ANALOGY AS A FACTOR IN SEMANTIC CHANGE SAMUEL KROESCH UNIVERSITY OF MINNESOTA It is now more than a half century since the memorable controversy concerning the application of analogy to linguistics began to be fought out between the old school on the one hand and the neo-grammarians on the other. Today, analogy is regarded as at least of equal if not greater importance than the phonetic laws themselves. Just as the basis of all phonetic laws is physiological, so the basis of all analogy is psychological. The association in the mind of one idea with another forms the basis of any analogical formation. Ideas are associated into groups in the mind, and so also are words which represent these ideas associated in groups. The tendency of analogy, then, is to counteract the great diversity in language and to bring the incongruous elements of speech into groups and systems, thereby simplifying them. In the process of simplification, however, the old form or system is modified so that, in the commonly accepted sense, analogy becomes synonymous with linguistic disturbance of form, function, or meaning. The first we may call morphological analogy, the second grammatical (Paul's formale Analogie), and the last semantic analogy.2 Under this classification will also be included certain types of associative interference discussed by Paul3 under the heading contamination (Kontamination), i.e. blend-forms. The above divisions apply to word-form alone. Analogy covers a wider field than this, however, for associative interference in the phrase or sentence, i.e. syntactical analogy, is equally frequent. Still another type concerns interference between the various sounds of a word. But analogy need not always imply associative disturbance, for analogical association may be creative as well as interfering. There is an association based upon similarity of idea and developing definite congeneric groups. In certain primitive languages, such as Bantu for example, all nouns are divided into a dozen classes by certain prefixes called classifiers. These classifiers furnish elements which connect 1 Cf. Wheeler, Analogy and the Scope of its Application in Language. Cornell University Studies in Classical Philology 1.5 (1887). 2 Cf. Oertel, Lectures on the Study of Language 154. verbs with their substantives in such a way that this congeneric classification of substantives governs the whole system of Bantu grammar. In the Indo-European languages there is a similar type of association, as has been shown by Bloomfield. Here the meaning of any word may be arbitrarily attached to some sound element, most commonly a suffix, contained in the word, and this element then become the center of a congeneric group of words. Although Bloomfield calls this adaptation, it is really a type of creative analogy. An example found in most of the IE. languages is the -r- suffix used exclusively in words for members of the family. Originally the -(t)er- -(t)or- -(t)r- found in words for father, mother, etc., had nothing to do with the idea of the family. Its use in one of these words, however, endowed it with its particular meaning, which became the starting point for the formation of all the other words of the congeneric group. For this type Wundt' uses the term begriffliche Angleichung or conceptual assimilation, by which he means analogy through similarity or contrast in meaning. The two important factors involved in linguistic analogy are the word form and the meaning. In the case of morphological and functional analogy discussed above, change of form has been brought about by association through the meaning. Semantic analogy, at least those phases of it that are discussed at all, is the reverse of this, in that change of meaning has been brought about by phonetic association. In general the types of semantic analogy parallel those of phonetic analogy. The first develops through associative interference, and often both form and meaning are affected. NE. shamefaced, originally shamfast has changed not only its form but also its meaning because of its association with face. These cases of analogy have given us the forms commonly known as "folk etymology," and since they are fairly well known, they need no further explanation. A similar type of analogy is that noted in bilingual districts where the calling to mind of an idea will lead to associative interference of the words for that idea in the two languages. Thus the German-American, associating NE. afford with NHG. erfordern says, "Ich kann es nicht erfordern." Here, of course, phonetic similarity plays an important part in the association. But semantic change may also be brought about in cases where the associative tie is purely semantic. When a word in a foreign language only partly corresponds in meaning with the native word there is a tendency to ascribe to the Oertel 158f. AJP 12. 1ff.; 16. 409f.; IF 4. 66f. Cf. Brugmann, Kz. vgl. Gram. §406. 7 Völker psychologie, Die Sprache 1. 439. former the meanings of the latter. In bilingual districts such analogically derived meanings easily become fixed in the language. Thus a word with two meanings, like the Slavic sramota, meaning both 'Schande' and 'Scham' interferes semantically in Slavo-Germanic districts, thereby accounting for such phrases as, "Habt ihr keine Scheu und Schande?"'8 The above comprise the types of semantic analogy usually discussed in linguistic works. No comprehensive study of them has as yet been made and the examples so far collected could easily be multiplied. But these, in our opinion, are by no means the only or the most important types of semantic change induced by analogy. That new forms have not been called to our attention before, especially in historical linguistic investigations, is not to be wondered at; for, except for changes of meaning actually in the process of creation today, direct proof of such analogical changes is extremely difficult. The existence of phonetic laws has made possible the recognition of the workings of analogy in word form, whereas the lack of any semantic laws and the difficulty of fathoming the infinite number of possible associations causing a change of meaning have made progress in this field baffling, to say the least. And yet undoubtedly every change of meaning is due to an association of some sort. Now this association may result in analogical suggestion leading to a change of meaning. Semantic changes due to the context in which the word finds itself may be the result of analogy, as the psychologist Stout suggests. 10 The extent to which change of meaning may be the result of analogy has been recognized by Vossler in his work, Sprache als Schöpfung und Entwicklung, 119 f. He believes that there is fundamentally only one type of semantic change, namely what he calls semantic transference (Bedeutungsübertragung) and by that he means semantic analogy. Figures of speech, for instance the metaphor, are good examples of this. When we use the expression the foot of the mountain, the word foot has taken on a new meaning but we are well aware of the old meaning of the word. The change of meaning has come about by the transference of an idea from one field (the parts of the body) to another (the parts of the mountain) and this transference is only another name for analogy. The cases of semantic analogy referred to above represent types of associative interference. Just as in the case of morphological analogy, Cf. Schuchardt, Slavodeutsches und Slavoitalienisches, Graz 1885; Paul, Prinzipien §382; Windisch, Sitz. Ber. d. kgl, sächs. Gesell. d. Wiss., 49. 101f. 'See Wundt, Die Sprache 2. 600. 10 Cf. Stout, Mind 16. 195f. there are types of semantic analogy which are not interfering but creative, and it is these which it is our purpose to discuss in this paper. It is a well known fact of semantics that theoretically the sources of a particular meaning are limitless; that is, a given meaning may develop from one of an infinite number of possible sources, depending upon the particular complex of associations calling that meaning to mind. An examination of all the sources of a particular meaning, however, will show a distinctly limited number of such sources and a grouping in some cases of a large number of words with a common source. A further examination of this grouping will show in some instances a tendency for words of one dialect to favor exclusively certain possible sources of a meaning. That is, no other dialect will show this particular development of meaning. This would not be surprising in the case of one word in one isolated instance, but it is noteworthy that there should be three, four, and often even more, words with a semantic source found in no other dialect. As illustrative material let us take the meaning 'deceive', whose sources the writer has investigated," although almost any other meaning, the sources of which have been worked out, would show similar results. Confining ourselves to the NE. and the NE dialects, we note the following cases of a development found only in these dialects: 'do up, finish: deceive' NEdial. dish, put through, NE. do, do for, put through. 'cast a spell over, bewitch: deceive' NEdial. beglammer, Scotch glamer, NE. cast a glamer over, NE., NEdial. overlook, NE. oversee. 'stuff: deceive' NEdial., NE. cram, stuff. 'confuse, bewilder: deceive' NE. amuse, NEdial., NE. arsy-versy, baffle, NEdial. flummox." In other cases one source is overwhelmingly predominant in one dialect with only scattering examples in others, as for example, 'blind, hoodwink: deceive' in NE. blear, blind, blenk, blend, bluff, glaik, hoodwink.13 Sometimes a particular semantic development apparently originates in one dialect and becomes very productive in another, as is the case with words showing the development 'befoul: deceive.' Here the oldest 11 Germanic Words for 'Deceive.' A Study in Semantics. In the series Hesperia, Schriften zur germanischen Philologie No. 13, Goettingen 1923. 12 Cf. op. cit. groups V, VIII, XXXIII, XXXVII. word, MHG. beschizen, represented a development found in both High and Low German dialects, which, however, with few exceptions became productive only in the Low German.14 Then there are groups which represent a development common to many dialects and about which we shall have more to say later. The question naturally arises: Do the individual words in these groups represent in each case a unique, spontaneous development of the meaning 'deceive,' or is this grouping to be accounted for through some form of association? We believe that in many cases the particular development of meaning is the result of analogy-semantic analogy. This may be of two kinds, an internal and an external. In the former, the change of meaning takes place within one dialect, whereas in the latter extraneous influence must be taken into account. Discussing first the former, let us represent the process graphically thus. A word X with a meaning A develops from this a meaning B. Thereupon a word Y also with a meaning A, a synonym, being associated with X likewise develops the meaning B. Applying this formula to some of the words. cited above, we note that NE. overlook was used in the meaning 'look with an evil eye upon, cast a spell over,' from which develops the meaning 'deceive.' According to the NED. the first record of this new meaning dates from 1596. Now NE. oversee, synonymous with overlook, also developed the meaning 'deceive,' the earliest instance of this new use dating from 1646. It is safe to say that the individual who first used oversee in this meaning knew the word overlook, 'deceive.' The two words, however, were closely associated in his mind, and the one had, no doubt, recalled the other on other occasions in other meanings which the two had in common. The use of the one for the other in the new meaning is, therefore, easily understood. The meaning 'deceive' in oversee is not developed directly from 'look with the evil eye,' but by analogy with overlook. An investigation of the words cram and stuff reveals that the meaning 'deceive' in the latter came in like manner by analogy with the former. That synonymity or similarity of meaning forms a most frequent basis of association in the mind has already been noted in the cases of morphological analogy. An examination of the psychological tests performed by Thumb and Marbe to determine the basis of analogical formations also shows that in certain parts of speech, notably verbs, the synonym is very frequently the associative reaction to the stimulus 14 Op. cit. XXXI. |