Word links with The Merry Wives of Windsor

Notes & Queries 22/4 (220), April 1975, p. 169-71

Alfred Hart's enumeration [1] of the total number of different words in Shakespeare's plays and poems is 17,677. The number of different words in each of the plays (see Table) varies from 2037 in Errors to 3882 in Hamlet, with an average of 2799. It is generally about the same as the number of lines in the play. The number of words per 100 lines varies from 127 in Macbeth and Tempest to 90 in Richard III, with an average of 101.7. Every play has a number of words appear­ing in it uniquely, varying from 70 in Julius Caesar to 396 in Hamlet. They total 6738, with an average of 182 per play.

    Hart makes it clear that Shakespeare's vocabulary changed over the course of time, and can be regarded as having grown with the writing of each new play. This assumes that a word once part of his vocabulary was never lost. However, some words at least, the "peculiar" words in Hart's terminology, once used were never used again. While it is conceivable that Shakespeare had the whole of his 17,000 to 18,000 words available throughout, from start to finish, making the selection pro­voked by his theme, his characters and their setting, this seems unlikely‑particularly in view of his neologisms, inventions and com­pound constructions. It is more probable that at any given time a part of his total vocabulary was more readily available than the rest; and that the change from play to play was a gradual transition, with a large measure of community between one play and the next.

    This line of thought suggests that the degree of contemporaneity of two plays might be measured by the degree of com­munity between their vocabularies. The present note is the result of a trial of this hypothesis in relation to The Merry Wives of Windsor. In the order of the plays pro­posed by Sir E. K. Chambers, 1600‑01 is the suggested date for Wives, in the sequence Twelfth Night, Hamlet, Wives, Troilus, All's Well. This dating has not gained universal assent. The Cambridge editors, Quiller‑Couch and Dover Wilson, consider the play an earlier creation, "somewhere about the years 1598‑1600", i.e. written about the time of Henry V and Julius Caesar. Hart, apparently for reasons unconnected with his word counts, takes a similar view, and places it in the sequence Much, Ado, Henry V, Wives, Julius Caesar, As You Like It. The Arden editor, Mr. H. J. Oliver, argues forcibly in favour of the theory that 2 Henry IV and The Merry Wives were being written at the same time‑the composition of the former having been interrupted for the hasty writing of the latter … ". [2] Adapting the Chambers chronology to this placing we would have the sequence 1 Henry IV, 2 Henry IV, Wives, Much Ado, Henry V.

    To test for communities of vocabulary between Wives and other plays of the canon, we need not follow Hart in making a complete count, but can make do with a representative selection, preferably chosen for us by an independent authority. We can then use a concordance to list the other plays in which these words have been used. For the present inquiry the words chosen were those words in the glossary to The Merry Wives in the Cambridge edition which provided not more than forty cita­tions from other plays in the Bartlett Concordance.'

    In deciding whether two words were the same or not Hart followed the principles adopted by the editors of the Oxford English Dictionary; and they were taken as guides here. Different parts of speech (noun, verb, adjective, adverb, etc.) were taken to be different words even though identically spelt. Different inflexions of a word, e.g. declension of nouns, conjugations of verbs and comparison of adjectives were not taken to constitute a difference. Without entering into subtleties and metaphori­cal uses, words with a highly specific sense as jven in the glossary were matched only with identical usages. These words of specific sense are so marked in the list in the footnote.

     Having found the number of word links between Wives and each of the other plays, it is necessary to test whether they exceed expectation, and if they do whether the excess is significant statistically. If links were randomly distributed, the number of links with any one play should be propor­tionate with the vocabulary of that play as estimated by Hart and as shown in the third column of the Table. As an alterna­tive the length of the play, i.e. the number of lines as shown in the second column might have been taken, but would have been less precise for an estimate of relative probabilities. The figures of both columns 2 and 3 have been taken from Hart's Table 1 (p. 132).

To show how the calculation of expected numbers was made, the links with 2 Henry IV may be taken as an example. The vocabulary of that play, by Hart's enumeration, is 3130 different words. The vocabularies of all plays together, excluding Wives, is 101042; and the total number of link citations counted is 1094. The null hypothesis (of only random association) predicts, then, 1094 x 3130 ÷101042 link citations in 2 Henry 1V = 33.8891. The number found, 59, in percentage propor­tion to 33.8891 is 174%, an excess so large that the probability of its appearance as a random event is negligibly small. There are four plays with which the observed number of links is in significant excess of expectation:

     The statistically significant association with 2 Henry VI may be regarded as a somewhat improbable event which hap­pened to occur. In a list of 36 pairings we would expect two, by random chance alone, to achieve a p value of 0.05 or less, i.e. one in twenty, and one such to be a deviation in the positive direction. No meaningful­ness can be attached to the association since links with 3 Henry VI and I Henry VI are in statistically significant deficiency. There are, as was to be expected, other statistically significant deficiences, i.e. with John, Merchant of Venice and Winter's Tale. But they do not have any particular relevance to problems of dating, as the positive associations do.

    These findings agree with the hypothesis advanced by Mr. Oliver. If, following his suggestion, we place The Merry Wives in the series 1 Henry IV, 2 Henry IV, Wives, Much Ado, Henry V, its four neighbours have a 40% excess of observed word links over expectation (χ² 31.6), beyond the reach of chance except as an astronomically infinitesimal probability. On the other hand, in the Chambers series, Twelfth Night, Hamlet, Wives, Troilus, All's Well, links with the neighbours are only in excess to 6% (x2 0.5), which has a probability of 0.46, which, one can attain with a single toss of a coin.

Notes

[1] "Vocabularies of Shakespeare's plays ", 1943, The Review of English Studies, xix (1943), 128‑140, and "The growth of Shakespeare's vocabulary", 1943, ibid, xix, 242‑254.

[2] The Merry Wives of Windsor, Arden edition, edited by H. J. Oliver, London (Methuen), 1971, p. l v.

[3] Keeping to the order in which they appear in the glossary they were: Actaeon, additions, aggra­vate, ailicholy, Amaimon, anchor, authentic, Barba­son, bead, bell‑wether, bilbo, board, bread, buck, bully, buttons, canary (ss), careers, carry't, carve (ss), cashier, Cataian, charactery, cheater, cheese, clapper‑claw, clerkly, cog, come off, cony‑catch, costard, council, in counsel, coxcomb, cry aim, curtal dog, diffused, distance, divinity, draff, Dutch, dish, eid, Eohesian, Ethiopian, fallow, farthingaic, fee'd, fee‑simple, recovery, fencing, flax, foin, foppery, fretted, gallimaufry, weaver, beam (ss), good even, grated, great chamber, groat. hack, dole, have with you, having (ss), haviour, hawk, bush, elder (ss), Herod, Jewry, high and low, hole (ss), horn‑mad, instalment, intolerable, jack, jack‑an­apes, jays, kibes, knot (ss), liquor, luce, lunes, lurch, mechanical, metheglins, mince, mummy, nay‑word. nurse, nuthook, obsequious, oeillades, o'erlooked (ss). ox, oyes, rack (ss), peaking, neck, pensioners, period, pernend. Phrygian, Turk, pinnace, plummet, pottle, predominate, preparations, primero, pud­dines, ounto, quean, quittance, rag, ragg'd, ravens, red‑lattice, register, ronyon, sack (is). sadness, sauce, sea‑coal, season, shaft, bolt (Ss), shent, shower, simples, slighted, stale, stewed, prunes. swine, strain, submission, sufferance. surge. swinge, tallow, tester, trow, tuckie‑bed, turtles, unkennel, urchins, weathercock, wink, by yea and no, yellows. yoke. In this list (ss) signifies the special sense of the word as given in the glossary and no other use.