The Case for Sound Symbolism

by Janis B. Nuckolls
The Case for Sound Symbolism
Janis B. Nuckolls
Annual Review of Anthropology
Start Page: 
End Page: 
Select license: 
Select License


JanisB. Nuckolls

Department ofAnthropolo~, UniversiQ ofAlabama-Birmingham, Birmingham, Alabama 35294-3350; e-mail: 

Key Words: iconicity, semiotics, linguistics, language and culture

Abstract The proposal that linguistic sounds such as phonemes, features, syllables, or tones can be meaningful, or sound-symbolic, contradicts the princi- ~lesof arbitrariness and double articulation that are axiomatic to structural lin- guistics. Nevertheless, a considerable body of research that supports principles of sound svmbolism has accumulated. This review discusses the most widelv at- tested forms of sound symbolism and the research programs linked to sound symbolism that have influenced linguists and anthropologists most. Numerous reports of magnitude sound symbolism in the form of experimental studies and comparative surveys have been integrated into a biologically based theory of its motivation. Magnitude sound symbolism also catalyzed a number of experi- mental studies by psychologists and linguists in search of a universal sound- symbolic substrate underlying all languages. Although the search for a sound- symbolic substrate has been abandoned, the success rates of these studies have never been satisfactorily explained. Sound-symbolic processes have had a de- finitive impact on morphological analyses of phonesthemes and on historical linguists' understandings of diachronic processes. A typologically widespread form of sound symbolism occurs as a kind of lexical class known as the ideo- phone, which is conspicuously underdeveloped in standard average European languages, and highly perplexing for linguists and anthropologists. Although it has always been a respectable domain of inquiry in ethnopoetics and interpre- tive ethnography, the case for sound symbolism has of late been argued with re- newed vigor on the part of psychological anthropologists and philosophers who see a paradigm shift under way.

Introduction ...................................................226
Magnitude Sound Symbolism ..................................... 229
Experimental Work Using Invented Words .............................. 230
Comparative Surveys of Magnitude Symbolism. .......................... 231

The Frequency Code Hypothesis ...................................... 233

The Impact of Sound Symbolism on Psychology and Linguistics ........ 234
The Search for a Sound-Symbolic Substrate ............................. 234
Sound Symbolism in Morphological and Diachronic Analysis. .............. 237

The Lexicalization of Sound Symbolism: Research on Ideophones ....... 239
Formal and Semantic Anomalies ......................................240
Theoretical and Cultural Import of Ideophonic Research. .................. 243

Sound Symbolism in Ethnopoetics, Symbolic, and Cognitive Theory ..... 244 Reactions Against Arbitrariness. ...................................... 244 Sound Symbolism, Iconici~, Connectionism ............................. 245

Conclusion ................................................... 247


This review discusses and evaluates evidence for sound symbolism-a highly controversial subject that continues to provoke empirical study and debate among linguists, anthropologists, and philosophers. The problematic nature of sound symbolism arises from its conflict with the structural linguistic axiom that sounds do their work through contrastive relations with other sounds rather than through their intrinsic sound qualities. Consider a pair of words such as tip and dip. We cannot assign semantic responsibility for their different meanings to their con- trasting alveolar stops, Id1 and It/. It is the fact that one is pronounced with vibrat- ing vocal cords and the other with nonvibrating vocal cords that is conventionally understood as significant. This significance is not natural but learned. How, after all, can the relatively small number of contrastive sounds in a language's phone- mic inventory participate in the creation of infinite varieties of meaning if those sounds are themselves meaningful? If we take the perspective of a linguistic engi- neer constructing a code of unlimited communicative possibilities, as does Hock- ett (1960), the sounds of a language must be arbitrary. Linguistic sounds do more than communicate an unlimited number of messages, however. They express our emotional states, aesthetic apperceptions, and the alignments and interrelations we have with other members of our social world, none of which can be neatly separated from denotational reference. From a speaker's point of view, some- times called the ontological argument (Friedrich 1979), a language without sound symbolism would be as impossible as an existence without culture. Nevertheless, linguists' claims for sound symbolism are often weak and highly qualified, as when a descriptive report of sound-symbolic patterning merely concludes that the

principle of arbitrariness is not absolute.

The caution with which linguists have approached the subject of sound sym- bolism contrasts strikingly with much anthropological linguistic work concerned with conceptions of sound and meaning in small-scale societies. The language philosopher and literary critic Kristeva (1981:50) articulates what a number of anthropologists have observed: "While the practice of language really presup- poses for primitive man a distance with respect to things, language is not con- ceived of as a mental elsewhere, or as an abstract thought process. It participates as a cosmic element of the body and nature, and is joined with the motor force of the body and nature." Studies of the Navajo language and culture by Reichard (1944, 1950) and Witherspoon (1977) are some ofthe earliest ethnographic mod- els of such a conception of language. The Navajo view linguistic sound as analo- gous to the very forces of life itself. For the Navajo, assumptions about "sound power," a term described by Reichard (1944:51), follow logically from their beliefs about the interrelations between speech, breath, and movement (Wither- spoon 1977). Air is the ultimate source of all life and the medium through which sounds are patterned. Sounds have the power to attract good and exorcise evil (Reichard 1950:257). To speak and to sing in ritual contexts is to control air and, at the same time, create order, harmony, and beauty, thus making "contact with the ultimate source of life" (Witherspoon 1977:61). More recently, work by Basso (1985, 1995) among the Kalapalo of Brazil reveals belief in the possibility of unifying oneself with a greater reality by means of sound. Basso describes a cultural hierarchy of animacy structured by relative potency in sound production, with linguistic and musical sound constituting the most encompassing levels. The ability of humans to speak is subordinate to the ability of powerful beings to cre- ate sung sounds that humans imitate in ritual contexts, thereby unifying them- selves with each other and with the greater universe.

Work by Feld (1982, 1996) examines interconnections between linguistic and musical sound, emotional expression, and ecology among the Kaluli of Papua, New Guinea. Feld (1982) demonstrates linkages between linguistic sound sym- bolism and core cultural values of emotional and aesthetic expression. The Kaluli language features an abundant inventory of lexicalized sound-imitative words called ideophones, discussed in detail below, which figure prominently in poetic evocations of bird cries, waterfall sounds and movements, and environmental phenomena generally. Their expression overlaps with a number of interlocking Kaluli values, practices, and cultural frames of understanding. The sounds of birds, for example, are more ecologically salient than are their visual characteris- tics because, reasons Feld (1982), dense rain-forest foliage often obscures their appearance. Moreover, birds generally are significant as spirit reflections of the deceased. As such, their sounds have an affective importance, which is particu- larly evident in the myth about a little boy whose persistent requests to his older sister for food are repeatedly ignored. The result is that he is transformed into a muni bird, a type of dove, whose cries represent the apotheosis of sadness, loss, and abandonment and, as such, provide the template for a mournful and evocative melodic song pattern. Feld (1996) expands on the linkages between sound words and concepts of place, directly focusing on water, generally, and water sound words for their potential to define and evoke peoples' experiences of places worked in, lived in, and traveled through. Places and spaces are defined by the bodies of water that flow through them, thus defining their contours. The lexical- ized sounds of the waterfalls running through places are prominent components of the paths that are composed in song texts designed to appeal to peoples' senti- ments. At the heart of this exposition is a core analogy between water as a flow- ing, resounding movement, connecting multiple branches of spatial segments, and the voice as a resounding flow through the human body. Waterfall terms are routinely used as metaphors for discussing song structure (Feld 1982: 164, 1996: 127, 134). Claims by Feld for the unity of sound, bodily experience, and the envi- ronment are most powerfully illustrated by the description of a Kaluli woman's performance of a song as she sits beside a creek. She is described as singing near the water, about the water, like the water, and with the water (Feld 1996:132).

A number of anthropological accounts, then, have found that the interrelations between sound and meaning are anything but arbitrary and unmotivated. It is not even necessary to restrict our perspectives to small-scale societies in order to find concepts of sound symbolism. There have been attempts to rationally systematize language that support a concept of sound symbolism in a powerful way. In Hindu linguistic tradition, the problem of sound and meaning in relation to ultimate real- ity is highly foregrounded. Some schools in this tradition conceive of reality itself as composed of sacred sound (Beck 1993). Anthropological linguistic analyses repeatedly uncover conceptions of sound and meaning that are congenial with sound-symbolic principles. The time has come, then, to build a bridge between the vigorously thriving conceptions of sound symbolism nurtured by anthropolo- gists and the careful, parsimonious concessions made by linguists admitting to its existence. This review outlines the main trends in sound-symbolism research in an effort to assess its significance for changing conceptions of linguistic structure. One of the most widely attested semantic domains-that of magnitude sound symbolism-has fueled an increasingly influential, biologically based theory. Magnitude studies also energized a protracted search by psychologists and lin- guists for a sound-symbolic substrate underlying all languages. Sound-symbolism data has influenced a range of subfields, including morphology and morphosyntax, as well as diachronic studies. A typologically widespread form of sound symbolism concentrated within a kind of lexical class known as the ideo- phone is attested in many language families, although it is conspicuously absent in Western European languages and continues to perplex linguists, anthropolo- gists, and psychologists. Finally, the ramifications of sound-symbolism research for symbolic and cognitive theory are explored. Recent discussions of the princi- ple of arbitrariness have proposed that it is implicated in an information-processing model of thought, which is about to be usurped by a major paradigm shift.

The term sound symbolism is used when a sound unit such as a phoneme, syl- lable, feature, or tone is said to go beyond its linguistic function as a contrastive, non-meaning-bearing unit, to directly express some kind of meaning. Discus- sions of sound symbolism are indebted to the semiotic scheme outlined by Peirce (1955). A component of this scheme concerns three logically possible relations between symbols and their referents. These three logical distinctions, the icon, index, and symbol, are stock-in-trade terminology for semiotic analysis and are heuristically useful as long as one remembers that such clear-cut types are practi- cally nonexistent. When a unit such as a phoneme seems highly motivated or natural with respect to some articulatory or acoustic criteria, it may be called an icon, which refers, denotes, or communicates by its resemblance to a sensation, feeling, or idea. There are also subtypes of iconic signs, depending on whether they communicate through their simple qualities as images, through analogous relations in their component parts as diagrams, or through parallelism with some- thing else as metaphors. The second logically possible relation between a sign and its referent is the index, which communicates through physical contingency or symptomatic signaling. The third logically possible sign type is termed the sym- bol, by which is meant a conventional link between a sound and an idea, without any apparent motivation. An ideal type of sound symbolism would seek to iden- tify iconic and indexical relations, but all types of sound symbolism necessarily involve conventionally symbolic relations as well.

To illustrate the overlap between these three possible relations, consider into- nation, a subsystem that is not discussed in this review because it is one domain where claims for sound symbolism are relatively uncontroversial. Bolinger (1985) describes intonation as a relatively autonomous system, based in emo- tional and gestural expression, which communicates iconically, indexically, and conventionally through rises and falls in pitch. Consider, for example, that the tenseness of the vocal cords is the most important factor determining voice pitch (Ladefoged 1975:224). Greater vocal cord tension produces higher pitch, which can be indexical of a heightened emotional state (Huttar 1968). Consider, further- more, that the pitch of a sound and its frequency are in general related (Ladefoged 1975:162-63) and that high-frequency sounds such as Iil in heed, which have shorter wavelengths (Ultan 1978:545), are often associated with ideas of small- ness. What has just been described is an iconic relationship between shorter wavelengths of sound and ideas of smallness. An example of metaphorical iconic- ity is found in the relations between intonational profiles and pitch. The increase in tension used for higher pitch and the decrease in tension used for falling pitch have a universal tendency to be associated, metaphorically, with the contrasting ideas of incompleteness (high pitch) and completeness (falling or low pitch). Breaks in utterances, and also questions, are indicated in most of the world's lan- guages by rising pitch, whereas terminal boundaries of units such as clauses and sentences are marked by falls in pitch (Bolinger 1978). Consider, however, that at a certain point, lowering pitch actually requires more effort and tension on the part of the stemo-hyoid muscle, which in order to lower the pitch draws the larynx down (Bolinger 1986:387). Nevertheless, the association of tension and effort with high pitch and relaxedness with low pitch has achieved a kind of normative, conventional status, which is left intact, despite the fact that more effort and ten- sion are required to achieve the lowest pitches. In this instance, the associations between high pitch and low pitch are conventionally symbolic.


A basic issue in sound-symbolism studies is whether or not it is possible to find any universal connections between sounds such as phonemes and meaning. A major difficulty in attempting to make such a claim is that structure in Saussurian terms emphasizes the purely relational nature of signs. A phoneme can only exist as it configures into relations with other phonemes in a linguistic system. Because phonemic systems of every language are patterned differently, no cross-linguistic comparisons of sounds should be possible in principle. Nevertheless, as Anderson has stated (1985:34-35), the positive character of the sounds themselves cannot be strictly excluded from linguistics proper. If it is in fact possible to make any claims for universality in sound symbolism, then those claims will have to begin with magnitude sound symbolism. Magnitude sound symbolism encompasses a diverse group of studies focused on contrasts between big and small and on asso- ciated concepts. It is a widespread feature of Native American languages (Sherzer 1976), where alternations of consonants in roots function to express augmenta- tiveldiminutive contrasts. Beyond that, reports of magnitude symbolism usually concern vowel sounds, and it is most common to encounter reports that diminu- tive concepts are expressed by a vowel such as lil in heed articulated in the front of the mouth and close to the roof, making it high and front. Although there are also reports of back vowels such as la1 in father symbolizing augmentative concepts, most notably Sapir (1949a), the overwhelmingly abundant evidence is for diminutive symbolism. Concepts most often associated with diminutive symbol- ism include the following: physical smallness (Jesperson 1933, Ohala 1994); brightness or lightness (Newman 1933, Nichols 197 1, Fischer-Jorgensen 1978); quickness (Jesperson 1933, Berlin 1994, Nichols 197 1); singularity, proximity (Ultan 1978, Woodworth 1991); and attitudes such as affection, intimacy, dis- dain, and acquiescence (Jespersen 1933, Ohala 1994, Sapir 1949b, Silverstein 1994, Ultan 1978). The motivation underlying magnitude sound symbolism has been explained through a combination of articulatory, acoustic, and biological


Experimental Work Using Invented Words

Sapir (1949a) conducted a provocative set of experiments on magnitude symbol- ism using invented word pairs featuring contrasting vowels. A sequence such as ma1 and mil was assigned the meaning "table" and subjects were asked to decide which sequence of sounds meant the larger and which the smaller table. Sapir hypothesized that differences in vowel sounds might pattern according to natural parameters of magnitude, based on the size and configuration of the vocal- articulatory tract. When pronouncing la1 as in German Mann and then li/ as in French fini, the vocal tract changes from open and expansive to small and con- tracted. Some pairs of sounds, however, such as /el in ete and Iil in fini or lael in hat and /a/ in Mann, are less different from each other, i.e. they are more alike in their articulatory configurations. Sapir discovered a high degree of sensitivity on the part of his subjects to these relational differences. A word containing la1 was at least 80% likely to be judged with a larger meaning if paired with a word contain- ing lil. When paired with vowels that were closer in the vocal articulatory tract,

i.e. less different, however, the likelihood of a word with the vowel la/ being judged larger falls as low as 56%. Sapir's study was elaborated by S Newman (1933), who computed scales for vowels and consonants based on subjects' judgements of relative magnitude and also brightness. Newman discovered that articulatory position and acoustic resonance are the most significant objective factors affecting subjects' judgements of vowel magnitude. As the tongue recedes in articulating vowels from the front to back of the mouth, and as acoustic fre- quency becomes lower, the vowels are judged to be larger and darker (S Newman


The significance of the studies of both Sapir (1949a) and Newman (S Newman 1933) is that speakers attribute magnitude symbolism to articulated sounds according to principled, objective criteria. Neither Sapir nor Newman, however, was willing to concede that such patterning was related to denotational expres- sion. A number of scholars, however, have made the stronger claim that high front vowels such as Iil do tend to occur more frequently in words denoting smallness (Jespersen 1933, Taylor 1963, Thorndike 1945). Additional criticism of the work of Sapir and of Newman came from Bentley & Varron (1933), who argue that the design of the studies forced subjects to make judgements regarding magnitude and brightness. They find only moderate success rates when subjects are given the option ofnot finding two forms comparable with respect to some attribute. Never- theless, the positive responses did confirm the findings of Sapir (1929,1949a) and Newman (S Newman 1933) that la1 sounds were judged larger than iil sounds. Bentley & Varron (1933) are also critical of interpretations by both Sapir and Newman of their results, stating that the magnitude and brightness differences attributed to linguistic sounds as phonetic symbols are actually the result of rela- tional, proportional differences that are inherent to a variety of natural and musi- cal sounds. Although the criticism by Bentley & Varron raises interesting questions about the status of magnitude symbolism as a uniquely linguistic phe- nomenon, their critique of the Sapir and the Newman studies fails to acknowledge the significance of the fact that subjects are highly susceptible to suggestions of the meaningfulness of sounds. In any case, their critiques did little to diminish a subsequent wave of enthusiasm for more experimentation and research on sound symbolism (discussed below).

Comparative Surveys of Magnitude Symbolism

Jespersen (1933) conducted an informal survey of diverse languages regarding words denoting "little" and associated semantic concepts such as child, small ani- mal, etc. They found that the high front vowel lil prevailed. The fitness between lii and ideas of smallness was explained with a combination of articulatoryigestural and pitch-related factors. When pronouncing lii, the tongue is raised and moved forward, which creates a small space between the front of the tongue and the roof of the mouth. This is in contrast to the large space created by lowering the tongue for la/, law/, and 101. This was not, as Jespersen conceded, a systematic search through each language's lexicon, but rather a collection of examples that fit the author's belief that diminutive concepts are sound-symbolically suited to expres- sion with high front vowels (Jespersen 1933). A different sort of survey was con- ducted by Nichols (1971), who considered the magnitude symbolism of consonants in indigenous languages of western North America. Many of these languages alternate consonants, chiefly in noun and verb roots, to express diminu- tive, augmentative, and related ideas. Nichols identified two main principles underlying diminutive consonant shifts: tonality and hardness. Tonality involves the raising of consonantal frequency by a more forward articulation or by palatali- zation, which raises a consonant's second formant. Hardness shifts involve a more forceful manner of articulation, involving more tension and muscle activity rather than a change in point of articulation. The articulatory rationale for corre- lating higher tonality with diminutiveness seems clear: There is a reduced reso- nance chamber, which is congenial with ideas of smallness. The relationship between hardness and diminutiveness is less clear. It may be, as Nichols (197 1) suggests, that extensions of diminutive concepts, such as quickness, brightness, and hardness, can be reconciled with articulatory hardness.

A number of significant observations emerge from this study. Nichols states that the perceptual properties of tonality and hardness account successfully for diminutive shifting in a way that formal representations using distinctive features cannot. Furthermore, for tonality shifts at least, the traditional consonantal land- marks of the vocal tract, namely labial, dental, and velar, are highly salient. A given tonality shift will occur among dental, alveolar, and palato-alveolar conso- nants, or among velar, postvelar, and palato-velar consonants, but will not cross over from the anterior to the posterior regions of the vocal tract. Nichols (1971) hypothesizes that diminutive shifting is an areal feature that has spread by bor- rowing. A number of facts support this hypothesis. Specific types of shifts have discernible patterns of geographical distribution. Glottalization, for example, is a northern shift characteristic of the Salish languages. Dental resonant shifting, a type ofhardness shift, is found further south, extending as far as Mexico. Borrow- ing is also evident by the fact that unrelated languages such as Nez Perce, North- ern Paiute, and Southern Sierra Miwok all share the same shift of s > c or E. Diminutive shifts are linked to more general morphophonemic processes. Glot- talization, for example, has the status of a morphophonemic rule in a number of dialects. Only dialects that lack this rule, e.g. Wishram, will use glottalization for diminutive symbolism. A language may execute a diminutive shift using analo- gous segments from its own phonemic inventory or create a new nonphonemic segment, which ultimately becomes phonemicized. In fact, the elaborateness of the consonantal phonemic systems of these languages may be attributable to the widespread borrowing of shifting. Aoki (1994: 18) suggests that diminutives may be a source of new lexical items as well, at least for Nez Perce. Diminutive shift- ing also functions to distinguish special genres of recitation or speech play from ordinary language, or to distinguish particular characters in myths in Nootka (Sapir 1949b; but see Hymes 1981), Sahaptin (Jacobs 193 I), and Nez Perce (Aoki


Ultan (1978) compared 136 languages for size ablauting, i.e. the alternation of a sound such as a consonant, vowel, or tone within a root, the function of which is to indicate a contrast in size. The goal of this study was to test whether diminutive phonemes would exhibit marked features, whether vowels expressing diminu- tiveness would be high and/or front, and whether diminutive consonants would be palatal or fronted. Although data were insufficient to confirm the first two hypotheses, languages using vowels to express diminutive concepts did make use of high front vowels and high tones. Hypothesis three was confirmed but weak- ened by the distributional skewing in favor of North American Indian languages, which comprised more than 65% of those languages illustrating size sound sym- bolism. More significant than the inconclusiveness of the original hypotheses is the discovery of interconnections between ideas of magnitude as physical size and other semantic and grammatical concepts. A third of the languages surveyed feature distance symbolism, which correlates with size symbolism. Languages with proximal/distal contrasts in their pronominal systems or locative adverbs exhibit a preference for front or high vowels correlating with proximity to speaker, an observation first made by Jespersen (1964:402-3). Number symbol- ism is also attested, with correspondences between singular and diminutive. Regarding distributional symbolism, small uncountable quantities correspond to diminutive types. A related study by Woodworth (1991) used a more random sample of languages to test for the relevance of acoustic criteria in relation to proximal/distal concepts associated with diminutive. She found that when deictic pronouns, place adverbs, and directional affixes were used to express proximal meanings, they made use of higher-pitch vowels than when they expressed distal meanings. One final survey to be mentioned (Berlin 1994) found magnitude sym- bolism in ethnozoological nomenclature. Bird and fish names from Huambisa, a Jivaroan language, were compared. Some comparative data from the nonrelated languages Wayampi (Tupian) and Apalai (Cariban) is also supplied. Birds and fish that are physically small, i.e. 10 in. or less in length, tend to be named with high-frequency sounds, especially high, front /i/, whereas larger birds and fish tend to have the low-frequency segments /a/ and /u/ in their names.

The Frequency Code Hypothesis

Any uttered sound, such as a vowel, has an overall or fundamental pitch, also called fundamental frequency or FO, in addition to the characteristic overtones that distinguish it from other vowels. Fundamental frequency figures in a biologi- cally based theory of magnitude sound symbolism called the frequency code hypothesis (Ohala 1994). Central to this theory is the claim that intonation and acoustic frequency represent a privileged domain for sound-symbolism research, with the potential to unify claims for magnitude symbolism in vowels, conso- nants, tones, and intonation. At the same time, Ohala defends the principle of arbi- trariness as a "productive working principle" that should not be abandoned. Essential to his argument is evidence from research on intonation (Bolinger 1964, 1978; Brown & Levinson 1987), which reveals that rising intonational profiles, or high FO, are used to signal questions, politeness, deference, and submission, whereas falling intonation is used to mark statements, assertiveness, aggressive- ness, and authority. Ohala (1982) demonstrated that dominance could be corre- lated with low FO, by testing speakers' perceptions of short speech fragments that had been stripped of their linguistic content, as well as evidence of speakers' gen- der. Subjects overwhelmingly judged the low-frequency fragments, and also those with steep terminal falls, as more dominant. To explain these correlations, Ohala cites Morton (1994) and others who believe that a primary function of the vocalizations of nonhuman terrestrial vertebrates is the signaling of size in face- to-face competitive encounters. Larger vibrating membranes will have lower fun- damental frequencies. A lower FO will therefore signal a larger vocalizer, as well as related concepts such as aggression and dominance, whereas a higher FO will signal appeasement and related concepts.

FO is also related to sexual differences in vocal tract anatomy. In many species, the male's larynx is significantly larger, longer, and, in humans at least, lower than the female's, which lowers the FO. Because this differentiation takes place during sexual maturity, it must function, reasons Ohala (1982), to enhance the acoustic frequencies of male vocalizations, giving them a basis, namely size, for competing with other males for the favors of females. There exists abundant evi- dence that could be used to support the frequency code (Berlin 1994, Childs 1994, La Polla 1994, Matisoff 1994, Miron 1961, Orr 1944, Samarin 1965). There have been objections to the frequency code, however. Diffloth (1994) argues against its universality, citing evidence from Bahnar to argue that there is articulatory justifi- cation for the reverse of the frequency code as well. In Bahnar there is a correla- tion between high vowels and bigness, which can be justified, according to Diffloth, on the basis of the tongue's relative volume within the articulatory tract. For high vowels, the tongue occupies more space within the mouth. Although Diffloth's assertion that iconicity can be justified on a number ofbases, due to the "rich sensation package" of articulatory movements, is logically correct, it begs the question of why certain justifications seem to have so much salience. Indeed, given that numerous exceptions to the frequency code could in fact be cited, why do so many linguists find the frequency code, whether or not they refer to it as such, explanatorily appealing? Equally problematic is the matter of how terms such as biologically driven and innate are to be interpreted. Ohala uses innate to mean "genetically predisposed" but at the same time, requiring "post-natal stimu- lation" (1994:343-44). Recent work by psychologists has shed some light on whether the frequency code is innate, learned, or both. Marks & Bornstein (1987) state that a very small number of sensory analogies are psychologically primitive and probably innate. However, the perception of a similarity between high pitch and small size occurs developmentally later, not until 11 years of age (Marks & Bornstein 1987:55). This suggests that the frequency code, however biologically driven, must be learned from experience.


The Search for a Sound-Symbolic Substrate

The research by Sapir (1949a) and S Newman (1933) catalyzed numerous experi- ments designed to test for sound symbolism in actual words, and in semantic domains other than magnitude. Brown et a1 (1955) boldly stated that the ability of subjects to guess meanings of foreign words beyond chance rates of success pointed to traces of phonetic symbolism in all languages, arising from universal intersensory connections. They also suggest that speech itself may have origi- nated from such imitative links between sound and meaning. Brown et a1 (1955) tested the ability of native English speakers to match meanings of antonyms from Chinese, Czech, and Hindi with comparable antonyms from English. Overall, their guesses were correct twice as often as they were incorrect. Maltzman et a1 (1956) reproduced the experiments of Brown et a1 using Japanese and Croatian, with the novel condition that both the stimulus words and the response pairs were from languages unknown to the subjects. The rate of error was significantly higher when both stimulus and response words were unknown. Maltzman et a1 (1956) concluded that successful matching had been based on something other than a universal phonetic symbolism. Brackbill & Little (1957) object to the assertions by Brown et a1 (1955) of universal phonetic symbolism on procedural and ideological grounds. Conceding that subjects are successful at guessing meanings of unknown words at above-chance expectancy rates, they attempted to identify factors that might influence success rates. They composed a random sam- ple of concepts generally agreed to be of high-frequency usage and translated them into English, Hebrew, Japanese, and Chinese. They did not restrict their data to word pairs featuring contrasting, antonymic concepts. They also attempted to determine whether audio or visual presentation influenced the subjects' success rates. Contra Maltzman et a1 (1956), they found that subjects who were asked to judge word pairs as the same or different did not guess at significantly below- chance rates when both sample words were from unknown languages. Nor was mode of presentation, i.e. visual or auditory, a conclusive factor.

In attempting to isolate characteristics of word pairs judged correctly as to their sameness by over 70% of subjects, Brackbill & Little (1957) identify five factors of similarity, including connotation, number of letters or syllables, pres- ence or lack of hyphenation, and types of consonants. Although the authors inter- pret their results as contrary to a universal phonetic symbolism, it is not difficult to imagine how these results might be interpreted as confirming some type of uni- versal intersensory connections (cf Weiss 1964:455-56). Taylor (1963) and Tay- lor & Taylor (1965) favor a modified view of phonetic symbolism, which would allow for limited kinds of associations between phonemic segments and mean- ings that are language specific and learned, rather than universal. They also sug- gest that different linguistic cultures may be more predisposed than others to judge sounds as symbolic, and they argue for the need to conduct more investiga- tions of different linguistic communities. Some investigators (Gebels 1969, Miron 1961) who conducted cross-cultural experiments, however, have sided with Brown et a1 (1955). In addition to the foregoing criticisms, two others, at least, need to be mentioned. One is that investigators themselves may be biased, unconsciously, to choose from nonrelated languages words that share similar sound segments. An additional serious consideration raised by Osgood (1960: 166) for a different experimental project concerns the translatability of opposites from one language to another. The terms energetic and lazy are logically opposed in English, but their closest translation equivalents are not opposite in Japanese. Similarly, for the Navajo, a more salient opposition than the English pair fast- slow would be moving-stationary.

Brown may ultimately have been his own severest critic. Brown (1958) and Brown & Nuttall (1959) concede that studies of phonetic symbolism are highly variable, depending on testing methods. Brown & Nuttall (1959) would restrict claims for phonetic symbolism to antonymic pairs that might evolve in such a way as to exhibit phonetic contrasts that reflect their semantic contrasts. They reject the earlier claim of Brown et a1 (1955), that phonetic symbolism might be linked to the origin of language in physiognomic representation. In fact, the origin ques- tion itself becomes transformed from an empirical problem to a set of specula- tions attempting to legitimize varieties of myths about the nature of language and meaning. Brown (1958: 130) concludes that the evidence for a universal sound symbolism is "totally problematical" and rejects the earlier view that sound sym- bolism is based on universal synesthetic connections. He argues, instead, for a limited phonetic symbolism whereby intersensory connections are made from learned sense experiences, rather than innately determined. Despite Brown's dis- claimers, sound symbolism continues in recent work (Dixon 1997:64) to be rele- vant, although not of exclusive importance, in discussions of possible scenarios for the origin of language. A final comment concerning the contributions of this experimental period is that none of the numerous and legitimate criticisms of this body of research has ever offered a satisfactory explanation for the above-chance success rates of subjects presented with the task of matching words from unknown languages.

The concept of synesthesia, i.e. the notion that the physiological properties of perception are designed to view different perceptual stimuli as the same, contin- ues to interest psychologists and psycholinguists with an interest in sound sym- bolism. Reichard et a1 (1949) found salient synesthetic correspondences between sounds and colors for some adults. Jakobson & Waugh (1979:129-30) discuss evidence for intersensorial correspondences between the acoustic features of speech sounds and chromatic distinctions of brightness and darkness. The acous- tic distinction between acute and grave consonants and vowels that specifies, respectively, the higher tonality (acute) versus lower tonality (grave) sounds cor- responds, respectively, to the chromatic distinctions of light and dark. Interest in sound-symbolic synesthesia has also overlapped with psychoanalytic explana- tions, such as that of Jakobson (1962), who analyzed the data of Murdock (1959) from a comprehensive survey of world cultures. Murdock tested whether there was any basis for the innumerable assertions by linguists and anthropologists that from children's nursery forms unrelated languages tend to develop similar words for mother and father. What Jakobson considered "spectacular" (1962:542) was the correlation between nasal consonants and maternal terms and oral consonants and paternal terms. In comparing words referring to mother, it was apparent that 55% have nasal consonants, such as the m, n, and g consonant classes, while only 15% of terms for father use these sounds. Nasals correlate with words denoting

mother, according to Jakobson, because infants can only produce nasal murmurs while sucking and such sounds become symbolic of the mother herself, as the "great dispenser" of all the child's emotional and nutritional needs. A recent treat- ment of synesthesia attempts to combine principles of phonology acquisition from Jakobson (1968) with a limited psychoanalytic framework and apply them to literary analysis (Tsur 1992).

Sound Symbolism in Morphological and Diachronic Analysis

Bolinger (1950) identified conventional sound-symbolic principles in the stmc- turing of English morphology and stimulated a number of similar analyses (Mar- chand 1959, Markel & Hamp 1960) that also pondered the morphemic status of phonesthemes, a term first used by Householder (1946). Bolinger argued that avowed principles of descriptive linguistic methods had accounted for only a small segment of morphology when they could have been pushed farther. Using principles of phonemic and semantic similarity, and ignoring etymological facts, Bolinger called attention to vast numbers of sound-meaning correspondences interrelated through rime and assonance. Word initial gl-, in roughly half of all commonly used English words, implies something visual, as in glance, glare, gleam, glimmer (Bolinger 1950:132-33). Another set, words with the letters "ash," implies fragmentation, collision, or impact, as in ash, bash, dash, gash. To relate his rimes and assonances to recognized English language morphemes, Bolinger sets up a continuum of types with morphemes on one end, grading into rimes and assonances, the more productive of which are as legitimately mor- phemic as any morpheme, and then grading into submorphemic differentials, which are the relatively meaningless bits of word remaining after a phonestheme has been analyzed. An initial "tw," such as in twist, twirl, tweak, and tweeze, is an assonant form that suggests a twisting or pinching motion. The "ist" left after "tw" is analyzed out would be considered a meaningless, submorphemic differen- tial, serving only to distinguish twist from other "tw" words. By contrast, the "irl" in twirl also occurs in curl, furl, whirl, and swirl, all of which suggest a circular or round movement or shape. The word twirl, therefore, can be analyzed as consist- ing of two phonesthemes "tw" and "irl." Some sets of words are interconnected in such bewildering ways that multiple analyses are possible. Should, for instance, the "oil" in roil, boil, and oil be analyzed as a phonestheme meaning "something liquid" or should we factor out the discontinuous r-1, as a phonestheme meaning "to stir liquid," in roil and rile because of their etymological relatedness?

Bolinger (1950) comes close to saying that phonesthemes create more prob- lems than they solve. He states that they should be described but are too fluid and contradictory to lend themselves to rule-governed formulation. Despite the mis- givings, his article raises important questions about the historical processes resulting in phonesthemic patterning. Why do etymologically related forms, such as defect, meaning a deficiency, and defection, meaning the act of leaving without permission, have such a thin semantic thread connecting their current usages? How, by contrast, do groups of forms with no etymological relatedness come to cluster together? In some instances there are haphazard and accidental conver- gences. In many cases there are similarities in form and potential similarities in meaning, which, by analogy to physical forces such as gravitation and attraction, seem to influence each other. In yet other instances there may be some combina- tion of both. According to Bolinger (1950), the word pick, for example, has, in addition to its meaning "to select," developed the related meaning "to select for unfavorable attention" because of its attraction to the forms peck, poke, and pike. The development of phonesthemic patterns is powerful evidence for a kind of sound-symbolic creativity that is protracted through generations, below the threshold of awareness for most, yet assented to and thereby engineered by entire communities of speakers.

The work of Bolinger (1950) continues to stimulate reanalyses of English mor- phology. Rhodes & Lawler (1981) and Rhodes (1994) argue that phonesthemic monosyllables consisting of an assonance and a rime have an internal structure that is analogous to modifier-head relations between syntactic constituents. Within the word-initial assonance, they identify semantic patterns suggestive of a classifier system. The "st" in stick, staff, and stem, for example, is a one-dimensional, rigid classifier. Besides their classifier functions, assonances also function as adverbial modifiers of a verb-like head or adjectival modifiers of a noun-like head. Snatch is analyzed as an adverbial assonance "sn," meaning quickly (cf snap, snag, snip) modifying a verb-like "atch," meaning come to hold (cf catch, latch). These analyses have been extended to rimes and assonances con- cerned with aural imagery (Rhodes 1994), which are structured to communicate basic and attendant properties of sound, the basic properties occurring in the rime, and the attendant properties occurring in the assonance. The study of rimes and assonances has changed the face of morphology studies. The problems first expounded by Bolinger have not been eliminated. Concepts have been expanded to accommodate the untidy facts of synchronic analysis. That only part of a word may by analyzable is no longer considered a problem. Rhodes & Lawler (1981) compare rimes and assonances to unitary entities such as gestalts, which are ana- lyzable into subparts. Waugh & Newfield (1995:202-3) consider segmental mor- phology typologically biased for agglutinative languages and suggest a paradigmatic view of lexical structure that would start with the word and then consider its interrelatedness with larger sets of words through phonesthemes, morphemes, and submorphemes. Other paradigmatically based approaches have revealed patterns in the structuring of lexicons. Comparisons of large sets of Eng- lish nouns and verbs used with high frequency revealed that nouns were more likely to have stressed back vowels and verbs were more likely to have stressed front vowels (Sereno 1994).

Submorphemic sound symbolism has broadened linguists' range of hypothe- ses regarding diachronic processes that are otherwise defiant to analysis. In words from Romance languages, Malkiel(1987, 1994) uses the phenomenon of phono- symbolic orchestration as a unique category of sound change, separate from regu- lar sound laws, to explain a number of changes that would otherwise seem haphazard, unpredictable, and wild. Kaufmann (1994) finds that in Huastec, sound symbolism has prevented some phonemes from undergoing regular changes. De Reuse (1986) examines instances of palatal Is1 that resisted change according to regular sound laws in Santiago del Estero Quechua and finds that they occur in words with similar affective connotations. Mithun (1982) states that the failure of sound-symbolic words in Iroquoian to undergo regular shifts in their point of articulation can be explained by their imitative function. Mannheim (1991) argues that in Southern Peruvian Quechua, ejective and aspirate sounds were borrowed from JaqiiAru languages and then spread through associated lexi- cal networks where they are iconic of various processes involving the expulsion of air. The presence of ejectives and aspirates, which are otherwise rare in the Quechua languages, is evidence for their spread through the lexicon through asso- ciative lexical influence, a process akin to the gravitational forces at work in the creation of the phonesthemic networks analyzed by Bolinger (1950). Antilla & Embleton (1995) point to the significance for Cockney English of rhyming sound as a form of linguistic play that enters into diachronic processes by contributing lexical items to colloquial British speech. The slang phrases "trouble and strife" or "Duchess of Fife" may be substituted (i.e. used as iconic indexes) for "wife," and over time, come to be reduced through ellipsis to "trouble" or "strife."


A typologically widespread form of sound symbolism occurs as a kind of lexical class. Ideophones (Doke 1935), expressives (Diffloth 1972), and onomatopoetics (Emeneau 1969) are just a few of the terms in current use. The most abundant and accessible materials have been published by linguists of African languages, where the term ideophone has been adopted. This term is therefore used here even when non-African languages are discussed. There are vast numbers of published works on ideophones in Asian languages as well. They are found in the Dravidian languages of South India (Emeneau 1969) but have spread to Indo-Aryan lan- guages, including Sanskrit (Hoffmann 1952) and Bengali (Dimock 1957). Ideo- phones exist in Korean (Martin 1962, Kim 1977), Vietnamese (Durand 1961), and Japanese (Kakehi et a1 1996). Not nearly as voluminous is the evidence for ideophones in South American language families, including Carib (Koehn & Koehn 1986), Ge (Popjes & Popjes 1986), Tupi-Guarani (Crofts 1984, Langdon 1994), Arawakan (Weidman de Kondo 1976), and Quechua (J Nuckolls 1996). Evidence is also accumulating on ideophones in Australian languages (Alpher 1994, McGregor 1999, Schultze-Berndt 1999). Space does not permit a list of North American language families where ideophones have also been attested, but are difficult to investigate due to the endangered status of so many of these lan- guages.

Doke (1935: 11 8) first defined the ideophone as "a vivid representation of an idea in sound. A word, often onomatopoeic, which describes a predicate, qualifi- cative or adverb in respect to manner, color, smell, action, state, or intensity." Although useful, Doke's definition does not even hint at the theoretical interest of ideophones or at their wide-reaching ramifications for formal linguistic descrip- tion. Ideophones are not totally absent from western European languages such as English, as Wescott (1977) has argued. Onomatopoeic words and expressions such as click, thwack, moo moo, baa baa, etc, are a few examples. Such utter- ances, however, are restricted in their usage to whimsical and childish styles of speaking. In languages where ideophones are a well-developed class, their use will be required in a variety of contexts where artful (Evans-Pritchard 1962: 145, Fortune 1962:6, Noss 1975), emotive (Cole 1955:370, Doke 1938:353), or collo- quial (Hamano 1994: 148) language is called for. To say that ideophones exist in English brushes aside the overwhelming inclination on the part of linguists to regard them as "a totally different kind of linguistic animal" (Diffloth 1976:251) with respect to their own linguistic culture. Most linguists would concur with Childs (1994: 180-8 I), whose comprehensive survey of African ideophones con- siders them a prototype category with a constellation of defining features. One approach is to note similarities between the functions of ideophones and other parts-of-speech notions, such as noun, verb, adverb, adjective, and interjection. Childs (1994: 18 1) and Samarin (1 97 1 :3 1) both state that in African languages ideophones are most often comparable to adverbs, although they may have multi- ple functions. Newman (1968) found that ideophones in Hausa functioned as adjectival intensifiers, verbal intensifiers, and descriptive adverbs and in Tera as adjectives and as several subcategories of adverb. Alpher (1994:170) states that ideophones in Yir-Yoront are roughly comparable to adverbials, nouns, and modal particles.

Formal and Semantic Anomalies

The salient difference separating ideophones from comparable word classes include phonological, morphological, syntactic, semantic, and pragmatic criteria. Courtenay (1976:24) states that syllabic nasals are found only in Yoruba's ideo- phones. Samarin (1971 :136) summarizes numerous reports of phonological fea- tures unique to ideophones in a variety of Bantu languages, including nasalization, devoicing, the presence of affricates, ejectives, and syllabic trills. More common than unusual phonemes are phonotactic combinations unique to ideophones or in violation of existing constraints. In Japanese, the palatalization of consonants is used in mimetic words to communicate a set of related concepts such as childishness, immaturity, instability, etc, and it appears to follow a set of constraints that do not operate outside the mimetic vocabulary (Hamano 1994: 154). Many investigators will simply note a phonotactic pattern that is unusual for a language without commenting on possible iconic motivations (Lanham 1960, Newman 1968: 115). Ideophones are morphologically special by their reduplica- tion, that is the complete repetition of a root, which often conveys repetition over time and extension in space. They may also occur only once, to express momen- tary, instantaneous actions (Basso 1985:66). Because many African languages are prone to agglutinative morphology, which means that any root may be subject to numerous affixations, linguists of these languages have often remarked that ideophones are noninflected. This contrasts with the situation in Asia, however, where ideophones may receive affixes. In Semai and Sre, ideophones may be modified by processes of infixation and prefixation, to convey concepts of simul- taneous plurality (Diffloth 1976:253, 1980:53). Ideophones in Japanese are suf- fixed with several different forms depending on their functions within an utterance (Kita 1997:383-85).

Regardless of language family, two contradictory observations have been made concerning the uniqueness of ideophonic morphology. On the one hand, it is said that there is a "relative freedom of replacement of a segment by some other segment without altering the meaning of a word (Newman 1968:107)." See also Childs (1994: 182-83), and Mithun (1982:54-55). On the other hand, optimism has been expressed over the possibility of correlating at least some ideophonic phonemes with meanings (Samarin 1965: 1 19-20, Childs 1994: 194). At times, both contradictory observations may be made by the same author concerning the same language. In an analysis of ideophones in Semai, for example, Diffloth (1976) has stated that the uniqueness of their morphology is in large part due to their lack of lexical discreetness. Put simply, lexical discreetness implies that a root is a linguistic unit that, if changed by a single phoneme, will change into a completely different, semantically unrelated root, as is true for the following set: sip, ship, chip, and tip. Diffloth (1976:260), however, provides examples from Semai of paradigm-like sets of ideophones in which differences in vowel pho- nemes create a set of semantically interrelated forms: lgri:pl, the noise of chewing small, brittle things; /gr :p, the noise of chewing large, somewhat soft things; /gr8:pl, the noise of chewing large, hard things; /gr3:p/, the noise of chewing large, crispy things. In the foregoing paradigm, differences in the vowels corre- spond to differences in meaning, although the exact nature of the motivation underlying these meaning differences is not explicitly stated. Nevertheless, it is clear that the forms and meanings are interrelated formally and semantically and that lexical discreteness does not apply. However, Diffloth also notes (1976:260) that in another set of ideophones, initial consonants may disappear, without changing meaning: lgli:ll, the gait of a very fat person "throwing around" a lot; lli:l/, the gait of a very fat person "throwing around" a lot. What is perplexing about ideophonic morphology, then, is that in certain contexts, changes in form correspond to seemingly motivated changes in meaning, whereas in other con- texts, changes in form correspond to no changes of meaning whatsoever. Diffloth believes that for ideophones as a class, traditional concepts from morphology such as root and morpheme will have to be discarded. Ultimately, he favors the view that ideophones are decomposable into phonemes or distinctive features that may be acoustically or articulatorily symbolic. Feld (1982) found that vowels in Kaluli ideophones carry phonestheme-like, conventionalized sound-symbolic meanings related to location, directionality of movement, and other sonic qualities.

Syntactically, ideophones are typically nonintegrated with respect to other sentential constituents. Noss (1975) cites examples from Gbaya, Sango, and Zande, in which ideophones have an appositional function, supplementing what the predicate states. Ideophones may function as complete predicates, without verbal inflections, such as person or number markers, and they may occur in "ver- bless contexts" (Alpher 1994: 168). The syntactic idiosyncrasies of ideophones are further evident by constraints that restrict them to certain sentence types. Most ideophones occur in declarative sentences (Childs 1994:188). A number of reports state that ideophones will be restricted from undergoing negation and question transformation when they function as descriptive adverbs, although not when they hnction in nominal or adjectival capacities (Kita 1997, Newman 1968). Although statements concerning ideophones' lack of integration within an utterance are common, some recent work has attempted to relate their functions to grammatical subsystems such as aspect, a universal grammatical concept that encodes the ongoingness or completiveness of an action, event, process, or state (Alpher 1994: 163, J Nuckolls 1996:58-59). Perceptions of an ideophone's lack of syntactic integration can also be related to pragmatic factors. Ideophones can be set apart by pauses (Childs 1994: 185), expressive intonation (Newman 1968, Childs 1994), and prosodic peaks (Alpher 1994: 16, Kita 1997:395). A major issue is the matter of ideophonic definition. Many investigators report great diffi- culties due to their highly specific and nuanced meanings. Samarin (1965, 1967) advocates the study of ideophonic meaning through a combination of compara- tive and inductive methods, which would relate them to a relatively small number of semantic categories encoding 15 types of sensory experience.

One analysis of ideophonic semantics, however, found that concrete, sensorily grounded meanings might grade into abstract, grammatical concepts. Nuckolls (1999) analyzed one widely used Quechua ideophone, tak, and found that it expresses a range of meanings, from sound imitation to aspectual grammar. At its most concrete and sensorily grounded, tak imitates an idea of the sound of contact between two firm surfaces, one of which, typically, is manipulated by an animate agent or force: A gourd can be tapped tak to test for ripeness; a pot can be placed tak onto a cooking fire of logs; and a stalk of sugar cane can be hit tak with a machete. Beginning with the idea of contact, it is possible to view the variety of tak's meanings as a series of increasingly abstract usages. For example, tak is commonly used to describe soundless contact, as when someone grabs hold of a machete. Yet further removed from concrete contact is a kind of contactless con- tact, for example when tak describes a positioning within a definite point in a spa- tial field, as when a snake about to strike places its tail tak at a point in midair. Whether describing the sound of contact, a soundless contact, or a contactless contact, the foregoing usages of tak are all aspectually punctual because they encode the momentaneousness of an otherwise aspectually neutral verb. In another associated set of usages, tak's meanings are expanded from the point-like idea of one-dimensional contact to all-encompassing three-dimensional contact, describing experiences of being completely surrounded, impeded, or contained. These usages, in turn, are unified by the aspectual concept of completiveness. Related to the issue of ideophonic semantics is the matter of how ideophones come to join a language's stock of ordinary verbs, adverbs, and nouns, a process attested in numerous languages (McGregor 1999, Voeltz 1971), and also the reverse processes whereby verbs become reanalyzed as ideophones (Childs 1989, Hutchison 1981).

Theoretical and Cultural Import of Ideophonic Research

The theoretical implications of ideophones have only begun to be explored. Fol- lowing Jakobson (1971), who equated iconicity with expressiveness, Diffloth (1972) states that the iconicity of ideophones makes their incorporation into a transformational-generative grammar impossible in principle and advocates the incorporation of an aesthetic component to be somehow plugged into the formal system. Kita (1997) postulates a two-dimensional semantic framework consisting of an affecto-imagistic dimension for ideophonic meaning and an analytic dimen- sion for non-ideophonic meaning. His argument draws from experimental evi- dence demonstrating an overwhelming correlation between ideophones and gesture (Kita 1993). Subjects asked to watch a cartoon were videotaped retelling the events in Japanese. When ideophones occurred in retellings, they were accompanied by gesture strokes 94% of the time. By contrast, verbs were accom- panied by gestures only 40% of the time. Such findings substantiate the many cas- ual and intuitive observations concerning similarities between ideophones and gesture and indicate the pressing need for more work along the lines of McNeill (1992), Kendon (1997), and others on the interrelatedness of language and ges- ture. The stylistic restrictedness of ideophones in European languages has led to the default assumption, not always clearly articulated, that they are highly signifi- cant of different perceptual modalities and different cultural worlds. Levy-Bruhl (1926) read Westermann's work (1930) on the Ewe language and considered ideophones illustrative of a primitive mentality that was incapable of, or disin- clined to, think abstractly. The existence of numerous ideophones to describe ways of walking suggested to Levy-Bruhl that "the conception of walking in gen- eral never presents itself alone" (1926: 166).

A more palatable theory borrowed from the developmental psychologists Werner & Kaplan (1963) would consider ideophones one type of expressive lan- guage grounded in dynamic schematizing processes, termed physiognomic per- ception, whereby similarities between sign vehicle and referent are actively constructed. Although it lacks legitimacy within abstract and logical conceptual styles privileged by scientific culture, people everywhere engage in physiog- nomic perception. Fortune noticed it in ideophonic use among the Shona, which he described as "a concern to give concrete and adequate expression to what they see of the forces and energies of the world" (Fortune 1962:41). The privileged connection between ideophones and special perceptual styles might be relevant to the numerous reports of ideophonic usage diminishing according to degree of Westernization or degree of urbanization (Childs 1994: 195, Fortune 1962:41). Yet many Asian languages with long-established traditions of literacy and com- plex urban civilizations feature vast inventories of ideophones. It is likely that ideophones are cognitively special but that their singularity will have to be inter- preted in culturally specific ways. An important issue, as yet unexplored by lin- guistic anthropologists, is ideophones' cultural, stylistic, and psychological significance. A startlingly original approach to Japanese ideophones treats them as analogous to imageic expression. Gomi (1989) wrote and illustrated a diction- ary featuring cartoon drawings of situations illustrating the meanings of Japanese ideophones. It is tempting for a Western-educated reader accustomed to viewing comics and cartoons as aesthetically insignificant to regard such an effort as whimsical. Pictorial representation in cartoons and comic books is, however, an elaborated art form in Japan. In his preface, Taro Gomi explains that ideophones are at least as important to the Japanese language as traditional classical art forms such as kabuki and bunraku are to Japanese culture.


Reactions Against Arbitrariness

It is not surprising that some of the most articulate defenders of sound symbolism

have been linguists with a deep interest in poetics and non-Western poetic tradi-

tions. Researchers in ethnopoetically oriented linguistics have discovered that

sound is a highly salient principle of discourse structure. Jakobson (1960) consid-

ered parallelism, the repetition in a stretch of discourse of sounds such as pho-

nemes, syllables, rhythmic patterns, and intonational and prosodic contours, in

conjunction with syntactic units, to be the defining feature of all poetic language,

even the most mundane cliches. To the extent that equivalences in meaning are

created by equivalences in sound, parallelism can be viewed as a special type of

discourse-structuring sound symbolism. A related body of research has con

cerned the identification of sound features that function to divide narratives into

lines and larger groupings, such as verses and stanzas. Controversy has ensued

over the relative importance of sound features, such as voice quality, loudness,

and pausing, that are associated with Tedlock (1983) as opposed to the syntactic

phrases and particles favored by Hymes (1981). Bright (1979), Sherzer (1990),

and Woodbury (1985), among others, have argued that both kinds of patterning

must by studied. Besides the issue of sound's discourse-structuring potential,

there is also work attempting to link the expressive power of aesthetic forms to the

repeated use of certain phonemes, features, or syllables (Hymes 1960, Fonagy

1961, Priestly 1994). Of great interest also is work on the sound symbolism of

speech styles. Urban (1985) presents a complex argument concerning the key

sound features of speech styles that index salient cultural values for the Shokleng

of Central Brazil. Part of his analysis concerns the practice of origin myth telling,

which is performed by two men, one of whom repeats in rapid-fire succession,

syllable after syllable, what the other has just recited. Origin myth performances

are expressively restricted in relation to ordinary speech in that a number of dis-

tinctive sound features, such as pause boundaries and contrastive stress, are elimi-

nated, while intonation is flattened. Such performances are also marked by

constricted pharyngeal articulation and abrupt diaphramic explosions, all of

which index the gruffness of Shokleng didactic style, thereby pointing to the

overall significance of the performance as an act of cultural teaching.

Philosophical defenses of sound symbolism often proceed by attacking the principle of arbitrariness. This is a traditional polemic about the nature of lan- guage and reality that can be traced to Plato's Cratylus, where the conventional- ity or naturalness of names is pondered. Recent incarnations of this debate often defend sound symbolism by attacking the logical inconsistency of the reasoning of Saussure about arbitrariness. Saussure (1959:67-68) states that arbitrariness characterizes the relationship between a signifier (i.e. sound image) and a signified (i.e. concept) and has nothing whatsoever to do with any object or reality denoted by a linguistic sign. Saussure defends this principle, however, by stating that the reality referred to by words such as boef and ochs is the same no matter what labels different languages use. Benveniste (197 1) noticed that by unconsciously bringing the world of objects and entities into his reasoning, he contradicted his original formulation. A number of scholars have noted that Saussure's inability to leave the world out of his equation points to the necessary connection (Hymes 1960, Jakobson 1971) and the ontological reality (Frie- drich 1979) of a symbol and what it refers to for the speakers of a language. Another contentious point concerns how the word arbitrary is to be interpreted. Friedrich states that it sometimes implies "some kind of randomness" and at other times is used to describe the wide divergences between different symbolic systems (1979:6). Durbin (1973) argues against the notion that different sym- bolic systems imply arbitrariness, saying that the existence of sound symbolism may be universal although it will manifest itself differently in various lan- guages because phonological and cognitive classificational systems themselves are varied.

Sound Symbolism, Iconicity, Connectionism

The idea that sound symbolism may be unique to a particular cognitive and lin- guistic system is problematical, as it conflicts with a fundamental premise in anthropological thought-that of psychic unity. Shore (1996) believes that this principle reflects a cognitively impoverished concept of culture that has mar- ginalized anthropologists from the cognitive science revolution that is shifting the very paradigms within which our cultural and linguistic models have grown. He advocates a view of culture as a collection of multifunctional models, both in the mind and empirically observable. One such model, the modularity schema, is considered foundational because of its far-reaching consequences for lin- guists, anthropologists, and psychologists. Essential to the modularity schema is that a wide range of phenomena are viewed as "assemblages, subject to decompo- sition and recombination" (Shore 1996: 118). Modularity is privileged in scien- tific discourse, where knowledge is conceived as information processing, and it underlies technologies based on digital encoding and production. It is also related to the linguistic principles of atomism and arbitrariness, because both imply the interchangeability of units. Shore attacks the principle of arbitrariness, stating that it has undermined attempts to understand the complexity of cultural and lin- guistic symbols. He believes that recent findings from psychology and cognitive science have the potential to illuminate unresolved issues about the relationship between neural structures and mental representations. Shore argues that connec- tionist models are superior because they allow for analogical reasoning and analogy formation generally, processes that cannot be accounted for by information-processing models of thought. He suggests that connectionist models may even someday be able to illuminate the intentionality of mental states. Bouis- sac (1995) also expresses optimism for connectionist models, although he focuses his argument on linguistic iconicity. He states that formal linguistic para- digms since Saussure and including Chomskyean generative grammar are essentially Platonic, because they view language as an expression of universal ideal forms originating from the mind. According to such a view, any analogies, such as iconic properties between language and the world, are anomalous and accidental. Bouissac argues that connectionist models are advantageous because they relate what is perceived with what linguistic processes actually produce. In such a framework, then, imitation could be accounted for neurologically. Bouis- sac believes that such models are most likely to help create a paradigm shift, in which iconicity would be the norm for human language and arbitrariness would be seen as an evolutionary adaptation arising from the need for deception and obfuscation.

The attribution of moral values such as truth and honesty to iconic communica- tion is also found in Bateson (1972), who hypothesizes that iconic, analogically encoded communication has survived alongside the development of digitally encoded verbal language, because it is transparent and not easily falsifiable. The ideas of Bateson (1972) and Bouissac (1995) reflect a restricted view of iconic- ity-one that equates iconic communication with gestures and intonation. In fact, iconicity is now widely acknowledged to be a significant factor at many levels of linguistic structure. In morphology, examples from degree adjectives, such as long, longer, and longest, show that the most-extreme degree of length is iconi- cally represented by the word with the greatest number of phonemes (Wescott 197 1). Jakobson (197 1) cites examples ofword order mimicking the natural order of ideas. Currently, work on diagrammatic iconicity in syntax is flourishing (Givon 1985, Haiman 1985, Landsberg 1995). It is possible, then, that our lin- guistic, cognitive, and symbolic theories are shifting into a more sound-symbolically congenial paradigm, one which would allow for motivated relations between form and meaning generally. Another possibility, drawing from recent work in culture theory (C Nuckolls 1996), would consider arbitrariness and sound symbolism as a single paradoxical construct and that they have in common with many systems of thought their status as irresolvable contradictions that catalyze dialectical thought processes and are important sources of creative conflict, resynthesis and reinterpretation. Throughout the exhaustive dissections and criti- cisms of the principle of arbitrariness, there has never been a serious suggestion that it be totally abandoned. The evidence discussed in this review points to a view of language as a system structured by competing tendencies. It is a mistake to use quantitative analogies to make claims about the relative importance of arbi- trariness versus relative motivation or iconicity.


Whether it is a paradigm shift that is under way or a prolonged sojourn into iconi- cally congenial territory is uncertain. Whatever the case, there is great potential for sound symbolism to contribute to unresolved understandings of linguistic structures and processes. If we concede that it is a necessary linguistic principle rather than an occasional, exceptional one, then an enormous task lies ahead. In their comprehensive survey of sound symbolism, Jakobson & Waugh (1979:186-87) implore linguists to construct sound-symbolic typologies of lan- guages in the interest of discovering what, if anything, is universal. This review has sketched only some of the most widely reported sound-symbolic processes and research trends. Many others have yet to be charted.


Sincerest thanks to David Basilico, G Tucker Childs, Sotaro Kita, and Annual Review ofAnthropology Editor Bambi B Schieffelin for many helpful comments and suggestions on an earlier draft. I alone am responsible for any remaining problems. I am also grateful to Mr. Eddie Luster of the Interlibrary Loan depart- ment of Steme Library, University of Alabama, Birmingham, for considerable assistance in tracking down and obtaining references.

Visit the Annual Reviews home page at www.AnnualReviews,org.


Alpher B. 1994. Yir-Yoront ideophones. See Steps to an Ecology of Mind, pp. 41 1-25. Hinton et a1 1994, pp. 161-77 New York: Ballantine Anderson SR. 1985. Phonology in the Twenti- Beck GL. 1993. Sonic Theology: Hinduism eth Century. Chicago: Univ. Chicago Press and Sacred Sound. Colombia: Univ. S. C.

Antilla R, Embleton S. 1995. The iconic index: Press from sound change to rhyming slang. In Bentley M, Varron EJ. 1933. An accessory Iconicity in Language, ed. R Simone, pp. study of phonetic symbolism. Am. J. Psy87-1 18. AmsterdamIPhiladelphia: Benja- chol. 45:76-86 min~ Benveniste E. 197 1. The nature of the linguis-

Aoki H. 1994. Symbolism in Nez Perce. See tic sign. In Problems in General Linguis- Hinton et a1 1994, pp. 15-22 tics, pp. 43-48. Transl. ME Meeks. Coral Basso EB. 1985. A Musical View of the Uni- Gables: Univ. Fla. Press

verse. Philadelphia: Univ. Penn. Press Berlin B. 1994. Evidence for pervasive synes-

Basso EB. 1995. The Last Cannibals. Austin: thetic sound symbolism in ethnozoological Univ. Texas Press nomenclature. See Hinton et a1 1994, pp.

Bateson G. 1972. Redundancy and coding. In 76-103

Bolinger D. 1950. Rime, assonance, and mor- pheme analysis. Word 6: 117-36

Bolinger D. 1964. Intonation as a universal. In Proc. 9th Int. Congr. Linguists, 1962, ed. HG Lunt, pp. 83344. The Hague: Mouton

Bolinger D. 1978. Intonation across languages. See Greenberg et a1 1978, pp. 47 1-524

Bolinger D. 1985. The inherent iconism of in- tonation. In Natural Syntax: Iconicity and Erosion, ed. J Haiman, pp. 97-108. Cam- bridge, UK: Cambridge Univ. Press

Bolinger D. 1986. Intonation and Its Parts: Melody in Spoken English. Stanford, CA: Stanford Univ. Press

Bouissac P. 1995. Syntactic iconicity and con- nectionist models of language and cogni- tion. See Landsberg 1995, pp. 393-417

Brackbill Y, Little KB. 1957. Factors deter- mining the guessing of meanings of foreign words. J. Abnorm. Soc. Psychol. 54: 312-18

Bright W. 1979. A Karok myth in measured verse: the translation of a performance. J. Gal$ Great Basin Anthropol. 1:1 17-23

Brown P, Levinson SC. 1987. Politeness: Some Universals in Language Usage.

Cambridge, UK: Cambridge Univ. Press Brown RW. 1958. Words and Things. Glencoe, IL: Free Press Brown RW, Black AH, Horowitz AE. 1955. Phonetic symbolism in natural languages.

J. Abnorm. Soc. Psychol. 50:388-93

Brown RW, Nuttall R. 1959. Method in pho- netic symbolism experiments. J. Abnorm. Soc. Psychol. 59:441-45

Childs GT. 1989. Where do ideophones come
from? Stud. Linguist. Sci. 19(2):57-78
Childs GT. 1994. African ideophones. See

Hinton et a1 1994, pp. 178-204 Cole DT. 1955. Introduction to Tswana Gram- mar. London: Longmans, Green

Courtenay K. 1976. Ideophones defined as a phonological class: the case of Yoruba. Stud. A?. Linguist. 6(Suppl.): 13-26

Crofts M. 1984. Ideofonos na narracao Mun- duruku. Ser. Linguist. 1 1 :207-18 Derbyshire DC, Pullum G, eds. 1986. Hand

book of Amazonian Languages, Vol. 1. Berlin: Mouton de Gruyter

De Reuse W. 1986. The lexicalization of sound symbolism in Santiago del Estero Quechua. Int. J. Amer. Linguist. 52, no. 1:54-64

Diffloth G. 1972. Notes on expressive mean- ing. In Pap. 8th Reg. Meet. Chicago Lin- guist. Soc., ed. P Peranteau, JN Levi, G Phares, 8:44047. Chicago: CLS

Diffloth G. 1976. Expressives in Semai. In Austroasiatic Studies. Oceanic Linguistics Spec. Publ. 13,1:24944. Hawaii: Univ. Press

Diffloth G. 1994. i: big, a: small. See Hinton et a1 1994, pp. 107-14 Dimock E. 1957. Symbolic forms in Bengali. Bull. Deccan Coll. Res. Inst. 18:22-29

Dixon RMW. 1997. The Rise and Fall ofLan- guages. Cambridge, UK: Cambridge Univ. Press

Doke CM. 1935. Bantu Linguistic Terminol- ogy. London: Longmans, Green

Doke CM. 1938. Textbook of Lamba Gram- mar. Johannesburg: Witwatersrand Univ. Press

Durand M. 1961. Les impressifs en Vietnamien, etude preliminaire. Bullktin de la Societe des Etudes Indo Chinoises 36, I: 5-5 1

Durbin M. 1973. Sound symbolism in the Ma- yan language family. In Meaning in Mayan Languages: Ethnolinguistic Studies, ed. MS Edmonson, pp. 2349. The Hague: Mouton

Emeneau MB. 1969. Onomatopoetics in the Indian linguistic area. Language 45: 274-99

Evans-Pritchard EE. 1962. Ideophones in Zande. Sudan Notes Rec. 43: 14346 Feld S. 1982. Sound and Sentiment: Birds, Weeping, Poetics, and Song in Kaluli Ex- pression. Philadelphia: Univ. Penn. Press

Feld S. 1996. Waterfalls of song: an acouste- mology of place resounding in Bosavi, Papua New Guinea. In Senses ofplace, ed. K Basso, S Feld, pp. 91-35. Santa Fe, NM: Sch. Am. Res.

Fischer-Jorgensen E. 1978. On the universal character of phonetic symbolism with spe- cial reference to vowels. Stud. Linguist. 32:80-90

Fonagy I. 1961. Communication in poetry. Word 17:194-218 Fortune G. 1962. Ideophones in Shona. London: Oxford Univ. Press

Friedrich P. 1979. The symbol and its relative nonarbitrariness. In Language, Context, and the Imagination: Essays by Paul Frie- drich, ed. AS Dil, pp. 1-61. Stanford, CA: Stanford Univ. Press

Gebels G. 1969. An investigation of phonetic symbolism in different cultures. J. Verbal Learn. Verbal Behav. 8:3 10-1 2

Givon T. 1985. Iconicity, isomorphism and non-arbitrary coding in syntax. In Iconicity in Syntax, ed. J Haiman, pp. 187-19. Am- sterdam: Benjamins

Gomi T. 1989. An Illustrated Dictionary of Japanese Onomatopoeic Expressions.

Transl. J Turrent. Tokyo: Japan Times

Greenberg JH, Ferguson CA, Moravcsik EA, eds. 1978. Universals ofHuman Language. Vol. 2: Phonology. Stanford, CA: Stanford Univ. Press

Haiman J, ed. 1985. Natural Syntax: Iconicity and Erosion. Cambridge, UK: Cambridge Univ. Press

Hamano S. 1994. Palatalization in Japanese sound symbolism. See Hinton et a1 1994, pp. 148-57

Hinton L, Nichols J, Ohala JJ, eds. 1994. Sound Symbolism. Cambridge, UK: Cam- bridge Univ. Press

Hockett C. 1960. The origin of speech. Sci. Am. 203(3):88-96

Hoffmann K. 1952. Wiederholende Onomato- poetika im Altindischen. Indoger. Forsch. 60:254-64

Householder F. 1946. On the problem of sound and meaning, an English phonestheme. Word 2183-84

Hutchison JP. 1981. Kanuri word formation and the structure of the lexicon. In Nilo-Saharan Proceedings, ed. ML Bender, TC Schadeberg, pp. 2 17-37. Cinnaminson, NJ: Foris

Huttar GL. 1968. Relations between prosodic variables and emotions in normal Ameri- can English utterances. J. Speech Hear. Res. 11:481-87

Hymes DH. 1960. Phonological aspects of style: some English sonnets. See Sebeok 1960, pp. 109-3 1

Hymes DH. 1981. How to talk like a bear in Takelma. In In Vain I Tried to Tell You, pp. 65-76. Philadelphia: Univ. Penn. Press

Jacobs M. 193 1. A Sketch of Northern Sahap- tin Grammar. Seattle: Washington Univ. Press

Jakobson R. 1960. Closing statement: linguis- tics and poetics. See Sebeok 1960, pp. 350-77

Jakobson R. 1962. Why "mama" and "papa"? In Roman Jakobson Selected Writings I, Phonological Studies, pp. 538-45. The Hague: Mouton

Jakobson R. 1968. Child Language, Aphasia, and Phonological Universals. The Hague: Mouton

Jakobson R. 1971. Quest for the essence of language. In Roman Jakobson Selected Writings 11, Word and Language, pp. 345-59. The Hague: Mouton

Jakobson R, Waugh L. 1979. The SoundShape of Language. Bloomington: Indiana Univ. Press

Jespersen 0. 1933. Symbolic value of the vowel i. In Linguistics; Selected Papers in English, French, and German, pp. 283-

303. Copenhagen: Levin & Munksgaard

Jespersen 0. 1964. Language: Its Nature, Development and Origin. New York: Nor- ton

Kakehi H, Tamori I, Schourup L. 1996. Dictionary oflconic Expressions in Japanese.

Berlin: Mouton de Gruyter

Kaufmann T. 1994. Symbolism and change in the sound system of Huastec. See Hinton et a1 1994, pp. 63-75

Kendon A. 1997. Gesture. Annu. Rev. Anthro- pol. 26: 109-28 Kim K-0. 1977. Sound symbolism in Korean.

J. Linguist. 13:67-75 Kita S. 1993. Language andthought interface: a study ofspontaneous gestures and Japa- nese mimetics. PhD thesis. Univ. Chicago, Chicago, IL. 146 pp.

Kita S. 1997. Two dimensional semantic analysis of Japanese mimetics. Linguistics 35:379-15

Koehn E, Koehn S. 1986. Apalai. See Derby- shire & Pullum 1986, pp. 33-127 Kristeva J. 1981. Language the Unknown. New York: Columbia Univ. Press Ladefoged P. 1975. A Course in Phonetics. New York: Harcourt Brace Jovanovich

Landsberg ME, ed. 1995. Syntactic Iconicity and Linguistic Freezes: The Human Di- mension. Berlin: Mouton de Gruyter

Langdon M. 1994. Noise words in Guarani. See Hinton et a1 1994, pp. 94-103

Lanham LW. 1960. The comparative phonol- ogy of Nguni. PhD thesis. Univ. of Witwa- tersrand, Johannesburg

La Polla R. 1994. An experimental investiga- tion into phonetic symbolism as it relates to Mandarin Chinese. See Hinton et a1 1994, pp. 130-47

Levy-Bruhl L. 1926. How Natives Think. New York: Knopf

Malkiel Y. 1987. Integration of phonosymbol- ism with other categories of language change. In Pap. 7th Int. Con$ Hist. Lin- guist., ed. AG Ramat, 0 Canuba, G Bern- ini, pp. 373-406. Amsterdam: Benjamins

Malkiel Y. 1994. Regular sound development, phonosymbolic orchestration, disambigua- tion of homonyms. See Hinton et a1 1994, pp. 207-21

Maltzman I, Morrisett L Jr, Brooks LO. 1956. An investigation of phonetic symbolism. J. Abnorm. Soc. Psychol. 53:245-5 1

Mandelbaum DG, ed. 1949. Selected Writings of Edward Sapir in Language, Culture, Personality. Berkeley: Univ. Calif. Press

Mannheim B. 199 1. The Language of the Inka Since the European Invasion. Austin: Univ. Texas Press

Marchand H. 1959. Phonetic symbolism in English word-formation. Indoger. Forsch. 64:146-68

Markel NN, Hamp EP. 1960. Connotative meanings of certain phoneme sequences. Stud. Linguist. 15 :47-6 1

Marks LE, Bornstein MH. 1987. Sensory simi- larities: classes, characteristics, and cogni- tive consequences. In Cognition and Sym- bolic Structures: the psycho lo^^ of Meta- phoric Transformation, ed. RE Haskell, pp. 49-65. Nonvood, NJ: Ablex

Martin SE. 1962. Phonetic symbolism in Ko- rean. In American Studies in Altaic Lin- guistics, ed. N Poppe, pp. 177-89. Bloom- ington: Indiana Univ. Press

Matisoff JA. 1994. Tone, intonation, and sound symbolism in Lahu: loading the syl- lable canon. See Hinton et a1 1994, pp. 1 15-29

McGregor W. 1999. Ideophones as the source of verbs in Northern Australian languages. Presented at Int. Symp. Ideophones, Jan. 25-27, Inst. Afrikan, Univ. Koln

McNeill D. 1992. Hand and Mind: What Ges- tures Reveal about Thought. Chicago: Univ. Chicago Press

Miron MS. 1961. A cross-linguistic investiga- tion of phonetic symbolism. J. Abnorm. Soc. Psychol. 62:623-30

Mithun M. 1982. The synchronic and dia- chronic behavior of plops, squeaks, croaks, sighs, and moans. Int. J. Amer. Linguist. 48:49-58

Morton E. 1994. Sound symbolism and its role in non-human vertebrate communication. See Hinton et a1 1994, pp. 348-65

Murdock GP. 1959. Cross-language parallels in parental kin terms. Anthropol. Linguist. 1:1-5

Newman P. 1968. Ideophones from a syntactic point of view. J. W.Afr. Lung. 2:107-17

Newman SS. 1933. Further experiments in phonetic symbolism. Am. J. Psychol. 45: 53-75

Nichols J. 1971. Diminutive consonant sym- bolism in western North America. Language 47:826-48

Noss PA. 1975. The ideophone: a linguistic and Literary device in Gbaya and Sango with reference to Zande. In Directions in Sudanese Linguistics and Folklore, ed. SH

Hurreiz, H Bell, pp. 142-52. Khartoum,

Sudan: Khartoum Univ. Press

Nuckolls C. 1996. The Cultural Dialectics of Knowledge and Desire. Madison: Univ. Wis. Press

Nuckolls J. 1996. Sounds Like Life: Sound- symbolic Grammar, Performance, and Cognition. New York: Oxford Univ. Press Nuckolls J. 1999. The structural indetermi- nacy of ideophones. Presented at Int. Symp. Ideophones, Jan. 25-27, Inst. Afrikan., Univ. Koln Ohala J. 1982. The voice of dominance. J. Acoust. Soc. Am. 72:S66 Ohala J. 1994. The frequency code underlies the sound-symbolic use of voice pitch. See Hinton et a1 1994, pp. 325-47 Orr J. 1944. On some sound values in English. Br. J. Psychol. XXXV 1:1-8 Osgood CE. 1960. The cross-cultural general- ity of visual-verbal synesthetic tendencies. Behav. Sci. 5: 146-69 Peirce CS. 1955. Logic as semiotic: the theory of signs. In Philosophical Writings of Peirce, ed. J Buchler, pp. 99-1 19. New York: Dover Popjes J, Popjes J. 1986. Canela-Kraho. See Derbyshire & Pullum 1986, pp. 128-99 Priestly TMS. 1994. On levels of analysis of sound symbolism in poetry with an appli- cation to Russian poetry. See Hinton et a1 1994, pp. 237-48 Reichard GA. 1944. Prayer: The Compulsive Word. Am. Ethnol. Soc. Monogr. 7. Seattle: Univ. Wash. Press Reichard GA. 1950. Navajo Religion. Bollin- gen Ser. 18. New York: Pantheon Reichard GA, Jakobson R, Werth E. 1949. Language and synesthesia. Word 5:224-33 Rhodes R. 1994. Aural images. See Hinton et a1 1994, pp. 276-92 Rhodes R, Lawler JM. 198 1. Athematic meta- phors. Chicago Lingist. Soc. 17:3 18-42 Samarin W. 1965. Perspective on African ideophones. Afi-. Stud. 24: 1 17-2 1 Samarin W. 1967. Determining the meanings of ideophones. J. West Afr. Linguist. 4(2): 3541

Samarin W. 1970. Field procedures in ideo- phone research. J. Afi. Lang. 9(1):27-30

Samarin W. 1971. Survey of Bantu ideo-phones. Aj?. Lang. Stud. 12: 130-68

Sapir E. 1929. A study in phonetic symbolism.

J. Experimental Psychol. 12:225-39 Sapir E. 1949a. A study in phonetic symbol- ism. See Mandelbaum 1949, pp. 61-72

Sapir E. 1949b. Abnormal types of speech in Nootka. See Mandelbaum 1949, pp. 179-96

Saussure F de. 1959. Course in General Lin- guistics, EC Bally, A Sechehaye, A Ried- linger. Transl. W Baskin. New York: McGraw-Hill

Schultze-Berndt E. 1999. Traces of ideo-phones in complex predicates of Northern Australia. Presented at Int. Symp. Ideo- phones, Jan. 25-27, Inst. Afrikan., Univ. Koln

Sebeok TA, ed. 1960. Style in Language. Cambridge, MA: MIT Press Sereno J. 1994. Phonosyntactics. See Hinton et a1 1994, pp. 263-75

Sherzer J. 1976. An Areal-Tyological Study of American Indian Languages North ofMex- ico. Amsterdam: North-Holland

Sherzer J. 1990. Verbal Art in Sun Blas: Kuna Culture Through Its Discourse. Cambridge, UK: Cambridge Univ. Press

Shore B. 1996. Culture in Mind. New York: Oxford Univ. Press

Silverstein M. 1994. Relative motivation in denotational and indexical sound symbol- ism of Wasco-Wishram Chinookan. See Hinton et a1 1994, pp. 40-60

Taylor IK. 1963. Phonetic symbolism reexamined. Psychol. Bull. 60(2):200-9

Taylor IK, Taylor MR. 1965. Another look at phonetic symbolism. Psychol. Bull. 64(6): 4 13-27

Tedlock D. 1983. The Spoken Word and the Workoflnterpretation.Philadelphia: Univ. Penn. Press

Thorndike EL. 1945. On Orr's hypthesis con- cerning the front and back vowels. Br. J. Psychol. 36:SlO-14

Tsur R. 1992. What Makes SoundPatterns Ex-

pressive? DurharnILondon: Duke Univ. Press

Ultan R. 1978. Size-sound symbolism. See Greenberg et a1 1978, pp. 525-28

Urban G. 1985. The semiotics of two speech styles in Shokleng. In Semiotic Mediation: Sociocultural and Psychological Perspec- tives, ed. E Mertz, R Parmentier, pp. 3 1 1-29. Orlando, FL: Academic

Voeltz EFK. 1971. Towards the syntax of the ideophone in Zulu. In Papers in African Linguistics, ed. CW Kim, HS Stahlke, pp. 14 1-52. Edmonton, Can: Linguist. Res.

Waugh L, Newfield M. 1995. Iconicity in the lexicon and its relevance for a theory of morphology. See Landsberg 1995, pp. 189-21

Weidman de Kondo R. 1976. Onomatopeya en guahibo. Articul. Linguist. Campos Afines 2:21-32

Weiss JH. 1964. Phonetic symbolism reexamined. Psychol. Bull. 6 1 (6):454-58

Werner H, Kaplan B. 1963. Symbol Forma- tion: An Organismic-Developmental Ap- proach to Language and the Expression of Thought. New York: Wiley

Wescott R. 1971. Linguistic iconism. Language 47(2):416-28 Wescott R. 1977. Ideophones in Bini and Eng- lish. Forum Linguist. 2(1): 1-3

Westermann D. 1930. A Study ofthe Ewe Lan- guage. Transl. AL Bickford-Smith. Lon- don: Oxford Univ. Press

Witherspoon G. 1977.LanguageandArt in the Navajo Universe. Ann Arbor: Univ. Mich. Press

Woodbury AC. 1985. The functions of rhetori- cal structure: a study of Central Alaskan Yupik Eskimo discourse. Lung. Soc. 14: 1 5 3-90 Woodworth NL. 199 1. Sound symbolism in proximal and distal forms. Linguistics 29: 273-99

  • Recommend Us