Linguistics. An International Review. 88/1972 p. 32-37 15.8.1972
Mouton & Co. The Hague. The Netherlands

The Word

Among the many articles in the collection Linguistics Today (ed. by Archibald A. Hill, [New York, 1969]) is the interesting "Lexicology and Semantics" by Sydney M. Lamb, professor of linguistics at Yale University.

I am inclined to regard the section dealing with the word (pp. 40-41) as the essence of the article, which seems to be the opinion of the author, too. The rest of the article is an adaptation of the solution presented. As I have expressed thoughts similar to Mr. Lamb's, ¹ I make bold to give some supplementary or clarifying remarks concerning the key matter of the article.

Lamb asks his reader, "Is  table  a word?" and judges, as is natural, that the answer generally is in the affirmative. The next question, "Is  tables (the plural of table) a word?" to which the answer is likewise affirmative, is interesting because of the added question, "If so, is it a different word from table ?" Then he states, "many people would say that table  and tables  are two different forms of the same word. But some linguists have defined 'word' [why are quotes used here?] as a minimal unit that can be said in isolation. According to this definition, tables is a word, since the plural suffix -s cannot be said in isolation; and it is thus a different word from table”. (Let it be noted here that this last-mentioned fact does not prevent tables from being a form of the word table.) The author then asks still a third question (p. 41). "Consider also another question: Since table has two quite different meanings, as in a book on the table and a table in the book, are these two words — table 1 and table 2 — or just one?"

I am in complete agreement with Mr. Lamb that an answer to his
_______
¹ The first time in an article in Finnish "The Ambiguity of the Word word" (Virittäjä [1929]), and again slightly later in German with Uuno Saarnio the article "Einige grundlegende Tatsachen der Worttheorie" (Erkenntnis [1934]). Uuno Saarnio in his study "Untersuchungen zur symbolischen Logik" (Helsinki, 1936) has treated the word extensively.

33
questions is not to be sought for in a definition of the word which, as a definition, would solve them, for "In fact each of these different units is important in linguistic structure, and each deserves to be recognized." Lamb continues proposing a different solution which indubitably aims at a correct answer, "Let us therefore distinguish the MORPHOLOGICAL WORD, the LEXICAL WORD and the SEMANTIC WORD. Then we may answer our questions easily. Table and tables are two different morphological words but they are two forms of the same lexical word. Table 1, as in the book on the table, and table 2, as in the table in the book, are two different semantic words corresponding to a single lexical word [here one might also say 'corresponding to a single morphological word']." Further on he uses the term LEXEME (Benjamin Lee Whorf) for lexical words and SEMEME (Adolf Noreen) for semantic words.

To this solution I should like to make the following clarification:

(1) The explanation of the polysemy of the term word is possible in another (and quite easily comprehensible) way: by counting the number of words of a text. The number can be counted correctly in three different ways, arriving at three DIFFERENT results. It is obvious that each time the formations counted are different, for otherwise the correct numbers of the words could not be a the first time, b the second, and c the third time.

It is very profitable to examine this count as one of the three different words (which one must learn to differentiate) reveals itself more clearly and correctly understood than it is understood in the light of aforementioned questions. I mean those observable words which primarily lend themselves to counting and which a receiving telegrapher, or an editor who has ordered a 500-word comment, in fact counts. For simplicity, I limit myself to considering only written language. One who has the task of determining the number of words in a written text will most often proceed to count visible groups of letters which he understands as forming separate words. These groups of letters, the number of which is determined by counting them according to the first principle, can be read and understood. Therefore they can be said to have a meaning (by no means can the meaning of these letter groups be isolated from them nor can anything beyond their physical characteristics be observed). The fact that these word groups are understandable (for anyone who knows the language) is by no means the only characteristic of observable words. There is no reason to name them on the basis of only one characteristic, that is of their meaning, and call them sememes. Furthermore, if table appears in the text more than once in the meaning of a certain piece

34
of furniture, it should be counted as a different word in this count of observable words, and not only in the case that it means something other than a certain kind of furniture. The term sememe is useful and belongs among the technical terms of linguistics, but is not necessary in the analysis of the concept of the word. On the other hand, it is absolutely necessary that empirical language research direct its attention to observable words and be able to distinguish them from other kinds of words. The starting point for all language research is its VISIBLE or AUDITORY realistic signs of communication. Here the term OBSERVABLE WORD (proposed by Uuno Saarnio) is a designation more suitable than sememe for the basic objects. They have numerous physical properties, and can be understood singly (e.g., table) or grouped (e.g., on account of ) — as has been repeatedly mentioned heretofore.

(2) Not all observable words are totally different — need this be mentioned? — but many are similar to one another. In the case that the ways of understanding similar observable words are similar or at least etymologically linked, these observable words are held to be the same, that is, belonging to the same set, to be elements of the same set (for instance, a book on the table and a table in the book). If on the contrary the observable words in question are considered as elements of different sets, they are homonyms (for instance, pallett < lat. palea, pallett < French pallette ). But when the number of words in a text is determined, a second counter might depart from the method of the first counter, and, on seeing the observable word table for the first time, count it; but on encountering table again — that is, precisely the same lettergrouping, he may think that this word has already been counted, and omit the second, and third and fourth occurrence from the count. In proceeding thus, the observable words are no longer counted, but rather the groups formed of observable words similar to each other — the so-called word forms, e.g., the set of all nominative singulars (table ), the whole set of nominative plurals (tables ), etc. These sets are the formations that Lamb names morphological words.

(3) There is still the third method of our word-counter. Lamb has considered that, having met in the text a certain observable word — e.g., table — a second, a third and more times, additional appearances do not increase the count of words in the text. Furthermore, on encountering tables, he considers — as is quite possible and usual — that it does not add to the number of words, either, for it is one of the forms of table, as is the nominative singular. The numerical result of the third counter is again smaller than that of the previous counter, whose count was

35
smaller than that of the first counter. He has counted, not the observable words, not their sets, nor word forms, but the sets of word forms, that is, the sets of some sets, the inflected or the uninflected words (in the last-mentioned set there is only one element). Lamb (and many others) called such sets LEXEMES. This term adds nothing new, for WORD is suitable to the purpose, just as exact, insofar as WORD FORM and OBSERVABLE WORD are correctly used.

(4) A small but necessary elucidation is the mention that, in spite of general belief, there are no words (or lexemes, or lexical words) in dictionaries. There are only observable words, as a glance in any dictionary will show, certain ones of which (the headwords) indicate words (lexemes), others word forms, others again are not cases of suppositio materialis (for example, the different meanings given).

(5) The aforementioned points are well known and recognized in part, but only in part. A number of authors state that in addition to observable words, there is another important group of words. Bertrand Russell has emphasized this several times. In his work An Inquiry into Meaning and Truth (New York, 1940) he writes (p. 26), "The spoken word 'dog' is not a single entity: it is a class of similar movements of the tongue, throat, and larynx. Just as jumping is one class of bodily movements, and walking another, so the uttered word 'dog' is a third class of bodily movements. The word 'dog' is a universal, just as dog is a universal. We say, loosely, that we can utter the same word 'dog' on two occasions, but in fact we utter two examples of the same species. There is thus no difference of logical status between dog and the word 'dog'; each is general, and exists only in instances. The word 'dog' is a certain class of verbal utterances, just as dog is a certain class of quadrupeds. Exactly similar remarks apply to the heard word and the written word." Another example, from John Perry's study of language, "The Same F" (Mind [1970], 181), p. 187, "Consider the following list of words:

A. Bull
B. Bull
C. Cow

36
Ingarden (Norsk filosofisk tidsskrift [1968], 104ff.), p. 115 [translation]: "When I write twice 'hund', I take care that the written forms are the same, and only these signs of the same form together make a certain element of the language. Thus we have gone beyond the purely physical, and come to the type, to what is properly meant by the word 'hund'. There is in German one word, one German word 'Hund'. It can be pronounced many times on this model but 'hund', this pronunciation, is something non-physical. It is an abstraction ... By no means is it anything physical."

(6) Of expositions of the difference between the observable word and the word there is a sufficiency. An essential addition is, nevertheless, necessary: in addition to the observable words there are sets of similar observable words that are elements of inflected or non-inflected words, in other words, there exist three kinds of formations. This triad Mr. Lamb has discussed creditably in his article, if not correctly.

(7) I have used the terms observable word, word form and word. This precision of terminology is not yet sufficient, for it must be possible to record exactly individual observable words, word forms and words.

In the aforementioned study by A. Penttilä and U. Saarnio, Einige grundlegende Tatsachen der Worttheorie each observable word has been presented for indexing as follows: table 0t, tables 0t, men 0t, words 0t etc., each word form table 1 t, tables 1 t, house 1 t, man 1 t, men 1 t, words 1 t, etc. and the word table 2t, house 2t, man 2t, words 2t, etc.

If the order of the members of the index is changed (t0, t1, t2), the referent changes radically (there is no longer question of suppositio materialis). The referent of table 0t is a visible or audible observable word table, table t0 signifies an actual piece of furniture; the referent of word 0t is a visible or audible word word, but the referent of word t0
is a visible (or audible) observable word (e.g. word or words or house or men etc.), the referent of table 1t is the set of visible or audible table words, but the referent of {table t0} is a set of certain pieces of furniture, word 1t is a set of observable words word, but to the set word t1 belong the elements — in addition to word 0t — house 0t, men 0t etc.; word 2t has the elements, among others, word 1 t, words 1 t, but {word t2}, besides word 2t also house 2t, man 2t, etc.

(8) One may question whether such precision of terms is of advantage. Neither it nor the terminology are absolutely necessary, as many assertions regarding words are often correctly understood although the manner of expression is not exact. We are accustomed to the same expression meaning now a set, now elements of a set (a procedure characteristic of

37
everyday usage). But in linguistics we also meet sets of sets, that is, formations in the third degree of the hierarchy. Here everyday usage is neither sufficient nor right. Without the delineated triple division, for example it is impossible to discuss e.g. the question of the parts of speech successfully. Obviously, there must be a clear idea of what is to be grouped — words, word forms or observable words. The grouping is different each time, since the formations to be grouped differ sharply one from another. But once a distinction is made between the observable words, the word forms and the words, the classification is on solid ground. Furthermore, the clarity of the system of words affects grammatical analysis. It gives grammar a design. In my paper Észrevételek a nyelvtan rendszeréröl (=A Magyar Tudományos Akadémia nyelv- és irodalomtudományi osztályának közleményei XV [1959]) I have explained this.

These are valuable findings, even these, but they are by no means the only ones. Certain of the questions brought up and treated by Mr. Lamb in his study will be very clearly benefited through the lucidity achieved.

Helsinki