An investigation of iconic language development in four datasets

Abstract

Iconic words imitate their meanings. Previous work has demonstrated that iconic words are more common in infants’ early speech, and in adults’ child-directed speech (e.g., Perry et al., 2015; 2018). This is consistent with the proposal that iconicity provides a benefit to word learning. Here we explored iconicity in four diverse language development datasets: a production corpus for infants and preschoolers (MacWhinney, 2000), comprehension data for school-aged children to young adults (Dale & O'Rourke, 1981), word frequency norms from educational texts for school aged children to young adults (Zeno et al., 1995), and a database of parent-reported infant word production (Frank et al., 2017). In all four analyses, we found that iconic words were more common at younger ages. We also explored how this relationship differed by syntactic class, finding only modest evidence for differences. Overall, the results suggest that, beyond infancy, iconicity is an important factor in language acquisition.

Not just form, not just meaning: Words with consistent form-meaning mappings are learned earlier

“…Moreover, we showed the robustness of the reported patterns by replicating the analyses using RF regressions, which remove possible adverse effects of collinearity ( Tomaschek et al, 2018 ). Finally, we showed that the relation between systematicity and AoA is robust to the inclusion of iconicity ( Perry et al, 2015 ; Sidhu et al, 2021 ), suggesting that systematicity and iconicity tap into different aspects of acquisition dynamics.…”

“…This dataset provides different AoA estimates for different word senses of the same word form; however, since our approach does not discriminate word senses, we took the earliest AoA score for each word form as our target variable. Next to providing objective AoA norms for a large set of words, this dataset was also recently investigated by Sidhu and colleagues (2021) who reported an effect of iconicity on AoA. It is thus interesting to investigate the effect of systematicity on AoA on the same dataset, to better situate any finding in the current literature on non-arbitrariness and language learning.…”

“…The factors that influence how children learn their first language from exposure have been widely studied ( Bates & MacWhinney, 1987 ; Pinker, 1984 ). The spate of studies which addressed this question highlighted several factors which predict Age of Acquisition (AoA), including word frequency ( Braginsky et al, 2019 ; Ghyselinck et al, 2004 ), phonological neighbourhood density (PND; Jones & Brandt, 2019 ), semantic neighbourhood density (SND; Fourtassi et al, 2020 ), contextual diversity ( Hills et al, 2010 ), perceptual features ( Peters & Borovsky, 2019 ), valence, concreteness, and word length ( Braginsky et al, 2019 ), salience in time, space, and, linguistic context ( Roy et al, 2015 ), iconicity ( Laing, 2019 ; Perry et al, 2015 ; Sidhu et al, 2021 ), sound symbolism ( Imai & Kita, 2014 ; Kantartzis et al, 2011 ), and systematicity in form-meaning mappings ( Monaghan et al, 2012 , 2014 ). Building on recent studies that quantified the degree of form-meaning systematicity for individual lexical items (Hendrix & Sun, 2020; Marelli & Amenta, 2018 ), in this work, we assess the effect of systematicity on AoA, after controlling for known predictors of AoA itself.…”

“…To sum up, this work leverages and extends two different yet connected lines of research. On one hand, we build on studies which investigate the factors influencing language acquisition ( Braginsky et al, 2019 ; Ghyselinck et al, 2004 ; Hills et al, 2010 ; Jones & Brandt, 2019 ; Peters & Borovsky, 2019 ; Roy et al, 2015 ; Sidhu et al, 2021 ), particularly on studies showing that sound symbolism helps language learning ( Imai et al, 2008 ; Imai & Kita, 2014 ; Kantartzis et al, 2011 ) and that more systematic words are overrepresented among early acquired words ( Monaghan et al, 2014 ). In detail, we investigate whether systematicity predicts AoA while controlling for neighbourhood density in both form and meaning, since recent studies have shown that words that are highly entrenched in the phonological and semantic networks are learned earlier ( Fourtassi et al, 2020 ; Jones & Brandt, 2019 ).…”

Subjective age of acquisition norms for 1604 English words by Spanish L2 speakers of English and their relationship with lexico-semantic, affective, sociolinguistic and proficiency variables

“…1 Also, words with a low AoA are more prevalent ("prevalence" being a measure of a population's word knowledge; Brysbaert & New, 2009, Brysbaert et al, 2016. Iconicity (the resemblance between the form and the meaning of a word) helps language acquisition (Imai & Kita, 2014;Massaro & Perlman, 2017;Perniss & Vigliocco, 2014) and is related to AoA (Perry et al, 2015), so words with high iconicity scores are acquired earlier and are more frequent in infancy than words with low iconicity scores in both oral and sign languages (Caselli & Pyers, 2017;Hinojosa et al, 2021;Sidhu et al, 2021;Thompson et al, 2012;Vinson et al, 2008). In addition, imageability (a measure of how easy it is for a person to create a mental image of something) scores are higher for words with early AoA (Citron et al, 2014).…”