LINGUIST List 31.1528

Tue May 05 2020

Review: Cognitive Science; Linguistic Theories; Psycholinguistics: Winter (2019)

Editor for this issue: Jeremy Coburn <>

Date: 15-Oct-2019
From: Joyce Cheung <>
Subject: Sensory Linguistics
E-mail this message to a friend

Discuss this message

Book announced at

AUTHOR: Bodo Winter
TITLE: Sensory Linguistics
SUBTITLE: Language, perception and metaphor
SERIES TITLE: Converging Evidence in Language and Communication Research 20
PUBLISHER: John Benjamins
YEAR: 2019

REVIEWER: Joyce Oiwun Cheung, Lingnan University


Bodo Winter’s Sensory Linguistics: Language, Perception and Metaphor, which was based on his PhD thesis, attempts to bridge the research gap between human perception and the language used to convey these perceptual experiences. Though substantial studies have been done in neuroscience in terms of calibrating human’s perceptual-cognitive correlation, the author sets out to propose and test a new hypothesis, by scrutinising patterns and frequency of sensory words and the occurrences of such words in corpora of naturally occurring data; he claims that ‘language mirrors perception’. Word lists and participant ratings from previous studies (e.g. Lynott and Connell, 2009; 2013) are adopted to run through a variety of statistical analyses to justify the hypothesis, as well as to support or debunk some myths in the five senses folk model (i.e. sight, sound, smell, taste and touch), the hierarchy of these senses (e.g. Ullmann 1945; Williams, 1976), ineffability (i.e. the difficulty or impossibility of expressing some perceptual experiences), and metaphors (e.g. whether some sensory words are actually sources ‘loaned’ to describe experience in another realm of sense). This book would be particularly useful to students and researchers who are interested in psycholinguistics, neurolinguistics, corpus linguistics, lexicosemantics and any interdisciplinary studies under the umbrella of cognitive studies. Starting with an introduction to the theories concerned and then following up with different case studies, the book is designed as a textbook to inform readers of existing literature and theoretical grounds before playing with statistics. It would be a nice read for audiences who have basic knowledge of statistical methods or corpus linguistics, as the book does not involve extensive calculation but only occasional comparative graph reading.

The first chapter gives an overall introduction to the whole book, defining sensory linguistics to be ‘studies of the way language is related to senses’ and forecasting what is to be expected in the rest of the book. The rest of the monograph is divided into two major parts – theory and case studies with each chapter detailing the key issues and providing comparative studies with extant literature, data sets or examples, arguments and limitations, as well as conclusions.

Part I begins with the second chapter which introduces the five senses folk model, by discussing the potential danger of neglecting finer-grained distinctions and defending the adoption of the five senses model throughout the rest of the book. Chapter Three deals with a box of semiotic tools in which words (or a basic sign unit conveying a meaning) are invented, including ‘icons’ that resemble part of the perceptual experience (e.g. ‘kiwi’ which mimics the sound emitted by the kiwi birds), ‘indices’ which often refer to the sources of meanings (e.g. the colour term ‘orange’ is taken from its origin, the ‘orange’ fruit), ‘symbols’ which are arbitrarily assigned to represent a meaning, ‘technical language’ which stems from the scientific system (e.g. hours and minutes), and ‘metaphors’ which borrow signs from the source to the target domain. Chapter Four provides possible accounts of ineffability. For example: 1) some senses are more perceptual and effable in the first place, 2) some ‘common sensibles’ involve several senses, 3) subjective experiences are idiosyncratic, or 4) simply that certain perceptual experiences outnumber linguistic labels allowed in human languages. Chapter Five puts forward the central idea – Embodied Lexicon Hypothesis - which proposes that perception influences language, in terms of ‘asymmetries’ (i.e. one sense dominates the perception and/or language), ‘associations’ (i.e. one sense is particularly close to another, and such proximity is mirrored in language), and ‘activations’ (i.e. the sensory-motor process is also reflected in language use). Chapter Six first brings up the concepts of ‘synesthesia’ (i.e. one perceptual sense involuntarily triggers/associates another sense at the same time) and ‘transfers in synesthetic metaphors’ (i.e. borrowing expression in one sense to represent perception in another), and then differentiates the two concepts i.e. that ‘synesthetic metaphors’ are not synesthetic at all. Additionally, Chapter Seven carries on to suggest that ‘synesthetic metaphors’ are not metaphorical either, by first recapping the concepts of Conceptual Metaphor Theory and then asserting that ‘synesthetic metaphors’, despite their title, are often either ‘supramodal’ (i.e. some neural areas highly associated with each other) or multisensory in nature, devoid of intersensory transfer. Following the traditional senses folk model, Chapter Eight continues by introducing the hierarchies proposed by scholars like Ullmann (1945) and Williams (1976) to organise the five or more senses based on intersensory asymmetries; Chapter Nine attempts to explain the hierarchies from a global (i.e. an overarching principle dictating the whole hierarchy) and local (i.e. accounting for asymmetry between different pairs of senses) perspective. After going through possible implications, the author determines that sensory vocabulary, emotion and iconicity are the key factors pertaining to explaining hierarchies.

Part II Case Studies are structured as follows: Chapter Ten lays a foundation for the methodologies used in the existing literature, to try out different approaches, to account for multiple facts, and to make the research design reproducible for other researchers. This chapter also emphasises the research-method ‘norm’ which invites participants to rate sensory words prior to any interpretation. This method is further explained in Chapter Eleven, where the ratings of the five senses of each words are weighted. It guides readers in calculating the exclusivity of a word belonging to a particular sense, and assessing the cosine similarity of a sensory word’s ratings in the two datasets. Chapter Twelve uses this normalising technique to gauge the distribution of words in each sense, as well as how much a sense’s exclusivity (i.e. the degree to which a word solely used for that sense) deviated from the average baseline. Chapter Thirteen compares these two components each time to arrive at intersensory correlation and experiments with cluster analysis to invent twelve new categories. Chapter Fourteen further investigates the correlation between adjective and noun pairs, by finding the overall cosine similarity and cross correlating different pairs of sense adjectives and sense nouns (e.g. sight adjective + touch noun). Chapter Fifteen adopts three models to compare the sensory word frequency, varieties of dictionary meanings and iconicity. Chapter Sixteen evaluates the absolute valence (i.e. being either very positive or negative) of sensory words and their contextual words. Penultimately, Chapter Seventeen cross tabulates sensory aspects with each other to update the sensory hierarchy, in which the author believes no single paradigm shall rule over (and account for) all sensory hierarchy, but the asymmetries depend on the lexicons, emotional valence and iconicity. The last chapter serves as a conclusion.


The book refers to extensive literature in sensory language, neuropsychology and metaphor studies; for example, in Chapter Five, reviewing extensive perceptual simulation studies, the author aims to relate the comprehensive findings to linguistic phenomena, and in Chapter Nine the author makes strong reference to neuropsychological studies and notes that it would be hard and ineffective to attempt to rank senses in a clear cut way. Adapted from previous literature, the traditional five senses folk model and monolithic hierarchy were initially referred to (but are later questioned as oversimplified) as a bridging gap between previous studies and Winter’s new approach. However, since existing research has long adopted the five senses folk model, following this tradition renders research findings comparable. The overall structure is well designed to brief readers about potential domains to consider in sensory linguistics, to propose the author’s hypothesis and his disagreement on a universal sensory hierarchy, and finally to support his claims with vast statistical analyses on the perceptual strength ratings incorporated in corpus data. Every chapter starts with a succinct introduction of concepts and literature, and concludes with a concise conclusion. For example, at the end of Chapter Thirteen, the author gives a clear summary about the necessity to draw more than five distinctions when viewing the microstructure, and fewer than five when viewing the macrostructure. In addition, the author provides detailed explanations alongside neuroscientific jargon, such as ‘amygdala’. The author also thoughtfully remarks on the etymology of certain technical terms, e.g. ‘common sensibles’ (Chapter 4.3.2). By tabulating the differences between canonical synaesthesia and synesthetic metaphors in a table format, the author also helps readers grasp the key differences collectively.

Contrary to the title, the book is not ‘metaphorical,’ as the book centrally discusses relations between senses as revealed in language. Although the Conceptual Metaphor Theory is well introduced in the book, the author himself makes it clear that synesthetic metaphors are not metaphorical. The title itself might disappoint readers who are only interested in primary, typical and embodied metaphors. Minor presentation issues are noticeable: in Chapter Three, only the phonological iconicity is exemplified in section 3.2, 'depicting sensory perceptions with icons’. As a result, readers might be misled to think that iconicity is exclusive to sound. In Chapter Five, the addition of ‘emotion meaning’ in the final section should be a key factor in sensory word analysis; therefore, I would suggest visualising the framework with a triangulation diagram in which three points are captioned with perceptual processing, linguistic system and emotional meaning respectively. Chapter Twelve and Thirteen are statistically dense. Chapter Twelve is a bit difficult to follow though the guidelines and explanations are useful. However, Chapter Thirteen jumps straight in the tables and charts with little reminder of what the data and components represent. For example, in Table 7, readers are clueless about what C1 & C2 represent. Moreover, data such as +0.3 in Table 7 is not properly introduced. Readers would be puzzled as to how the figures were computed. Likewise, Figure 7 should be explained in greater detail. Additionally, in Chapter Fifteen, the difference between categorical and continuous models is not conspicuous. It would be logical to refer each model to their respective chapter origin. The book should attend to the choice of words as well. In Chapter Three, when stating the fact that ‘iconicity’ varies across modality of expression, ‘speaking versus signing’ would be better phrased as ‘verbal versus signing’ since readers, without having read the remaining chapters, would take ‘speaking’ as ‘speech’ rather than an umbrella term for ‘spoken and written words’. On p.115 while discussing the reasons for sense asymmetry, in the line ‘using metaphors in poetry’, readers would be disconcerted by the sudden disjointed mention of metaphors (not even synesthetic metaphor). In Chapter Fourteen, although only adjective-noun pairs are studied, the author makes a broad claim on p. 179 that ‘sensory words tend to combine with words that have similar perceptual characteristics’ which may not be the case in noun-verb and other types of pairs. The research design is neither unproblematic. For example, in Chapter Two, perceptual ratings are collected from British native speakers, which unfortunately neglects the variations of world Englishes today. As the author makes it clear that the model is culturally specific, the model could be unusable for non-British English or even other languages. This is exemplified in Chapter Three when the author notes that different languages have different ideas of what aspect should be selected to be represented iconically.

Furthermore, there are still many ambiguities left unaddressed by the end of the book: while talking about ‘arbitrary signs’ on p.17, the author mentions ‘arbitrary symbols’ on p.36. It is confusing whether the author referred to signs or symbols, primarily because de Saussure (1959) was cited as making explicit distinctions between the two. The author recognises these two terms as distinct from each other and should not have used them interchangeably. In Chapter Eleven, since the author offers verbal instruction instead of clear-cut equations for the exclusivity on p.143, I attempted to calculate the exclusivity but the number did not match (e.g. I got 92.5% for 95.1%, and 11.5% instead of 11.6%). If round up or variance is employed, readers should be informed. Likewise, simple equations such as ratio = total source count / total target count (in Chapter Seventeen) should be provided. Also, it is unclear whether participants in Lynott and Connell’s (2009, 2013) experiments were given options to pass on words about which they were uncertain. As pointed out by the author himself on p.157, participants might simply be guessing if they did not know the words well and thus this is not an accurate reflection of their choices. Besides, the degree of certainty may be a factor influencing the overall perceptual strength of each sense of each word. Were certainties calibrated and weighted beforehand? Regarding the categorical modality classification on p.155, what is the threshold of categorising each word into a sense, given that many words are not exclusive, especially if the maximal perceptual strength rating on one sense is almost negligibly above the others? In Chapter Fourteen, why was the Corpus of Contemporary American English (COCA) used instead of the British National Corpus (BNC), given that the raters in Lynott and Connell (2009; 2013) were all British? How are dictionary meanings counted in Chapter Fifteen? Do the 12 clusters work well for the 13,915 English lemmas in Chapter Sixteen as well, given that the clusters were generated from only 423 adjectives? Is new clustering necessary? Also,given that five content words preceding and following the headword would be considered in the average valence measure, were words positioned closer to the headword weighted heavier? Finally, in Chapter Seventeen, the Lynott and Connell (2013) noun modality rating was criticised for bias; if the noun list was so biased, why didn’t the author use Strik Lievers (2015) data/findings in all previous chapters?

The book has successfully achieved some ambitious goals: firstly, the five senses folk model serves as a window for researchers to probe into studying the multisensory nature of sensory words; secondly, the case studies such as the intersensory correlation and semantic preference support the hypothesis that perceptual associations and asymmetries relate to the corresponding linguistic association and asymmetries, (though I cannot see the ‘result in’ causal relation promised); and thirdly, the local accounts of different asymmetries under different circumstances have provided challenges to the one hierarchy since a single principle can no longer account for the subtle differences observed between senses.

Lastly, the book demonstrated appreciable contributions to the field. Reproducibility is highlighted as the author recognises the significance to share raw data for open access by which different researchers can make use of and support/challenge the claims and design. As sensory language research often relies on linguistic intuition, the author brought in merits from corpus linguistics as a counterintuitive approach. It is also an excellent move to combine the strength of human ratings with corpus data, contextual corpus and non-contextual corpus. As corpus linguistics often yields scattered patterns of language use which linguists find hard to generalise, and human rating alone is prone to be subjective, using human ratings as an anchor to narrow down and study how different aspects of (multi)sensory words perform in corpora of natural linguistic data would be hugely beneficial. Several future directions are suggested by the author, such as looking at ratings in other word classes (e.g. verbs), sensory language of deaf and blind people, or looking at which explanatory factor matters the most for the sense hierarchy. These suggestions, as a result of a well-informed, and mostly well-argued paper, would be worth pursuing and make it possible for various students in order to build on this research.


Lynott, Demot & Louise Connell. 2009. Modality exclusivity norms for 423 object properties. Behavior Research Methods 41. 558-564.

Lynott, Demot & Louise Connell. 2013. Modality exclusivity norms for 400 nouns: The relationship between perceptual experiences and surface word form. Behavior Research Methods 45. 516-526.

de Saussure, Ferdinand. 1959. Course in general linguistics. New York, NY: The Philosophy Library.

Strik Lievers, Francesca. 2015. Synaesthesia: A corpus-based study of cross-modal directionality. Functions of Language 22(1). 69-95.

Ullmann, Stephen. 1959. The principles of semantics. Glasgow: Jackson, Son & Co.

Williams, Joseph M. 1976. Synesthetic adjectives: A possible law of semantic change. Language 52. 461-478.


I am a research fellow at the School of Graduate Studies at Lingnan University. I hold an MA in English Language Studies and BSSc in Communication. My research vision sees the need to contrastively study what cross-modal and cross-linguistic big data can inform us. By comparing the patterns driven from corpora of various modes (e.g. text, visuals, gestures, sound) or testing hypotheses in corpora of different languages, we can exploit the systematicity of scrutinising the naturally-occurring data to make a better sense of the world. In other words, I am mostly interested in corpus linguistics, multimodality and discourse analysis. My publications can be found on Social Semiotics and ReCALL.

Page Updated: 05-May-2020