Periodic Reporting for period 1 - LexPex (The Lexicalisation of Perceptual Experience)
Période du rapport: 2019-10-28 au 2022-10-27
Western thought has long held that the perceptual senses fall along a hierarchy of relative primacy, with vision at the top, followed by audition, touch, taste and smell. Intriguingly, the hierarchy has been proposed to also impose itself on language. In a seminal study, the linguist Åke Viberg compared basic perception verbs (e.g. ‘see’, ‘hear’, ‘smell’) across a sample of the world’s languages and found that their semantic and structural properties reflected the sense hierarchy in various ways. This work led to the idea that the hierarchy is universal to human languages, grounded in our shared biology. From a different disciplinary perspective, anthropologists of the senses have questioned the possibility of universals, arguing that the relative dominance of the senses is cross-culturally variable.
Against this backdrop, LexPex had three main objectives. The first (“Typology”) was to quantitatively test whether perception verb lexicons pattern hierarchically and to assess the extent of cross-linguistic regularity in this lexical domain. The second objective (“Phylogeny”) was to extend the search for regularities across time, by examining whether perception verbs are replaced at different rates in accordance with the sensory hierarchy. The third objective (“Cognition”) was to probe the potential link between recurrent typological patterns in perception verb lexicons and cognitive biases of language learners. The results of the project support the conclusion that while vision stands apart from the other senses with respect to the typological patterning and acquisition of perception verbs, the notion of a biologically-grounded fixed hierarchy of senses is not well-founded.
Against this backdrop, LexPex had three main objectives. The first (“Typology”) was to quantitatively test whether perception verb lexicons pattern hierarchically and to assess the extent of cross-linguistic regularity in this lexical domain. The second objective (“Phylogeny”) was to extend the search for regularities across time, by examining whether perception verbs are replaced at different rates in accordance with the sensory hierarchy. The third objective (“Cognition”) was to probe the potential link between recurrent typological patterns in perception verb lexicons and cognitive biases of language learners. The results of the project support the conclusion that while vision stands apart from the other senses with respect to the typological patterning and acquisition of perception verbs, the notion of a biologically-grounded fixed hierarchy of senses is not well-founded.
The project was organised around three scientific work packages (WPs). WP “Typology” involved the manual compilation of a genealogically and geographically stratified sample of basic perception verb lexicons in 100 spoken languages. This database served as the empirical basis for two large-scale studies. The first analysed the mappings between perception verb forms and sense meanings. This established that there is a strong cross-linguistic bias for the visual modality to be expressed with a dedicated perception verb, but that the non-visual modalities do not pattern hierarchically. The study also revealed that certain cross-modal combinations (e.g. hear-feel) are commonly expressed in polysemous verbs across languages, while others, most notably smell-taste, are rare. The bias against smell-taste polysemy goes against the commonly held idea that meanings that are conceptually close are more likely to be co-expressed in words, and may instead be driven by general communicative pressures (ambiguity avoidance). This research was published in Language, the flagship journal of linguistics (Norcliffe & Majid 2024a), and won the award for Best Paper in Language for that year. A second study extended the focus to patterns of word formation in perception verb lexicons. Perception verbs sometimes extend their meaning to a different sensory meaning through processes of word formation. This study found that across languages, semantic extensions of this type tended to proceed from higher to lower sense modalities. While this is consistent with Viberg’s original generalisation, the study showed that human biology does not need to be directly invoked to explain it. Previous research has shown that the directionality of semantic extensions in word formation is predicted by word frequency: words that are sources of extensions tend to be more frequent than the derivations. Lexical frequency therefore is likely the direct cause of the directionality finding, as it has been shown that vision and audition verbs tend to be more frequent than the lower senses. This does not rule out the possibility that the frequency patterns themselves are shaped by the relative dominance of the senses in human perception, though research suggests that the frequency patterns are more likely the outcome of convergent factors. This study was published in Linguistic Typology (Norcliffe and Majid 2024b).
WP “Cognition” examined the acquisition trajectories of perception verbs among English speaking children, using data from parent questionnaires and corpora of children’s spontaneous speech. This revealed that visual perception verbs are acquired earliest, mirroring the dominance of vision in lexical systems. Intriguingly, touch rather than audition verbs came second, in both age of acquisition and production frequency. This finding was replicated in acquisition data from a further 11 languages. The robust touch-before-hearing result possibly reflects the early relevance of the tactile modality for infants, for whom much of the experienced world is explored through the hands. It may also be driven by the speech of caregivers about behaviour regulation, especially prohibitions against touching, raising fascinating questions for future study about how individual sensory experience interacts with linguistic input to scaffold language learning. This study was published in Cognitive Science (San Roque, Norcliffe & Majid 2024).
Data collection and analysis for WP “Phylogeny” are ongoing. This WP is exploring whether perception verbs denoting the different senses evolve at different rates in accordance with the sense hierarchy, in the context of the Mayan language family.
The results of the project were disseminated at conferences, workshops and seminars straddling linguistics and cognitive science. A theme session, “Words and Meanings", was also organised at the International Cognitive Linguistics Conference. Wider publicity within the linguistics community was achieved following the publication of the Language paper. This was selected as one of the four ‘most interesting’ papers published in the journal that year and resulted in an invitation to speak at a ‘meet-the-author’ webinar organised by the Linguistics Society of America. Many of project results are summarised as part of a larger review article on ‘The Lexical Typology of Sensory Perception’ (Majid & Norcliffe 2026).
WP “Cognition” examined the acquisition trajectories of perception verbs among English speaking children, using data from parent questionnaires and corpora of children’s spontaneous speech. This revealed that visual perception verbs are acquired earliest, mirroring the dominance of vision in lexical systems. Intriguingly, touch rather than audition verbs came second, in both age of acquisition and production frequency. This finding was replicated in acquisition data from a further 11 languages. The robust touch-before-hearing result possibly reflects the early relevance of the tactile modality for infants, for whom much of the experienced world is explored through the hands. It may also be driven by the speech of caregivers about behaviour regulation, especially prohibitions against touching, raising fascinating questions for future study about how individual sensory experience interacts with linguistic input to scaffold language learning. This study was published in Cognitive Science (San Roque, Norcliffe & Majid 2024).
Data collection and analysis for WP “Phylogeny” are ongoing. This WP is exploring whether perception verbs denoting the different senses evolve at different rates in accordance with the sense hierarchy, in the context of the Mayan language family.
The results of the project were disseminated at conferences, workshops and seminars straddling linguistics and cognitive science. A theme session, “Words and Meanings", was also organised at the International Cognitive Linguistics Conference. Wider publicity within the linguistics community was achieved following the publication of the Language paper. This was selected as one of the four ‘most interesting’ papers published in the journal that year and resulted in an invitation to speak at a ‘meet-the-author’ webinar organised by the Linguistics Society of America. Many of project results are summarised as part of a larger review article on ‘The Lexical Typology of Sensory Perception’ (Majid & Norcliffe 2026).
Since Viberg’s seminal study, scholars have been intrigued by the possibility that the perceptual lexicon may have a ‘direct line’ to human biology. There has not, until now, been any attempt to test the claim against a balanced sample of the world’s languages. LexPex has moved the field of sensory linguistics forward by providing such a test and demonstrating that although perception verb lexicons show a strong visual bias, they do not exhibit consistent hierarchical patterning. Beyond the hierarchy, LexPex has uncovered new kinds of regularities in perceptual vocabulary, in this way contributing to the the study of lexical typology. The project’s findings on learning led to an important and unexpected discovery—that touch verbs are acquired before hearing verbs by young infants across diverse languages—opening up new and exciting research avenues regarding the interaction between sensory experience and linguistic input in the acquisition of sensory language.