Author's Note: The text which follows was written several months ago. This note is an effort to summarize and re-state the basic thesis in a shorter form. I'll re-state the basic theory first, and then illustrate with examples based on the word, poverty.

First, a basic distinction between words (as lexical units) and terms (as words with a defined meaning) is needed. Most words can represent several concepts -- it is only when one of these concepts is relevant that we can think of the word as constituting a term. Because both word and term are polysemes with many meanings, it is easy to become confused when using them. To avoid possible ambiguity, we may substitute technical equivalents, lexeme for word when used to refer to a lexical unit, and designator for term when used to mean any expression that represents a concept. To illustrate, poverty, as a lexeme, has no meaning -- it is just a three-syllable noun with various properties. As a term, however, poverty, may refer to the condition of being poor. However, the same word may also refer to a deficiency of ingredients as in "poor quality soil," or lack of resources, as in "poverty of medical supplies." To use poverty as a designator, one needs to know which of these three concepts is intended.

Second, any term (designator) is subject to possible misunderstanding for three different reasons: level of abstraction, metaphoric usage, and sheltering. The first two problems are well understood and have been much discussed, but the third has been neglected and is the focus of this paper.

The first problem was a focus of Giovanni Sartori's attention in his discussion of the ladder of abstraction. It occurs when, in a spectrum of generic to specific concepts, a term is applied as though it represented a more generic or a more specific concept For example, if fruit is used to mean apple, it lowers the level of abstraction, but if apple is used to mean fruit it raises the level. It typically involves adding to or subtracting from the set of essential features of a concept. This is a serious problem for scholars attempting to generalize concepts from Western contexts to different situations found in the non-Western world. I'll not try to discuss if here.

The second problem has been stressed by David Collier and others who talk about how an idea formed in one context is extended, as a metaphor, to different contexts. This usually involves the use of some but not all of the essential characteristics of a concept while adding new one's that do not fit into a ladder of abstraction. Using mouse to mean a computer gadget is a good example. Again, this is an important problem but there is no need to discuss it here.

A third problem involves sheltering concepts, and it needs our attention because it is almost completely unrecognized and yet extremely important. We may see it as a problem of how to operationalize concepts -- as defined in a dictionary, most terms (designators) identify very fuzzy concepts. To make such a concept precise (to operationalize it), additional criteria are needed, but every such refinement in meaning restricts the applicability of the concept. A shelter term can be used unambiguously, in context, for every concept it shelters, yet out of context, an additional unequivocal term is needed in order to express the intended idea clearly.

For example, poverty may be due to lack of money in a market economy, or lack of resources in a subsistence economy. In marketized contexts, we presuppose poverty to involve lack of cash income, whereas in non-marketized societies, subsistence farmers may live comfortably without cash. We could make the distinction by speaking of cash poverty, or material poverty whenever that distinction is relevant. However, in most contexts, poverty alone will be adequate because everyone involved will know whether it involves cash or non-cash relationships, or a mixture.

In the CROP project, a large number of definitions of poverty found in the literature have been collected. Many of them identify a precise sheltered concept that can be called "poverty" without ambiguity in many contexts, yet its availability permits ambiguity to be overcome in contexts where this term by itself would not be clear. By accepting the dictionary definition of a word like "poverty," one can create a shelter that permits one to use the word unambiguously for a variety of different concepts, always provided the contextual environment is clear. When the context is not clear, however, having a set of sheltered terms permits users to substitute a more precise term to replace the shelter term. The paper which follows takes a different line of reasoning to develop this basic notion.

Warning: some new concepts and terms introduced in this essay are defined in a glossary given at the end of this paper.

Fuzzy concepts have an identifiable core but indefinite boundaries. Most of the concepts identified by ordinary words are like that -- try to distinguish "scholars" from "non-scholars" and you will easily see a large margin of persons who may or may not be classed as scholars -- but there is no doubt about the core of the idea of a learned person. If you look up 'scholar' in your dictionary, you will see that it has several meanings (senses) of which one may read, "a learned or erudite person." In addition, the word has other senses, such as "an elementary school pupil," "the holder of a scholarship," or "a good learner." These are just different concepts -- focus on any one such concept, however, and you will see that it is probably "fuzzy" -- no sharp lines can be drawn between members and non-members of the category it names.

Whenever one considers words like "mother," "poverty," or "income," one sees that their dictionary definitions identify several fuzzy concepts. Any one of them when selected for close analysis may be thought of as a target concept. In this discussion, I will talk only about such target concepts, ignoring all the other concepts the same word can represent. An extended example can be found in a report on the many meanings related to globalization prepared for the World Sociology Congress in July 1998.

An easier example is provided by scholar. Let us assume that, among the various senses (concepts) that can be represented by this word, we target the one identified above as "a learned or erudite person." Note that this clearly establishes a core notion, but at the periphery it provides no criteria to distinguish between learned and un-learned persons, between those who are erudite and those who are not. In order to make a serious study of "scholars" as designated by this definition, we would need some criteria to distinguish scholars from non-scholars. A very simple expedient might involve including everyone listed in a global "who's who of scholars." Alternatively, we might prefer to make a selection based on criteria such as persons who hold graduate degrees from a recognized university. Knowing, however, that some erudite people hold no such degrees, we might offer a test and rank as "scholars" only those who are able to pass it. No such operational definition is right or wrong -- they can all be justified and considerations of cost and convenience may determine which one we will use -- generally, the more precise the definition, the more costly it will be to use.

Let us assume that we have developed several such operational definitions or sets of criteria to decide whom to class as a "scholar" for use in a survey or research project. For convenience, all of them can be subsumed (sheltered) under the word "scholar." I shall speak of this word as a shelter term Virtually any sense of a word, as defined in a dictionary, can be selected as a target and treated as a shelter for a set of operational definitions.

We need to distinguish the notion of a shelter from that of a generic concept. A genus links a set of more specific concepts, each of which differs from from the others: for example, plums, peaches and persimmons are all species of the genus, "fruit."

However, "fruit" can also be treated as a shelter for various definitions of what constitutes a fruit. One of its dictionary senses is that of a "ripened ovary of a seed-bearing plant." We think of fruits as edible and sweet, yet some are inedible and not sweet. Since fruits eaten in some places are considered inedible elsewhere, we can imagine different definitions of fruit which seek to include edibility as a characteristic. Apparently a tomato is a fruit, even though it is normally classed as a vegetable -- perhaps because it is not sweet. Moreover, tomatos used to be called "love apples" and were then thought to be poisonous, and hence inedible. Clearly more or less inclusive definitions of "fruit" can be written, for different purposes -- they are not species of a genus, however, but sheltered concepts created to accommodate overlapping categories of something usually represented well enough by an ordinary language word.

To summarize, although a word may represent a genus that includes several species, it can also be a shelter for different operational concepts designed to permit scientific analysis. Moreover, as I shall show later, a shelter term can also be used unambiguously, in context, to represent any one of the sheltered concepts it covers.

Having discussed the basic concept of sheltering, let us now inquire into the reasons for this phenomenon, and ask what should be done to handle the problems it creates. To place the matter in context, remember that scientists rely on ordinary language structures and words as tools for scholarly analysis. Sometimes, especially in natural sciences, they rely heavily on cryptic language, by which I mean the use of neologisms that laymen are not expected to understand. This enables them to coin terms for all the technical concepts they need and to minimize their reliance on ordinary language words. It is quite easy to distinguish between ordinary and cryptic languages and I shall not say more about this matter.

By contrast, in the social sciences, scholars rely heavily on ordinary language and try to avoid neologisms as much as possible. The underlying premise is that, since they are studying what human beings do, and since humans talk about their activities in well-developed languages, it should be possible to analyze and write about what they do in their own terms, without coining neologisms. Moreover, social scientists seek to communicate with people who are not specialists, both as their subjects of investigation and as audiences for their findings. If they can use ordinary language in their work, they assume they can succeed in this goal. However, since they often need to make distinctions that are not reflected in ordinary language or in the definitions of words found in dictionaries, they resort to "stretching" these words, to stipulating new meanings for them. The result is what may be called delphic language -- see Riggs, "Lexical Lucidity: The Intelligibility of Technical Communications," Theo Bungarten, ed., Wissenschaftssprache und Gesellschaft Hamburg, 1986, pp.113-132, for more details on this matter.

I do not think social scientists can or should copy natural scientists by developing cryptic languages, however. Rather, I think it is possible for them to make more efficient use of ordinary language to achieve their purposes and this paper is intended to explain one method that can help us achieve this goal. We can proceed by talking about stretching as the process by which, new concepts become attached to old words, and then take up sheltering as a conceptual framework designed to help us handle the results in a sensible way. In order to do that, however, I need to make some distinctions not easily made in the available ordinary language -- this involves coining some terms (mostly as phrases using only familiar words) and, for convenient reference, I have appended them to this paper in a short glossary. Suggestions by readers to help revise and clarify these thoughts will be much appreciated.

Stretching Terms and Sheltering Concepts

Precisely because it can be used metaphorically, a word often acquires new meanings that are more or less similar to those it already designates. When there is enough semantic difference, the new sense of the word is easily recognized as a separate concept -- my favorite example is 'mouse' whose new computer gadget meaning is easily distinguished from the idea of a kind of rodent. As this example shows, when a new concept is needed, we may avoid coining a new word for it by borrowing a familiar word and stipulating a new meaning for it. When enough semantic distance separates the new concept from all the old ones, it is easy to remember the difference. Moreover, when the users are quite different, they will easily see that what a word means for them is unambiguosly different from what it usually means. Computer users will have no problem remembering that, for them, a "mouse" is not a rodent. We may refer to the process involved here are "borrowing." A loan-word is a kind of borrowing in which a word used in one language is adopted in another -- "kindergarten" is a familiar example. In the same way, we can borrow a word to mean something quite different -- that is not what I am talking about here.

Instead, I have in mind situations in which a familiar word is used to represent a concept that is very similar to what it originally meant. I shall use stretching to refer to this process, giving credit to Giovanni Sartori who uses this word for this process. It is also involved in Wittgenstein's "family concept." He discusses how the original idea of a "mother" as the biological parent of a child has been extended to include the legal construct of a "step mother," plus other ideas like those of a surrogate mother, a foster mother, etc. Another example I often use involves the Chinese word for "pen" which originally referred only to a brush-tipped writing implement. The introduction of Western pens and pencils led to the stretching of the meaning of the Chinese word to include both brush and brush-less implements.

The question we must ask is how to handle "stretching." Sartori objects to the process, arguing that the original meaning of a word should be preserved and similar yet different concepts should be recognized as such and assigned new terms. He repeats this argument in a current article -- "Understanding Pluralism" -- in the current (Oct. 1997) issue of the Journal of Democracy, taking up a theme he also wrote about many years ago. As a "purist," Sartori argues that "pluralism" should continue to mean only what it originally meant in Western democracies, and its extension to cover multi-ethnic societies as found in authoritarian regimes of the Third World is confusing and should be avoided.

I think that, although this may be desirable, we cannot realistically prevent people from following the familiar metaphorical practice of extending the meaning of a familiar word to new situations where a similar concept is needed. As the Oxford English Dictionary demonstrates repeatedly, the ordinary language meanings of words have evolved continuously and this process will not stop -- indeed, it has probably escalated. We should not only acknowledge this reality but we can exploit it for scholarly purposes. We can do so if we reject the purist approach which, I think, is doomed to failure anyhow. A more realistic and feasible response to conceptual stretching requires us to accept, even welcome, stretched words as shelters which enable us to link and identify a set of closely related concepts. Moreover, we must recognize that a shelter cannot be purified into a clearly defined concept -- rather, it has to be accepted and exploited as a "fuzzy concept."

Wittgenstein saw family concepts as a problem to be accepted and dealt with, but I think we may welcome them as an asset. To do so, however, we need to embrace them as conceptual shelters that usefully cluster a set of closely related concepts even when they do not precisely identify any one of them. The notion of a "shelter" differs from that of a "conceptual family" in that it views the phenomenon as a potential asset rather than a liability. To make use of conceptual shelters, however, we need to do three things: first, we need to accept fuzzy concepts as legitimate; and second, we need to be receptive to neologisms for the precise concepts covered by a shelter that we want to use -- those we have no need for can, of course, just be ignored -- and third, we need to understand the necessary function of contexts. By this I mean that the name of a shelter can always be used as a precise term for any of its sheltered concepts provided its meaning is specified in context. I will deal with these three points in reverse order.

Context. The Chinese pen may be the easiest case to understand. Someone pointing to an object and saying, "Please hand me that pen", uses the word denotatively in a context that clearly shows what s/he has in mind. If one were to describe the way Chinese traditionally used "pens," it would be apparent that the reference was to brush pens, not ball-point pens.

Neologisms. In a non-contextual situation, anyone referring to the old meaning of "pen" in China would need a neologism, such as "brush pen," to make h/er meaning clear. Clearly a purist position would not work if one were to demand that, because "pen" used to refer only to brush-pens, it is redundant to coin a new term for this concept. Yet this is precisely the basis for Sartori's objection to John Rawls' use of "reasonable pluralism" because, he says, "pluralism as such is (must be) 'reasonable.' If not, pluralism is a misnomer." However, since the meanings of "pluralism" have been stretched to unreasonable situations, we can understand why Rawls decided to coin an apparently redundant phrase to specify his intended concept even though it was the original meaning of the word. To designate any well defined sheltered concept, we simply need another term -- it could be a newly coined word or a phrase composed of familiar words. The point is that if we are willing to expand our vocabularies, we can retain familiar words as conceptual shelters while also adopting a new and precise term for each of the relevant concepts covered by the shelter . I'm confident that Rawls himself would not insist on using "reasonable pluralism" whenever he had this idea in mind -- rather, he would continue to use "pluralism" without a modifier whenever, in context, his audience would understand that he had reasonable pluralism in mind. Incidentally, terms like this have to be understood as closed, not open, phrases. To explain this distinction, consider the difference between a "blue bird" and a "blue- bird." The former is an open phrase in which we understand that a bird is colored blue, but the latter refers to a bird of the genus Sialia. A blue bird may not be a Sialia, and a blue-bird (Sialia) is just partly blue, having reddish-brown breasts. Similarly, you can interpret the meaning of "reasonable pluralism" as a term for the original meaning of the word rather than as an attempt to contrast reasonable with unreasonable forms of pluralism. Actually, if people could remember it, any word could be used to create a new term for this idea -- "prime pluralism," for example, by contast with all its stretched forms as "secondary pluralism."

There is a corollary, however. Just as anyone using a term for its original meaning may need to add a neologism to designate it, so also anyone stretching a term has no right to claim that it's new meaning should be accepted by everyone. Instead, they should suggest a new term that can be used as a synonym for the new meaning assigned to an old word. The Chinese example is again relevant. When the word for "pen" was stretched to include imported non-brush implements, it became important to adopt new terms also -- in this case, "steel pen" and "lead pen (pencil)" were added to the vocabulary. No doubt in most situations after steel pens replaced brush pens in China, the word "pen" could be used unambiguously for the innovation. However, whenever the distinction was relevant, the use of new terms for both the older and the newer kinds of pens made it possible to be unambiguous.

To give another example, consider the word, income as used in the CROP glossary. Originally, income referred only to cash inflow based on salary and wages. It was easy enough to expand it to include interest from lending, and capital gains. And later to home-made things -- garden produce which could be consumed or sold. When sold, it clearly becomes income, but if consumed, is it then not also income? How about gifts? Or bribes? And benefits from public policies and infrastructures paid for by taxation? Where should the line be drawn between "income" and non-income benefits or gains?

It is pointless, I think, to suppose that there was an original meaning of "income" that should be preserved and all of the stretched meanings of this word should not to be viewed, properly speaking, as "income." Instead, why not accept the fact that in ordinary language usage, "income" will be used for a variety of related concepts -- it is, therefore, a "conceptual shelter" and, as such, a "fuzzy concept." Anyone specifically talking about the original meaning of "income" will need a neologism -- perhaps "earned income," or "economic income." Similarly, all the concepts that include non-earned or non-economic income in the stretched concept of "income" will need to adopt more neologisms such as "comprehensive income," or "substantive income."

Having said that, let us quickly recognize that the stretched term, although a shelter, can also be used unambiguously for any of its sheltered concepts whenever, in context, the intended meaning is clearly understood. In country A, for example, everyone might understand that "" refers to "-A," whereas in B, it refers to "-B." This practice is safe provided users of these concepts remember, whenever there might be ambiguity, to say "-B" instead of "" if that is the sheltered concept they have in mind. Anyone thinking that "income" means only "recurrent benefits, in cash or kind, earned by a person or household" should not protest when the word is used for broader concepts since it will then be easier just to say that he concept s/he had in mind was that of "economic income" or "earned income." Having said that, it will be easy enough to continue without protesting, knowing that h/er audience will understand "income" to mean "earned income," or any other specified meaning of this word.

Fuzzy Sheltering Concepts

This brings us to the first point mentioned above, the need to accept fuzzy concepts and the advantages they offer as concept shelters. To distinguish between fuzzy and operational concepts consider the distinction between core and periphery. When the periphery is well bounded, it becomes possible to decide what cases fall within or outside the boundaries of a concept -- what is or is not a "pen" or "income," for example. By contrast, as each of these words suggests, it may be possible to identify the core meaning of a concept even if the periphery is vague and indefinite -- we know that all pens are writing instruments and all income is regularly received benefits. The shape of the pen or the source of the income remains in doubt and there can be many variations.

Fuzzy concepts are useful for two reasons. First, the terms used for them are typically familiar and simple -- pen, income, poverty, pluralism, etc. -- and we easily accept them as loan words from ordinary language. Second, they easily serve as conceptual shelters: there are different kinds of pens, income, poverty, pluralism, etc. and each can be described precisely and assigned unambiguous terms. What is more, the shelter term can be used without ambiguity for any of its sheltered concepts in context. Whenever the context tells us what kind of pen, poverty or pluralism is to be examined or used, the shelter term is clear and sufficient. We may easily distinguish the concept of a shelter from the term used to represent it -- and I shall write "shelter term" for this concept -- note that it is a neologism, but easy to learn because its meaning is transparent.

I have introduced "shelter term" in order to simplify consideration of how it differs from three other notions: a genus, a polyseme, and a homonym. They are all useful and could, perhaps, be confused with shelter terms so it seems useful to comment on these distinctions.

Genus. A generic concept is a precisely described concept that links more specific concepts -- they have a genus/species (broader/narrower) relationship. Fish, as a genus, includes salmon and trout. The first sense of fish in my American Heritage Dictionary comes from zoologists who identify a species of aquatic animals who are cold-blooded aquatic vertebrates: they use "Pisces" as a technical term (neologism) to identify this generic concept it includes a large number of different species.

A second sense of "fish" is defined in the AHD as including "any aquatic animal, such as a jellyfish, shuttlefish or crayfish." Since these examples are invertebrate, they cannot be included in the class of Pisces, but they are a type of "fish" considered as a shelter term. Could this term also cover aquatic mammals? In his famous novel, Moby Dick , Melville explains why he views whales as "fish." Anyone who contradicts Melville by claiming that whales are not fish engages in a purely semantic game. If we accept fish , in its second dictionary sense, as a shelter term, we can see that it is a fuzzy concept that can include not only members of the Pisces family, but also shellfish and even aquatic mammals like the whale and dolphin.

We could formalize the distinction by saying that if a genus is defined as containing A, B, and C, it includes all varieties that share these properties -- i.e. ABC+. However, when the term for this concept is stretched to include things that have A but not B or C (e.g., all aquatic animals) it becomes a "shelter" for Pisces (ABC) and also for other kinds of aquatic animals -- including (AB) and (AC). The stretching of "fish" to include warm-blooded as well as cold-blooded aquatics and invertebrate as well as vertebrate species can be compared to the stretching of "pluralism" or "income" to include/exclude characteristics inconsistent with those found in the prototype concept. We can see that all sheltered concepts are related to each other in some way, but they are not necessarily species of a genus. No doubt many related concepts are not sheltered under the same term. However, all the concepts covered by the same shelter term are linked as sheltered-concepts -- they may all be referred to by the same word.

Another useful distinction can be made: the species of a genus are mutually exclusive whereas members of a conceptual shelter are not. Apples, oranges and bananas are all fruits, and the boundaries between them are sharp. By contrast, sheltered concepts overlap and may be more or less inclusive: thus income may include earned income, taxable income, gifts, bribes, welfare payments, interest on loans and various mixtures of these categories. This also means that a general concept is precise, providing clear criteria for inclusion/exclusion, whereas shelter concepts are fuzzy. When "fish" is used to identify all and only cold blooded vertebrate aquatic animals, it precisely designates a genus, but when it lumps together non-Pisces and Pisces aquatic animals it becomes a fuzzy (shelter) concept.

Homonyms: The concept of a "homonym" is important for lexicographers but irrelevant to our concerns here. I mention it only because it is often confused with "polyseme," a concept that can easily be confused with that of a shelter, as explained below. For the record, two words are homonyms if they have the same form (spelling and pronunciation). For example, "fast(1)" refers to speed, as a fast horse, boat, or runner. By contrast, "fast(2)" involves control as seen in fastening a boat, to hold it fast. There are three words spelled "b-a-r-k" -- they refer to the sound of a dog, the covering of a tree, and a kind of sailing ship. No two homonyms, despite their shared form, can ever serve as a shelter-term.

Polysemes. A word that has several meanings is known by linguists as a "polyseme" -- it is often referred to, erroneously, as a homonym, perhaps because of its unfamiliarity. To illustrate: "mouse" is a polyseme that can mean a rodent, a computer device, or a timid person. Although the word, 'mouse', was used as a metaphor to establish each of its different senses, we should not think of this as a kind of conceptual stretching. Actually, there are some rodents who do not belong to the Muridae family, such as the "jumping mouse", a member of the Zapodidae family that looks like a "mouse". This example suggests the use of "mouse" as a shelter term for any small animal with mouse-like characteristics.

Parallels can be found in the meanings of many words used by social scientists: "party" for example, is a polyseme that can mean not only a political party, but a social gathering, a person, an accessory, or an informal group. In any of these senses, the word could probably be used as a shelter term but let me mention its use in political science. One sense of "party" is that of a "political party." In Western democracies, a party is an organized group that supports candidates for election to a public office. Implicitly, election to such offices affects the distribution of power.

We might summarize these three properties as A=elections, B=competition, and C=power. All political parties in Western democracies are thought, by definition, to possess these characteristics. However, there are countries dominated by a single "party" where electoral competition does not prevail. There are also parties in military dictatorships where elected members of parliament merely rubber-stamp the decisions of the ruling group. We may, then, recognize "political party" as a shelter term covering three different concepts: A, AB, and ABC. They are related to each other by sharing A -- they support their nominees for election to public office. AB also shares competition with rival parties but lacks power; but A lacks competition and its victorious nominees lack power. Only ABC has all three properties. To identify the sheltered concepts we need a special term for each: e.g., "single-party" for A, "quasi-party"for AB, and "true party" for ABC. These are neither accepted nor, perhaps, acceptable terms, but they illustrate the function of assigning neologisms to represent each sheltered-concept found in a shelter.

Shelter terms should not be picked or defined randomly -- a safe rule to follow would be use only terms (words or phrases) entered in an ordinary language dictionary and, normally, to use only one of their senses. If the term is a polyseme -- as it almost always is -- all but one of is senses can be ignored. If "mouse" is used as a shelter for all rodent-like animals, its use for a gadget can be excluded. If "income" is used as a shelter for non-cash and unearned benefits as well as economic income, it should not include irrelevant senses of this word, such as the influx of .migrants to a town or country. The meaning of "mother" as a non-human source of creativity clearly falls outside the shelter concept of "mother" covering various types of parent/child relationships. For political scientists, consider that "party" can represent five different concepts, according to the American Heritage Dictionary, but only one of them -- defined as "a permanent political group organized to promote and support its principles and candidates for public office" -- is appropriate for the shelter concept of a political party.

Subject to such a caveat, however, we can safely provide fuzzy generic definitions for general language words to use in scientific and precise discourse provided we also have more precise descriptions for each of the concepts included in that "family", together with precise terms for each member of the family.

Note that this definition clearly includes A, but makes no reference to B and C. Yet many political scientists would claim that, without B and C, a party cannot be a "true party." Lexicologically speaking, one could treat sheltered concepts as sub-senses of one sense of a polyseme. However, I propose adding a conceptual distinction: a shelter concept is fuzzy, clear at its core but vague at its periphery; by contrast, all the sheltered concepts belonging to such a shelter should be precise or operational. Normally, any sense of an ordinary language polyseme could be used as a shelter-concept -- sometimes two or more senses could be linked in one shelter, but this may be rare. An example is "fish" as mentioned above -- its first sense in the AHD is precisely that of Pisces, and its second sense admits aquatic animals that may be warm-blooded or spineless. This means that, as a shelter, "fish" includes both of these senses -- it shelters both members and non-members of the Pisces family, provided they live in water.

In ordinary language, polysemes can be used unambiguously because the context shows which of its possible meanings is intended: for example, the phrases "cocktail party," "a sailing party," or the "injured party" refer quite clearly to particular senses of party that are not "political parties." The same principle applies to shelter terms. We can use "party" unambiguously to designate different types of political party provided the context shows which of its possible meanings is intended. In a discussion of politics in democracies, for example, one may know that "party" refers to a "true political party," while an analysis of military dictatorships may contain references to parties that are "quasi-parties."

As this example suggests, any ordinary language word can be used unambiguously. As a polyseme, its context should reveal which of its dictionary senses is intended. In a special language, however, a deeper refinement is needed -- among political scientists, for example, "party" will normally refer to a political party, but it may not be clear what kind of political party is meant. The distinction between types of parties mentioned above gives one the opportunity to use a new (technical) term for one kind of party, or to use "party" alone for the same concept provided it is contextualized. We refer to any such term as "equivocal" if, within its special field, it has several meanings. Thus "party" is an equivocal term for political scientists. That is very different from saying it is a polyseme, a term relevant to ordinary but not special language contexts. A political scientist writing about politics will never have a "cocktail party" in mind when talking about "parties."

All shelter terms are ordinary language words, typically used in only one of their dictionary senses and standing for a fuzzy concept that can be used to designate several precise concepts. In addition, however, shelter terms can also be used unambiguously, in context, for any one of the sheltered concepts they cover. Normally, therefore, these terms are both shelter terms and equivocal terms. More formally, if A is a shelter term for concepts A1, A2, and A3, it is also an equivocal term that can be used, in context, to designate A1, A2, or A3. Less formally, a term -- fish, income, poverty, party, etc. -- is used as a shelter to refer to all members of a class, but as an equivocal term to represent only one member at a time.

When the context does not clearly indicate how a term is being used, it is important to provide clarification -- thus one may say "income" in all forms, or "income" as cash only -- see the discussion of this matter above. In contexts where unearned and non-cash income -- including interest on loans, non-cash benefits, bribes, gifts, home-made products, etc. -- is under consideration, new terms may be needed to designate each of the useful covered concepts. However, if in any context, only one of them is considered, then the shelter term, "income," can be used as an unambiguous though equivocal term to represent it.

Moreover, if the new meanings of a word are likely to cause ambiguity, it will become necessary to coin a new term for its original meaning -- e.g., "economic income" to replace "income" as originally conceived. Critics might object that all income is, by definition, economic. However, this point is be irrelevant if the redundant phrase is needed to pinpoint the original meaning of this word after new meanings have undermined it.

A Terminological Example: "Term"

To conclude this discussion, consider term as an example based on its special meaning in terminology. As everyone knows, "term" is an ordinary language word, a polyseme with a host of meanings, some closely linked semantically, and others quite distant -- e.g, a semester, the end of a gestation period, when a debt becomes due, the duration of a public office, or footing (as to be on speaking terms). Among all these senses of "term," consider the one that is defined in the AHD as "a word having a precise meaning." We may think of it as a shelter for several concepts that are used by lexicographers and terminologists alike. However, in each of these fields it is operationalized in a different way. We need not argue about which is the "correct" meaning of 'term' -- it is surely correct in any of its more precise meanings.

Lexicographers often use "term" to mean "a word or a phrase." Although not usually defined formally as a sense of the word, I see it as an important member of the fuzzy family of concepts represented by 'term'. I shall refer to it as a "lexical unit."

Two other concepts called "terms" are particularly relevant here:

For lexicographers, whose primary task involves entering lexical units in dictionary entries which describe them and define their main meanings, a "term" may mean a lexical unit used in a special language. Medical terms, for example, include the technical vocabulary used by physicians. These often include unequivocal expressions used as replacements for ordinary language words that are polysemes. For example, "cancer" is defined, technically, as "a malignant neoplasm..." However, this word has some other popular meanings. It can be replaced by "carcinoma" which is unequivocal. We may use "technical term" to distinguish this sense of "term" from its use to mean a "lexical unit." Since lexicographers need to distinguish between the words and phrases entered in a general dictionary and those listed in a specialized glossaries, this sense of "term" is important for them. Since both technical and non-technical terms are lexical units, lexicographers can establish their own policies to decide what to include or exclude from their dictionaries. The Oxford English Dictionary, for example, excludes technical terms whereas, by contrast, Webster's Unabridged Dictionary includes them. As a shelter term, "term" is often used to mean both technical terms and lexical units, a practice that is confusing only in situations where a distinction between them is needed.

For terminologists, "term" also has a third (onomantic) meaning: it refers to a lexical unit chosen to represent a concept. Their starting point is conceptual, involving the description of concepts needed in some field of work, whether or not there are any established designators for them. Many specialized concepts already have well established terms and lexicographic methods can suitably be used to identify and define them. Terminography is the branch of Terminology that involves the lexicography of special languages: technical terms in use already are entered and defined in glossaries by means familiar to lexicographers. The distinction between the two fields is contextual rather than conceptual -- lexicographers concentrate on the lexical units found in ordinary language whereas terminographers use similar methods to process technical terms found in a special language.

However, some terminologists also concern themselves with the problems involved in finding suitable "terms" for new concepts -- concepts that can be described but still lack a designator. Of course, the process of finding (or coining) convenient and unambiguous expressions to represent any new concept is a problem for everyone involved in developing a new field of research -- in the social as well as the natural sciences. Terminologists can help them by offering professional guidelines for describing concepts and selecting designators to represent them. Should such designators be called "terms"? After they have been accepted, they become the technical terms used in a special language. But before that, what are they? Perhaps we could call them "suggested terms." Elsewhere, I have called them tags -- for a fuller discussion see "Onomantics and Terminology" in International Classification, as also available, in draft on my WebPage.

The relevant dictionary definition of a "term" specifies that it is a word that "has" a meaning -- if it is only a suggestion for a word that might be used to represent a concept, it would not yet be a term, although it might soon become one. To call such an expression a "term" involves a metaphorical expansion of the original idea -- like expanding "mother" to include "step mothers" or "income" to include non-cash benefits. A suggested term is not yet a technical term in a special language, but it is clearly a lexical unit. As an equivocal term, "term" can, therefore, mean a lexical unit, a technical term, or a suggested term -- and, as a shelter term, "term" can also be used to include all three of these concepts. In most context these distinctions don't matter and it is unnecessary to be more specific -- but that is generally true of all shelter terms. The shift from ordinary to special language contexts do not, therefore, involve the character of the lexical units employed so much as the degree of precision required -- in general, the distinctions needed in any special language are specific to that language and require the use of technical and suggested terms to supplement the familiar lexical units found in ordinary language.

Summary

We have three possibilities for dealing with "mother," "income," "term" and a host of similar ordinary language words. I shall refer to them as the "purist," "stipulative," and "sheltering" opproaches.

The first position is that of Giovanni Sartori who argues that the original meanings of a word should be retained, and new terms ought to be found for any new concepts. According to this position, "term" should only refer to established words or phrases used to represent well defined concepts. To apply it to candidates for acceptance as the designators of a new concept should not, therefore, be called "terms."

The second approach could be called "stipulative" -- it permits metaphors that assign new meanings to familiar words. Such metaphoric terms would be accepted as "terms" even though they had not yet been accepted in a special language. Since this approach leads to the proliferation of meanings for established words, its acceptance calls for heavy reliance on contexts to determine which of several meanings a word has, even when they involve only slight modifications of an established sense of the word..

A third approach which we may call sheltering hinges on acceptance of fuzzy concepts.

While acknowledging the proliferation in ordinary language of metaphors that represent a variety of related concepts, it calls for their supplementation by a set of precise designators for each of the needed concepts in this set. Normally, such metaphoric meanings apply to different subject fields so it is not difficulty to sort them out -- one need only remember a relevant sense of a word that applies to one's only field of interest. Consider 'term' as an example: college students think of it as semester, pregnant women as the time of parturition, lawyers as duration of a contract, or when a payment is due, politicians for the time during which one holds an elected office, and lexicographers as a word or phrase with a special meaning.

For specialists developing a special language for use in their field of specialization, a similar problem arises but at a more precise level. They often need to make fine distinctions between different sub-senses of a word -- for example, different kinds of "term" understood as a "lexical unit." They may need to use all these sub-senses and, therefore, find means to distinguish between them. If they accept the stipulative approach, they may find it hard to distinguish between the different sub-senses of a word. They will be torn between using the word for a new sense that someone has recently stipulated and their preference for using the same word in an earlier sense. The sheltering approach permits both solutions and combines them.

Consider the meanings of poverty" as a shelter term in the CROP glossary: this word has been assigned specific definitions for statistical analysis in Australia, Brazil, the Philippines, the U.S., the World Bank, etc.. A dictionary definition of "poverty" as a "state of indigence, having little or no money, goods or means of support" could suffice as a fuzzy shelter concept. Among Australians, its specific Australian definition might be taken for granted, and comparativists could specify that particular meaning of "poverty" whenever they wanted to use it. We could recognize "poverty" as the shelter term for a number of different concepts and also, in context, use the word precisely to designate any one of them. In case of doubt, we could speak of the "Australian definition of poverty" to remind folks of the precise notion of poverty we might have in mind. For more examples, see the CROP glossary in hypertext .

Sometimes we might be puzzled to know whether to use a familiar but equivocal term, or an unfamiliar but unequivocal one -- e.g. "fish" or "Pisces," "pluralism" or "reasonable pluralism." The answer may well be that we should use both. I call this the pleonastic solution -- it simply couples an ordinary language words with a neologism. The ordinary words convey a general sense of what one has in mind, and the neologism provides a precise tag for anyone who wants to make finer distinctions. We may write "fish (Pisces)" to let readers know that we have in mind only cold-blooded aquatic vertebrates, and we can write "pluralism (reasonable)" to show that we are using this word in its original sense, excluding the situations in which "pluralism" is applied, reasonably or not, to unreasonable relationships.

We may conclude, therefore, that a workable solution to the problem raised by the use of ordinary language words to designate specialized concepts can be found if we recofgnize them as shelter terms and coin neologisms for each of the more precise and operational concepts they shelter. Rather than rely only on the equivocal or unequivocal terems produced by this means, we can use the pleonastic method to combine them and pave the way for the unambiguous use of the former, i.e. simple ordinary language words to represent new technical concepts.

Return to top of this page or click here for Home Page links:

Personal Autobio

PubAd GRD

Globalization Concepts

Ethnicity ETHNIC-L

COCTA Onoma

COVICO Choices

Impeach SITES

Search Engines

---

Updated: 10 June 2001