Entity, event, and sensory modalities: An onto-cognitive account of sensory nouns

The paper addresses the long-standing debate on the nature of grammatical categories, particularly the noun-verb dichotomy. It challenges the traditional approach of defining grammatical categories based on a priori knowledge, arguing instead for an approach grounded in embodied cognition and the ontological distinction between beings and their representation. The study explores the fundamental ontological concept that underpins the conceptualization of shared human experiences and the formation of basic grammatical categories like nouns and verbs. It focuses on the interaction between human bodies and the environment through the five sensory modalities (visual, auditory, gustatory, olfactory, and tactile), using sensory lexicon as a lens to understand how humans conceptualize sensory experiences and translate them into language. The Generative Lexicon (GL) theory is adopted as the analytical framework due to its formal qualia structure and its ability to represent experiential knowledge, linking cognitive studies, linguistics, and ontology. The research seeks to substantiate the cognitive motivation for the noun-verb categorization, initially proposed by Strik Lievers and Winter (2018) for English, and to extend it to Mandarin Chinese. The study hypothesizes that the ontological spatio-temporal continuum, rather than the nature of an entity, provides the most robust explanation for the concrete-abstract dichotomy.

The paper reviews existing linguistic theories of parts-of-speech (PoS), focusing on the binary features [±N] and [±V] used to account for the four major PoSs (Noun, Verb, Adjective, and Adverb). It points out the circularity and limitations of these systems due to their reliance on a priori knowledge of grammatical categories. The study contrasts the entity-eventivity dichotomy commonly used to classify nouns and verbs, highlighting its inadequacy in capturing the nuances of noun-verb bifurcation. It then presents Aristotle's distinction between nouns (without temporal reference) and verbs (with temporal reference) as a more promising basis for understanding the noun-verb distinction. This distinction is further supported by the endurant-perdurant bifurcation in formal ontologies like DOLCE and BFO, where entities are categorized as either enduring continuants (time-independent) or perduring occurrents (time-dependent). The paper also discusses previous proposals regarding the cognitive origins of PoSs, such as Givón's (2001) temporal stability, Gentner's (1982) and Ahrens's (1999) mutability, and Strik Lievers and Winter's (2018) eventivity. These are contrasted with the proposed time-dependency approach.

The study employs a corpus-based approach, leveraging the Chinese Web 2011 corpus in the Sketch Engine. It utilizes basic sensory verbs (kàn 'to look,' jiàn 'to see,' tīng 'to listen,' cháng 'to taste,' wén/xiù 'to smell,' mõ/chū 'to touch,' gǎnjué 'to feel') to extract sensory events and identify the nouns that co-occur as objects of these verbs. Data cleaning focuses on selecting nouns that elicit perceptual information, excluding extended or metaphorical meanings. The Generative Lexicon (GL) theory provides the analytical framework, focusing on qualia structures (CONST, FORMAL, TELIC, AGENTIVE) and lexical typing structures (natural types, artefactual types, complex types) to analyze the semantic representations of the selected nouns. Nouns are classified based on their lexical typing structures and analyzed in terms of their endurant (time-independent) or perdurant (time-dependent) properties. The process involves extracting nouns collocated with sensory verbs, annotating them with qualia structures, classifying them into the three lexical types, and analyzing their characteristics based on their qualia values. The analysis examines the distribution of lexical typing structures for each sensory modality, determining the time-dependency of each noun based on its interaction with the predicate verb.

The analysis of sensory nouns in Mandarin Chinese revealed distinct patterns across sensory modalities. Visual nouns (310) showed a relatively even distribution between natural types (perceiving through sight) and artefactual types (understanding/appreciating through sight), with a smaller proportion of complex types. Auditory nouns (385) showed a strong preference for complex types, indicating an integration of sound and information or events, with few natural types. Gustatory nouns (42) were predominantly artefactual types (food items), while olfactory nouns (52) were exclusively natural types (odors). Tactile nouns (58) were mostly natural types (body parts, substances), with a smaller proportion of artefactual types (technology products). The study found that tactile perception predominantly involves endurant objects, whereas auditory perception strongly prefers perdurant objects. Visual perception displayed versatility, encompassing both endurant and perdurant properties. The findings are discussed in relation to the endurant-perdurant dichotomy, suggesting that tactile sensory properties are more likely to be represented as nominal elements, while auditory properties are more likely to be expressed as verbal elements. The results corroborate Strik Lievers and Winter's (2018) findings on English sensory language. The study also notes the parallelism between its findings and those of Sanchez et al. (2020) on late latency in sensory processing, suggesting a possible link between endurant/perdurant properties and conscious/non-conscious perception. The relatively scarce data for gustatory and olfactory senses is noted as a limitation.

The findings support the hypothesis that the endurant-perdurant dichotomy, reflecting the ontological spatio-temporal continuum, is a fundamental basis for the noun-verb bifurcation in language. The results show how the time-dependency of sensory experiences is encoded differently in the nouns of different sensory modalities. The differences in the distribution of endurant and perdurant properties across sensory modalities reflect the cognitive properties of these senses and their relation to embodied cognition. The study addresses the challenge of tense-marked nouns, arguing that tense markers highlight the variations of an endurant entity, rather than contradicting the endurant nature of nouns. The findings suggest a connection between the time-dependency of sensory experiences and the linguistic representation of these experiences as either nominal or verbal elements. The study’s results are compared to those of other studies using different methodologies, such as those involving MEG measurements of brain activity, suggesting possible interpretations in terms of conscious versus non-conscious perception and the access to qualia roles.

This study establishes the cognitive motivation for the noun-verb bifurcation without relying on a priori knowledge of grammatical categories. It proposes that the manipulation of ontological perspectives to achieve time-(in)dependent conceptualization underpins human cognition and the grammatical system. The findings support the hypothesis that the endurant-perdurant dichotomy is fundamental to language and cognition. The study highlights the different encoding of time-independent/endurant and time-dependent/perdurant concepts in sensory nouns across modalities. Future research could explore other languages, investigate the role of other grammatical categories, and further investigate the relationship between sensory modality, time-dependency, and linguistic representation.

The study primarily focuses on Mandarin Chinese, limiting the generalizability of findings to other languages. The data for gustatory and olfactory senses were relatively sparse, potentially affecting the reliability of comparisons. The analysis relies on the Generative Lexicon theory, and alternative theoretical frameworks might yield different results. The interpretation of the findings regarding conscious/non-conscious perception remains speculative and requires further investigation. Finally, the study relies on existing corpus annotations which might have some errors.