LDL | Department of Linguistics | HKULDL | Department of Linguistics | HKUIconic words are known to exhibit an imitative relationship between a word and its referent. Many studies have worked to pinpoint sound-to-meaning correspondences for ideophones from different languages. The correspondence patterns show similarities across languages, but what makes such language-specific correspondences universal, as iconicity claims to be, remains unclear. This could be due to a lack of consensus on how to describe and test the perceptuo-motor affordances that make an iconic word feel imitative to speakers. We created and analyzed a database of 1,888 ideophones across 13 languages, and found that 5 articulatory properties, physiologically accessible to all spoken language users, pattern according to semantic features of ideophones. Our findings pave the way for future research to utilize articulatory properties as a means to test and explain how iconicity is encoded in spoken language.
Chinese Ideophone Database (CHIDEOD) is an open-source dataset coded in a user-friendly format, which collects 3453 unique onomatopoeia and ideophones (mimetics) of Mandarin Chinese, as well as Middle Chinese and Old Chinese (based on Baxter & Sagart 2014). These are analyzed according to a wide range of linguistic features, including phonological, semantic, as well as orthographic ones. CHIDEOD was created on the basis of data collection and analysis conducted by Arthur Thompson in our lab in collaboration with Thomas Van Hoey (then in National Taiwan University). For individual sources and files relevant to the database, please visit https://osf.io/kpwgf/
Iconicity is when linguistic units are perceived as “sounding like what they mean,” so that phonological structure of an iconic word is what begets its meaning through perceived imitation, rather than an arbitrary semantic link. Fundamental examples are onomatopoeia, e.g., dog”s barking: woof woof (English), wou wou (Cantonese), wan wan (Japanese), hau hau (Polish). Systematicity is often conflated with iconicity because it is also a phenomenon whereby a word begets its meaning from phonological structure, albeit through (arbitrary) statistical relationships, as opposed to perceived imitation. One example is gl- (Germanic languages), where speakers can intuit the meaning “light” via knowledge of similar words, e.g., glisten, glint, glow, gleam, glimmer. This conflation of iconicity and systematicity arises from questions like “How can we differentiate or qualify perceived imitation from (arbitrary) statistical relationships?” So far there is no proposal to answer this question. By drawing observations from the visual modality, this paper mediates ambiguity between iconicity and systematicity in spoken language by proposing a methodology which explains how iconicity is achieved through perceptuo-motor analogies derived from oral articulatory gesture. We propose that universal accessibility of articulatory gestures, and human ability to create (perceptuo-motor) analogy, is what in turn makes iconicity universal and thus easily learnable by speakers regardless of language background, as studies have shown. Conversely, our methodology allows one to argue which words are devoid of iconicity seeing as such words should not be explainable in terms of articulatory gesture. We use ideophones from Chaoyang (Southern Min) to illustrate our methodology. Published here.
Studies on English and Spanish use ratings to identify words speakers consider iconic. Our study replicates this for Japanese but, owing to additional variables, yields more nuanced findings. We propose that ratings reflect a word’s relationship to sensory information rather than iconicity.
In this project, we explore how people interpret imitative quotatives “and then he was like…” in several conditions.
Ideophones are marked words that depict sensory imagery and occur in many languages. It has been found that these words are easier to learn which might be due to their depictive properties. In this project we investigate whether visual cues such as lip rounding and mouth opening help in learning ideophones.
Some languages have more forms of conventional spoken iconicity than others. Japanese, for example, has more ideophones than English. So how do speakers of a language with limited semantic categories of ideophones depict percepts? One possibility is demonstrations: unconventional, yet depictive, discourse. Demonstrations follow quotatives (e.g., I was like ___) and perform referents as opposed to describing them. In English, a language with arguably restricted sets of ideophones, speakers may enact/create demonstrations using their hands, voice, and body. This paper examines which visual and spoken components are vital to comprehending demonstrations in English with features from Güldemann’s (2008) observations: enacted verbal behaviour, non-linguistic vocal imitation, ideophones, and representational gesture. 28 videos containing demonstrations of 11 celebrities engaging in impromptu storytelling on USA talk shows were our critical stimuli. 145 native speakers completed forced multiple-choice judgement tasks to qualify each demonstration. To see which forms of visual and spoken communication contributed to comprehension, videos were presented in visual (muted), audio (pixelated and darkened), and audio–visual (left as is) conditions. Our results show that if arbitrary speech (e.g., I was like I can’t go over the ocean!) is in a demonstration, then it is vital to comprehension. The visual condition rendered these demonstrations uninterpretable. If sound imitations (e.g., I was like prfff!) or ideophones coupled with hand gesture (e.g., I was like yay! + hands opening and closing in unison) are in a demonstration, then the interpretability of that demonstration across our experimental conditions depends on whether its components (gesture, sound imitation) can unambiguously express meaning in isolation. These findings allow us to make several conjectures about the wellformedness of demonstrations. Our findings are in line with studies on enactments in deaf signed languages whereby the more unconventional a form of iconic depiction is, the more it requires conventional framing to be interpretable.
This project establishes new measurements of phonological distance by incorporating lexical tone through experimental approaches and modeling, using Hong Kong Cantonese as a case study. Results show correspondences between the experimental data and predictions from information-theoretic measures, including entropy measure and functional load, suggesting that lexical components playing a more crucial role in phonological distance judgments are also lexically less predictable. The measurements and their iImplications on phonological distance measures are discussed in our paper.
Chinese tone sandhis can be broadly classified as left-dominant or right-dominant according to the position of the syllable preserving the citation tone. Right-dominant Chinese sandhis are more common than left-dominant sandhis, which might be attributed to phonetic naturalness such as duration. In studies of phonetic naturalness bias and structural simplicity bias, researchers normally control one factor to investigate another factor. When a certain effect is found, it could be the combinatorial effect. In this project we investigate the role of phonetic naturalness vs. structural simplicity in learning tone sandhi.
This project investigates the role of learning biases in acquiring sound patterns of languages among children from 4 to 7 years of age, who learn Cantonese, English, and Korean as their first language. Various work on language acquisition show cases where the outcome of language learning does not perfectly reflect what was in the input: learners may fail to acquire some input patterns or they may make some assumptions when input is ambiguous. When a discrepancy between the input and the learning outcome is observed, a conclusion has been made that the learning was biased in a certain way. This project explores types of learning biases children bring to the task of learning sound patterns. We also investigate the causal relationship of learning biases to phenomena of natural language acquisition and to the shapes of different languages.
This project investigates the role of production in learning unnatural sound patterns. Evidence for the naturalness bias is weak and we believe that this might be due to the methodological settings of most studies. The majority of artificial learning studies employ perception-only methods and not production. We plan to investigate the role of production by comparing learning in both settings: perception-only and perception and production.
This project investigates the existence of a link between phonotactics and alternation in phonological learning. We address learning of a vowel harmony pattern through the use of three artificial languages; one with a harmony pattern both within and across stems, another with a harmony pattern only across stems; and the other with a disharmony pattern within stems but harmony across stems. We find that evidence for such link may not be universally available.
This project explores a method of creating confidence bounds for information-theoretic measures in linguistics, such as entropy, Kullback-Leibler Divergence (KLD), and mutual information. We show that a useful measure of uncertainty can be derived from simple statistical principles, namely the asymptotic distribution of the maximum likelihood estimator (MLE) and the delta method. Published here.
The goal of this project is to develop a computational software tool designed to simulate phonotactic patterns of languages with metric information.
This project aims to model wordlikeness judgements when tone is incorporated. We first show how the two major determinants of wordlikeness judgements, i.e. phonotactic probability and neighbourhood density, can be measured when tone is involved. Our wordlikenss judgement data show that phonotactic probability, but not neighbourhood density, influences wordlikeness judgements. We also show that phonotactic probability affects the tendency to judge items as absolutely perfect or more or less wordlike, while it does not affect judgements that an item is absolutely not wordlike. Our study and the implications of our results for phonotactic modelling and processes involved in wordlikeness judgements are discussed in our paper.
This project explores a psycholinguistic reality of the hierarchical structure of words with multiple morphemes. In generative morphology, it has been assumed that individual morphemes in a complex derived word are represented in a hierarchical fashion with binary branching. We investigate its reality in L1 as well as L2 speakers” mind.
This project tests the Shallow Structure Hypothesis (SSH), which claims that compared to L1 processing, L2 language processing generally underuses grammatical information, prioritizing nongrammatical information. Specifically, this cross-modal priming study tests SSH at the level of morphology, investigating whether late advanced L2 learners construct hierarchically structured representations for trimorphemic derived words during real-time processing as native speakers do. Our results support SSH. In lexical decision on English trimorphemic words, L1 recognition of the targets was facilitated by their bimorphemic morphological-structural constituent primes, but not by their bimorphemic nonconstituent primes , which were only semantically and formally related to the target. In contrast, L2 recognition was equally facilitated by both constituent and nonconstituent primes. These results suggest that unlike L1 processing, L2 processing of multimorphemic words is not mainly guided by detailed morphological structure, overrelying on nonstructural information. Published here.
We are running a follow-up study of Song & Do (2016) ‘Cross-linguistic syntactic priming in bilinguals: priming of the subject-to-object raising construction between English and Korean.’ Bilingualism: Language and Cognition. [link]
This cross-modal priming study is one of the first to empirically test the long-held assumption that individual morphemes of multimorphemic words are represented according to a hierarchical structure. The results here support the psychological reality behind this assumption: Recognition of trimorphemic words (e.g., unkindness or [[un-[kind]]-ness]) was significantly facilitated by prior processing of their substrings when the substrings served as morphological constituents of the target words (e.g., unkind), but not when the substrings were not morphological constituents of the target words (e.g., kindness). This morphological structural priming occurred independently of the linear positions of morphological constituents. Published here.