No CrossRef data available.
Published online by Cambridge University Press: 22 April 2026
[The frequencies of derived forms (such as possessives in -'s, negatives in un-, adverbs in -ly, and verb forms in -ed, -ing, and -s) are computed from certain word counts as yet unpublished, in which a record was kept of all distinguishable forms of every word. The ratio of each such derived form to the basic form of the word is computed; and the distribution tables showing the variation in each such ratio are presented and discussed.]
1 The samples of words used were: for A, nouns from permission to renown and from sa to tea; for B, adjectives from a to cy; for C, adjectives from a to clear, from la through sy, and some in h. To insure sufficient freedom from chance variations, only those cases are used where the sum of the two numbers in the ratio is large. In A, only the 125 nouns in the sample examined were used that occurred at least 100 times in the Lorge count. In B, the 269 adjectives were used that occurred at least 40 times, including the negative formations. In C, the 510 adjectives were used where x + xly was at least 50. But the ratios for cases of lower frequencies of x were studied also. They support all the general facts of Table 1.
2 I have inspected the Kaeding count far enough to know that the facts would hold if ratios were computed for larger samples.
3 There is also one occurrence of philosophy's, but I conjecture that this may have been a contraction of philosophy is.
4 The ratios for proper nouns naming persons are far above those for common nouns. For example, we have Perry 21/256 or .08, Philip 9/114 or .08, Ralph 28/128 or .22, Robert 12/269 or .04, Roosevelt 21/189 or .11, Roy 17/206 or .08, Sam 19/192 or .10.
5 Much in language has to do with scales or gradients. Very cold, cold, cool, lukewarm, warm, hot, and very hot refer to one such scale. Indubitable, certain, probable, unlikely, very unlikely, and impossible refer to another. Impassive, calm, disturbed, excited, and maniacal refer to a third. When an adjective or adverb x does or can refer to a point or segment of such a scale or gradient, unx or in x may mean ‘not near that point or segment’, but it may also mean ‘not near that point or segment nor any further toward that extreme of the gradient’ and it may even mean ‘not near that point or segment but at or toward the other extreme of the gradient’.
Even where words seem to refer to discrete classes, they may really cover a continuous gradient. This is often the case with x and un x or x and in x pairs. For example, accurate does not in practice mean ‘errorless’, and inaccurate does not mean ‘making one or more errors’ or ‘having one or more errors’. Accurate means ‘being better than a certain mediocre status’, and inaccurate means ‘being worse than it’. The mediocre status, though it may include a large percentage of the persons or pieces of work, has no name. Similarly for active: inactive, adequate: inadequate, attentive: inattentive, and many others.
6 Or of verb and noun, in cases where the word is both verb and noun. But words in the commonest 500 are not included.
6a [On the other hand, the use of adverbs in -lily seems to be a favorite mannerism of certain poets. Francis Thompson was one of these; cf. the well-known lines from his Ode to the Setting Sun: ‘Whatso looks lovelily | Is but the rainbow on life's weeping rain.‘ BB]
7 The 1050 includes salt as a substantive.
8 The 5067 includes same as a substantive.