Ideophones, expressiveness and grammatical integration, or: how morphosyntax can depend on mode of communication

Ideophones —vivid sensory words found in many of the world’s languages— are often described as having little or no morphosyntax. That simple statement conceals an interesting puzzle. It is not often that we can define a word class across languages in terms of its syntax (or lack thereof). After all, most major types of word classes show intriguing patterns of cross-linguistic variation. There is no particular reason to expect that the morphosyntactic position or degree of embedding of, say, noun-like or verb-like words will be similar across unrelated languages. Indeed that is why typologists define comparative concepts primarily by reference to semantic rather than grammatical or morphosyntactic properties (Croft 2003; Haspelmath 2007). 

What does the morphosyntactic profile of ideophones look like across languages, and how can we explain it? We have a new paper out on these questions in the Journal of Linguistics, entitled “An inverse relation between expressiveness and grammatical integration: on the morphosyntactic typology of ideophones, with special reference to Japanese” (Dingemanse and Akita 2016). That’s by far the longest title of any of our publications — and we don’t intend to better that record. The benefit is that it gives a fairly complete account of the contents. So let me just break it down.

Expressiveness and grammatical integration

Words and phrases can differ in the degree to which they are susceptible to performative prosody: their expressiveness. This can be measured in terms of (i) intonational foregrounding, when words are produced with an intonational profile that deviates from the surrounding speech; (ii) phonational foregrounding, when words are produced with types of phonation that deviate from the normal phonation evident in the surrounding speech, and (iii) expressive morphology, when words are modified on the fly by means of playful morphological processes like lengthening, free reduplication and iteration (Zwicky and Pullum 1987).

Words and phrases can also differ in the degree to which they are embedded in the morphosyntactic structure of the utterance: their grammatical integration. Three reasonable indices of grammatical integration are (i) linear position, where peripheral items are less integrated, (ii) syntactic optionality, where optional items are less integrated, and (iii) embedding in morphosyntactic structure, where less deeply embedded items are less integrated.

An inverse relation (and a proposal about its cause)

Our paper is about the expressiveness and grammatical integration of ideophones, and the starting observation is this: there appears to be an inverse relation between them such that a higher degree of expressiveness tends to go with a lower degree of integration. Why would this be so?

We propose this is the case because ideophones are depictions rather than descriptions — they are more akin to performances than to propositions. This status aparte is signalled and supported by their expressiveness and grammatical independence, much like the whitespace around an image shows it to be distinct from the surrounding text. Our explanation builds on a characterisation of ideophones that goes back over a century: ideophones are “vocal gestures” (Westermann 1907), they act out rather than describe (Kunene), and they perform rather than merely inform (Nuckolls). In our analysis, these observations all point to the same fundamental distinction between two modes of communication: depiction versus description (Clark 2016).

The ideophone zabɯ:n-zabɯ:n with intonational foregrounding

The ideophone zabɯ:n-zabɯ:n ‘splashing’ with intonational foregrounding

Combining two modes of communication in one linearized speech signal is a non-trivial challenge, as the required mode of interpretation is different for each. We propose that the optimal solution to this linearization problem is precisely the combination of prosodic foregrounding and syntactic independence that ideophones show. (In the paper, we have more details about other kinds of words and other combinations of semiotic resources.) So beneath the observable facts about expressivess and integration lies a semiotically motivated, psycholinguistically grounded account.

Morphosyntactic typology of ideophones

There is a twist to the story which turns out to support this line of argument. Even though ideophones tend to be characterised as having “little in the way of morphosyntax” (Childs 1994:179), matters are more complex than usually advertised. In particular, grammatical descriptions careful enough to go into some detail about the morphosyntax of ideophones (and there are preciously few of those) often observe that these words can occur in a range of constructions with different degrees of grammatical integration. And grammars that describe the prosody of ideophones (again, preciously few) may observe that ideophones are not always expressive. Rather than brushing these observations aside as irrelevant exceptions, we use them to shed light on the relation between expressiveness and grammatical integration.

In Bambara (a Mande language spoken in Mali), ideophones tend to occur at the right edge of the utterance, but “non-final position of the ideophone causes it to lose its high intonation and its status as expressive adverb” (Dumestre 1998:327). In Awetí (Tupian, Brazil), ideophones can occur either as independent clauses or embedded in a light verb construction. In the first case, they are “always prosodically marked”, whereas in the second, “they may lose this feature and thus also their status as ideophones” (Reiter 2012:576). These languages, and eight more for which we cite evidence in the paper, provide evidence for a situation that is exactly opposite to the one sketched above — one in which ideophones are more deeply grammatically integrated and less prosodically expressive.

Most of the grammars we cite don’t connect these sets of observations; they just catalogue the facts. It is when evidence from multiple languages is brought together that the similarities become visible.  So typological evidence suggests that ideophones are at their most expressive when free, but lose this expressiveness when more deeply integrated. I write “suggests”, because the available evidence from grammatical descriptions is relatively thin and not systematic. That’s where Japanese comes in.

…with special reference to Japanese

We take our observations and proposals to the test using a video corpus of Japanese, a language rich in ideophones. Japanese ideophones (also known as mimetics) can appear in a range of constructions with different degrees of morphosyntactic integration. We collected 692 ideophone tokens (~200 types) and coded them for three measures of expressiveness and three degrees of morphosyntactic integration. The prediction should be clear: we expect the least grammatically integrated ideophone tokens to show the most expressive features, and the most integrated ones to behave more like normal words.

That is indeed what we find. There is a strong negative correlation between expressiveness and grammatical integration; so strong that it is almost pointless to provide statistical tests of significance (which we do nonetheless because reviewers asked for it). All three expressive features show a clear stepwise pattern such that they decrease when grammatical integration increases.

Clinching evidence from gesture

Now, it’s nice to find corpus data behaving exactly as theoretically predicted, but so far it is just a correlation. We want to get closer to a causal account. Here’s how we do that. We look beyond the speech signal to co-occurring iconic gestures — hand movements that depict aspects of the reported events. Such gestures provide us with an independent measure of depiction, or the degree to which a speaker enacts (rather than merely describes) the events. So if ideophones are indeed depictions when at their most free, we should see that in the relative frequency of gestures.

Again, that is indeed what we find. The most expressive and grammatically free ideophones are accompanied by iconic gestures in 82% of cases; the least expressive and most grammatically bound ideophones come with gestures in only 12% of cases. This dramatic contrast supports the causal account and provides even more evidence for the depictive nature of ideophones in their most common mode of occurrence.


Gesture accompanying the ideophone ba:n 'incoming wave'

Gesture accompanying the ideophone ba:n (depicting an incoming wave)


We started with a very general question about the relation between expressiveness and grammatical integration. Checking this for ideophones in ten diverse language led to a possible pattern and a plausible explanation. A Japanese corpus provided quantitative support, while gesture provided clinching evidence for the causal account proposed. In the final section of the article, we zoom out to the bigger picture and enumerate some implications for ideophone typology, ideophone creation (Dingemanse 2014), and broader phenomena like reported speech in spoken language and perspective shift in sign language.

The findings also have implications also for how we approach linguistic structure generally. Many of us are used to a neat division between words and images — with words doing the serious work and images at best playing a secondary role in the form of supportive-but-redundant gestures and other things normally shelved away as paralanguage. Ideophones, a major lexical class in  many of the world’s languages, greatly complicate this neat picture. They show that speech, too, can be depictive. They show how morphosyntax can depend on mode of representation. They suggest that a complete account of linguistic form and function will have to take into account both description and depiction.

The paper:

Dingemanse, Mark, and Kimi Akita. 2016. “An inverse relation between expressiveness and grammatical integration: on the morphosyntactic typology of ideophones, with special reference to Japanese.” Journal of Linguistics FirstView: 1–32. doi:10.1017/S002222671600030X (download PDF).

Data and R code for this paper is available on Open Science Framework:


  • Childs, G. Tucker. 1994. African Ideophones. In Sound Symbolism, ed by. Leanne Hinton, Johanna Nichols, and John J. Ohala, 178–204. Cambridge: Cambridge University Press.
  • Clark, Herbert H. 2016. “Depicting as a method of communication.” Psychological Review 123 (3): 324–347. doi:10.1037/rev0000026.
  • Croft, William. 2003. Typology and Universals. 2nd ed. Cambridge: Cambridge University Press.
  • Dingemanse, Mark. 2014. “Making new ideophones in Siwu: creative depiction in conversation.” Pragmatics and Society 5 (3): 384–405. doi:10.1075/ps.5.3.04din.
  • Dumestre, Gérard. 1998. “Les Idéophones: Le Cas du bambara.” Faits de Langues: Revue de Linguistique 11: 321–333.
  • Haspelmath, Martin. 2007. “Pre-established categories don’t exist: Consequences for language description and typology.” Linguistic Typology 11 (1): 119–132. doi:10.1515/LINGTY.2007.011.
  • Reiter, Sabine. 2012. Ideophones in Awetí. PhD thesis, Kiel: University of Kiel.
  • Westermann, Diedrich Hermann. 1907. Grammatik der Ewe-Sprache. Berlin: Dietrich Reimer.
  • Zwicky, Arnold M., and Geoffrey K. Pullum. 1987. Plain Morphology and Expressive Morphology. In Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society, ed by. John Aske, Beery, Natasha, Laura Michaelis, and Hana Filip, VII:330–340. Berkeley: Berkeley Linguistics Society.