Traditional schools of thought in linguistics have stated that arbitrariness is a fundamental property of language. Arbitrariness is the principle that linguistic forms have no inherent correspondence with their meanings. Opposing this concept is iconicity, the idea that semantics and the forms used to denote them are related in some way. Many studies have shown that not only does iconicity exist across languages, but that it may be an essential part of language acquisition and development in children. This paper explores current research on iconic mappings in both auditory-oral and visual-manual modalities and suggestions for future explorations on the topic.
Keywords: iconicity, language acquisition, language evolution
Iconicity in Spoken and Signed Language: Origins, Acquisition and Development
In recent years, iconicity has become more prominent and accepted as a fundamental property of language than it used to be. It has become clear that even though it is not always obvious, there are non-random mappings between a word or sentence’s form and the meaning it conveys. The existence of these iconic mappings and the effect they have on language acquisition and processing may have important implications for how language began. If the human body is wired to associate certain stimuli with specific concepts, it could provide insight into how origins of language as we know it.
What is Iconicity?
A large reason arbitrariness is believed to be the fundamental driving force of how words come to mean something is because language needs to have clear distinction between words, be it vocal or signed. Human language has essentially unlimited expressive power and a substantial number of possible referents. Arbitrariness can be thought of as a linguistic “force” pushing language to have clearly distinguishable forms. By this token, it wouldn’t make sense if form was intrinsically linked to meaning. For example, consider the semantic domain of trees which has many types and attributes to communicate about. Iconicity implies that words in the tree domain would sound or look similar and would therefore make it difficult to differentiate between them when being addressed.
Besides the unintuitive nature of iconicity, resistance to its acceptance is a historical artifact of the linguistic field. For one thing, much of early research was focused on Indo-European languages. These languages do not boast many linguistic categories that are obviously iconic, as opposed to languages such as Japanese which contain whole lexicons of mimetic words. A more holistic view at the dawn of linguistics could have led to earlier acceptance of iconicity’s role in language.
Similarly, sign languages have not always been considered worthy of being attributed the complexity of a full human language. They were thought to be pantomime or gestures without grammars, phonemes or morphemes. However, due to the visual and manual nature of signs, sign languages generally have a larger proportion of symbols that map to physical characteristics of the real world. It would have been hard to deny that iconic form-meaning mappings exist if this entire class of languages had been studied before arbitrariness became the prominent belief.
There are two separate areas of study on iconicity that deal with the major modalities of auditory-oral and visual-manual languages, or spoken and signed, respectively. The first class studies the extent of sound-symbolic mappings in speech and what its effects are in human language processing. Oral speech, as might be expected, has a relatively small collection of words that can be counted as inherently iconic. It has even less that are noticeably so at the conscious level. On the other hand, signs make use of movement, location and other features that allow more direct mappings of signs to their referents. The below sections detail studies for each modality and explain some of the similarities and differences between them. Lastly, there are a few experiments that attempt to gain knowledge of how gestures and speech integrate into a cohesive whole that can be understood by the human mind.
The subtle nature of iconicity in spoken language naturally gives rise to the question of whether evidence for it actually exists. In that regard, a study by Blasi et al. shows a statistical analysis of sound-meaning association biases in over two-thirds of the world’s languages. Their analysis used a 40-item subset of the Swadesh list from 6,447-word lists, with 328 lists having the full 100-word Swadesh list. The results of their analysis convey that there are indeed non-arbitrary sound-meaning associations across languages. Specifically, 74 associations were found between 30 concepts and 23 symbols. An interesting note is that some of these iconic mappings were well known in previous literature, but some were not known beforehand. For instance, the concept of round was found to have a strong association with the “r” sound (Blasi, Wichmann, Hammarström, Stadler, & Christiansen, 2016).
Much of the literature dealing with iconic associations use front and back vowels as typical examples of phonemic units that tend to elicit the semantic connotations of small and large, respectively. One well-known mapping is from vowel articulation location to round or jagged object shape, called the Bouba-Kiki effect. When asked which of two shapes was “kiki” and which was “bouba,” over 95% of participants in a famous study said the rounder shape was “bouba” (Perniss, Thompson, & Vigliocco, 2010). More recently, Jan Auracher wrote about a study that expands this idea of vowel sound to meaning mappings to other abstract concepts such as social dominance or power (Auracher, 2017). Native Japanese speakers were presented with random pseudo-words made up of solely front (i.e. /i/ and /e/) or back vowels (i.e. /o/ and /u/). They were then asked to categorize these words, as well as illustrations of small and large animals using the Implicit Association Test. Reaction times of the participants confirmed the hypothesis that the articular placement of vowels has an association with small or large. Even more interesting is that the nature of the relevant experiments indicate that this categorization occurs at the semantic level, not the perceptual level (i.e. a larger picture of the same animal did not affect the results). To test whether the results hold true for associations with social dominance, illustrations of a person in dominant or submissive poses were displayed instead. This confirmed that back vowels were indeed associated with power and front vowels with weakness (Auracher, 2017).
Several other kinds of non-arbitrary, or motivated, language forms have been observed. This includes rate of speech, pitch and repetition, to name a few. An instance of speech rate being used to encode meaning is shown when participants of an experiment were asked to describe the motion (e.g. left, right, etc.) of a dot on a computer screen. The rate of speech was consistently higher for higher dot speeds. Speed semantic encoding was observed for listeners as well. In the same experiment, those who were asked to listen to the dot’s description were generally able to guess the ball’s speed (Perniss et al., 2010).
Another study about iconicity and other non-arbitrary linguistic tendencies (Dingemanse, Blasi, Lupyan, Christiansen, & Monaghan, 2015) describes how iconicity may go hand in hand with another type of sound-meaning mapping called systematicity. Systematicity is a “statistical relationship between the patterns of sound for a group of words and their usage” (Dingemanse, et al.). One such group words is those that relate to lighting and begin with the “gl” sound (e.g. glare, glimmer, glow). Importantly, these groups of words are distinct from iconic words, but are useful and non-arbitrary in their own way. The classes of words known as ideophones and onomatopoeia are also brought up as quintessentially iconic and found across the world’s languages. I point this out to note that iconicity is not the be all end all of nonrandom form to meaning mappings and that iconicity and systematicity may have different effects on language acquisition and processing. A more thorough explanation of the effects of iconicity will be provided, but systematicity is relevant and important enough to make a digression. Specifically, the sounds found in systematic categories of words may facilitate categorical word learning in learners and is more unique per language than iconic words (Dingemanse et al., 2015).
In terms of language processing, experiments have shown participants’ responses to stimuli is clearly sensitive to iconic form-meaning mappings. One example studied how age of acquisition and lexical category relate to iconicity in English and Spanish (Perry, Perlman, & Lupyan, 2015). Part of what makes this and other studies possible is the understanding that iconicity is not a binary concept, but more of a continuum. Degrees of how direct a mapping is to a referent’s sensory experience varies from word to word.
The experiment set up a rating system from -5 to 5 with -5 sounding like the opposite of its meaning, 0 sounding arbitrary and 5 sounding highly iconic. Around 600 words were presented to participants in different combinations of written and auditory forms. The participants then had to rate how much the word’s form mapped to its meaning. Over five similar experiments in English and Spanish, the lowest average iconicity rating was 0.75, which is significantly above 0. The resulting data indicated that earlier age of acquisition corresponded with higher iconicity ratings. This held true for all five experiments and, along with other studies, shows that iconicity of a word’s form may have facilitate learning and comprehension in children (Perry et al., 2015).
Another important finding was that iconicity ratings varied consistently between lexical categories. Onomatopoeia and adjectives were rated as more iconic than nouns and function words by most participants of the experiments. This makes sense because onomatopoeia are words that imitate the object that it describes, which is almost by definition iconic. An example is the word “meow” used to represent the sounds a cat makes. Relative levels of iconicity in different lexical categories is a useful insight indicating that non-arbitrary form-meaning mappings are a key property of language (Perry et al., 2015).
Sign languages contain a much larger proportion of units that mimic their referent. For example, the British Sign Language (BSL) form for airplane is made by holding the pinkie and thumb out while the middle three fingers curl down, making what looks like airplane wings. The hand then moves through space. As you can see, signs have several properties that make them particularly suited to bridging the gap between language and our sensory experience.
Sign phonemes can be split into the following sub-lexical units; handshape; orientation; location; and movement (Ortega, 2017). The ability to use 3D space allows signs to have a more direct mapping than speech phonemes that, if taken as purely oral, are restricted to subtle acoustic and physical mappings with the human body’s vocal instruments (e.g. lips). Despite this property, there are still different degrees of iconicity. Ortega describes four levels of iconic mappings. Transparent mappings are obvious and clearly mimic an object or concept. Signs of this type may even be understood by non-signers. Translucent mappings are not immediately clear but convey some attributes of the referent. Obscure mappings are evident only after the connection between the form and referent has been explained. Lastly, opaque mappings have no apparent iconicity (Ortega, 2017).
Another distinction is that between perception-based and action-based form-meaning associations. This would be the difference between a sign for deer and the sign for drinking. Deer is conveyed with the hands spread next to the head in an imitation of antlers. On the other hand, drinking is signed with the hand curled as if around a cup and brought up to the mouth. The first example plays to how deer are visually perceived, while the second represents the act of drinking.
These degrees and types of iconicity are important because how words and signs are processed is affected by it. Ortega points out that initial research concluded that iconicity does not play a role in language acquisition. However, the reason may be that those studies did not take into account the several types of iconicity when designing experiments and interpreting the results. More direct, transparent mappings have a greater effect on comprehension and learning. More recent studies that use the MacArthur Bates CDI found that the first signs produced by children tended to be more directly iconic than later signs. It was also found that direct symbol-referent mappings were not understood by children until about the age of three (Ortega, 2017).
Iconicity does not just affect comprehension and production in children, though. It seems to improve response time and memory in adult hearing non-signers when asked to state the meaning of iconic signs two weeks after exposure. Surprisingly, it had a negative effect for hearing proficient signers. This indicates that iconicity helps during the initial learning phases for mapping a sign to its referent. However, once an established lexicon exists within one’s mental model, other linguistic processes such as frequency and lexical competition can dominate.
One conclusion described by Ortega that I found particularly interesting was about production of signs. Phonologically accurate reproduction of learned signs was negatively affected by level of iconicity. Essentially, non-signers payed less attention to the exact form of iconic signs and instead produced similar sign phonemes that still encapsulated the semantic mapping of the sign. For example, different fingers or motions were used in the sign for airplane, but still in a way that imitated the wings and motion of the object being referred to. Since learners have access to semantic knowledge of iconic signs, they pay less attention to the phonological aspects of the sign. This leads to decreased accuracy in sign production (Ortega, 2017).
Besides acquisition, sign iconicity may play a role in priming for word retrieval. Thompson published a study on what signed languages reveal about language processing and acquisition. In this study, Thompson describes a picture-sign matching task where deaf early learners and hearing late learners of American Sign Language (ASL) were told to state whether a picture matches a sign. The experimental condition was the salience of the picture regarding the sign. For example, the sign for candle depicts a flame as part of its form, so a picture of a lit candle would be salient while an unlit candle would not be. Both early learners and late learners were faster to response “yes” when the picture shared features with the sign. This effect was reproduced with BSL (L., 2011).
While separating language into isolated modalities of auditory-oral and visual-manual seems intuitive at first glance, in reality language is a complex and relies on many cognitive processes and abilities (Özyürek, 2014). In addition, Perniss et al. (2014) have suggested that language should be treated primarily as a method of face-to-face communication, which entails speech in combination with co-speech gestures. These systems are separate but integrate to convey semantic information to the addressed party (Perniss & Vigliocco, 2014).
Perniss et al. (2014) have also discussed how iconicity may play a role in language evolution through displacement, ontogenesis through referentiality and language processing through embodiment. Displacement is the ability of language to refer to that which is not present. This would be key to the evolution of language faculty in early humans where conditions for displacement to arise could have been ideal. Particularly, large group sizes with division of labor and transmission of skill would encourage displacement. Skills and information would have to be shared between members of the group but would need to refer things in the world that were not immediately present. Gestures and iconicity would be a valuable tool in initiating the communication process be helping to provide referential insight to objects that were out of sight. Transparent mappings would arise first as a way to mimic properties of objects, with more abstract forms coming later as forms were conventionalized (Perniss & Vigliocco, 2014). Imai and Kita (2014) note that this process would have helped build an initial shared lexicon in our ancestors.
They also describe a sound symbolism bootstrapping hypothesis that helps explain how children can map symbols to specific referents, despite the huge amount if input stimuli they receive. A shortened explanation of the hypothesis is that infants are born with a biologically endowed ability to integrate multi-modal inputs. This ability allows them to isolate sound symbols and their referents in complex scenes. As an example, iconicity in combination with gestures (e.g. eye gaze or pointing) would help an infant focus on the relevant object in the scene that relate to the sound symbol (Imai & Kita, 2014).
Areas of Focus
Research on the area of iconicity as a fundamental property of language has gained much attention in recent years. It has progressed from being dismissed to having in-depth studies about its role in language evolution, acquisition and processing. A lot of that research focuses on one modality, but I believe keeping a more complex and nuanced view of language is important for future insights. The cognitive processes involved may differ between modality, degree and type of iconicity, and other features.
In that regard, my admittedly narrow search for experiments and data turned up only one study that involved brain measurements (Imai & Kita, 2014). Response times to various stimuli have provided important insights but may be affected by unknown variables. In addition, I feel that brain scans such as EEGs that measure brain responses can give concrete data into what is happening when one processes iconic mappings from its form to its semantics.
Iconicity, the idea that there is an inherent meaning associate with certain language forms, has become an accepted feature of language. It has been shown that speech in both the auditory-oral and visual-manual modalities has iconic mappings that cross language barriers. Examples include an association of front vowels to small object back vowels to large objects. However, there are differences between the modalities. For one, sign languages can make use of movement and space in ways that lend itself to imitating the real world. This is shown in the higher number of iconic forms found in sign languages compared to spoken languages.
Language acquisition and processing is clearly affected by form-meaning mappings. Children may use iconicity to gain referential insight during the beginning phases of language learning. Studies show that transparently iconic words are learned earlier. Even in adults, motivated forms increase comprehension and retention over extended time periods. In addition, iconic forms appear to be mapped at the semantic level and not the phonological level.
Lastly, iconicity may have helped bootstrap language as we know it. With a need to refer to objects and concepts that were not present, early humans could have used mimetic gestures and sounds to form a sort of protolanguage. It would have used direct mappings at first, with more abstract and arbitrary forms appearing as more referents were needed. Overall, iconicity is a key aspect of language and likely played a role in the development of language and our ability as humans to understand and produce it.
Auracher, J. (2017). Sound iconicity of abstract concepts: Place of articulation is implicitly associated with abstract concepts of size and social dominance. PLOS ONE, 12(11), e0187196. https://doi.org/10.1371/journal.pone.0187196
Blasi, D. E., Wichmann, S., Hammarström, H., Stadler, P. F., & Christiansen, M. H. (2016). Sound–meaning association biases evidenced across thousands of languages. Proceedings of the National Academy of Sciences, 113(39), 10818 LP-10823. Retrieved from http://www.pnas.org/content/113/39/10818.abstract
Dingemanse, M., Blasi, D. E., Lupyan, G., Christiansen, M. H., & Monaghan, P. (2015). Arbitrariness, Iconicity, and Systematicity in Language. Trends in Cognitive Sciences, 19(10), 603–615. https://doi.org/10.1016/J.TICS.2015.07.013
Imai, M., & Kita, S. (2014). The sound symbolism bootstrapping hypothesis for language acquisition and language evolution. Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1651). Retrieved from http://rstb.royalsocietypublishing.org/content/369/1651/20130298.abstract
L., T. R. (2011). Iconicity in Language Processing and Acquisition: What Signed Languages Reveal. Language and Linguistics Compass, 5(9), 603–616. https://doi.org/10.1111/j.1749-818X.2011.00301.x
Ortega, G. (2017). Iconicity and Sign Lexical Acquisition: A Review. Frontiers in Psychology, 8, 1280. https://doi.org/10.3389/fpsyg.2017.01280
Özyürek, A. (2014). Hearing and seeing meaning in speech and gesture: insights from brain and behaviour. Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1651). Retrieved from http://rstb.royalsocietypublishing.org/content/369/1651/20130296.abstract
Perniss, P., Thompson, R. L., & Vigliocco, G. (2010). Iconicity as a General Property of Language: Evidence from Spoken and Signed Languages. Frontiers in Psychology, 1, 227. https://doi.org/10.3389/fpsyg.2010.00227
Perniss, P., & Vigliocco, G. (2014). The bridge of iconicity: from a world of experience to the experience of language. Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1651). Retrieved from http://rstb.royalsocietypublishing.org/content/369/1651/20130300.abstract
Perry, L. K., Perlman, M., & Lupyan, G. (2015). Iconicity in English and Spanish and Its Relation to Lexical Category and Age of Acquisition. PLOS ONE, 10(9), e0137147. https://doi.org/10.1371/journal.pone.0137147