Este trabajo apunta a contribuir a la comprensión del sentido como inherentemente impreciso e inestable excepto cuando se lo estudia como el proceso activo en el cual el discurso establece relaciones léxicas. La imprecisión y la inestabilidad semántica en cuanto a referencia fue investigada por Eleanor Rosch (1973, 1978). Rosch descubrió que las categorías conceptuales no poseen límites definidos. Esta misma falta de estabilidad en las categorías es lo que adjudica Hanks (2008) a las palabras. Hanks sostiene que el sentido de las palabras proviene de componentes semánticos que se activan en contexto, donde el contexto sirve para desambiguar la composición semántica de una palabra en uso. Aunque algunos componentes se activan mediante disparadores contextuales, definir qué componentes se encuentran activos rara vez resulta sencillo. Esa es precisamente la tarea que las relaciones léxicas llevan a cabo en contexto. Las relaciones léxicas han sido estudiadas desde diversas perspectivas: desde Lyons hasta Cruse (Geeraerts, 2010, p. 82) y desde la lingüística generativa (Radford et al., 2009, p. 170) hasta la lingüística funcional (Halliday, 2014, p. 644). Todos estos lingüistas analizan los cuatro tipos clásicos de relaciones léxicas: hiperónimo-hipónimo, holónimo-merónimo, sinonimia y antonimia. El análisis tradicional de estas relaciones léxicas ha sido fuera del contexto discursivo. Una excepción es el libro Opposition in Discourse de Jeffries (2010). Jeffries sostiene que pares de palabras pueden crear relaciones de oposición en virtud de su contexto lingüístico. La postura del presente trabajo es que todas las relaciones léxicas se construyen en los textos, no solo la antonimia. Esta visión permite apreciar con mayor especificidad tanto la flexibilidad como la indefinición del sentido, pero a la vez contribuye a la comprensión del proceso mediante el cual se desambigua el sentido en tanto la instauración de relaciones léxicas discursivas.

This paper contributes to the understanding of meaning as inherently vague and indeterminate except when seen as the active process in which discourse sets up lexical sense relations. Vagueness and indeterminacy in reference was studied by Eleanor Rosch (1973, 1978). She found that conceptual categories do not have sharp boundaries. This same indeterminacy in categories has been ascribed to words by Hanks (2008). He contends that word meaning is made up of components which are activated by the context, where the context may help disambiguate the semantic composition of a word in use. Even though some components are activated by contextual triggers, discriminating which components are active is rarely straightforward. This is precisely the kind of task lexical relations effect in context. Lexical relations have been studied from various perspectives: from Lyons to Cruse (Geeraerts, 2010, p. 82) and from generative linguistics (Radford et al., 2009, p. 170) to functional linguistics (Halliday, 2014, p. 644). All discuss the four classic types: superordination-hyponymy, holonymy-meronymy, synonymy and antonymy. Traditionally, these relationships have been analysed in isolation, i.e. not as part of discourse. One exception is Jeffries’s Opposition in Discourse (2010). Jeffries argues pairs of words may enter into oppositional relationship by virtue of their textual surroundings. The contention here is that all sense relations are constructed in texts, not just antonymy. This view helps appreciate more accurately both the flexibility and the fuzziness of meaning, but it also contributes to understanding the processes of meaning disambiguation as instantiated by lexical relations in discourse.

This paper contributes to the understanding of meaning creation seen as the active process by which discourse sets up lexical sense relations in context. Sense relations at the lexical level have been analysed from various perspectives: from Lyons to Cruse (Geeraerts, 2010, p. 82) and from generative linguistics (Radford et al., 2009, p. 170) to functional linguistics (Halliday, 2014, p. 644). All studies include the four classic types: superordination-hyponymy, holonymy-meronymy, synonymy and antonymy. Traditionally, these relationships have been studied in isolation, i.e. not as part of discourse. Exceptions are some studies on antonymy, notably Jeffries’s Opposition in Discourse (2010). Jeffries argues pairs of words may enter into oppositional relationship by virtue of their textual surroundings, and she calls this relationship “constructed opposition” (p. 1). The contention here is that all sense relations are constructed in texts, not just antonymy, and that this view helps appreciate more accurately both the flexibility and the fuzziness of meaning.

In order to understand lexical relations more thoroughly, it is instructive to draw on categorisation studies. To find out if a word, or rather a concept, is a superordinate, a hyponym, a meronym, a synonym or an antonym, it is necessary to make a categorisation decision, or basically to decide what something is, such as “love” or “fruit”. Thanks to the studies conducted by Eleanor Rosch (1973, 1978) on conceptual categories, we now understand that categories do not have sharp boundaries, and that this decision might prove tricky. Her discovery entails that categories are organised around prototypes, or “best” or “most typical” examples. This is why, for example, when people think of a penguin they rarely believe it is the most frequent representative of the category of “birds”, but sparrows and robins are. Seen from a different perspective, this means that when language users try to interpret a conceptual category they are capable of handling a range of possibilities that go from very good examples, such as apple or orange as “fruit”, to less good examples, such as tomato or lemon that fit the conceptual category in question more or less accurately. This range naturally contributes to fuzziness as, when interpreting the concept “piece of furniture,” a language user may be considering options such as chair and sofa, but also TV set or lamp. Indeed, Rosch’s studies go to show how vague concepts can be. This view is reinforced by findings of linguists specialized in lexicography. Patrick Hanks (2008), in his revealing article “Do Word Meanings Exist?”, contends words in use have “meaning potentials” (p. 130) and that “text meanings arise from combinations, not from any one word individually”. In this context, it becomes tempting to postulate language is an excessively tricky tool for communication. If the representation of concepts and meaning is so elusive, how do we make sense of language? A partial answer to this question may lie in lexical sense relations. While it seems to be quite evident that language is inherently vague and fuzzy, this indeterminacy may be countered by something that functions as part of a veritable system of checks and balances, otherwise known as linguistic context.

As stated in the introductory paragraph, sense relations have been studied mostly in isolation, but they may also be successfully studied in linguistic context. To do this, it is necessary to describe each category in greater detail and focus on the basic features of each sense relation to draw certain conclusions.

The basic feature of the superordinate-hyponym relationship is that the superordinate has to be an entity of a higher order that includes any number of members that can be referred to by the more general superordinate, for example, “tree”. The set of hyponyms are referred to as co-hyponyms and are consistently perceived as being an open-ended category. On the other hand, co-hyponyms, such as apple tree and pear tree can be conceived as examples that realise the superordinate “tree” more specifically. In most cases, then, an entity will be identified as a co-hyponym if it shares some basic features with other co-hyponyms so that it may be subsumed under the same superordinate. This discussion seems quite close to the one about conceptual categories and fuzzy boundaries. More to the point, is there a definite number of features which are necessary for a conceptual category to be included in a larger one? According to Rosch’s studies, there is no single answer. Rather, there will be a range of options, with some fitting the conceptual category more closely and others less closely. It is precisely this flexibility in considering “contenders” for a certain category that can be seen in texts quite frequently. Consider the following examples taken from the second paragraph of this paper: “Thanks to the studies conducted by Eleanor Rosch (…) on conceptual categories, we now understand that categories do not have sharp boundaries (…). Her discovery entails that categories are organised around prototypes, or ‘best’ or ‘most typical’ examples.” (Italics added.)

The word “discovery” in the first example is used to recover the previous sentence in its entirety. It provides a sort of label for the studies and the results, and effectively creates a relationship of inclusion, or superordination, between the referent and “discovery”. The interesting point here is not just that one of the lexical relationships has been successfully set up, but also that the choice of superordinate carries added meaning, since “discovery” conveys positive evaluation. More factual terms would have been finding, investigation or research. In referring to Rosch’s studies as a hyponym to the superordinate “discovery”, the text is equating her discovery with other discoveries the reader may have in mind, and resort to, in order to make sense of why Rosch’s studies merit such a distinction. Naturally, the reader may find fault with the choice of term and decide Rosch’s findings should be taken with a pinch of salt and not given such an important status, but in doing so the reader will also have linked the two portions of text as superordinate and hyponym. Here’s another example taken from the same paragraph: “In this context, it becomes tempting to postulate language is an excessively tricky tool for communication.” (Italics added.)

In this case, the superordinate-hyponym relationship is the same, with the hyponym “language” being subsumed under the superordinate phrase “an excessively tricky tool for communication”. In this example, there seems to be a more complex cognitive leap between the two. Given the wider reference of the superordinate phrase, the language user is directed to interpret “language” in a new light, as a more “normal” superordinate would have been “means of communication” rather than “tool”. The choice of a novel superordinate forces the language user to treat “tool” figuratively and map certain features of the term that may be applied to “language”, as the concept of “tool” here is clearly not that of a machine operated by hand but rather something with very specific design features meant to fulfil a definite purpose.

It should be noted that in both cases, that of “discovery” and “an excessively tricky tool for communication,” the type of co-hyponyms that may be called up are quite varied. But if we view them as the instantiation of a superordinate-hyponym relationship, we are induced to find the cognitive relationship between the two in order to make sense of the meaning. Despite the complexity or the novelty, comprehension does take place thanks to the superordinate-hyponym link that was set up.

This lexical relationship, though similar to that of superordinate-hyponym in that it entails a hierarchical imbalance, is somewhat different. This is because the holonym includes a limited number of co-meronyms which together make up the whole. Such is the relationship that hammer, pedal, wheels, keys, pins, strings have with “piano”. In this sense, the parts of the whole can be quite different from one another, as opposed to co-hyponyms which share several features. Because meronyms are parts of a whole, they are frequently associated with having a specific role or function as regards the holonym. Moreover, since the class is initially closed, it is understood that adding a new “part” may involve possible minor changes to the whole, without essentially changing its entity, such as the similar but different co-meronyms for “telephone” and “cordless telephone”. This peculiar link can be seen in the following example: “While it seems to be quite evident that language is inherently vague and fuzzy, this indeterminacy may be countered by something that functions as part of a veritable system of checks and balances, otherwise known as linguistic context.” (Italics added.)

Average language users are aware of what the system of checks and balances is in very general terms, and they understand it is a system by which the three branches of government stop one another from exerting excessive power. This system of three is part of the constitution in many democratic countries, and it involves a limited number of actions that each branch has at its disposal to resort to in case the need arises. This knowledge, though vague, allows language users to understand “linguistic context” is not one of those checks and balances as the text is not about government. However, the relationship between “linguistic context” as a likely part of the holonym “system of checks and balances” has been set up, and it is possible to make sense of it by understanding that linguistic context is capable of exerting limits on the interpretation of meaning.The fact that the text is silent as regards which could be the other two systems that exert checks and balances apart from linguistic context is irrelevant. The meaning that has been activated through the holonym-meronym relationship is clearly restricted to the idea that indeterminacy in language is difficult to deal with, but that at least there are a few systems, and one of them is “linguistic context”, that can come to the aid of the unwary language user.

Synonymy in texts is usually restricted to a stylistic feature, and as such it is frequently known as elegant variation. This phrase points to a change of words intended to make a text more attractive, and it is therefore based on the concept that repetition is unattractive and thus should be avoided. Irrespective of the interest in tastefulness, the cognitive link set up between synonyms is that they are to be perceived as being interchangeable, and very much in the same way superordinates and holonyms do, synonyms may entail another cognitive leap. Here’s another example taken from this paper:

This is why, for example, when people think of a penguin they rarely believe it is the most frequent representative of the category of birds but sparrows and robins are. Seen from a different perspective, this means that when language users try to interpret a conceptual category they are capable of handling a range of possibilities that go from very good examples, such as apple or orange as ‘fruit’, to less good examples, such as tomato or lemon that fit the conceptual category in question more or less accurately. (Underlining added.)

The reader of this text is led to infer that “people” and “language users” have the same referent. How is this inference possible when, in fact, the two expressions are not precisely equivalent? It is the text that has set up a relationship of equality because for the purposes of understanding the message, the concept “people” is only considered in terms of their ability to communicate through language, and all other abilities and qualities and properties that define people as people are irrelevant in the context.

This lexical relationship is one of opposition, which means it can be explained negatively. Antonymy is basically the opposite of synonymy, so if synonymy entails a relationship of sameness, antonymy is based on difference. Sameness involves a perfect match, and any variation, whether small or large, entails difference. Antonymy thus covers a much wider spectrum. Significantly, the literature gives greater attention to this lexical relation. In fact, so varied is the topic of antonymy that studies do not agree on a single taxonomy (see, for example, Cruse, 2000, p. 197; Geeraerts, 2010, p. 85 and Jeffries, 2010, p. 19). Despite the complexity of antonymy, some cases are rather straightforward and involve an either/or relationship, as can be seen in the following example: “As stated in the introductory paragraph, sense relations have been studied mostly in isolation, but they may also be successfully studied in linguistic context.” (Italics added.)

There are possibly several of ways of studying something, but as regards lexical sense relations the two settings, in isolation or in context, are contrasted as seemingly the two only options, particularly as yielding different, contrasting, results. An important factor as regards meaning that should not be neglected when analysing antonymy is that while it may be vague as regards what exactly the difference is, it is very clear as to what it is not. The basic meaning to be made of opposition is “not that”, but the main job is to work out what it specifically entails.

The previous and rather brief analysis has attempted to draw attention to the significance of studying sense relations in current language use in order to view context in a new light as crucial to meaning. For long, applied linguists have maintained word sense is an illusion (Kilgarriff, 2003; Hanks, 2008). In the field of cognitive linguistics, Jean Aitchison (2012) says the vast majority of psychologists have abandoned the idea that word meaning involves any straightforward imagery and that underspecification is a basic quality of language. On the other hand, language users have developed the ability to extend the application of words so that they can map stored meanings on novel situations and infer senses as they interact. But out of the richness and variety of options available for interpretation, how is it possible to home in on the right interpretation? It seems exploring lexical relations in text may be truly worth our while.

Geeraerts, D. (2010). Theories of Lexical Semantics. Oxford: Oxford University Press.

Halliday, M. A. K. (2014). Halliday’s Introduction to Functional Grammar (4th ed.). London: Routledge.

Hanks, P. (2008). Do Word Meanings Exist? In T. Fontenelle (Ed.), Practical Lexicography. A Reader (pp. 125-134). Oxford: Oxford University Press.

Kilgarriff, A. (1997). I Don’t Believe in Word Senses. Computers and the Humanities, 31(2), 91−113. Retrieved from http://www.jstor.org/stable/30204773

Radford, A. (1990). Transformational Grammar. Cambridge: Cambridge University Press.

Radford, A. et al. (2009). Linguistics: an introduction. Cambridge New York: Cambridge University Press

[1]* Traductora Literaria y Científico-Técnica, especialista en traducción técnica y científica, y Magíster en Lingüística Aplicada. Correo electrónico: avaleriacarbone@gmail.com

Ideas, IV, 4 (2018), pp. 1-9