AUTHORS: Gerd Carling (Lunds University), Niklas Erben Johansson (Lund University), Johan Frid (Lund University) KEYWORDS: iconicity, sound symbolism, historical linguistics, phylogenetics, ABSTRACT: Vocal iconicity, i.e., a motivated connection between form and meaning in language, is a property of language that most linguists believe to be cognitively fundamental. According to one theory (Wescott 1971), iconic words were more frequent in early human language, but the iconicity was replaced by arbitrariness over time, by means of conventionalization of meaning and sound change. While iconicity and synesthetic cross-modal mappings are present in the early stages of human ontogenetic development (Walker et al. 2010) and go at least as far back as the ancestor we share with chimpanzees (Perlman 2017), they do not seem to disappear but rather gradually decrease with age, language competence and vocabulary size. However, iconicity is still frequently occurring throughout the lexicon, particularly in certain semantic domains (e.g., kinship terms, birds, deictic terms, color words) (cf. Dingemanse et al. 2015). The current paper looks at the principles of emergence and decay of iconicity in the Indo-European family from a comparative phylogenetic perspective. We start from a cross-linguistic genetically and areally balanced study (Erben Johansson et al 2020), which identifies iconic concepts in basic vocabulary as well as the most overrepresented phonemes in these concepts. Using these results as a basis, we take a cognacy coded dataset of basic vocabulary and analyze the lexemes into phonemic annotation, which we tokenize and align into sequences, using specialized historical linguistic software for automatic cognate detection (List et al. 2019). The tokenized data set we segment into syllable structure. Thereupon, we simplify the dataset by converting existing phonemes into phoneme prototypes, based on salient phonetic and sound symbolic features (Erben Johansson et al 2020, Joo 2019)). Finally, we use a phylogenetic comparative model with a Continuous-Time Markov Reversible inference, running phonemes and phoneme prototypes of cognates over a phylogenetic reference tree (Hammarström, Forkel, and Haspelmath 2017), to test the change (gain/loss) rates for corresponding phonemes and phoneme prototypes, both independently and by syllable structure. We measure the results by concepts that are iconic and concepts that are not (i.e., arbitrary), concluded from a global sample (Blasi et al. 2016). We test whether phonemes/phoneme prototypes, which are motivated by a form-meaning association in iconic concepts, behave differently than those in arbitrary concepts. We also test whether motivated phonemes/phoneme prototypes have higher change rates (gain/loss) and are thus more likely to appear (or disappear), in iconic concepts. To conclude, the aim of the paper is to investigate, using one large and well-studied language family, the mechanisms that retain, create (and re-create) vocal iconicity, in spite of conventionalization and language change. SCHEDULE: Tuesday 11.20-11.30