Polysemy in dictionaries

Despite the seeming simplicity, the concept of polysemy is complex and involves a certain number of problems. Many of these problems are primarily to do with how polysemous words are represented in dictionaries. The challenges that dictionary compilers most often encounter with respect to polysemy are of three kinds. The first type involves establishing whether they deal with polysemy or whether it is a case of homonymy. Dictionaries have to decide whether a particular item is to be handled in terms of polysemy or homonymy because a polysemous word will be treated as a single entry, while a homonymous one will have a separate entry for each of the homonyms. This, however, does not necessarily mean that this ‘rule of thumb’ works consistently. The policy that a dictionary adopts depends on different factors, the size of the dictionary being one of them. Smaller-sized dictionaries, for reasons of limited space, tend to bracket homonyms together under the same headword. As a result, there is no way of telling whether one deals with the senses of a polysemous word or distinct homonyms. Even some average-sized dictionaries may lump homonyms together in one entry. LDOCE is a case in point. For example, while the Oxford English Reference Dictionary (OERD) or Random House Webster’s Unabridged Dictionary (RHWUD) treat such lexical items as punch in the meaning of “a hit”, punch as “a metal tool for cutting holes” and punch in the sense of “a type of drink” as homonyms and, therefore, as separate items, LDOCE puts them indiscriminately in one entry, as different senses of the same word. Other similar examples include band in the sense of “a group of musicians” and band as “a flat, narrow piece of material with one end joined to the other to form a circle”, bat in the sense of “a flying mammal” and bat as “a special stick that is used in some sports”, bank as “land along the side of a river” and as “a financial institution”, etc. These differences are clearly a result of certain policies adopted by dictionary-makers. Considering the more-than-average size of OERD and WUD and the much smaller size of LDOCE, such policies are arguably motivated by space constraints.

If lexicographers do establish that they deal with polysemy, another potential problem is the sequencing, or ordering of the senses that the word has. It is obvious that different senses of the same word are not ordered randomly inside a lemma. The sequencing of the senses is by no means arbitrary and there is usually a very distinct regularity to it. The kinds of ordering may vary from dictionary to dictionary, but commonly they include:

(1) Historical or etymological – the older before the newer. Following a historical order, some lexicographers give the oldest recorded senses first, even if these are now obsolete and largely unknown. The historical principle used to be a universal practice until the end of the 19^th century. (2) Frequency – the common before the rare. Most modern dictionaries intended for the learners of the language use a corpus to establish which are the most frequent uses of a word in a large quantity of text, and, consequently, list senses of a word in order of frequency. This has been the primary criterion in most present-day dictionaries. (3) Logico-semantic, the general before the specialized, the concrete before the abstract, and the literal before the figurative.

Apart from the approaches listed above, a compiler may order the senses in a way that makes the defining easier and more concise. Although no objective criterion lies at the heart of this approach, it is probably of help to the reader. For instance, the word season is commonly used in phrases like the football season, the rainy season, the holiday season, the tourist season, a season ticket, out of season, etc. These uses taken together probably outnumber what many people may think of as the basic meaning of season as “time of year; one of the main periods into which a year is divided, one of the four seasons: spring, summer, autumn and winter”. Nevertheless, the dictionary compiler may judge it sensible to open the entry with this particular sense not only because this is what most readers expect, but also because the subsequent definitions of season, like “the usual time for something”, examples of which have been cited above, may seem easier to grasp if they are preceded by the supposedly basic sense (Yallop, 2004: 26).

It sometimes seems to be mere convention to list certain meanings first. For example, definitions of the word have in most dictionaries begin with the sense of “possess” or “own”. Indeed, many people traditionally consider these as the fundamental or ordinary meanings of have. At the same time, extensive corpus data indicate that the uses of have as an auxiliary verb (as in She has always been like that) and in constructions expressing modality, like have / have got to (as in You have to keep an eye on them) are a lot more frequent than the supposedly fundamental uses like They have three pets. These uses of have are a vivid illustration of the fact that the most frequently used sense of a word is not always the one that strikes most people as the core meaning. This traditional approach may well start to change as increasing amounts of objective data from computerized corpora have more influence on the practice of dictionary making.

A third challenge for lexicographers is deciding how many ‘meanings’ or ‘senses’ of a word to recognize. The decision depends on whether the lexicographer is, according to Allen, a ‘lumper’ or a ‘splitter’ (Allen 1999:61). By “lumpers” he means dictionary compilers who tend to lump meanings together and leave the user to extract the nuance of meaning that corresponds to a particular context. On the other hand, “splitters” seek to discriminate between even the minutest senses and list them separately within an entry. The distinction corresponds roughly to that between summarizing and analyzing. Most lexicographers, however, tend to overdifferentiate senses, since they seek to give a better understanding of the multiple meaning of words. If, for example, we compare how complain (v) is treated in different dictionaries, we will find that LDOCE (2003)and СOBUILD (2006) list two senses of the word, while others – RHWUD (1996) and OERD (1996) – register 3. Interestingly, these meanings do not coincide even in the first two that list only two meanings. They both list the main sense, that of “express annoyance, dissatisfaction, resentment etc.” first, but while СOBUILD has “tell of your pain or illness” as its second sense, LDOCE records an entirely different meaning in its second spot: I/he/they can’t complain – “used to say that a situation is satisfactory, even though there may be a few problems”. The meaning listed second in СOBUILD is not treated as an independent sense in LDOCE, but rather as a phrasal verb: complain of (an acute pain, severe headaches, insomnia etc.). The three senses in the other two dictionaries do not seem to coincide either. Apart from the two main senses cited above, OERD lists additionally “make a mournful sound; groan; creak under a strain”, which does not appear anywhere else, while RHWUD has “make a formal accusation” as its third sense, like in They eventually complained to the police. It is apparent that the sense of “make a formal accusation” is a result of the first – and probably the main – sense (“express annoyance, dissatisfaction…”) splitting up into two. To illustrate the variation in presenting polysemy across dictionaries, the table below demonstrates differing listings of meanings for overwhelm, squint and crush in LDOCE, COBUILD, OALD, and RHWUD [Havronska, 2010: ]

The table and the discussion above demonstrate that we cannot assume that the wording of a dictionary definition and, importantly, how dictionaries go about listing different senses of a word are, if not entirely arbitrary, then certainly not an ideal representation of what the word means.