference Levin and Hovav1995, Levin Reference Levin2012) inanimate DPs also obey such physical ‘possession’ constraints and “it seems clear that although trains and tables are not animate entities, they have properties internal to their construction that makes them appropriate or typical whistlers and squeakers. Trains, in fact, are built with whistles in them, and tables that squeak do so by virtue of their physical characteristics.”

With an eventive perspective similar to the one developed here, they further assume that in such cases “Agents are entities which can produce particular events by themselves: they are sufficient on their own to initiate and carry out the entire event denoted by the predicate” (Folli and Harley Reference Folli and Harley2008: 192). Thus, given their dual nature of owners on the one hand and initiators/undergoers on the other, it is conceivable that there might be languages that use the light verb for ‘have’ to introduce ideophonic complex predicates. The ‘do’ strategy is, in any case, far more common.

Interesting evidence for a link between possession and ideophones can be found in languages like Great Andamanese (Abbi Reference Abbi2011: 765) where ideophonic sound/light emission items are treated as equivalent to a product of the body, appearing obligatorily with a ‘body class marker’; that is, a specialized classifier that classifies items based on the part of the body that they occupy. An analogous pattern seems to hold in Belhare (Eastern Kiranti, Tibeto-Burman) where body-products such as hildkpa ‘hiccup’ or gauppa ‘burp’ are derived from or interpreted as ideophones (Bickel Reference Bickel and Bradley1997: 140). While body-part terms are commonly treated as possessed, the possessor may also be promoted to be the external argument of the sentence (Wierzbicka Reference Wierzbicka1999), as in the Vietnamese examples in (74). Italian works similarly, as shown in (75).

  1. (74)

    1. a. bụng  tôi  (ḅi)  đau

      stomach  1SG  suffer  sick

      ‘My stomach aches’

    2. b. Tôi  (ḅi)  đau  bụng

      1SG  suffer  sick  stomach

      ‘I ache/am sick in the stomach.’ (lit. ‘I stomach-ache’)

      (Clark Reference Clark, Chappell and McGregor1995: 544, cited in Vittrant Reference Vittrant and Williams2013)

  2. (75)

    1. a. La testa mi fa male.

      ‘My head aches.’ (lit. ‘the head does pain to me’)

    2. b. Ho mal di testa.

      ‘I have headache.’ (lit. ‘I have pain of head’)

This example illustrates an alternation between fare ‘do’ in (75a) and avere ‘have’ in (75b), a further indication that the possession hypothesis may be on the right track. Further exploration of this issue is left for future research.

4. Conclusion

I have shown that verbal compound constructions containing an ideophone and a light verb are crosslinguistically widespread, and that phono-symbolic morphemes cannot be treated as bare direct objects in such constructions. Ideophones appearing adjacent to the light verb form a semantic unit with the verbal predicate, although in some languages they can be syntacticized as bare nouns and appear in argumental position. Specifically, I have argued that ideophones in complex predicates are part of the verbal domain, forming a single predicate with the verb through a mechanism of conflation, along the lines of Hale and Keyser (Reference Hale, Keyser, Hale and Keyser1993, Reference Hale and Keyser2002), and building on Ramchand (Reference Ramchand2008). I tentatively sketched in section 3.2 a possible characterization of the syntax of nominal ideophones in terms of a possessum-possessor relation.

Ideophones are an intriguing topic and many aspects of their lexical and syntactic behaviour require further investigation. Recent experimental work, for instance, has shown that three-year-old children learned non-words words better when the form and meaning of the words were ideophonically related. Such results were obtained both with children learning Japanese, a language with a rich inventory of sound symbolic expressions, and English, which has a smaller number of ideophones (Imai et al. Reference Imai, Kita, Nagumo and Okada2008, Kantartzis et al. Reference Kantartzis, Imai and Kita2011, among others). This kind of sound-symbolic ‘assistance’ has been thought to be an evolutionary device. That is, children's ability to use sound symbolism in acquisition would be the vestige of a protolanguage (Bickerton Reference Bickerton1990, Jackendoff Reference Jackendoff1999, Hurford Reference Hurford2011, among others) consisting largely of ideophones. On this, McGregor (Reference McGregor2002: 335ff) (see also Heat Reference Heath and Dixon1976, Givón Reference Givón2009), investigating diachronic processes involving ideophones in Northern Australian languages, proposes that the uninflecting parts of complex predicate constructions have consistently originated from ideophones. Surely, phono-symbolic expressions also deserve more attention in the realm of formal linguistics.

Footnotes

*

I wish to thank audiences at CLUNL, Universidade Nova de Lisboa, and CISCL, Università di Siena, where parts of this article were presented — in particular, Adriana Belletti, Joao Costa, Oana Lungu and Luigi Rizzi. I also wish to thank Antonio Civardi, Rita Manzini, Leonardo Savoia and two anonymous reviewers of CJL for their extremely helpful comments and suggestions. All errors are my own. Special thanks are due to the editors of this special issue, Solveiga Armoskaite and Päivi Koskinen, for their encouragement. I gratefully acknowledge the Portuguese National Science Foundation, Fundação para a Ciência e a Tecnologia (FCT), for supporting this work with the research grant IF/00846/2013.

1 The following abbreviations are used (note that original glosses have been retained in examples from different sources): 1sg: 1st person singular; 3: 3rd person; 3p: 3rd person plural; 3pl: 3rd person plural; 3sg: 3rd person singular; 3sm: 3rd person masculine singular; 3sr: 3rd person singular; 7,10: noun class markers; a: agent-like marker; acc: accusative; Appl: applicative; art: article; auto: autobenefactive/spontaneous; aux: auxiliary; cl: clitic ; clf: classifier; cnv: converb; cons: consecutive; cont: continuous; conv: converb; cvant: anterior converb; cvb: converb; d: gender agreement marker; dat: dative; decl: declarative; des: desiderative; dem: demonstrative; dim: diminutive; dom: differential object marking; DP: determiner phrase; dur: durative; emph: emphasis; erg: ergative; f: feminine; fact: factual; fp: far past; fs: feminine singular; fut: future; g: gender; gen: genitive; idph: ideophone; if: immediate future; inch: inchoative; ind: independent; init: initiator; ipfv: imperfective; lit: literally; lnk: linker; loan: loanword; loc: locative; lv: light verb; m: masculine; med: medial; n: neuter; nar: narrative; nmz: nominalizer; nom: nominative; NP: noun phrase; pass: passive; past: past; per: personal-factual evidential; pf: perfective; pfv: perfective; pl: plural; poss: possessive; possn: possession; PP: prepositional phrase; proc: process; prog: progressive; prs: present; prx: proximal; pst: past; ptc: participle; ptcp: participle; rel: relative pronoun; rep: reported evidential; res: result; sbj: subject; sc: subject class marker; sbrd: subordinator; seq: sequential; Spec: specifier; sg: singular; ss: same-subject marker; TAM: tense/aspect/mood; testim: testimonial; todp: today past; top: topic; vis: visual-sensory evidential; VP: verb phrase; wp: witnessed past tense; yestp: yesterday past.

2 The data reported in this section are drawn from comprehensive descriptive grammars (e.g., Croom Helm/Routledge Descriptive Grammars, Mouton Grammars, PhD dissertations, etc.) and articles with a typological focus, published up to 2014. For many languages, the relevant data are not available, as ideophones are often not included in the average reference grammar. A language is included in the survey only if unambiguous data were available concerning the ideophonic nature of the items entering complex predicates. All the items glossed as idph in this paper are glossed/interpreted as such in the original sources.

3 A light verb for do (wo:-) merges with ideophones also in the cognate language Udihe, as shown in (i), taken from Nikolaeva and Tolskaya (Reference Nikolaeva and Tolskaya2001: 642).

  1. (i) modor modor wo:-ini.

    idph:slobber slobber  make-pst.3sg

    ‘He slobbered.’

4 Note that a very similar pattern has been reportedfor the Trans New-Guinea language Abui, in which the verb ba ‘say’normally follows onomatopoeic lexemes (Kratochvíl Reference Kratochvíl2007).

5 Another isolate language of Central Bolivia, Yurakaré (cf. Van Gijn Reference van Gijn2006, Reference van Gijn2010: 283) presents a large inventory of ideophones accompanied by the light verb ta roughly meaning ‘say’. The same morpheme appears, interestingly, in Yurakaré as a middle ‘intransitive’ marker, contrasting with the causative morphemes li/che. Consider the examples in (i):

  1. (i)

    1. a. porrok  porrok  ta-ø=w=ya.

      idph  idph  say-3=pl=rep

      ‘“porrok porrok”, they went (hoofs of tapirs).’

    2. b. dürrüm  ta-ø=ya

      idph  say-3=rep

      ‘“Brooom”, it went.’ (something heavy falling on the ground)

6 According to Sakel (Reference Sakel2004, Reference Sakel, Miestamo and Wälchli2007), in Mosetén, verbness markers can appear with unmarked stems, (exactly as in (25)) with the ‘bare’ item wai’ ‘boil’, as well as with adjectives, adverbs, nouns and demonstratives. As Sakel (Reference Sakel, Miestamo and Wälchli2007: 321) points out, Mosetén unmarked stems usually occur with a verbness marker, forming a complex predicate, but they can also appear on their own as ideophones. This pattern is not permitted with the other lexical classes, which merge with the verbness marker.

7 For a detailed list of Catalan ideophones see Hualde (Reference Hualde1992: 413–415).

8 Another recently described Iranian language that employs the light verb–ideophone construction is Juhuri, spoken in Azerbaijan, Dagestan and Israel (Authier Reference Authier2013: 242). Ideophones (intransitively) combine with the dummy verbs soxde ‘do’ or zere ‘hit’, as in (i).

  1. (i)

    1. a. seg  ħovħov  sox-de.

      dog  idph  does

      ‘the dog barks’

    2. b. qiloq  qärqär  ze-re.

      raven  idph  hits

      ‘the raven croaks’

    3. c. xuruz  ʕüʕü ze-re.

      cock  idph  hits

      ‘the cock says cock-a-doodle-doo’

9 Karimi (Reference Karimi1997: 276) argues that the number of verbs in Persian does not exceed 115, and many of them are infrequent. Haig (Reference Haig2000) shows that verbs in Kurdish form a closed class, based on a corpus study in which about 60 verbs account for over 96% of all verb tokens (see also Schultze-Berndt Reference Schultze-Berndt2000).

10 Plank notes that it has been suggested that the historical source of the delocutive verbs employed with ideophones might be a verb meaning ‘to do, make’. For instance, in Hindi-Urdu and other modern Indo-Aryan languages, ideophones normally enter the verbal domain by means of a suffix, such as Hindi jhaTa-k ‘to make the sound jhaTat’. This suffix has been interpreted as deriving from the light verb kr- in Sanskrit (see also Deo Reference Deo2002), which originally meant ‘do/make’ and ‘generally served to verbalize non-verbs’ (Plank Reference Plank2005: 480).

11 Regarding possible split intransitivity patterns (i.e., unergative vs. unaccusative), we may say that – at least for the languages for which we have primary data – ideophonic complex predicates appear to be consistently unergative. For instance, in Italian, they select the auxiliary avere ‘have’ as unergatives do, whereas unaccusative predicates commonly select essere ‘to be’ (Burzio Reference Burzio1986, among many others).

12 On this, see (33c) above and the discussion in section 3.2 below.

13 Megerdoomian (Reference Megerdoomian2012:189) has shown that the interpretation of the verb and of the preverbal item is very different with heavy verbs and with light verbs, even though certain verbs can be either heavy or light. Considering xordæn (eat), the preverbal item in sentence (38a) “corresponds to an entity that undergoes the action denoted by the verb (i.e., they are being consumed).” These nouns must satisfy the selectional restrictions of the verbal element. The verb xordæn in (38a) is thus a thematic verb meaning ‘to eat’ and it can appear with “any noun that refers to edible entities.” In contrast, this is not true of the preverbal element in (38b), which has an ‘idiomatic’ reading.

14 The grammaticality judgements were provided by six native speakers of Persian (age range 14–67).

15 These judgements were provided by native speakers of Italian.

16 An anonymous reviewer suggests that passivization could be another useful test to show that ideophones are not internal arguments. Actually, passivization is not included here because many authors (Moyne Reference Moyne1974, Karimi Reference Karimi2005, Folli et al. Reference Folli, Harley and Karimi2005, among others) have suggested that Persian lacks passive constructions entirely. Nevertheless, in Romance languages, the passivization of ideophones adjoined to light verbs as bare roots does not appear to be grammatical, as expected. Consider the Italian examples in (i).

  1. (i)

    1. a. l'acqua ha fatto glu glu.  active voice

      ‘the water flowed’ (lit. the water has done idph)

    2. b. *glu glu è stato fatto dall'acqua  passive voice

17 Italian ideophones are also very unlikely to appear with markers of specificity/definiteness; they cannot normally be introduced by definite determiners when they are merged with light verbs (e.g., *il cane fa il bau bau, ‘the dog barks’, lit. ‘the dog does the idph’; *il treno fa il ciuf ciuf, ‘the train chugs’, lit. ‘the train does the idph’).

18 According to Ramchand (Reference Ramchand2008), when part of the structure in a lexical item is unused for the spell-out of a given syntactic structure, that unused piece is “underassociated”.

19 In recent work, Haugen (Reference Haugen2009) has provided evidence that Incorporation and Conflation are two distinct ways of forming denominal verbs. He shows that Incorporation is conceived of as head movement (Baker Reference Baker1988), and is instantiated through the syntactic operation of Copy, whereas Conflation is instantiated directly through Merge (compounding). Under this view, it may be assumed that ideophone roots are directly merged with light verbs via Conflation.

20 This is a very productive process in Hindi and elsewhere in Indo-Aryan languages, not only with ideophones but also with other lexical categories.

21 These properties represent a possible linguistic universal, (Childs Reference Childs, Hinton, Nichols and Ohala1994).

22 Interestingly, as shown in Ibarretxe-Antuñano (Reference Ibarretxe-Antuñano2015), some Basque ideophones (e.g., arrast ‘drag’) are likely Spanish loanwords, but due to their phonological markedness, they pattern with ideophones in Basque. Conversely, some Basque ideophones have been borrowed into Spanish: for instance, sirimiri ‘drizzle’ from Basque zirimiri (txirimiri) or pilpil ‘sound of boiling water’ from Basque pil-pil.

23 Pylkkänen (Reference Pylkkänen2008) claims that the presence of a causee – the specifier of VInit in (57) – in languages such as Finnish (which allows causativized unergatives), is possible because in such contexts it is realized in the specifier position of a projection CauseP. According to her analysis, this [Spec, Cause] position is available only when Voice and Cause appear in the syntax of a given language as independent heads. In the case of English, this is not possible because Voice and Cause are bundled on a single head, making it impossible for Cause to have its own Spec (see also Tubino Blanco Reference Tubino Blanco2011).

24 Ramchand (Reference Ramchand2008) assumes that the addition of the -aa morpheme to unergative items leads to underassociation of the init feature of the root. Thus, the structure of causativized transitives would still involve a configuration with both init and proc. The simplified representation in (62), however, is sufficient to illustrate the basic facts introduced here.

25 See Manzini and Franco (to appear) for an alternative implementation by which the ‘dative’ preposition a ‘to’ has as its primitive predicational content an ‘inclusion’ (part-whole) relation. See the discussion below.

26 An anonymous reviewer suggests that it is unlikely that a light verb meaning ‘hit’, used in combination with ideophones in Iranian languages (see (32) and fn. 9), could appear in the structure in (73). Nevertheless, the grammaticalization path from ‘hit’ verbs to light verbs and auxiliaries is widely attested. Heine and Kuteva (Reference Heine and Kuteva2002: 51) describe such cases in Swahili and Ewe, where a frequently used action verb turns into a semantically empty predicate marker. A Swahili example is given in (i).

  1. (i) Swahili ku-piga ‘to beat’, ‘to hit’,        verb > empty predicate marker

    a. ku-piga picha b. ku-piga kelele

    to-hit picture to-hit noise

    ‘to make a photo’ ‘to make noise’

References

Abbi, Anvita. 1992. Reduplication in South Asian languages. New Delhi: Allied Publishers Limited.Google Scholar
Abbi, Anvita. 2011. Body divisions in Great Andamanese: Possessive classification, the semantics of inherency and grammaticalization. Studies in language 35(4): 739–792.CrossRefGoogle Scholar
Abbott, Miriam. 1991. Macushi. In Handbook of Amazonian languages 3, ed. Derbyshire, Desmond C. and Pullum, Geoffrey K., 23–160. Berlin: Mouton de Gruyter.Google Scholar
Alpher, Bary. 1994. Yir-Yiront ideophones. In Sound symbolism, ed. Hinton, Leanne, Nichols, Johanna, and Ohala, John, 161–177. Cambridge: Cambridge University Press.Google Scholar
Alpher, Barry. 2001. Ideophones in interaction with intonation and the expression of new information in some indigenous languages of Australia. In Ideophones, ed. Erhard Voeltz, F. K. and Kilian-Hatz, Christa, 9–24. Amsterdam: John Benjamins.Google Scholar
Alsina, Alex, Bresnan, Joan, and Sells, Peter, eds. 1997. Complex predicates. Stanford, CA: CSLI Publications.Google Scholar
Amberber, Mengistu. 1996. Transitive alternations, event-types and light verbs . Doctoral dissertation, McGill University.Google Scholar
Amha, Azeb. 2010. Compound verbs and ideophones in Wolaitta revisited. In Complex predicates: Cross-linguistic perspectives on event structure, ed. Amberber, Mengistu, Baker, Bret, and Harvey, Mark, 259–290. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Armoskaite, Solveiga, and Koskinen, Päivi. 2014. Finnish nominal ideophones as evaluatives. Paper presented at “Structuring sensory imagery: Ideophones across languages & cultures”, University of Rochester, Rochester, NY.Google Scholar
Authier, Gilles. 2013. Grammaire Juhuri, ou Judéo-Tat, langue iranienne des Juifs du Caucase de l'est. Wiesbaden: Reichert Verlag.Google Scholar
Baker, Mark. 1988. Incorporation: A theory of grammatical function changing. Chicago: University of Chicago Press.Google Scholar
Barwise, Jon, and Cooper, Robin. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4(2): 159–219.Google Scholar
Belvin, Robert, and den Dikken, Marcel. 1997. There, happens, to, be, have . Lingua 101(3/4): 151–183.Google Scholar
Benveniste, Émile. 1966. Problèmes de linguistique générale. Paris: Gallimard.Google Scholar
van den Berg, René. 1989. A grammar of the Muna language. Dordrecht: Foris.Google Scholar
Berro, Ane. 2012. Three levels of root insertion in Basque intransitive verbs. Journal of Portuguese Linguistics 11(1): 7–21.Google Scholar
Bhatt, Rajesh, and Embick, David. 2004. Causative derivations in Hindi. Ms., University of Texas at Austin and University of Pennsylvania.Google Scholar
Bickel, Balthasar. 1997. The possessive of experience in Belhare. Papers in Southeast Asian linguistics 14, ed. Bradley, David, 135–155. Camberra: Pacific Linguistics.Google Scholar
Bickerton, Derek. 1990. Language and species. Chicago: University of Chicago Press.Google Scholar
Bosque, Ignacio, and Gallego, Ángel J.. 2014. Reconsidering subextraction: evidence from Spanish. Borealis: An International Journal of Hispanic Linguistics 3(2): 223–258.Google Scholar
Bowern, Claire. 2008. The reconstruction of Nyulnyulan complex predication. Diachronica 25(2): 186–212.Google Scholar
Bradshaw, Joel. 2006. Grammatically marked ideophones in Numbami and Jabêm. Oceanic Linguistics 45(1): 53–63.Google Scholar
Burzio, Luigi. 1986. Italian syntax: A government and binding approach. Dordrecht: Reidel.CrossRefGoogle Scholar
Cagri, Ilhan. 2007. Persian accusative case: A lexicalist approach. Paper presented at Second International Conference on Iranian Linguistics (ICIL2). University of Hamburg.Google Scholar
Childs, Tucker. 1994. African ideophones. In Sound symbolism, ed. Hinton, Leanne, Nichols, Johanna, and Ohala, John, 178–209. Cambridge: Cambridge University Press.Google Scholar
Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris.Google Scholar
Chomsky, Noam. 2001. Derivation by phase . In Ken Hale: A life in language, ed. Kenstowicz, Michael, 1–54. Cambridge, MA: MIT Press.Google Scholar
Cinque, Guglielmo. 2004. Restructuring and functional structure. In Structures and beyond: The cartography of syntactic structures 3, ed. Belletti, Adriana, 132–191. New York: Oxford University Press.CrossRefGoogle Scholar
Clark, Marybeth. 1995. Where do you feel? Stative verbs and body-part terms in Mainland Southeast Asia. In The grammar of inalienability, ed. Chappell, Hilary and McGregor, William, 529–564. Berlin: Mouton de Gruyter.Google Scholar
Creissels, Denis. 1997. The auxiliarization of re ‘say’ in Setswana. In Proceedings of the Twenty-Third Annual Meeting of the Berkeley Linguistics Society: Special Session on Syntax and Semantics in Africa, 59–70. Berkeley: Berkeley Linguistics Society.Google Scholar
Creissels, Denis. 2001. Setswana ideophones as uninflected predicative lexemes. In Ideophones, ed. Erhard Voeltz, F. K. and Kilian-Hatz, Christa, 75–85. Amsterdam: John Benjamins.Google Scholar
Cuervo, María Cristina. 2003. Datives at large . Doctoral dissertation. MIT, Cambridge, MA.Google Scholar
Delfitto, Denis, and Melloni, Chiara. 2009. Compounds don't come easy. Lingue e Linguaggio 8(1): 75–104.Google Scholar
Deo, Ashwini. 2002. A diachronic perspective on complex predicates in Indo-Aryan. Paper presented at the Workshop on complex predicates, particles and subevents, Universität Konstanz.Google Scholar
Derbyshire, Desmond C. 1977. Discourse redundancy in Hixkaryana. International Journal of American Linguistics 43(3): 176–188.Google Scholar
Dhoorre, Cabdulqaadir Salaad, and Tosco, Mauro. 1998. 111 Somali ideophones. Journal of African Cultural Studies 11(2): 125–156.Google Scholar
den Dikken, Marcel. 1995. Particles: On the syntax of verb-particle, triadic, and causative constructions. New York: Oxford University Press.Google Scholar
Dingemanse, Mark. 2012. Advances in the cross-linguistic study of ideophones. Language and Linguistics Compass 6(10): 654–672.CrossRefGoogle Scholar
Emeneau, Murray B. 1980. Language and linguistic area: Essays by Murray B. Emeneau, selected and introduced by Dil, Anwar S.. Stanford, CA: Stanford University Press.Google Scholar
Enfield, Nick J. 2007. A grammar of Lao. Berlin: Mouton de Gruyter.Google Scholar
Evans, Nicholas, and Levinson, Stephen C.. 2009. The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and Brain Sciences 32(5): 429–48.Google Scholar
Fedden, O. Sebastian. 2007. A grammar of Mian: a Papuan language of New Guinea . Doctoral dissertation, University of Melbourne.Google Scholar
Fedden, Sebastian. 2011. A grammar of Mian. Berlin: De Gruyter Mouton.Google Scholar
Fillmore, Charles J. 1968. The case for case. In Universals in linguistic theory, ed. Bach, Emmon and Harms, Robert T., 1–88. New York: Holt, Rinehart, and Winston.Google Scholar
Fleck, David. 2003. A grammar of Matses . Doctoral dissertation, Rice University.Google Scholar
Folli, Raffaella, Harley, Heidi, and Karimi, Simin. 2005. Determinants of event type in Persian complex predicates. Lingua 115(10): 1365–1401.Google Scholar
Folli, Raffaella, and Harley, Heidi. 2007. Causation, obligation, and argument structure: On the nature of little v. Linguistic Inquiry 38(2): 197–238.Google Scholar
Folli, Raffaella, and Harley, Heidi. 2008. Teleology and animacy in external arguments. Lingua 118(2): 190–202.Google Scholar
Franco, Ludovico, Zampieri, Elisa, and Meneghello, Francesca. 2013. Prepositions inside (and at the edge) of words: A view from agrammatism. Language Sciences 40: 95–122.Google Scholar
Franco, Ludovico. 2015. The morpho-syntax of adverbs of the carpone/i type in (old and modern) Italian. Probus 27(2): 271–306.Google Scholar
Freeze, Ray. 1992. Existentials and other locatives. Language 68(3): 553–595.Google Scholar
Gallego, Ángel J. 2012. A note on cognate objects: cognation as doubling. Nordlyd 39(1): 95–112.CrossRefGoogle Scholar
Ghomeshi, Jila, and Massam, Diane. 1994. Lexical/syntactic relations without projections. Linguistic Analysis 23: 175–217.Google Scholar
Ghomeshi, Jila. 1997. Non-projecting nouns and the ezafe construction in Persian. Natural Language and Linguistic Theory 15(4): 729–788.Google Scholar
Ghomeshi, Jila. 2003. Plural marking, indefiniteness and the noun phrase. Studia Linguistica 57(2): 47–74.Google Scholar
van Gijn, Rik. 2006. A grammar of Yurakaré . Doctoral dissertation, Radboud University Nijmegen.Google Scholar
van Gijn, Rik. 2010. Middle voice and ideophones, a diachronic connection. The case of Yurakaré. Studies in Language 34(2): 273–297.CrossRefGoogle Scholar
Givón, Talmy. 2009. The genesis of syntactic complexity. Amsterdam: John Benjamins.Google Scholar
Guasti, Maria Teresa. 1996. Semantic restrictions in Romance causatives and the incorporation approach. Linguistic Inquiry 27(2): 294–313.Google Scholar
Güldemann, Tom. 2008. Quotative indexes in African languages: A synchronic and diachronic survey. Berlin: Mouton de Gruyter.Google Scholar
Haig, Geoffrey. 2000. Anatomy of a closed word class: Frequency, regularity and productivity of verbs in Kurdish. Paper presented at the 22nd Jahrestagung der Deutschen Gesellschaft für Sprachwissenschaft, Phillips-Universität, Marburg.Google Scholar
Hale, Kenneth, and Keyser, Samuel Jay. 1993. On argument structure and the lexical expression of syntactic relations. In The view from Building 20: Essays in honor of Sylvain Bromberger, ed. Hale, Kenneth and Keyser, Samuel Jay, 53–109. Cambridge, MA: MIT Press.Google Scholar
Hale, Kenneth, and Keyser, Samuel Jay. 2002. Prolegomenon to a theory of argument structure. Cambridge, MA: MIT Press Google Scholar
Hamano, Shoko. 1998. The sound-symbolic system of Japanese. Stanford, CA: CSLI Publications.Google Scholar
Harley, Heidi. 1996. Subjects, events, and licensing . Doctoral dissertation. MIT, Cambridge, MA.Google Scholar
Harley, Heidi. 2002. Possession and the double-object construction. Linguistic Variation Yearbook 2: 29–68.Google Scholar
Harley, Heidi. 2005. How do verbs get their names? Denominal verbs, manner incorporation, and the ontology of verb roots in English. In The syntax of aspect: deriving thematic and aspectual interpretation, ed. Erteschik-Shir, Nomi and Rapoport, Tova, 42–64. Oxford: Oxford University Press.Google Scholar
Harris, Alice C. 2008. Light verbs as classifiers in Udi. Diachronica 25: 213–241.Google Scholar
Haugen, Jason D. 2009. Hyponymous objects and late insertion. Lingua 119(2): 242–262.Google Scholar
Heath, Jeffrey. 1976. Topic E: Simple and compound verbs: Conjugation by auxiliaries in Australian verbal systems. North-east Arnhem Land. In Grammatical categories in Australian languages, ed. Dixon, R.M.W., 735–740. Canberra: Australian Institute of Aboriginal Studies.Google Scholar
Heine, Bernd, and Kuteva, Tania. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Hinton, Leanne, Nichols, Johanna, and Ohala, John J., eds. 1994. Sound symbolism. Cambridge: Cambridge University Press.Google Scholar
Hirose, Masayoshi. 1981. Japanese and English contrastive lexicology: The role of Japanese “mimetic adverbs” . Doctoral dissertation, University of California, Berkeley.Google Scholar
Horvath, Julia and Siloni, Tal. 2011. Causatives across components. Natural Language and Linguistic Theory 29(3): 657–704.Google Scholar
Hualde, José Ignacio. 1992. Catalan (Descriptive Grammars). London: Routledge.Google Scholar
Hurford, James. 2011. The origins of grammar. Oxford: Oxford University Press.Google Scholar
Ibarretxe-Antuñano, Iraide. 2004. Tipi-tapa, tipi-tapa korrika!!! Motion and sound symbolism in Basque. LAUD Series A: General and Theoretical Papers, No. 629.Google Scholar
Ibarretxe-Antuñano, Iraide. 2015. Basque ideophones from a typological perspective. Ms., University of Zaragoza.Google Scholar
Imai, Mutsumi, Kita, Sotaro, Nagumo, Miho, and Okada, Hiroyuki. 2008. Sound symbolism facilitates early verb learning. Cognition 109(1): 54–65.Google Scholar
Jackendoff, Ray. 1990. Semantic structures. Cambridge, MA: MIT Press.Google Scholar
Jackendoff, Ray. 1999. Possible stages in the evolution of the language capacity. Trends in Cognitive Sciences 3(7): 272–279.Google Scholar
Jacques, Guillaume. 2013. Ideophones in Japhug (Rgyalrong). Anthropological Linguistics 55(3): 256–287.Google Scholar
Kachru, Yamuna. 2006. Hindi. Amsterdam: John Benjamins.Google Scholar
Kahnemuyipour, Arsalan. 2014. Revisiting the Persian ezafe construction. A roll-up movement analysis. Lingua 150: 1–24.Google Scholar
Kantartzis, Katerina, Imai, Mutsumi, and Kita, Sotaro. 2011. Japanese sound symbolism facilitates word learning in English speaking children. Cognitive Science 35(3): 575–586.Google Scholar
Karimi, Simin. 1997. Persian complex verbs: Idiomatic or compositional. Lexicology 3(2): 273–318.Google Scholar
Karimi, Simin. 2005. A minimalist approach to scrambling: Evidence from Persian. Berlin: Mouton de Gruyter.Google Scholar
Kayne, Richard. 1975. French syntax. Cambridge, MA: MIT press.Google Scholar
Kayne, Richard. 1993. Toward a modular theory of auxiliary selection. Studia Linguistica 47(1): 3–31.Google Scholar
Kayne, Richard. 2009. Antisymmetry and the lexicon. Linguistic Variation Yearbook 8: 1–31.Google Scholar
Kilian-Hatz, Christa. 2001. Universality and diversity. Ideophones from Baka and Kxoe. In Ideophones, ed. Erhard Voeltz, F. K. and Kilian-Hatz, Christa, 155–163. Amsterdam: John Benjamins.Google Scholar
Klamer, Marian. 1998. A grammar of Kambera. Berlin: Mouton de Gruyter.Google Scholar
Klamer, Marian. 2002. Semantically motivated lexical patterns: A study of Dutch and Kambera expressives. Language 78(2): 258–286.Google Scholar
Kratochvíl, František. 2007. A grammar of Abui, a Papuan language of Alor . Doctoral dissertation. Universiteit Leiden.Google Scholar
Krishnamurti, Bhadriraju. 2003. The Dravidian languages. Cambridge: Cambrige University Press.Google Scholar
Larson, Richard, and Cho, Sungeun. 2003. Temporal adjectives and the structure of possessive DPs. Natural Language Semantics 11(3): 217–247.Google Scholar
Levin, Beth. 2012. Slap, give a slap, slap a slap: Crosslinguistic diversity in hitting event descriptions. Hand-out of a talk given at Tenth Biennial Conference of the High Desert Linguistics Society, University of New Mexico, Albuquerque.Google Scholar
Levin, Beth, and Hovav, Malka Rappaport. 1995. Unaccusativity. Cambridge, MA: MIT Press.Google Scholar
Levin, Beth, and Hovav, Malka Rappaport. 2005. Argument realization. Cambridge: Cambridge University Press.Google Scholar
Levinson, Lisa. 2011. Possessive with in Germanic: have and the role of P. Syntax 14(4): 355–393.Google Scholar
Loughnane, Robyn. 2009. A grammar of Oksapmin . Doctoral dissertation, University of Melbourne.Google Scholar
Lundquist, Björn. 2008. Nominalizations and participles in Swedish . Doctoral dissertation, University of Tromsø.Google Scholar
Mahootian, Shahrzad. 1997. Persian. London: Routledge.Google Scholar
Manzini, M. Rita, and Franco, Ludovico. 2016. Goal and DOM datives. Natural Language and Linguistic Theory 34: 197. doi:10.1007/s11049-015-9303-y.Google Scholar
Manzini, M. Rita, and Savoia, Leonardo M.. 2011a. Grammatical categories. Cambridge: Cambridge University Press.Google Scholar
Manzini, M. Rita, and Savoia, Leonardo M.. 2011b. Reducing ‘case’ to denotational primitives: Nominal inflections in Albanian. Linguistic Variation 11(1): 76–120.Google Scholar
Marantz, Alec. 1997. No escape from syntax: Don't try morphological analysis in the privacy of your own lexicon. University of Pennsylvania Working Papers in Linguistics 4: 201–225.Google Scholar
Mateu, Jaume. 2002. Argument structure: Relational construal at the syntax-semantics interface . Doctoral dissertation, Universitat Autònoma de Barcelona.Google Scholar
Matić, Dejan and Pakendorf, Brigitte. 2013. Non-canonical say in Siberia: Areal and genealogical patterns. Studies in Language 37(2): 356–412.Google Scholar
Matisoff, James. 1994. Tone, intonation, and sound symbolism in Lahu: Loading the syllable canon. In Sound symbolism, ed. Hinton, Leanne, Nichols, Johanna, and Ohala, John, 115–129. Cambridge: Cambridge University Press.Google Scholar
McGregor, William B. 2002. Verb classification in Australian languages. Berlin: Mouton de Gruyter.CrossRefGoogle Scholar
Megerdoomian, Karine. 2012. The status of the nominal in Persian complex predicates. Natural Language and Linguistic Theory 30(1): 179–216.Google Scholar
Merlan, Francesca. 2001. Form and context in Jawoyn place names. In Forty years on: Ken Hale and Australian languages, ed. Simpson, Jane, Nash, David, Laughren, Mary, Austin, Peter, and Alpher, Barry, 367–383. Canberra: Pacific Linguistics.Google Scholar
Mihas, Elena I. 2012. Ideophones in Alto Perené (Arawak) from eastern Peru. Studies in Language 36(2): 300–344.Google Scholar
Mithun, Marianne. 1982. The synchronic and diachronic behavior of plops, squeaks, croaks, and moans. International Journal of American Linguistics 48(1): 49–58.Google Scholar
Moravcsik, Edith. 1975. Borrowed verbs. Wiener Linguistische Gazette 8: 3–30.Google Scholar
Moyne, John Abel. 1974. The structure of verbal constructions in Persian . Doctoral dissertation. Harvard University.Google Scholar
Muysken, Pieter. 2000. Bilingual speech. A typology of code-mixing. Cambridge: Cambridge University Press.Google Scholar
Nedjalkov, Igor V. 1997. Evenki. London: Routledge.Google Scholar
Nedjalkov, Vladimir P., and Otaina, Galina A.. 2013. A syntax of the Nivkh language: The Amur dialect. Amsterdam: John Benjamins.Google Scholar
Nichols, Johanna. 2011. Ingush grammar. Berkeley: University of California Press.Google Scholar
Nikolaeva, Irina, and Tolskaya, Maria. 2001. A grammar of Udihe. Berlin: Mouton de Gruyter.Google Scholar
Nuckolls, Janis B. 1996. Sounds like life: Sound-symbolic grammar, performance, and cognition in Pastaza Quechua. New York: Oxford University Press.Google Scholar
Pāṇini, 1962. The Ashṭādhyāyī. Trans. Srisa Chandra Vasu. Delhi: Motilal Banarsidass.Google Scholar
Partee, Barbara H., and Borschev, Vladimir. 2003. Genitives, relational nouns, and argument-modifier ambiguity. In Modifying adjuncts, ed. Lang, Ewald, Maienborn, Claudia, and Fabricius-Hansen, Cathrine. Berlin: Mouton de Gruyter.Google Scholar
Pesetsky, David, and Torrego, Esther. 2001. T-to-C movement: Causes and consequences. In Ken Hale: A life in language, ed. Kenstowicz, Michael, 355–426. Cambridge, MA: MIT Press.Google Scholar
Pesetsky, David, and Torrego, Esther. 2004. Tense, case, and the nature of syntactic categories. In The syntax of time, ed. Gueron, Jacqueline and Lecarme, Jacqueline, 495–537. Cambridge, MA: MIT Press.Google Scholar
Plank, Frans. 2005. Delocutive verbs, crosslinguistically. Linguistic Typology 9(3): 459–491.Google Scholar
Potts, Christopher. 2007. The dimensions of quotation. In Direct compositionality, ed. Barker, Chris and Jacobson, Pauline, 405–431. Oxford: Oxford University Press.Google Scholar
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT press.Google Scholar
Ramchand, Gillian. 2008. Verb meaning and the lexicon. Cambridge: Cambridge University Press.Google Scholar
Reiter, Sabine. 2013. Ideophones in Awetí . Doctoral dissertation, University of Kiel.Google Scholar
Richa, . 2008. Unaccusativity, unergativity and the causative zlternation in Hindi: A minimalist analysis . Doctoral dissertation, Jawaharlal Nehru University, New Delhi.Google Scholar
Rizzi, Luigi. 1978. A restructuring rule in Italian syntax. In Recent transformational studies in European languages, ed. Keyser, Samuel J., 113–158. Cambridge, MA: MIT press.Google Scholar
Sakel, Jeanette. 2004. A grammar of Mosetén. Berlin: Mouton de Gruyter.Google Scholar
Sakel, Jeanette. 2007. The verbness markers of Mosetén from a typological perspective. In New challenges in typology, ed. Miestamo, Matti and Wälchli, Bernhard, 315–336. Berlin: Mouton de Gruyter.Google Scholar
Samvelian, Pollet. 2001. Le statut syntaxique des objets nus en persan. Bulletin de la Société de Linguistique de Paris 96(1): 349–388.Google Scholar
Samvelian, Pollet. 2007. A (phrasal) affix analysis of the Persian ezafe . Journal of Linguistics 43(3): 605–645.Google Scholar
Schultze-Berndt, Eva. 2000. Simple and complex verbs in Jaminjung. A study of event categorization in an Australian language . Doctoral dissertation, Katholieke Universiteit Nijmegen.Google Scholar
Schultze-Berndt, Eva. 2001. Ideophone-like characteristics of uninflected predicates in Jaminjung (Australia). In Ideophones, ed. Erhard Voeltz, F. K. and Kilian-Hatz, Christa, 355–373. Amsterdam: John Benjamins.Google Scholar
Shibatani, Masayoshi. 1990. The languages of Japan. Cambridge: Cambridge University Press.Google Scholar
Svenonius, Peter. 2008. Complex predicates and the functional sequence. Nordlyd 35(1): 47–88.Google Scholar
Torrence, Harold. 2013. A promotion analysis of Wolof clefts. Syntax 16(2): 176–215.Google Scholar
Travis, Lisa. 1984. Parameters and effects of word order variation . Doctoral dissertation. MIT, Cambridge, MA.Google Scholar
Travis, Lisa. 2000. Event structure in syntax. In Events as grammatical objects, ed. Tenny, Carol, and Pustejovsky, James, 239–282. Stanford, CA: CSLI Publications.Google Scholar
Tubino Blanco, Mercedes. 2011. Causatives in minimalism. Amsterdam: John Benjamins.Google Scholar
Vahedi-Langrudi, Mohammad-Mehdi. 1996. The syntax, semantics and argument structure of complex predicates in modern Farsi . Doctoral dissertation, University of Ottawa.Google Scholar
Vittrant, Alice. 2013. Psycho-collocational expressives in Burmese. In The aesthetics of grammar: Sound and meaning in the languages of mainland southeast Asia, ed. Williams, Jeffrey P., 255–279. Cambridge: Cambridge University Press.Google Scholar
Wälchli, Bernhard. 2005. Co-compounds and natural coordination. Oxford: Oxford University Press.Google Scholar
Wichmann, Sören, and Wohlgemuth, Jan. 2008. Loan verbs in a typological perspective. In Aspects of language contact: New theoretical, methodological and empirical findings with special focus on Romancisation processes, ed. Stolz, Thomas, Bakker, Dik, and Palomo, Rosa Salas, 89–121. Berlin: Mouton de Gruyter.Google Scholar
Wierzbicka, Anna. 1999. Emotions across language and cultures: Diversity and universals. Cambridge: Cambridge University Press.Google Scholar
Zariquiey Biondi, Roberto. 2011. A grammar of Kashibo-Kokataibo . Doctoral dissertation, La Trobe University.Google Scholar
Figure 0

Table 1. Ideophonic Complex Predicates