Multiplex lexical networks reveal patterns in early word acquisition in children

Abstract

Network models of language have provided a way of linking cognitive processes to language structure. However, current approaches focus only on one linguistic relationship at a time, missing the complex multi-relational nature of language. In this work, we overcome this limitation by modelling the mental lexicon of English-speaking toddlers as a multiplex lexical network, i.e. a multi-layered network where N = 529 words/nodes are connected according to four relationship: (i) free association, (ii) feature sharing, (iii) co-occurrence, and (iv) phonological similarity. We investigate the topology of the resulting multiplex and then proceed to evaluate single layers and the full multiplex structure on their ability to predict empirically observed age of acquisition data of English speaking toddlers. We find that the multiplex topology is an important proxy of the cognitive processes of acquisition, capable of capturing emergent lexicon structure. In fact, we show that the multiplex structure is fundamentally more powerful than individual layers in predicting the ordering with which words are acquired. Furthermore, multiplex analysis allows for a quantification of distinct phases of lexical acquisition in early learners: while initially all the multiplex layers contribute to word learning, after about month 23 free associations take the lead in driving word acquisition.

Estimating Semantic Networks of Groups and Individuals from Fluency Data

“…Some possible semantic relations include causal relations (e.g., Causes(moon, tides)), featural similarity (e.g., Similar Color(sky, water)), subordinate or superordinate relations (e.g, Is A(oak,tree)), or temporal cooccurrence (e.g., Precedes (Rain, Lightning)). While some semantic networks make the relation type between nodes explicit (e.g., Stella et al 2017), many leave it implicit (e.g., Steyvers and Tenenbaum 2005).…”

Multiplex networks quantify robustness of the mental lexicon to catastrophic concept failures, aphasic degradation and ageing

“…Indeed, free associations identify recalls from semantic memory 11 that might be related not only to semantics but also to sound similarity or visual features. Despite a small overlap with semantic features and phonological relationships, as highlighted in 28,31 , a layer similarity analysis performed in a previous work 26 identified free associations as structurally dissimilar from the phonological layer and more similar, for its connectivity layout, to other semantic layers like conceptual generalisations or synonyms. This quantitative result identified the layer of free associations as being mainly semantic, which justifies the main approach of this study.…”

“…This work adopts a minimal representation of semantics and phonology in the human mental lexicon as a multiplex lexical network [26][27][28][29] , where concepts are represented by nodes and are linked on two network layers by:…”

“…Although it is appealing to investigate robustness of single aspects of the mental lexicon to degradation, this approach would neglect the extensive evidence indicating how failed lexical retrieval crucially depends on the interplay between semantic and phonological conceptual similarities 3,6,8,26,27 . In order to embrace multiple aspects of the mental lexicon in a single network representation, this work suggests studying lexicon robustness through a minimal representation of semantic and phonological associations for 16000 English words as a multiplex lexa) Electronic mail: massimo.stella@inbox.com ical network 26,[28][29][30][31] . Such representation is used as a quantitative framework for exploring the large-scale robustness of the mental lexicon under progressive word failure through the formalism of network-based attacks 20,22,23,25 .…”

Modeling bilingual word associations as connected monolingual networks

“…Despite their 'localist' approach in which a word is simply represented by a node (rather than using distributed representations), such models are a useful tool in the study of lexical access and acquisition. In particular, they have successfully replicated patterns of human verbal behavior in free word association (Enguix et al, 2014;Gruenenfelder et al, 2015), semantic fluency tasks (Abbott et al, 2015;Nematzadeh et al, 2016), lexical growth/acquisition (Stella et al, 2017;Bilson et al, 2015), assessment of semantic similarity (Jackson and Bolger, 2014;De Deyne et al, 2016), etc.…”