¹ The most comprehensive work on English is G. Stern, Meaning and Change of Meaning with special reference to the English language, Göteborgs Högskolas Årsskrift 38 (Göteborg, 1931).

² See Bloomfield's criticism of Paul and Wundt, Language 435, and his own explanation of the process, 440–3. O. Springer reviews modern work in Probleme der Bedeutungslehre, Germanic Review 13.159.

³ Bloomfield 396–8; for examples in English see now Edna R. Williams, The Conflict of Homonyms in English, Yale Studies 100 (New Haven, 1944), and for general comment my earlier article, Lang. 12.229 (1936).

⁴ On the two adjectives light see Lang. 12.242.

⁵ On the two adjectives strait, straight see Miss Williams 103–11.

⁶ Die Bedeutung des Wortes² 13–4 (Leipzig, 1910).

⁷ M. Bréal, Essai de sémantique²157 (Paris, 1899), pointed out that the context of situation prevents misunderstanding in polysemy, but he failed to consider adequately the kind of polysemy where confusion might occur.

⁸ A. Darmesteter's term was rayonnement, La Vie des Mots² 73–6.

⁹ Cf. Jaberg, Aspects de la géographie linguistique 55 (Paris, 1935): ‘La coexistence de deux significations parait tolerable quand l'une représente le sens propre, l'autre le sens figure d'un mot, et que ce rapport est encore sensible.‘

¹⁰ Words and their Ways in English Speech 268 (New York, 1912). I have modified the development indicated by the authors in the light of more recent investigations; see NED s.v. parson, final note.

¹¹ Even ‘abstract’ words of this sort are subject to interference from historical shifts in mores, as Jost Trier has shown in the Middle High German words for ‘wisdom’ and ‘intelligence’, which in feudal times had social connotations which they have since lost (Der Deutsche Wortschatz im Sinnbezirk des Verstandes; Heidelberg, 1931). A second kind of interference, as Trier (7–10) points out, comes from the interdependence of a group of words within a given field, since a shift in one may affect the semantic limits of another. Just as the phonemicists have taught us that one cannot study the development of a particular word without considering its relations to the general pattern of sounds, so semanticists will have to beware of explaining the change of meaning in any word without considering all the other words of similar meaning. In spite of these two kinds of interference the adjectives chosen for discussion seem to present clearer cases of the relation of obsolescence to multiple meaning than would more ‘concrete’ words.

¹² We shall have to wait many years for an historical dictionary that will provide information about dialects. Even the Middle English Dictionary in preparation at Michigan will not provide much information about the geographical distribution of words. For Scottish one may consult Sir William A. Craigie, Dictionary of the Older Scottish Tongues (Chicago, 1931 ff.), A–Dull; and William Grant, Scottish Historical Dictionary (Edinburgh, 1931 ff.), A–Covenant.

¹³ Aspects 43–4.

¹⁴ Aspects 44–5 and Map 3. The other studies in Chapter II are all relevant.

¹⁵ The numbers on this and other charts refer to the NED's numbered definitions unless otherwise noted. No attempt is made to list all NED's definitions in every word considered. NED records all words known to have been in use since the 12th century, but gives Old English examples, if they exist, of all words included, not always recording, however, the earliest possible example in Old English, perhaps because of the paucity of good examples and the difficulty of dating some texts. In the charts I have had to consider the period 700–1000 as one, since there is not enough evidence to be sure that a sense recorded only in 1000 did not exist in 700; but I have in Table I begun the line at 700, since in sad, which occurs in Beowulf, the sense is known to be early, indicating the time of NED's first example by a vertical line. If a sense persists after 1850, the date of the last recorded instance in NED is of little significance in the earlier volumes, since publication began in 1888 and much of the material had been collected in the '70's. If the sense is still well-known I simply continue the line to 1900.

¹⁶ Joseph Wright, English Dialect Dictionary (EDD) (Oxford, 1898 ff.) s.v. sad senses 3, 4, 5.

¹⁷ R. Jordan, Handbuch der mittelenglischen Grammatik² 54 §34.

¹⁸ This is NED's silly 3: ‘unlearned, unsophisticated, simple, rustic, ignorant’ (1547–1790), 3b: ‘lowly’, 3c: ‘humble’ (of things). Seely forms die out, it should be noted, in the first half of the 17th century.

¹⁹ NED's numbers for definitions of silly are in parenthesis after the numbers for seely.

²⁰ It is of course perfectly possible for words to have both good and bad meanings for many centuries according to circumstances. Thus bold (see NED's senses 1 and 4) as applied to a soldier refers to a virtue, but as applied to a child or a young woman refers to a fault. It is only when the bad sense becomes regularly more frequent than the good that the good tends to fall out of use.

²¹ A man of limited reading, when I remarked that something was a ‘happy accident’, laughed scornfully and said: ‘How can an accident be happy?’ Accident is popularly always ‘an unfortunate event’ and without qualification usually refers to an ‘automobile accident’. Cf. below on the limitations of uneducated people in the interpretation of words like happy.

²² This sense of fair apparently developed from that of ‘free from bias’ (NED's 10), ‘having an equal chance’ (10b, 1711–), and hence between good and bad.

²³ Applied to animals in the sense ‘abnormally furious’, ‘rabid’ (NED's 6) as in mad dog, mad bull, which dates from the 16th century, mad is, of course, still good literary English in England.

²⁴ Informants at 18, 33, 43, 51, 62, 103, 296, 428; an exception is 10.

²⁵ 8, 48.2, 212.1, 2, 239, 281.2, 284.1, 324.2, 333.2, 382, 407.1, 412–2, 416.1.

²⁶ 122.2, 202.1, 257, 259, 290, 310.2, 890.1, 2, 391.2, 403 (only during the last twenty years), 404, 408.1. See the commentary, Map 458.

²⁷ The word was recorded without definition by Sir Thomas Browne before 1682 (Tract VIII), as characteristic of Norfolk and East Anglia.

²⁸ ‘Clever may mean (a) good-natured; (b) good-natured but stupid, weak-willed, or clumsy; (c) good-natured but shiftless and irresponsible; (d) foolishly obliging, so good-natured as to be at the mercy of others; (e) gullible or spineless; (f) ignorant; (g) stupid, brainless; (h) foolish but harmless; (i) indolent, shiftless or worthless. . . . Clever in senses (a) to (d) has been entered on the map; when recorded in other senses it is given in the commentary. ‘ Commentary on Map 468 good-natured.

²⁹ In Maine, where the field-worker (Lowman) resorted to suggestion freely, 70% or 17 out of 24 informants who gave suggested responses reported both senses, but of the 53 informants to whom no suggestion of either sense was made, only 6 or 11% report both senses.

³⁰ Unlike the case of stout, where the historically newer sense penetrates dialect from Standard English, the historically older sense of clever, i.e. ‘skillful’, is replacing a newer sense that developed in East Anglian dialects and was brought thence to New England. This should teach historical scholars to beware of concluding that a new sense encroaching upon an old in any given period is necessarily a development of the old. The facts of lost antecedent history might, if recovered, show that the apparently new sense was really older historically than the one it is replacing, but merely happens to be newer in the particular locality where it is preserved.

³¹ Map 468 good-natured, commentary: 146.1, 160.1, 2, 266.2, 356, 374, 376.2, 377.2, 396.2; cf. 272.2.

³² Cf. the similar situation in homonyms, Bloomfield, Language 396; and my comment Lang. 12.231.

³³ A pious Irish fellow was very much embarrassed when asked in my hearing by a man reading a newspaper what the ‘Passion of Our Lord’ meant. It is curious that the NED in 1909 labelled passionate ‘affected by the passion of love’ obsolete (sense 4). An earlier example of the same process is lewd, originally ‘lay’ as opposed to ‘clerical’, then ‘unlearned’, ‘common’, ‘vulgar’, ‘lascivious’, the last of which ousted all other senses after 1700.

³⁴ Tabu may of course restrict the use of older senses among the learned as well as among the uneducated. An extreme instance is pervert, noun, which in reference to sexual abnormality was not yet cited in NED's 1909 volume, the Supplement's first instance being dated 1906. The rapid spread of the psychologist's term put a sudden stop to the old use of the word in the sense of ‘apostate’, and a modern Dean Farrar could not refer to St. Paul as ‘that audacious pervert’ (NED's last quotation, 1879).

³⁵ Some representatives of the Prague School believe that English and other analytical languages differ from Czech and the Slavic languages generally in making the sense of a word depend upon its context, and find in this tendency an explanation of the fact that English avoids homophony less than other languages. See B. Trnka, Bemerkungen zur Homonymie, Travaux du Cercle Linguistique de Prague 4.154, and V. Mathesius, Travaux 1.84. This principle would apply to polysemy as well as to homophony; but I suspect that these scholars underestimate the conflicts of homophony in English and exaggerate the differences between English and Czech in this respect.