Course:LING447/2014WT1/Assignments

Background

Ideophones are instances of sound symbolism wherein the relation between sound and meaning is iconic and non-arbitary. In formal linguistics, ideophones have largely been overlooked (Nuckolls 2004), but there is a now a surge of interest in this research area.

The perceptual basis of ideophones

Ideophones typically connote sensory percepts (e.g. visual, auditory, gustatory, olfactory, tactile precepts) as well as motor and kinaesthetic percepts. By yoking together one signal —auditory for spoken language, visual for signed languages — they are inherently synesthesthetic (Ramachandran and Hubbard, 2001; Dingemanse, 2011, 2012). For this reason, variation in the perceptual basis (as well as cognitive capabilities) shapes the types of non-arbitrary meanings that come to be attached to individual sounds.

Linguistic evidence

Experimental findings indicate that acoustic properties carry inherent meaning; this especially the case for vowel quality (Sapir 1929, Fischer-Jørgensen 1978). As summarized in the following table, vowels have been found to have expressive values relating to size (magnitude), hue (brightness), and shape (roundness).

Summary of Sapir's (1929), Bentley, Madison, Varon's (1933) and Fischer-Jørgensen’s (1978) results

Semantic feature	Vowel/Tone	Ranking
Size: large	[a]	[a]/ > { [i], [e] }
Size: large	[e]	[e] > [i]
Shape: round	[a]	[a] > [i]
Texture: soft	[a]	[a] > [i]
Size: big & lax	Lo tone
Size: narrow, tight, squeezed	[i]
Hue: dark	[i]

Sapir (1929) pioneered the study of phonetic symbolism by presenting subjects with minimal pairs of nonce words differing in one segment only, and asking them to choose the word that best conveys the meaning of ‘large’. The results indicate that the vowel [a] consistently conveys the meaning of ‘large’ better when compared to other vowels ([i], [e], [o], [u], etc.) Althought Sapir's the study included only English-speaking subjects, the consistency of judgements across subjects is striking. On the basis of these findings, Sapir argues that kinesthetic and auditory factors play a role in the interpretation of ideophones. This early works supports that the view that there is a perceptual basis for ideophonic meaning.

[+LARGE] <--------------------> [-LARGE]
   [a]

Newman (1933). Newman broadened the range and investigated a variety of vowels, consonants, and clusters. Volunteers’ interpretations of nonce words were plotted onto a gradient scale representing degrees of ‘magnitude’ and ‘darkness’ for each sound investigated. The study also looked at three different group ages, in order to control variables and to investigate whether age plays a factor in interpretation of ‘expressive’ lexicon. Newman’s study discovered five major points of interest regarding ideophones: phonetic elements are patterned on a non-linguistic symbolic scale; phonetic symbolism is fundamentally objective; different phonetic patterns are formed by different symbolisms; linguistic association plays no part in symbolism adjudication; age plays no part in symbolism adjudication. The claim that phonetic symbolism is objective is of support for the hypothesis here investigated, as it explains the consistency of the results related to the (biological) consistency of the perceptual paths used for interpretation. The results showed that adjudication of meaning is mechanical, and further feed the theory here presented that different types/paths of perceptual understanding are at play in ideophones, three of which described by Newman: kinesthetic, auditory, and possibly visual.

However, other research findings (Bentley, Madison, Varon 1933) indicate that vowel quality is not inherently meaningful. This argument is based on the fact that participants in vowel suggestiveness experiments require scales to measure the ideophone's meanings i.e. an evaluation according to e.g. roundness. Experiments that were not conducted in accordance with any given scales for evaluation do not show sufficient evidence to claim that sounds inherently carry meaning.

Bentley, Madison & Varon (1933) explore how subjects assign meaning to nonce words, and more specifically to vowel quality, in two different experimental conditions. In one task, subjects are asked to assign meaning in relation to a particular scale, e.g., to what extent the nonce word describes the quality of roundness. In another task, subjects are asked to assign meaning to a nonce word without reference to a scale. The results indicate that while speakers are consistent in adjudicating the meaning of vowel quality for a scale-dependent task, there is is a larger degree of between-speaker variation when the adjudication task is not scale-dependent. Bently et al. conclude that the constituency of the results is given by meaning expressed through kinesthetic factors in a non-arbitrary fashion, in accordance to the theory here presented.

Fischer-Jørgensen (1978). description and analysis of ideophones presented an important theory that correlates to the view hypothesized in the present paper. Fischer-Jørgensen take on ideophones and their perception is to propose a hierarchical structure in which some perceptual paths play a more prominent role in conveying meanings than others: the example proposed is that auditory perception plays a greater role than kinaesthetic perception in accordance to the meaning adjunct onto ideophones (in both natural languages and experimental data). Furthermore, it is proposed that sound-symbolism pairing takes place at a relatively deeper psychological/physiological level, one that involves activation of synesthetic processing.

The majority of authors seek for a different explanation of positive results in suggestiveness experiments – suggestiveness due to a subconscious linkage between articulatory phonetics and vowel quality or a linkage between the quality of everyday things and their attributes e.g. the sky is high and bright, therefore high voice suggests brightness etc. (Fischer-Jørgensen 1978). The lack of a sufficient and serious explanation for suggestiveness of certain sounds leads to the simplified conclusion that expressive value is always given and possibly can be explained by the general phenomenon of synesthesia.

Miron 1961. However, suggestiveness and a feeling for phonetic symbolism is suggested to present a universal tool that every speaker of a natural language may use as part of the expressive as well as the prosaic lexicon (Miron 1961).

Neurological Evidence

It is not until Osaka & Osaka (2005) that concrete proof of synesthetic processes relative to sound symbolism and ideophones is presented. In this study, Osaka et al. monitor brain activity of subjects when exposed to ideophones bearing the meaning of laughter (at different gradients). The brain activity is shown to mimic laughter behaviour and neurotransmissions by just exposure to the ideophone. This is an extremely important finding as it ground the theory here investigated to physiological behaviour. Furthermore, it can be argued that these findings can be transported onto other synesthetic processing involved in different types of ideophones.

Synesthesia and Ideophones

Iconicity - Dingemanse

In Dingemanse’s (2011&2012), the aspect of Iconicity of ideophones and their interpretations were investigated. In terms of Iconicity, he presents three types: Imagic, where a mapping between a word sound and a sound in the world is made; Gestalt, where a mapping between a word structure and spatio-temporal event structure is crated; and Relative, where a mapping between a relation of forms (sounds) and a relation of meaning is created. Relative and Gestalt Iconicity relate to the ideophones analyzed by the studies presented above (Bently et al., 1933; Newman, 1933; Sapir, 1929), while Imagic ideophones relate to onomatopoeias, a relatively overlooked aspect of iconicity in this literature review (as it is considered outside the scope of this paper).

Dingemane's Iconicity definitions:

Imagic: mapping between real world sound and word sound
Gestalt: mapping between spatio-temporal event and word structure
Relative: mapping between a relation of forms and a relation of meanings

An example of Imagic Iconicity is an onomatopeia: a reproduction of a real world sound. Below, an example from Italian

(1) chicchirichi
    kiʔkiriki(ː)
    sound.of.rooster
    cock-a-doodle-do

Cognitively, using an onomatopeia involves the 're-stating' of a memorized string of sounds that mimics a naturally occurring string of sounds. This process can be also explained in terms of data: a memorized sensorial data (e g a string of sounds) is revitalized and used to reproduce said memorized sensorial data. In the case of (1), auditory data is stored and used only as auditory data, with no other meaning possible. When applied to Ramachandran et al (2001)'s synesthetic paths, Sensory (auditory) data transfers onto the same Sensory (auditory) data. Arguably, Imagic iconicity can also refer to the transfer of Motor data, such as gestures that mimic a naturally occurring (or occurred) gesture.

Gestalt Iconicity is interpreted as forms that focus on their internal form. The following two examples are from Japanese:

(2) goro
    goro
    rolling
    heavy object rolling

(3) gorogoro
    goro~goro
    rolling.repeatedly 
    heavy object rolling repeatedly

With this kind of iconicity, the information that is transferred can morph in its sensory/motor nature. The iterative meaning of (3) is conferred by the reduplication of the base root seen in (2), goro; this reduplication can be said to be composed of Sensory (auditory) data as well as Motor (kinesthetic, such as the movement of the mouth area emphasized by the reduplication) data. As the string of sounds is heard, the meaning can be interpreted as Sensory (visual) or Motor (the iterative-ness can be loosely translated to spatial-motor?)

Relative Iconicity is peculiar in that it is mainly constructed around the relationship between different forms. Most famous is the bouboa vs kiki effect, which is represented here below:

Booba vs Kiki effect

Relative iconicity focuses on how change in the content (or nature) of the information is related to change in phonological segments/strings/features. Sapir's research (1929) focused on Relative iconicity, looking at how different sounds affected the interpretation of nonce words. For example, he noted that the vowel /a/ was better at representing heavy objects than /i/. Nonce words such as /fam/ or /vam/ were found to denote physically larger features than /fim/ or /vim/ would express. Relative iconicity takes the relation of form (/a/ and /i/) and compares it to the relation of meaning (large and small). The relational links between forms and meaning can then be placed onto gradient scales, offering a better tool for analysis.

Sensory-Motor Synaesthesia - Ramachandran et al

Closely connected to Dingemanse’s Iconicity is the study of Ramachandran & Hubbard (2001), in which synesthetic processes in language were analyzed in detail: Ramachandran et al. suggest that the variety of synesthetic processes found in ideophones perception can be categorized under two main types: sensory and motor. Each of these types has its own subtype, namely the respective paths that they encompass (the five senses for sensory, motor-mechanisms, etc.). If this study were to be tied together with Dingemanse’s Iconicity, one can notice the relation between synesthetic paths and the three Iconicity groupings. While Imagic (onomatopoeias) allow only one type of sensory perception path (namely, auditory), Gestalt Iconicity seems to suggest that a certain sensory or motor process is morphed into a different one, and such morphing is the basis of perception that distinguish ideophones from the rest of the lexical inventory of natural languages. Relative Iconicity is then the relation seen on gradient scales where ideophones’ interpretations are plotted: a similar process to Gestalt Iconicity rises only when two different forms are compared in a parallel-fashion, and this parallelism is then expressed in the (related) meaning associated to each form.

Ramachandran et al's attested synesthetic paths:

Sensory to Sensory
Motor to Motor
Sensory to Motor / Motor to Sensory

Taken together, the results show that the perception of ideophones in rooted in the different processes and paths activated when exposed to ‘expressive’ lexicon. From this literature it can be assumed that the role played by synesthesia is one of great magnitude in terms of ideophones and the meanings to the adjudicated, and that its different components can create variety in a quantal fashion. Further experiments in the area regarding Iconicity and its relation to synesthetic processes would help with typology of ideophones and, later, of languages. As shown in this literature review, the research that has been done on ideophones has still room for improvement and exploration, and a global understanding of ideophones is necessary for further hypotheses on the matter. From looking at each iconicity and the different sensory/motor paths they govern, a basic representation of the governance of iconicity over (synesthetic) pathways can help better differentiate the three types:

The prosaic lexicon and the expressive lexicon

(Christina, Allyssa)

Ideophones are part of the expressive lexicon, and as such contrast with regular words that are part of the prosaic lexicon. The expressive and the prosaic lexicons use the same phonological inventory, but they use it in different ways. In the expressive lexicon, the sound-meaning relation is iconic, but in the prosaic lexicon the relation is arbitrary. The expressive lexicon is often described as deviant and not affected by prosaic rules and restrictions of a language, however, ideophones are not as abnormal as they are regularly characterized.

The most notable difference between ideophones and regular words is that ideophones always have expressive force. In this respect, ideophones are part of the expressive lexicon; as such they are "signs" (Nuckolls 1995), that depict sensory images (Dingemanse 2012). These images have precise meaning and present the speaker's perspective of sensory precepts based on environmental and somatic properties.

Expressive lexicon		Prosaic lexicon
kànya amε tìtìrìtììì	'mouth inside sticky'	lε	'be good'
mouth inside sticky		be.good
ìgbèdi sinisinisinisini	'smooth cassava'	dzɔ	'be straight'
cassava smooth		be.good
nyãkãnyãkã	'sensation of grooved texture'	mɔ	'be big'
smooth.texture		be.big

Examples from Dingemanse (2011) in Siwu, Ghana

Ideophones are "regular words"

Samarin (2001) argues that ideophones are “regular words”: as such, they are collectable, they are subject to speaker variation, they are meaningful, they belong to the lexicon of the language, they are definable, and they are acquired at the same time as other words in a language.

Ideophones belong to the lexicon of a language. They are able to be described just like any other word in a language and they belong to a whole speech community, not just to individuals or small groups of individuals. There is intersubjective variation, and variation in both form and use (Samarin 326).

Ideophones are collectable. They occur and can be collected in a wide range of contexts, including traditional story-telling, performance of improvised skits, recounting of life-threatening experiences, descriptions of landscape. (Samarin 321).
Definitions of ideophones are replicable. Lexicographers will arrive at the same or very similar definitions with same data & dictionary-writers will come to same or very similar conclusions about what each means. Ideophones cannot be created by any member of a speech community because they do not become part of the “ideophonic repertoire of a community of speakers” (Samarin 329).

Ideophones are subject to variation and change.

Ideophones vary intersubjectively in a manner that resembles geo-linguistic variation. This means that we cannot assume the meaning of an ideophone or give examples; one context = many ideophones. In certain contexts there is variation in a range of meaning. Whatever the stimulus may be, there will always be many different responses (Samarin 325). In an experiment, subjects were asked to describe certain objects by touching or smelling them, and in one instance, they were asked to describe a sound. In another experiment, subjects were asked to run their fingers over the fibers of a scrub brush, which the experimenter (Samarin) assumed would elicit ideophones that describe the roughness of the fibers. However, subject volunteered a wide range of ideophonic forms that describe properties such as sharpness, scratchiness, yielding to touch, dryness. (Samarin 330).
Urbanized speakers of a language have a poorer inventory of ideophones than their more traditional co-ethnics and use ideophones less competently. The word urbanized meaning that a person who acquired his or her ethnic language in a rural village, but has lived most of his or her life in a city where the individual uses the language less frequently than the city’s lingua franca (Samarin 323).
Ideophones are vulnerable to attrition to a degree greater than any other part of a language’s lexicon in situations of rapid change, such as pidginization. In saying this, eliciting ideophones in pidgins – such as Sango – may be more difficult because in Sango there are very few of them (Samarin 324). The use of ideophones varies according to the competence of the speakers of a language. This is something that should be expected because competence in the practise of language is variable in speech communities throughout the world. The richness of one’s vocabulary, the range of complexity of one’s sentence structures, the organization of one’s discourse, and many other features of the use of the language are known to us without study (Samarin 333).

Children acquire ideophones at the same time that they acquire other words in their language. Ideophones constitute no challenge to the language acquisition competence of children for phonological, grammatical, or semantic reasons. Children can create and can use ideophones from the ages of 2-5 years old approximately (Samarin 322).
Ideophones contribute meaningfully to an utterance. Ideophones do not just add affect, they add meaning as well. They specify that nature of the object, event, or phenomenon. (Samarin 326).

Some ideophones are specific enough in meaning that it is possible to find for them synonyms and antonyms (Samarin 330).
The phonological shapes of ideophones associated with a certain meaning are similar to some degree (Samaring 324).
Ideophones linked with a given essence of meaning increase in number by natural evolution. In some instances, variant forms can be explained by a rule that reduplicated forms can occur in another form while still showing reduplication (Samarin 325).
In some languages ideophones can be modified phonologically and morphologically for stylistic or expressive purposes. This presupposes that ideophones in themselves do not have expressiveness as their sole or primary function. Ideophones manipulated for expressivity, such as, raising or lowering the height of the pitch above what might be considered normal in an utterance, or reducing the height of the pitch of high tone in small degrees (in steps) over a stretch of syllables) (Samarin 332).

  eg. zólóló in its normal lengthened form; adding syllables to a word that already had reduplicated ones
  (zólóló > zólóló́lóló ‘white’)
  eg. altering the dictionary-form of a complex ideophone (ngalang > ngáláng ngalang ngáláng) (Samarin 333).

Ideophones are definable in a lexicographic manner. By using the sentence-completion protocol, eliciting contrastive ideophones, and the appropriateness test, Samarian discovered that it is hard to get to same semantic essence (Samarin 326). In the sentence-completion protocol, the subject is required to supply an ideophone at the point where one stops in a reading sentence, both affirmative (1) and negative (2). Since it follows that if a stick is not straight, it is bent; in this way one elicits contrastive ideophones. One would therefore be given appropriate words for ‘bent’ in describing such an object (Samarin 327).

Sentence-completion protocol

(1) wesé     dɔ́-á      go bá      nu___
      sun     burn-SFX   so seize   earth ____
     ‘The sun is shining so brightly that the earth is hot ___’
(2) te-í      gan tán          ná,  go ɔ́ ___
      tree-DET  NEG-be straight  NEG  so be ___
     ‘The tree (or stick) is not straight, so it is ___’

In the Illustrative Object, the subjects are asked to name an object which would be described in each of the sentences. Each sentence would begin with the phrase ge mɔ́ gé ‘what thing?’. The purpose of this protocol was to arrive at some idea of semantic consensus (Samarin 327).

  (3)  ge   mɔ́       ge    a      yóó    wey    sésésé   ge   ndé
       NT   thing    INT  LINK   stand   fire   sésése   INT  CLT
       ‘What thing is hot sésése?’

In the Appropriateness Test, after finding a number of ideophones with similar meanings, the subjects are asked if he or she would use each of them (Samarin 327).

Ideophones are "marked words"

The notion of ideophones as being "marked words" (Dingemanse 2011) is put forward when investigating how expressives can be distinguished from regular words. It is important to note that characteristics of markedness are cross-linguistic tendencies and their implementation are language specific. What makes ideophones marked are liberties and restrictions concerning their phonology, morphology and morphosyntax. These peculiarities of ideophones can include phonotactic distributions, harmonies (especially vowel harmonies), tonal melodies, syllable structures, word forms, expressive morphology (especially reduplication and lengthening), and syntactic independence (Dingemanse 2012: 656-658) that does not occur in the prosaic lexicon. For more information regarding peculiarities of phonology, morphology and morphosyntax see the respective sections of this article.

Based on this assumption that ideophones make use of the same inventory as verbals, the ideophonic lexicon differs from the prosaic lexicon in several ways. Dingemanse points out that ideophones differ from nouns and verbs, as they are usually longer, have different word structure, are differently integrated in an utterance, they give emphasis, they can be altered prosodically and are non-derived basic terms (2011). In some languages the expressive language contains exceptions from rules and restrictions of the prosaic lexicon and vice versa – they “stretch the system of a language” (Newman 251). Their liberties and limitations result in the markedness of ideophones, that is easily overstated as making them unlike ordinary words.

Differences and similarities between expressive and prosaic lexicon in Hausa (Newman 2001):

Vowel inventory: prosaic and expressive lexicon make use of the same vowel inventory
Consonant inventory: prosaic and expressive lexicon make use of the same consonant inventory
Tone inventory: prosaic and expressive lexicon make use of the same tones
Phonological processes: words in the prosaic and expressive lexicon both undergo phonological processes
Segmental phonotactics: in the prosaic lexicon, most Hausa words end with a vowel; in the expressive lexicon, numerous ideophones end in consonants
Tonal phonotactics: in the prosaic lexicon, it is is for words to have a long final vowel end in LL ton; in the expressive lexicon long final vowels and LL tones occur frequently

Ideophones are "iconic words"

The idea that ideophones are in some way iconic is a key feature of many analyses. They are performative in that they present a re-enactment of an event such a way that the speech act cannot be distinguished from the narrated action, event or process. In order to achieve this effect, speakers draw upon the expressive dimension of the ideophonic vocabulary. Different authors have come to different conclusions about the sensory perceptions that ideophones can denote:

Samarin (1965): ideophones describe of appearance, arrangement, emotion, measure, motion, odour, quality, shape, sound, taste, state, temperature, time, touch and weight.
Alexandre (1966): ideophones depict auditory, visual, tactile, gustative, olfactive perceptions, and also illustrate aspects of animate entities and can describe movement, situation and position.
Nuckoll's (1995) proposes that sound symbolic words are iconic in that they connote auditory, rhythmic, visual or psychophysical sensations.
Dingemanse (2011, 2012) argues that ideophones are “marked words that depict sensory imagery” (Dingemanse 2011: 78), which is the knowledge that derives from an experienced sensory perception. These perceptions include sound, movement, visual patterns, other sensory perceptions, inner feelings and cognitive states. Dingemanse proposes that percepts are ranked according to the following hierarchy:

sound < movement < visual patterns < other sensory perceptions < inner feelings / cognitive states

According to this hiearchy, if a language has ideophones that depict perception of inner feelings and cognitive states, then it will likely also have ideophones that depict other sensory perceptions as well is very high. Therefore, the use of ideophones of one perception hightens the tendency of the use of ideophones for a lower ranked perception, but not vice versa. This predicts the following typology:

Language	Auditory percept	Movement	Visual percept	Other sensory percept	Inner feeling/Cognitive state
A	Yes	Yes	Yes	Yes	Yes
B	Yes	Yes	Yes	Yes	No
C	Yes	Yes	Yes	No	No
D	Yes	Yes	No	No	No
E	Yes	No	No	No	No

With a range of perceptions ideophones can depict, their meanings reflect different kinds of iconic imagery; for example, Dingemanse (2012) distinguishes between three types of iconicity:

Examples taken from Dingemanse (2012: 661) in Kwa, Niger-Congo

Type of iconicity	Description	Example	Gloss
imagic iconicity	sound of the ideophone mimics sound of sensory percept	glbogblogblo	'bubbling'
gestalt iconicity	word structure of the ideophone mimics event structure	kpɔtɔrɔ-kpɔtɔrɔ	'walking like a tortoise' (reduplication)
item relative iconicity	related forms map onto related meanings	ɣììì	'sensation of vertigo' (use of vowel space to suggest intensity)

Diagrammatic iconicity

Diagrammatic iconicity

Icons "which represent the relations of the parts of one thing by analogous relations in their own parts."

(Peirce 1955: 105)

Diagrammatic iconicity at the word-level

Watery and waterfall are related to each other diagrammatically by virtue their shared root water:

(1) water-y   water-fall
    water     water

Watery and rainy are related to each other diagrammatically by virtue their shared suffix -y:

(2) water-y   rainy-y
         -y        -y

Diagrammatic iconicity at the phrase-level

The intransitive form of eat (3a) and the transitive form of eat the cake (3b) are related to each other diagrammatically (4) by virtue of their shared head eat.

(3) a. Lucy will [_VP[_Veat]]]
    b. Lucy will [_VP[_Veat][_DPthe cake]]

(4)  Diagrammatic relation between eat and eat the cake
     [_VP [_V eat]]   [_VP [_V eat] [_DPthe cake ]
         [_V eat]        [_V eat]

Diagrammatic iconicity at the sentence-level

Affirmative (5a) and negative (5b) sentences are diagrammatically related by virtue of their propositional content (6).

(5) a. Lucy did so eat the cake.
    b. Lucy did not eat the cake.

(6) AFF[_S Lucy eat cake]     NEG[_S Lucy eat cake]
       [_S Lucy eat cake]        [_S Lucy eat cake]

Isomorphic iconicity

[introduce text]

One-form/one-meaning principle

Sameness of form from one sign to another signals sameness of meaning.

Difference of form from one sign to another signals difference of meaning.

[Describe patterns in the following table, and show how they illustrate isomorphic iconicity. Examples adapted from Waugh (1994:56f.)]

[α]	[α -β]	[ [a] [β] ]
water	water-y	water-fall
rain	rain-y	rain-drop
snow	snow-y	snow-shoe

Sub-morphemes

Sub-morphemes are based on partial shared structure. Although words are divisible into parts, this does not mean that they are totally divisible into morphemes. For Waugh (1994:58), word parts that are not morphemes have the status of sub-morphemes.

Lexical categories

Sub-morphemes are found with lexical categories. For example, while black and berry are both nouns, *cran, *boysen, and *rasp are sub-morphemes:

[[α] [β]]	α=morpheme?	β=morpheme?
blackberry	Yes; black	Yes; berry
cranberry	No; *cran	Yes; berry
boysenberry	No; *boysen	Yes; berry
raspberry	No; *rasp	Yes; berry

While day and -ly are both morphemes, *holi, *yester, *Mon and *to are sub-morphemes:

[[α]-[β]]	α=morpheme?	β=morpheme?
daily	Yes; day	Yes; -ly
holiday	No; *holi	Yes; day
Monday	No; *Mon	Yes; day
yesterday	No; *yester	Yes; day
today	No; *to	Yes; day

While rain and -y are both morphemes, *naught, *joll, *clums and *coze are sub-morphemes:

[[α]-[β]]	α=morpheme?	β=morpheme?
rainy	Yes; rain	Yes; -y
naughty	No; *naught	Yes; -y
jolly	No; *joll	Yes; -y
clumsy	No; *clums	Yes; -y
cozy	No; *coze	Yes; -y

While spit, run and -le are morphemes, *nozz is a sub-morpheme:

[[α]-[β]]	α=morpheme?	β=morpheme?
spittle	Yes; spit	Yes; -le
runnel	Yes; run	Yes; -le
nozzle	Yes; *nozz	Yes; -le

While bro and -ther are morphemes *mo and, *fa are sub-morphemes:

[[α]-[β]]	α=morpheme?	β=morpheme?
brother'	Yes; bro	No; -there
mother'	No; mo	No; -there
mother'	Yes; *mo	No; -there

Functional categories

Sub-morphemes are also found with closed-class items; i.e. functional categories.

[[α]-[β]]	α=morpheme?	β=morpheme?
the	Yes; th-	No; -e
this	Yes; th-	No; -is
that	Yes; th-	No; -at
they	Yes; th-	No; -ey
their	Yes; th-	No; -eir
thee	Yes; th-	No; -ee
thou	Yes; th-	No; -ou
thy	Yes; th-	No; -y
thine	Yes; th-	No; -ine
then	Yes; th-	No; -en
there	Yes; th-	No; -ere
thus	Yes; th-	No; -us
than	Yes; th-	No; -an
though	Yes; th-	No; -ough

Phonesthemes

Even in languages that have restricted idephonic lexicons, such as English, one can observe the emergence of sound symbolism as the segmental level. Such sound-symbolis segmental melodies are called "phonesthemes".

Header text	Header text	Header text
flap	fl-	-ap
flare	fl-	-are
flee	fl-	-ee
flick	fl-	-ick
flicker	fl-	-ick-er
fling	fl-	-ing
flip	fl-	-ip
flit	fl-	-it
flitter	fl-	-itt-er
flow	fl-	-ow
flutter	fl-	-utter
fly	fl-	-y
flurry	fl-	-urry
flask	*fl-	*-ask
flag	*fl-	*-ag

(5) a. snorkel
    b.  sniff
    c.  snour

(6) a. ??-s

     b. ??-s

(7) a. book < any written material < document < pieces of wood from tree < 'tree'

    b. bank1 (grassy bank) versus bank2 (savings bank)
    c. rain-y, craz-y, canny-, joll-y

Ideophonic continua

Ideophones can be placed on several continua.

+	0	–
fully iconic	many degrees of iconicity	fully non-iconic
isomorphic	partially isomorphic	not isomorphic
motivated	partially motivated	not motivated
morphemic	semi-morphemic	not morphemic
onomatopoeic	partially onomatopoeic	not onomatopoeic
sound symbolic	partially sound symbolic	not sound symbolic
full compositionality	degrees of compositionality	no compositionality
unitary meaning	polysemy	homonymy (accidental identity)
identity	degrees of similarity	no identity
full relatedness	partial relatedness	no relatedness

Bentley, Madison, and Edith J. Varon. “An accessory study of ‘Phonetic symbolism’.” The American Journal of Psychology 45.1 (1933): 76-86. Print.
Brown, Roger W., Abraham H. Black, and Arnold E. Horowitz. “Phonetic symbolism in natural languages.” Journal of Abnormal and Social Psychology 50.3 (1955): 388-393. Print.
Dingemanse, Mark. “Ideophones and the aesthetics of everyday language in a West-African society.” Senses & Society 6.1 (2011): 77-85. Print.
---. “Advances in the cross-linguistic study of ideophones.” Language and Linguistics Compass 6.20 (2012): 654-672. Print.
Fischer-Jørgensen, Eli. “On the universal character of phonetic symbolism with special reference to vowels.” Studia Linguistica XXXII I-II (1978): 80-90. Print.
Miron, Murray S. “A cross-linguistic investigation of phonetic symbolism.” Journal of Abnormal and Social Psychology 62.3 (1961): 623-630. Print.
Newman, Paul. “Are ideophones really as weird and extra-linguistic as linguists make them out to be?” Ideophones. Ed. Voeltz, F. K., and Christa Kilian-Hatz. Philadelphia: John Benjamins, 2001. 251-258. Web. 6 October 2014.
Nuckolls, Janis B. “Quechua texts of perception.” Semiotica 103.1/2 (1995): 145-169. Print.
Sapir, Edward. “A study in phonetic symbolism.” Journal of Experimental Psychology 12.3 (1929): Print.
Waugh, Linda R. “Degrees of iconicity in the lexicon.” Journal of Pragmatics 22 (1994): 55-70. Print.

Phonology

(Noriko, Masaki) Ideophones exhibit unique phonological properties, including:

consonantal melody
vocalic melody
tone
syllable structure
word size (e.g. minimality versus maximality effects)
intonation

Nembe

Maduka (1988) Size and shape ideophones in Membe: a phono semantic analysis

Nembe is part of the Ijoid sub-family of the New Benue-Congo family [link to Ethnologue?]

Data: Nembe-English Dictionary

Nembe phonological inventory

(1) Phonological inventory
   a. Consonants
      { p, b, ɓ, m, f, v, t, d, ɗ, r, s, z, n, y, k, g, kp, gb, w }
   b. Vowels: +wide vs -wide; [wide] correlates to pharynx size 
      { ??? }
   c. Tone
      { High, Low]

(2) Size restrictions for Nembe ideophones
    a. 2 syllables: CV.CVV
    b. 3 syllables: CV.CV.CVV
    c. 4 syllables: [CV.CV] + [CV.CV]
    d. 6 syllables: [CV.CV.CV] + [CV.CV.CV]

(3) Melodic restrictions for Nembe ideophones (??)
   a. L…H
   b. L...
   c. L…HL…  L…HL
   d. (HL)…, (HL…)…, (LHL)…

Nembe phono-semantic rules and ranking effects

Absolute size ideophone

Vowels produced with wide pharynx refer to large or neural size/Vowels produced with narrow pharynx refer to small size

Phonology		Semantics	Ranking
V [α wide]	→	[α LARGE]	o in isolation specifies [+LARGE] at level n, ɔ in isolation specifies [-LARGE] at level n-4
C3 [-α alv, α vel]	→	[α LARGE]
{r, t, d, ɗ}	→	[-LARGE]	r in isolation specifies [-LARGE] at level n-1
{k, g}	→	[+LARGE]	k in isolation specifies [+LARGE] at level n-3
C1 [α grave]	→	[α LARGE]
{ɗ}	→	[-LARGE]
{gb}	→	[+LARGE]

Example	Translation	Pattern
gbɔkɔgbɔkɔ	very tall or long	k [+LARGE] outranks ɔ [-LARGE]
sokosoko	tall or high	k [+LARGE] agrees with o [+LARGE]
sorosoro	bulgy, swollen	o [+LARGE] outranks r [-LARGE]

Reduplication

Ideophones are a highly flexible part of the expressive lexicon and do not depend – but certainly do make use of – phonetic symbolism, but it is mainly repetition, reduplication and other varieties of intonational elaboration can add meaning to an ideophone (Nuckolls 1995).

Examples taken from Nuckolls (1995: 154-56) in Quechua

Reduplication
N: Charapa wawa sa sa sa sa sa tiyumanda l^yukshinawn’?

The turtle hatchlings emerge from the sand sa sa sa sa sa?

Reduplication, lengthening and falling pitch (indicated with smaller font)
M: P^haa pa pa pa pa pa pa pa pawasha rig, kuti mandzhariracha.

Flying p^haa pa pa pa pa pa pa pa, it (a condor) went off because it was frightened.

Lengthning and reduplication
E: T^haaaaaay t^haaaay t^hay! Ay Janel^ya uyangi ma yanga t^haaaay t^baay ton ton ton ton ton ton ton ton ton ton

T^haaaaaay t^haaaay t^hay! Oh Janet, you would hear (them) just t^haaaay t^baay ton ton ton ton ton ton ton ton ton ton

t^hay – simulation of rifles firing

ton – thud-like sound of ammunition hitting a body

Reduplication can occur in a variety of ways, each with a slightly different meaning (Diffloth, 1976). Diffloth identifies major reduplication, minor reduplication, and two types of modified reduplication.

Major reduplication occurs in ideophone roots with CVC structure, so that the entire CVC string is reduplicated usually three times. Major reduplication is used to represent “repetition at intervals of time” (Diffloth, 1976).

For example, the root /tus/ is repeated three times, surfacing as /tustustus/ for ‘the repeated sound of running fast.’

Minor reduplication, which represents “prolongation or continuous repetition in time,” consists of taking the first and last consonant of the ideophone, and prefixing them to the ideophone.

For example, if the root /dyɔ:l/ undergoes minor reduplication, the first consonant “d” would be combined with the last consonant “l” at the beginning of the ideophone, so that it becomes “dldyɔ:l,” which means ‘appearance of an object which goes on floating down.’ In the first type of modified reduplication, the ideophone is completely repeated, but the major vowel is modified. The major vowel retains any nasality and lengthening, but front vowels become back vowels (i.e. /ɛ/ becomes /u/), while back and central vowels are advanced (i.e. /u/ and /a/ become /ɛ/). This type of reduplication is used to signify “irregularity distributed in time or space.”

Example of front vowel becoming /u/ from Diffloth (1976)

  /klcwɛc/ → /klcwuc klcwɛc/ 
  ‘irregular flapping circular movements

Example of back vowel becoming /ɛ/ from Diffloth (1976)

  /mŋu:y/ → / mŋɛ:y mŋu:y/
  ‘people in a crowd raising their heads here and there’

Example of central vowel becoming /ɛ/ from Diffloth (1976)

  /prada:k/ → /pradɛk prada:k/
  ‘noise of scattered large drops of rain falling on leaves or roof’

The second type of modified reduplication, which has the same connotation as the first type of modified reduplication, is more restricted and only occurs in specific dialects of West Semai located in the Kampar basin (Diffloth, 1976). This type of reduplication consists of only repeating the main syllable of the ideophone, and prefixing /ma-/ to the syllable.

For example, the root /r(ŋ)rŋʔaŋ/ has a main syllable of /ʔaŋ/, which is repeated. /ma-/ is then added to the repeated syllable, so that /r(ŋ)rŋʔaŋ/ becomes /r(ŋ)rŋʔaŋ maʔaŋ/. It means ‘appearance of irregular cracks (e.g., in earth or durian fruit).’

Maduka, O. N. (1988). Size and shape ideophones in Nembe: a phonosemantic analysis. Studies in African linguistics, 19(1).

Maduka, O. N. (1991). Phonosemantic antecedents of some verbs in Igbo. Journal of West African Languages, 21(1), 105-15.

Maduka, O. N. (1991). Phonosemantic rules and hierarchies: Evidence from roundness ideophones in Hausa. Afrikanistische Arbeitspapiere, 26, 167-175.

Maduka, O. N. (1992). A PHONOSEMANTIC ANALYSIS OF NEMBE REDUPLICATED SOUND IDEOPHONES. Frankfurter afrikanistische Blätter, (4), 71.

Fordyce, J. F. (1983). The Ideophone as a phonosemantic class: the case of Yoruba. Current Approaches to African Linguistics, 1, 263-278.

Akita, K. (2011). Toward a phonosemantic definition of iconic words. Semblance and Signification, 10, 3.

Hinton, L., Nichols, J., & Ohala, J. J. (Eds.). (2006). Sound symbolism. Cambridge University Press.

Shinohara, K., & Kawahara, S. (2012). A cross-linguistic study of sound symbolism: The images of size. In Proceedings of BLS (Vol. 36).

Masuda, K. (2007). The physical basis for phonological iconicity. Insistent Images. Amsterdam/Philadelphia, 57-71.

Schmidtke, D. S., Conrad, M., & Jacobs, A. M. (2014). Phonological iconicity. Frontiers in psychology, 5.

Sound Symbolism in the Languages of Australia

Diffloth, Gérard. (1976). Expressives in Semai. Oceanic Linguistics Special Publications, No.13, Austroasiatic Studies Part I, 249-264. Retrieved from http://www.jstor.org/stable/20019159

Nuckolls, Janis B.. (1999). The case for sound symbolism. Annual Review of Anthropology, 28, 225-252. Retrieved from http://www.jstor.org/stable/223394

Syntactic Category

Interjections

Ideophones are a part of speech that shares the same status as interjections, as they both express emotion and neither is subject to the rigid constraints other grammatical word classes are governed by (Doke, 1948). Other word classes are governed by rules of length, stress, and tone, whereas ideophones can have variable tone, length, and stress. "Ow!" in (1) is an interjection in English for expressing pain. "Ow" can be lengthened with a different prosodic melody to express a slightly different meaning, such as, "Owwwwwww" in a whiny tone, with the connotation of complaining and trying to get someone's attention. In terms of variable stress, “Ow” can be stressed, or exclaimed louder, to demonstrate a sharp, sudden, unexpected pain as in “OW!!” With the additional stress, it is perceived as more painful than “Ow.”

(1) "Ow!"

Nouns

"Slap" in (1-3) is an ideophone in English expressing a painful, intentional contact of the hand against a surface. The English ideophonic noun "slap" can be lengthened reduplicated (1), lengthened (2), and verbalized (3).

(1) reduplication of the ideophonic noun slap
    a. The slap-slap-slap of the water against the shore.
    b. The water went slap-slap-slap against the shore.

(2) expressive lengthening of the ideophonic noun slap
    a. The slaaap-slaaap-slaaap of the water against the shore.
    b. The water went slaaap-slaaap-slaaap against the shore.

(3) verbalization of the ideophonic noun slap
    a.   The water slapp-ed against the shore.
    b. ? The water slap-slapp-ed against the shore.
    c. ? The water slaaapp-ed against the shore.

Regular English nouns like bottle (4a) are not expressive, cannot be reduplicated (4b) or lengthened (4c), but they can be verbalized (4d).

(4) regular noun "bottle"
    a.   The bottle floated away.
    b. *The bottle-bottle floated away.
    c. *The booottle floated away.
    d.  They bottle-d three cases of wine.

Dhoorre and Tosco suggest that ideophones in Somali belong to a subclass of nouns (1998). Like nouns, they have a definite suffix or anaphoric determiner and have feminine gender.

In (5a), because there is a noun, the noun is focalized. In (5b), since there is no noun, the ideophone carries the focus, demonstrating how ideophones are nouns in Somali.

From Dhoorre and Tosco (1998)
(5)a. Affirmative, declarative sentence with both ideophone and noun
      irriddii ayaa qab tiri
      door.F.ANAPH FOC IDEO said.3F
      ‘the door banged’
   b. Affirmative, declarative sentence without the noun, only the ideophone
      big buu yiri
      IDEO FOC.3M said.3M
      ‘he left suddenly’

Adverbs

Syntactically, ideophones in Bantu languages are their own word class, but behave like descriptive adverbs, so they contain descriptive information about manner, colour, sound, smell, and so on. They do not carry information about time or place (Doke, 1948). Examples: The Shona ideophone for bright, “ngwe” (1) describes colour, and state. The intensity can vary depending on the length of the ideophone (i.e. longer vowel means brighter). “Ngwe” usuall references the moon, so that could be a possible reference to place.

  (1) Mwedzi wakachena kuti ngwe-e. 
      mu-edzi u-aka-chen-a ku-ti ngwe-e.
      C3-moon SM-PAST-V(to be bright)-FV SM-say IDEO-bright.
      The moon was extremely bright.

Without the vowel lengthening in “ngwe-e,” the moon as perceived as less bright with the ideophone “ngwe.” Without the ideophone completely, the sentence would mean, “The moon was bright.” This demonstrates how the ideophone “ngwe-e” acts as an adverb. The Shona ideophone for snatch, “biku” (2) describes manner, and action. It also sounds “fast,” possibly due to the vowel or the velar stop /k/, so it might describe duration/time as well.

  (2) Imbwa yakati nyama biku.
      C9-dog SM-PAST-say C9-meat IDEO-pick up/snatch.
      The dog snatched the piece of meat.

“Biku” can be verbalized (3), and though the sentence with the verbalized ideophone has the same meaning, is it perceived as less dramatic than the ideophonic, adverbial form.

  (3) Imbwa yakabikura nyama.
      C9-dog SM-PAST-snatch-verbalizer-FV C9-meat.
      The dog snatched the piece of meat.

In Chasu, ideophones are identical to adverbs in distribution and function (Mreta, 2012). The following example (4) demonstrates that ideophones and adverbs both occur post-verbally, both serve to modify the verb, and never co-occur (Mreta, 2012)

  From Mreta (2012)
  (4) Adverbs occur post-verbally
      a. é‐ m‐ bigh‐ ie kwerí
         SPM OM beat COMPL1‐ ADV
         “S/he beat him/her severely”
      b. Ideophones also occur post-verbally
         é‐ l‐ a pa‐pa‐pa
         SPM eat FV IDEO
         “S/he eats hurriedly”
      c. Ideophones and adverbs cannot co-occur
         *é‐ l‐ a pa‐pa‐pa na mnó
         SPM eat FV IDEO and ADV
         *“S/he eats very fast and in a non‐stop motion and very much”

Adjectives

Ameka argues that ideophones belong to the adjective class in Ewe (2001). Adjectives in Ewe appear immediately after the noun, and their function is to describe a property of the noun, which is exactly ideophonic adjectives behave (Ameka, 2001). Example (1) demonstrates that ideophonic adjectives are located in the same place and do the same things as simple, non-ideophonic adjectives.

  From Ameka (2001)
  (1)a. Ideophonic adjectives appear after the noun
        nenem nyɔnu kɔkɔ tralaa dzetugbe eve ma- wo koŋ ko
        ID N ADJ ADJ ADJ QT DEM PL INT INT
        such woman tall slender beautiful two that PL just only
        ‘only those same two tall slender and beautiful women’
     b. Simple, non-ideophonic adjectives also appear after the noun
        awu ɣi/ɣe la
        N ADJ DEF
        garment white DEF
        ‘the white garment’

Heterogeneous categories

Ideophones do not belong in their own category because they have different syntactic properties in each language (Newman 1968). In Hausa, they act as adjective-intensifiers, verb intensifiers, or descriptive adverbs, whereas in Tera, they act as adjectives or adverbs.

Hausa

Ideophones are adjective-intensifiers, verbal-intensifiers, or descriptive-adverbs in Hausa. In Hausa, ‘expletive ideophones’ strengthen the preceding adjective (1).

  From Newman (1968)
  (1) a. fari fat
         white IDEO-snow
      b. ja wur
         red IDEO-blood
      c. baki kirin
         black IDEO-coal

The adjective and the adjective-intensifying-ideophone must come after the head noun, whereas the adjective can precede the noun (2) (Newman 1968).

  From Newman (1968)
  (2)a. rago fari fat
        ram white IDEO-snow
        ‘a snow white ram’
     b. *fari fat rago
     c. farin rago
        ‘a white ram’
     d. rago fari
        ‘a white ram’

The adj+IDEO construction can appear in equational sentences, questions, and negative sentences, but they are always in the “noun, adj, ideo” order (3) (Newman, 1968). In all three sentence types, the ideophone has the same meaning and sound.

  From Newman (1968)
  (3)a. Equational sentence
        ragonsa fari fat ne
        His ram is snow white.
     b. Question
        ragonsa fari fat ne?
        Is his ram snow white?
     c. Negative sentence
        ragonsa ba fari fat ba ne.
        His ram is not snow white.

The intensifier and stabilizer “ne” can be optionally inserted between the adjective and the ideophone in affirmative declarative sentences, and the meaning remains the same as the equational sentence (4) (Newman 1968).

  From Newman (1968)
  (4) ragonsa fari ne fat.
      His ram is snow white.

In Hausa, ideophones can also immediately follow verbs as verbal-intensifiers (5) (Newman 1968). These can occur in a variety of syntactic constructions, and are not just limited to affirmative, declarative sentences.

  From Newman (1968)
  (5)a. Affirmative declarative sentence
        ya chika pal
        He filled it to the brim.
     b. Affirmative declarative sentence
        yak one kurmus
        It burnt to the ground.
     c. Question 
        yak one kumus?
        Did it burn to the ground?
     d. Negative 
        bai cika pal ba
        He didn’t fill it up completely.
     e. Imperative
        cika ta pal!
        Fill it full!

Ideophones can also follow verbs as descriptive-adverbs (6). These only appear in affirmative, declarative sentences, and do not have the freedom of appearing in different syntactic constructions (Newman 1968).

  From Newman (1968)
  (6)a. Affirmative declarative sentence
        ya tashi farat
        He got up in a flash
     b. Affirmative declarative sentence
        ya fadi sharap
        He fell headlong
     c. *Question
        *ya fadi sharap?
        *Did he fall headlong?
     d. *Negative
        *bai tashi farat ba
        *He didn’t get up in a flash.
     e. *Imperative
        *tashi farat!
        *Get up in a flash!

Tera

In Tera, ideophones fall into two classes: adjectival or adverbial. While adjectival ideophones are a sub-class of adjectives, adverbial ideophones constitute a distinct word-class.

Tera adjectival ideophones are adjectives

Ideophonic adjectives can be used anywhere a regular adjective can be used. This is illustrated in examples (1) to (3). Adjectives — whether non-ideophonic or ideophonic — occur in affirmative clauses (1), interrogative clauses (2), as well as negative clauses (3). On the basis of the fact that Tera ideophonic adjectives and regular adjectives have the same distribution, Newman (1968) concludes that ideophonic-adjectives, rather than being a separate word class from adjectives, are a subclass of adjectives with special phonological properties.

	Clause type	Non-ideophonic adjective	Ideophonic-adjective
(1)	Affirmative	wudi-a mbarə	wudi-a ndolndol
		milk-X good	milk-X cool/fresh
		'The milk is good.'	'The milk is cool (i.e. fresh).'
		INSERT REFERENCE	INSERT REFERENCE
(2)	Interrogative	ndogd-a tada mu?	ndogd-a 'kwakədak mu?
		stone-X heavy Q	stone-X hard Q
		Is the stone heavy?	Is the stone hard?
		INSERT REFERENCE	INSERT REFERENCE
(3)	Negative	wudi-a mbar ɓa	wudi-a ndolndol ɓa
		milk-X good NEG	milk-X cool NEG
		The milk is not good.	The milk is not cool.
		INSERT REFERENCE	INSERT REFERENCE

Tera adverbial ideophones are a separate word-class

In Tera, ideophones also appear as descriptive-adverbs, and like in Hausa, they are limited to affirmative, declarative sentences (2) (Newman 1968). They can also be general adverbs, and as general adverbs, they are not limited to affirmative, declarative sentences.

	Clause type	Adverb type	Example
(3a)	Affirmative	‘with’ + N	wa cə ɬəna ndə xaŋkal
			INSERT GLOSS
			'He worked carefully/with care.'
			INSERT REFERENCE
(3b)		‘like/as’ + N	wa pata ɓar kə ŋgəbəŋ
			INSERT GLOSS
			'He exchanged it foolishly/like a fool.'
			INSERT REFERENCE
(3c)		descriptive ideophone	wa vi nə dam cacaɬ
			INSERT GLOSS
			'He went out in a rushing manner.'
			INSERT REFERENCE
(3d)		general ideophone	wa xa ɣa tətək
			INSERT GLOSS
			'He sat quietly.'
			INSERT REFERENCE

	Clause type	Adverb type	Example
(4a)	Negative	‘with’ + N	nə cə ɬəna ndə xaŋkal
			INSERT GLOSS
			He didn’t work carefully.
			INSERT REFERENCE
(4b)		descriptive ideophone	nə vi n dam cacaɬ* ɓa
			INSERT GLOSS
			[Target: He did not go out in a rushing manner.]
			INSERT REFERENCE
(4c)		general ideophone	nə xa ɣa tətək ɓa
			INSERT GLOSS
			'He didn’t sit quietly.'
			INSERT REFERENCE

xxx

	Clause type	Adverb type	Example
(5a)	Question	‘like/as’ + N	wa pata ɓar kə ŋgəbəŋ ya?
			INSERT GLOSS
			'Did he exchange it foolishly?'
			INSERT REFERENCE
(5b)		descriptive ideophone	wa vi nə dam cacaɬ* ya?
			INSERT GLOSS
			[Target: Did he go out in a rushing manner?]
			INSERT REFERENCE
(5c)		general ideophone	wa ruɓa nda didiŋ ya?
			INSERT GLOSS
			'Did he injure him very much?'
			INSERT REFERENCE

Morphology

Verbalization

In many Bantu languages, ideophones are verbalized by adding verbalizing suffixes to ideophonic roots (Doke 1948). Three types of suffixes are found, according to whether they derive an intransitive verb, a transitive verb, or a causative verb. Below are representatives examples of these suffixes from three Bantu languages: Lamba, Zulu, and Shona. In Lamba, the suffix -k derives intransitive verbs, -l derives transitive verbs, and -sy derives causative verbs. Zulu likewise uses -k and -l to derive intransitive and transitive verbs respectively; causative verbs are formed with -z (Doke 1948:292). Similarly, Shona uses -k and -r to derived intransitive and transitive verbs, and -dz to derive causative verbs.

Language	Lamba	Zulu	Shona
Intransitive form	-k	-k	-k
(1)	wêwû-k-a	qhofo-k-a	samhu-k-a
	coming.off-INTR-FV	crushing.in-INTR-FV	growing-INTR-FV
	‘come off’	‘crush in’	‘grow’
Transitive form	-l	-l	-r
(2)	wêw-l-a	qhofo-l-a	saku-r-a
	snipping-TRANS-FV	crushing.in-TRANS-FV	cultivating-TRANS-FV
	'snip’	‘crush in’	‘cultivate’
Causative form	-sy	-z	-dz
(3)	awû-sy-a	FIND EXAMPLE	samhu-dz-a
	crossing-CAUS-FV		samhu-CAUS-FV
	‘take across’		‘cause to grow’

Lamba

[Comments from RMD: look at other references in the literature on Lamba to sort what's going on.]

In Lamba, the suffix -ma has the additional meaning of either ‘vibration’ or ‘state’. For example, when -ma is attached to the ideophonic root ‘yoko’, the meaning of the derived verb becomes ‘rattle’. (Doke, 1948, p. 292). The original meaning of the Lamba ideophone yoko is not provided in the paper.

(1)    -ma      'vibration'
    a. ŵimbi-ma 'quiver'
    b. yoko-ma  'rattle'

Other suffixes that can be added to ideophones in Lamba include -ma (glossed as a stativizing suffix) -ta, -mana and -njila (Doke 1948:292). The functions and meanings of these suffixes are not provided in the paper. Illustrative examples are given in (2) to (5).

(2) -ma 'state'  sana > sana-ma    'lie on the back
(3) -ta          lici > lici-ta    'rush'
(4) -mana        fulu > fulu-mana  'bend forward'
(5) -njila       fuka > fuka-njila 'creep along'

Ronga

In Ronga, verbalization of ideophonic roots is also achieved by certain suffixes, such as -ka and -kuta.

(6) a. dzuu
          red.IDEO
         'of redness'
    b. -dzu-k-a
          red-STATE-FV
           'be red'
(7) a.  ŝu
          noise of bird. IDEO
           'noise of bird-scaring'
    b.  ŝu-kuta
          'scare birds'

Japhug

light verb constructions

In Japhug, four light verbs can be used with ideophones, namely “the semantically empty stative verb pa, the transitive speech verb ti ‘say’, the manner deixis verb stu ‘so like…’, and its reflexive form ʑɣɤ-stu ‘act like…’ (Jacques, 2013, p.270)”. Identical construction is also evident in Shona, the light verb -ti (say) is always used in conjunction with and precedes the ideophones.

verbalizartion

There is a very rich and productive system of denominal prefixes, which is used to derive verbs from nouns, in Japhug (Jacques, 2014, p.278). A subgroup of these denominal prefixes can also be used to derive verbs from ideophones, namely ɣɤ-, sɤ- and nɯ- (Jacques, 2014, p.278).

(8) a. ɣɤ- derives intransitive dynamic verbs. These intransitive dynamic verbs are productive, 
      and can potentially be created from any ideophones that allow dynamic semantics
    b. sɤ-  derives transitive verbs with comparable semantics
    c. nɯ-  derives verbs used in idiomatic expressions whose meaning cannot be predicted

As shown by the examples below, the derivational pattern of the deideophonic prefixes (ɣɤ- and sɤ-) usually involves partial reduplication of the ideophonic root where the onset of the second syllable is replaced by the phoneme “ɭ” (Jacques, 2013, p.279). The example sentences below containing the specific verbalized ideophones are elicited, but were reevaluated by two speakers of Japhug and were deemed natural (Jacques, 2013, p.258).

When used in conjunction with ideophonic roots connoting concrete shapes such as zjɣɤ ‘tall’, the deideophonic prefix ɣɤ- derives verbs indicating an irregular motion without direction (Jacques, 2014, p.279). The example below is illustrated with the ideophonic root: zjɣɤ ‘tall’

(9) rŋɡɯ       nɯ    ɯ-zda                    rŋɡɯ      ra  kɯ-fse               ku-rɤʑi 
    boulder    TOP   3SG.POSS-companion      boulder    PL  NMLZ:S/A-be.like      IPFV-stay
    mɤ-kɯ-khɯ                ci,      tɕendɤre    tu-ɣɤ-zjɤɣɭɤɣ 
    NEG-NMLZ:S/A-be.able     INDEF    LNK         IPFV-DEIDEOPH:INTR-IDEO:DISORDER:tall
    nɤ   tu-ɣɤ-zjɤɣɭɤɣ                             kɯ-ra                 ci 
    LNK  IPFV-DEIDEOPH:INTR-IDEO:DISORDER:tall     NMLZ:S/A-have.to      INDEF
    pjɤ-ŋu
    EVD.IPFV-be
    ‘This boulder could not stay in place like the other boulders, it was always moving around, very huge.'

When the deideophonic root sɤ- is used in combination with partial reduplication in ɭ, the underlying meaning of the derived form represents a disorderly action (Jacques, 2013, p.280). The example below is illustrated with the ideophonic root: zjɣɤ ʼtallʼ

(10) ɭaʁjɯɣ ɲɯ-sɤ-zjaŋɭaŋ
     staff  CONST-DEIDEOPH:TR-IDEO:DISORDER:tall
     ‘He sways the staff in all directions.” (elicited)

No reduplication of the the ideophonic root is involved in the derivational pattern in the third deideophonic prefix nɯ-. nɯ- is usually used to derive verbs that express an action resulting in a state (Jacques, 2013, p.280). The example below is illustrated with the ideophonic root: ɕkrɤɣ ‘lying on a hard and cold surface’

(11) tɤ-aʑɯʑu-ndʑi   tɕe,   ɯ-zda 
     PFV-wrestle-DU   LNK   3SG.POSS-companion 
     pa-nɯ-ɕkrɣɤ                                              ʑo 
     PFV:3—>3’-DEIDEOPH:STATIVE-IDEO:lying.on.a.hard.surface  EMPH
     pa-tʂaβ
     PFV:DOWN:3—->3’-cause.to.fall.down
     ‘When they wrestled, he threw his adversary on the hard and cold ground.’ (elicited)

Ameka, F. (2001). Ideophones and the nature of the adjective word class in Ewe. In F. Voeltz & C. Kilian-Hatz (Eds.), Ideophones. Philadelphia: John Benjamins Publishing Company.

Doke, C. (1948). The Basis of Bantu Literature. Africa: Journal of the International African Institute, 18(4), 284-301. Retrieved October 10, 2014, from the JSTOR database.

Dhoorre, C., & Tosco, M. (1998). 111 Somali Ideophones. Journal of African Cultural Studies, 11(2), 125-156. Retrieved October 8, 2014, from the JSTOR database.

Jacques, G. (2013). Ideophones in Japhug (Rgyalrong). Anthropological Linguistics, 55(3), 256-287. Retrieved October 9, 2014, from the Project Muse database.

Mreta, A. Y. (2012). Ideophones in Chasu. 韓國아프리카學會誌¸æoeƒèªOE, 36(8), 3-50. Retrieved October 7, 2014, from the DBPia database.

Newman, P. (1968). Ideophones from a syntactic point of view. J.W.A.L., 2, 107-117. Retrieved October 9, 2014, from the Google Scholar database.

Semantics

Kevin, Virginia

Semantic fields

Since research began on sound symbolism and ideophones, the semantics of ideophones has been discussed. Sapir (1929) noticed that sound symbolic words denoted differences between small and large, which sparked a psychological interest in what can be called universal ideophonic intelligence. It was found through psychological studies that when testing subjects with nonce words, there is a striking similarity in the responses given for possible meanings of words. This similarity is dictated by the specific vowels and consonants that are used in the nonce words (Köhler 1929, Bentley & Varon 1993, McMurray 1960, Nielsen & Rendall 2011). These psycholinguistic findings are what truly revived the notion that some words may not be arbitrarily linked between form and meaning.

Semantic hierarchy

There is no agreed-upon cross-linguistic semantic classification for ideophones, but various scholars have tried to define semantic fields of ideophones for single languages (Awoyale 1983, Dhoorre & Tosco 1998, Dingemanse 2012).

Based on a sample of languages that have large ideophonic vocabularies, Dingemanse (2012) identifies five major semantic fields for ideophones: auditory percepts ("sounds"), followed by action ("movement", visual percepts ("visual patterns"), other sensory percepts, and inner feelings and cognitive states. Representative examples are given in the table below.

Semantic field	Ideophone	Gloss	Language
Sound	kpuk	a rap on the door	Ngbaka Gbaya
Movement	kiláŋ-kiláŋ	in a zigzagging motion	Ngbaka Gbaya
Visual patterns	tál-tál	pure white	Ngbaka Gbaya
Other sensory percepts	potil	soft and tender (surface)	Somali
Inner feelings & cognitive states	blbʔəl	painful embrassment	Semai

Dingemanse proposes that these semantic fields form an implicational hierarchy, as follows:

Sound > Movement > Visual pattern > Other sensory percepts >Iinner feelings & cognitive states.

The presence of ideophones from the lowest ranked (rightmost) semantic field, implies the existence of ideophones from more highly ranked semantic fields. This hierarchy predicts that sound-based ideophones will be the most common cross-linguistically, while ideophones that connote inner feelings and cognitives states will be the least common. This implicational hierarchy predicts the existence of five types of languages. Dingemanse specifies that some of these types of languages have been heavily documented, but some of the types we expect are not exemplified by Dingemanse. As is shown in the table, Dingemanse states that Japanese and Korean are strong examples of languages with all semantic fields represented. Then, Dingemanse broadly states that many African ideophone systems have ideophones illustrating many sensory percepts, but doesn't give specific language examples. Dingemanse then skips over the type of language that includes auditory, action and visual percepts. He does not exemplify this type. Dingemanse cites Pastaza Quechua and Upper Necaxa Totanac as languages with auditory and action ideophones. Lastly, Dingemanse cites Navajo as well as a general claim that many languages of the Americas only have ideophones for sounds. While Dingemanse does some of the work to support his proposed implicational hierarchy, he does not fully evidence his proposal.

Language type	Auditory percept	Action	Visual percept	Other sensory percept	Feelings & cognition
Japanese, Korean	Yes	Yes	Yes	Yes	Yes
African ideophone systems	Yes	Yes	Yes	Yes	No
?	Yes	Yes	Yes	No	No
Pastaza Quechua, Upper Necaxa Totonac	Yes	Yes	No	No	No
Navajo, languages of the Americas	Yes	No	No	No	No

Methods for investigating semantic fields

There have been a few methods put forth for identifying semantic fields within and across languages (Kohler, 1929, Samarin, 1967, Dingemanse, 2011). These are important to consider because a common critique of semantic analyses of ideophone inventories is of using theories that apply well to English but not well to other, richer, ideophonic languages (Dingemanse 2011). It is important to note that some sort of methodological approach must be outlined in order to preserve the systematic and replicable study of ideophones cross-lingustically as well as methodologies that reduce any biases for certain types of ideophonic languages over others.

Nonce words: Köhler 1929

Köhler (1929) found that, when presented with nonce words, subjects performed similarly in assigning a particular word to a semantic field relative to shape. Systematically variation of consonant and vowel melodies of nonce words were classified in similar ways: For example, participants tended to classify a nonce word such as takete as describing a jagged shape, while a word such as maluma tended to be classified as described a curved shaped. This has since come to be known as the takete/maluma phenomenon. These findings holds of other commonly tested nonce words; for example, bouba tends to described curved shapes, while kiki tends to describe jagged shapes. The jagged and curved shapes used in the original experiments by Köhler are shown in figure (1). This experimental protocol — where participants are asked to match nonce words with various shapes — has been replicated in many different ways with speakers of many different languages. The same results continue to be found (Nielsen & Rendall, 2011).

File:Kohlertaketemaluma.png

Figure (1) adapted from Nielsen & Rendall

Multiplication of synonyms: Samarin 1967

The use of ideophones is highly context dependent, so this makes elicitation difficult. Samarin (1967) solves this problem with an elicitation method which involves deliberately multiplying synonymies. With multiplication of synonyms, for each ideophone elicited, the elicitor tries to find all related contexts and related ideophones. For example, a speaker could be asked if their language has an ideophone to describe the brightness of the moon. Then the speaker could be asked if that ideophone can be used to describe the brightness of any other contexts, like a lamp or the sun. It can be noted if the original ideophone can be used in these different contexts. IT can also be noted if not, what different ideophones can be used in these different 'brightness' contexts. Then the speaker could be asked if there are other ideophones that can be used to describe the brightness of the moon, and start over again. In this way the related meanings and contexts of ideophones can be traced and mapped.

This approach acknowledges that the accurate description of the meaning of an ideophone can be arrived at only by examining it context-of-use. It also has the advantaged of tracking the possibly multiple meanings that an ideophone can have in different contexts.

Stimulus based elicitation: Dingemanse 2011

Dingemanse (2011) uses stimulus-based elicitation: this entails presenting many different stimuli, covering as many semantic fields as possible, to a speaker and inquiring if there are any ideophones in their language that could be used to describe the stimulus. For example, a speaker may be shown a rabbit pelt and asked what ideophones could be used to describe the colour or texture. The same can be done for actions and sounds.

The drawback to stimulus-based elicitation is this it cannot target ideophones whose context-of-use is difficult to reproduce in a lab or elicitation setting. The advantage of this method is that it can be used to identify cross-linguistic patterns by using the same stimuli to test speakers of many different languages.

Two dimensions: Kita 1997

Based on data from Japanese ideophones (also called mimetics), Kita (1997) develops a two-dimensional analysis. In particular, while ordinary lexical items occupy an ‘analytic’ dimension of meaning, ideophones occupy an ‘affecto-imagistic’ dimension of meaning. Kita argues that the existence of these two separate dimensions has psycho-linguistics implications for how ideophones are processed.

Analytic dimension

The analytic dimension is where the semantics of the regular lexicon, and names is derived. This is the dimension of decontextualized predication, and consists of a hierarchy of all items in a sentence combining to create the meaning. Kita stresses that this dimension is amodal, and not connected with any specific cognitions other than language processing. Below are a few examples of lexical items that exist in the analytic dimension:

Proper names: John, Mary
Verbs: run, eat
Nouns: dog, tree
Adjectives: green, furry

Affecto-imagistic dimension

The affecto-imagistic dimension is where the semantics of ideophones is derived. In this dimension, form are not syntagmatically related (as they are in the analytic dimensions); rather, they are spatio-temporally related. In this dimension, sensory experiences are re-experienced via a direct connection of an ideophonic word (in the auditory domain) with a sensory experience. Note that Kita's affecto-imagistic dimension is synaesthestic in that it involves the yoking together of distinct sensory percepts. This dimension includes ideophones primarily, and perhaps only.

Ideophones: whack, splat, sniff

Kita provides evidence for the affecto-imagistic dimension on the fact that there is a high correlation between ideophones and spontaneous gesturing as well as prosodic peaks. Specifically, Kita found that ideophones are often timed with a prosodic peak and moreover if an utterance contains both a prosodic peak and an ideophone then the prosaic peak always falls on the ideophone and not elsewhere in the utterance.

These findings are interpreted by Kita to support the existence of an affecto-imagistic dimension: he argues that paralinguistic content (such as gesturing) is derived from a dimension separate from the analytic dimension. Given that ideophones are often accompanied by spontaneous gesturing, he concludes that this because they occur the same dimension, namely they are both located in the affecto-imagistic dimension.

Interaction of the two dimensions

Kita argues that distributional differences between ideophones and regular lexical items support his claim that ideophones occupy a distinction dimension of meaning. The first diagnostic is related to semantic redundancy. As shown in (1), regular adverbs that denote similar meanings cannot co-occur: thus, the PP modifier isogi-ase de 'with hurried feet/hurriedly' cannot co-occur with the PP modifier haya-aruki 'walk hastily'. In contrast, as shown in (2), it is possible to combine an ideophonic adverbial modifier such as sutasuta 'hurriedly' with a regular adverbial modifier such as haya-aruki 'walk hastily', even though they contribute redundant information.

(1) a.  Taro wa [isogi-asi     de]  arui-ta.
         Taro TOP hurried-feet with walk-PAST
        ‘Taro walked hurriedly’
    b. *Taro wa [isogi -asi   de] [haya-aruki o]  si-ta. 
        Taro TOP hurried-feet with haste-walk ACC do-PAST
        [Target: ‘Taro walked hastily hurriedly’]
        [adapted from Kita 1997:pg388, (ex. 8)]

(2) a.  Taro wa  [sutasuta     to]   arui-ta.
        Taro-TOP hurrying.IDEO QUOTE walk-PAST
        'Taro walked hurriedly’
    b.  Taro wa   [sutasuta     to]   [haya-aruki o] si -ta. 
         Taro TOP hurrying.IDEO QUOTE haste walk ACC do-PAST
        ‘Taro walked hurriedly’
        [adapted from Kita 1997:pg388, (ex.8)]

A second distributional argument in support for Kita's claim that ideophones occupy a different dimension of meaning comes from their interaction with negation. In many languages, including Japanese, it is not possible for ideophones to occur in a sentence which is negated. This is illustrated in (3) for Japanese, which shows that a sentence with an ideophone resists negation.

(3) [PP [S tama ga  gorogoro to    korogat-ta no] de] wan a -i
           ball NOM IDEO     QUOTE roll-PAST 
   a. ≠ ‘it was not [the case that a ball rolled gorogoro]’ NEG>CP
   b. ≠ ‘it was not [rolling gorogoro] that a ball did’     NEG>[V+IDEOPHONE]
   c. ≠ ‘it was not [rolling] that a ball did gorogoro’     NEG>V
   d. = ?? ‘it was not [gorogoro] that a ball rolled’       NEG>IDEOPHONE
   e. = ?? ‘it was not [a ball] that rolled gorogoro’       NEG>DP
   [adapted from Kita 1997:pg390, (ex10)]

Kita argues that the incompatibility of ideophones with negation reflects the fact that the latter comes from the analytic dimension, while ideophones comes from the affecto-imagistic dimension, and that is not possible to negate both dimensions simultaneously. Note however, that while propositional scope negation (3a) and [V]-scope negation (3b-c) are unavailable, negation can scope over the ideophone (3d) or over DP (3e). Strictly speaking, if the impossibility of negation is sensitive to the two dimensions postulated by Kita, then this predicts (3a) and and (3b) will be ill-formed (which they are), but also predicts that (3c) will be well-formed, which it isn't. A possible confound is the fact that Japenese ideophones are introduced by quotative particles, which are part of the evidential system. On independent grounds, evidential cannot generally scope under negation, so this may the source of the ill-formedness of (3a-b-c). This remains to be confirmed.

Morpho-semantics of ideophones

Temporal Aspect in Shona Morphology

Shona morphology makes use of a diversity of forms. There are several instances of reduplication, vowel lengthening and vowel repetition. Ideophone arguably are the closest linguistic representation real-world things, actions, feelings, sensation, etc., so it seems likely that their morphological forms should relate to their semantics. This section of Shona ideophonic semantics examines temporal aspect and it’s relationship to morphological form. The purpose of the remainder of this section is to prove (and or disprove) the following table:

Temporal Aspect	Iterative	Durative
Morphology	Reduplication	Vowel-lengthening
Phonology	CV~CV	CV-V

Evidence for the table

Reduplication and repetition

One type of morphology that is of particular interest in terms of its relationship with semantics is reduplication. Many of the reduplicated forms in Shona evoke an iterative sense or aspect.

gabada gabada	walking noisily
ga-gadhí ga-gadí	galloping
gedé gedé	rattling
go go go	dripping continuously on the same spot
ga ga ga	clapping

All of the above ideophones denote an action that involves repetition to some degree. Walking requires moving one’s legs in a repetitive motion. Clapping is also repetitive and clearly so is dripping on the same spot. The repetition of syllables shows that the action is made up of smaller actions which repeat over and over. One could even posit that the smaller syllables denote smaller actions. For example, gabada gabada is walking and ga ga ga means clapping. Walking involves a repetition of more larger actions than does clapping.

Vowel lengthening exhibits duration

Other morphological forms of ideophones in Shona denote other types of ideophones. Vowel lengthening shows a relationship with a durative aspect. A longer linguistic form (longer vowel) best represents an action that takes longer to happen so again we can see here how the meanings have a direct relationship to the forms.. Some examples from the data are seen below.

ga-a	shining brightly
gidhi-i	resounding in distance
go-o	tasting strongly
gú-ú	rumbling (stomach)

Evidence against the table

Counter evidence is a necessary element to effectively support this hypothesis. (Un)fortunately there is plenty of data that counters the above suggestions.

Reduplication is not exclusively iterative

First let us investigate the notion that reduplication creates an iterative aspect. The above examples of reduplication all show an iterative aspect. Positing that all forms with reduplication have this iterative aspect may be too much of an ambitious feat. The following ideophones serve as counter evidence that reduplication always involves a feeling of repetition:

gó gó gó	asking permission to enter
gábha gábha	slopping over
gegerere	ripping strong cloth

The sheer number of reduplicated forms that do exhibit this iterative aspect is staggering in comparison to the those that do not exhibit it. Having said that, there is evidence that does suggest a relationship between reduplication and iterative aspect, though there still requires further and more in-depth investigation.

Long vowels do not always indicate a long action

There is also data that suggest that vowel lengthening does not have a one-to-one relationship with durative aspect. The following form supports this:

gú-ú	bursting out laughing

Laughing on its own does have a longer feeling, but this form seems to focus on the initial burst before the actual laughter itself. Again, there is not sufficient evidence in either case that proves or disproves that vowel lengthening relates to durative actions. However, Initial examination suggests it does.