Vocabulary yuppishness or puppy biscuit has not necessarily encountered these forms earlier, but may be inventing them at the moment of use. These two sorts of words — the established and the spontaneously produced — do not differ from each other in kind, and are not recognisably different in form. In syntax, a fairly clear distinction exists between grammatical patterns or rules, which are established as the product of past history, and sentences, which are spontaneously produced as the expressions of current compe- tence. In vocabulary, on the other hand, words are indifferently of either kind. As a consequence of the blurring of the diachronic—synchronic axis in vocabulary, lexicologists may use the same term, such as 'derivation', to refer to either the historical origin of a form or its current pattern of pro- duction. Yet the two do not always coincide. An excellent, thorough overview of the history of the study of word origins is Etymology by Yakov Malkiel (1993). Not limited to English, it pro- vides both a survey of the general subject and much information on etymological studies of English. Among widely used etymological diction- aries are those by C. T. Onions (1966), Ernest Klein (1966—7), and Robert Barnhart & Sol Steinmetz (1988). 2.1.2 A taxonomy of word origins The taxonomy of word origins used here is based on that defined and exemplified by Algeo (1978, 1980) and is most similar to those used by Cannon (1987) and Barnhart and Barnhart (1982—). It pays particular atten- tion to the relationship between a word and the sources from which it is constructed, its etyma. The primary factors are (1) whether a word has an etymon — is based on any earlier words; (2) whether the word omits any part of an etymon; (3) whether the word combines two or more etyma; and (4) whether any of its etyma are from a language other than English. The inter- section of those four factors defines six major etymological or historically derivational classes, as follows: 1 Creations: words not based on other words. Vroom, imitative of the sound of a car moving at high speed, is a noun for such a sound or a verb for such movement (1965). 2 Shifts: words that neither combine nor shorten etyma. Read, as in 'a good read', has been shifted from verb to noun use (1825); and weekend, as in 'to weekend in the country', from noun to verb (1901). 3 Shortenings: words that omit part of their etyma. Caff is a short- ening of cafe (1931), PC of police constable (before 1904), telly of 59 John Aigeo television (1940), and with changed part of speech, burgle of burglar (1872). 4 Composites: words that combine two or more etyma. Tower block is produced by compounding (1966), and privatise by affixation (1948). 5 Blends: words that combine two or more etyma and omit part of at least one. Chunnel blends two words, channel and tunnel (1928); and brekker, the word breakfast and the suffix -er (1889). 6 Loanwords or borrowings: words with at least one non-English etymon. Courgette is from French (1931), zucchini from Italian (1929), and strudel from German (1893); spring roll is a translation of a Chinese term for an egg roll (late 1960s). Classes (2)—(5) are varieties of word-formation proper, words made from other words in the language, as in the OED2 (1989, xxvii—xxviii), which also distinguishes between two processes of borrowing — adoption and adaptation. Adoption is said to be a popular process, borrowing words with minimum change, as sima (a geological term for 'the continuous basal layer of the earth's crust, composed of relatively heavy, basic rocks in silica and magnesia, that underlies the sialic continental masses and forms the crust under the oceans') was adopted from German (1909). Adaptation is said to be a learned process that alters the morphological shape of the bor- rowed word, as snorkel or schnorkel (an underwater breathing apparatus) was adapted from German Schnorchel (1944). The distinction between 'adoption' and 'adaptation' is a tenuous one and often, as in these two examples, cor- relates poorly with popular versus learned borrowing. The OED also identifies some foreign words as 'alien', not yet natural- ized in English. An example is %ori, a Japanese term used in English for what are also called thongs or flip-flops— a sandal with a thong. Like the adop- tion—adaptation dichotomy, the naturalized—non-naturalized one is unclear, being based on variable factors such as the italicisation of foreign words. Both these oppositions are continuums rather than discrete categorisations. Loanwords range from those like ngwee (a unit of Zambian currency, 1966) with exotic spellings, pronunciations, morphology, and reference to those like street (a prehistorical loan from Latin, doubtless made before the AnglorSaxon invasion of Britain) which few English speakers would think of as foreign. In addition to the preceding major six classes, there are two others used by etymologists, which are types of incomplete etymology. They are (7) native developments, words that are phonological and semantic developments of earlier words in English and are therefore not traced to Go Vocabulary another origin (like town, which developed from Old English tun 'an enclosed place') and (8) forms of unknown origin, words about whose earlier history we have insufficient information to make statements (like nitty-gritty, which appeared in 1961 with the spelling knitty-gritty but whose beginning is mysterious). 2.2 The growth of the vocabulary Change that is on-going in present-day English is easiest to see in the vocabulary, although it certainly exists in all aspects of language (Barber 1964; Foster 1968). In recent times, intercommunication between the UK and the US and between each of those countries and the rest of the English-speaking world has been so extensive, with consequent mutual influence of the two varieties, that an international form of English has arisen. Local and national accents remain highly distinctive, and to a small extent national grammatical differences can be identified. In vocabulary, there are national words little known elsewhere, and sometimes not even throughout the country to which they are native, for example, British bap 'a bread roll used for sandwiches' and American poor boy 'a sandwich made on a long roll of bread'. By and large, however, the vocabulary of the English-speaking world is so intertwined that it must be treated as a funda- mental unity, with only marginal national variation. 2.2.1 The si^e of the vocabulary The English vocabulary has grown much in size since 1776. Exactly how much is difficult to say even approximately because there are no accurate counts of the number of words used in English either in 1776 or today. Estimates of the size of the vocabulary based upon dictionaries are flawed by the highly selective contents of all word books. There are said to be about 616,500 forms in the second edition of The Oxford English Dictionary (1: xxiii). Yet it records chiefly literary vocabulary and primarily the English of England. It represents only spottily folk language, recent neologisms, colloquialisms, technical terms, and national varieties of the language other than English as spoken in England. A complete list of present-day English words would be impossible to make; but if we had an approximation, it would surely be many times longer than the 616,500 forms of the OED; indeed, it is potentially unlim- ited in size. In thinking of the size of the English vocabulary, we must be clear about what kind of vocabulary we have in mind: the words used by 6i John Aigeo almost every English speaker, the words used by an average person, the words understood by an average person, all the words used by any English speaker, all possible words, whether actually attested or not, the words most often used by many persons, and so on. Those various vocabularies differ not only in size but also in character. On e count (Finkenstaedt, Leisi & Wolff 1970) indicates that only about 5.4 per cent of the words in a dictionary are descended from Old English, whereas another (Neuhaus 1971: 39—40) indicates that, in a running text from newspapers, 74.5 per cent of the words derive from Old English. Clearly, the nature of the often used vocabulary is different from that of seldom used words. 2.2.2 Wordfrequency The frequency with which words are used has implications as a practical matter in stylistics, for example in setting an appropriate reading level for school books. The word frequencies in two standard corpuses of English, the Brown Corpu s for American and the LOB Corpus for British, are reported by Hofland and Johansson (1982). In the LOB Corpus, the 100 most frequent words are, with only 8 exceptions, grammatical words. The 10 most frequent words in that corpus are the, of and, to, a, in, that, is, was, it. The 8 non-gram- matical words among the 100 most frequent are said, time, Mr, made, new, man, years, people. The analysis made by Hofland and Johansson (1982) was of word shapes; so for example, say, says, saying, said were each counted as separate words, whereas time the noun and time the verb were counted as the same word. A subtler analysis appears in Johansson and Hofland (1989), which deals with the LOB Corpus only, but analyses a tagged version distinguishing various classes of words. That analysis presents the frequencies of word shapes and also of forms belonging to different word classes. In addition, it gives frequencies of typical combinations of words and of word classes. Magnus Ljung (1974) has made a study of the frequency of morphemes to be found in a list (Thoren 1959) adapted from the 8,000 most frequent words in the Thorndike—Lorge (1959) list. The last was compiled to show word frequencies for pedagogical use. 2.2.3 Gauging changes in the si^e of the vocabulary Given such fluctuation in what we mean by the Vocabulary' of English and the problems in counting it, any estimate of its increase in size since 1776 62 Vocabulary must be viewed sceptically. Yet it seems certain that the vocabulary has increased significantly. In a sample of words from the OED (the first shape or sense on each page of volume 1), 393 of 1,019 are first attested after 1776. Those figures suggest that the pre-1776 vocabulary (626 words in the sample) has increased by 63 per cent, but are suspect because of the selectivity of the OED and the sample. The most convenient source for estimating an increase in the size of the English vocabulary is the Chronological English Dictionary (Finkenstaedt, Leisi & Wolff 1970; reviewed byDerolez 1972, also 1975). However, that work must be used with caution because it is based on The Shorter Oxford English Dictionary, a selection from the OED, and the latter is not reliable for the earliest dates of use of words, although it is the best record we have. Of the 80,506 dated words the CED covers, 5.4 per cent originated in Old English, 18.9 per cent in Middle English, and 75.7 per cent in Modern English. Of the latter, about one-third originated after 1776 (a 34 per cent increase over pre-1776 vocabulary). An indication of the caution with which such figures must be viewed, however , is the fact that the Chronological English Dictionary also indicates that of the words originating after 1776,51 per cent were coined in the mid- nineteenth century (1826—75) and only 4 per cent in the early twentieth century (1901—50). Clearly what those figures show is not the growth of the vocabulary, but the extent of the lexicographer's sources. Such a caution is applicable to almost all statistical conclusions based on OED materials. Nevertheless, it seems intuitively obvious that the English vocabulary has grown and continues to do so. Objective support for that obvious intuition runs into problems of documentation, continuity, and identification. 2.2.3.1 Documentation. The problem of documentation is to find strong evidence for the origin of a word. Our major source for such documentation is the OED. However, the evidence of the OED has to be used cautiously because we know that its earliest date of attestation is frequently not the earliest documentable use of a word. The sources drawn upon by the OED are not evenly distrib- uted across the centuries. The OED is biased in favour of literature and particularly of canonically enshrined authors. Moreover, inescapably the OEDs readers were inconsistent in the thoroughness with which they gathered citations. The improved availability of scholarly sources (editions, bibliographies, indexes, concordances, and the like) since the work on the OED was done 6 3 John Aigeo enables us to see how much was missed by the compilers of that great dic- tionary and how cautious we must be in drawing conclusions from it (Schäfer 1980). We are now aware that the OEDs datings are often inade- quate by several decades or even more than a century. Thus, the adjectival abominate is first documented in the OED from 1850; but it was used at least as early as 1594 (Bailey 1978:1). As electronic texts become more available, it will be feasible to estimate more accurately how cautious we need to be in using the OEDs evidence, and it will become easier to correct that evi- dence. Several estimates of the rate of growth of the English vocabulary have been based on The Shorter Oxford English Dictionary, 1968 edition. There are, however, two problems with using that work as a basis for study. First, the principles on which it was abridged from the OED parent work are not clear; and second, the text of the parent work itself is seriously flawed, in the ways suggested above. In particular, excerpting of eighteenth-century books for the OED was to have been done in America, but citation sups for that century did not reach Murray, and so, despite efforts to cover the period, it is seriously under-represented in the OED. Comments upon the growth of the English vocabulary based (as they generally are) on OED evidence, often through the medium of the Shorter OED, show a significant decline in the production of new words in the eighteenth century (Finkenstaedt & Wolff 1973: 29; Neuhaus 1971: 31). The temptation is to explain that decline as a consequence of the conservative temperament of the Age of Reason, a neat instance of the effect of world view on language. In fact, what the 'decline' almost certainly shows is lack of evidence due to uneven gatherings of citations. It is a fact, not about the language of the mid-eighteenth century, but about the vicissitudes of lexicography in the late nineteenth. The neat and impressive-looking line graphs that have been drawn to show the peaking of word-making in the vigorous, language-intoxicated high Renaissance, its deep valley of decline in the eighteenth century, and its subsequent rise to a new, if lesser, high in the mid-nineteenth century show nothing about the language. What they show is the extent and assidu- ousness with which the OED volunteers read and excerpted books. Shakespeare was over-read; the eighteenth century under-read — that is what the graphs show. We have no reliable data on which to base generali- sations about the growth of the English vocabulary. To get such data we need, not a computerisation of the faulty OED sampling, but a wholly new approach. 6 4 Vocabulary 2.2.3.2 Continuity The problem of continuity is a more difficult and generally an unsolvable one. After a word is coined in English, we usually assume that all later instances of the word derive from the initial coinage. But clearly there is no reason why that should be the case for many words. A word may be inde- pendendy reborrowed or reformed many times. For example, cosmos 'the world' was used by Orm in the spelling cossmos about 1200 and identified as of Greek origin in the Middle English Dictionary (Kurath & Kuhn 1954- ). The first citation of the word in the OED is from 1650: As the greater World is called Cosmus from the beauty thereof, with the reference to 'beauty' echoing the Greek sense 'world, order, beauty' despite the Latinate form of the ending. The next citation is from an 1848 translation from German of Humboldt's Cosmos. Thereafter, the QEDhas citations illustrating several closely related senses from 1858, 1865,1869,1872,1874,1882, and 1885. This evidence suggests that cosmos has been borrowed into English at least three times, twice (1200 and 1650) from Greek or Latin, and once (1848) from German. The lack of evidence for continued use of cosmos between 1200 and 1650 and between 1650 and 1848 suggests that the two earlier borrowings were abortive; present-day use of cosmos begins with its 1848 borrowing from German. The OEDs 1865 citation, however, has the spelling Kosmos and refers to the Pythagorean concept of numerical order; it is at least influenced by Greek directly and may be another independent borrowing. It appears that the word in contemporary use is not descended from an early Middle English borrowing from Greek, but from a late Modern bor- rowing from German reinforced by Greek. 2.2.3.3 Identification The Latinate vocabulary is a particular problem for both analysis and ety- mology. English has borrowed so many Graeco-Latin words that it has imported much of the morphemic and morphophonemic patterning of those languages, thereby creating difficulties in analysing English morphemically (Ellegard 1963) and also in identifying the etymology of new classically based words. Because the Graeco-Latin vocabulary has been influential also on other European languages and is the basis for much scientific terminology, it is often difficult to be sure of the origin of a particular new word formed from ultimate Graeco-Latin sources. Without detailed knowledge of its history, we cannot predict the origin of a word like hopioid. American Heritage (1969) derives it from Greek haploeides; World Book (1988) derives it from 6 5 John Aigeo Greek haplous and English -oid; Random House (Flexner 1987) and Webster's New World (Neufeldt 1988) derive it from the English formatives haplo- and -old; The Oxford English Dictionary Supplement (OEDS) derives it from German haploid. The ultimate Greek source is not in doubt, but the immediate English source is a matter of disagreement. To meet this problem, the editors of Webster's Third (1961: 7a) coined the etymological label 'ISV for 'International Scientific Vocabulary', that is, words of uncertain origin used in several languages. A comparable label was used in the OED (1989: xxviii): 'mod. £' standing for 'modern forma- tion'. These labels avoid a misstatement when exact information is lacking, but they are an acknowledgement of ignorance rather than an etymology. 2.3 Creating as a source of new words Words that are coinages ex nihilo are extremely rare, if they exist at all. Words that seem to be of that type are usually words about whose history we merely have insufficient information. An apparent exception to that generalisation is the use of computer- generated trade names, but that exception is more apparent than real. When a new name for a product is sought from a computer program, the candidates are unlikely to be randomly generated stings of letters. Instead the computer has been programmed to produce only certain patterns of letters (CVCVC, CVCCVC, etc.) and certain final sequences are prominent in the trade names selected from such lists: -an, -ar, -el, -ex, -on. It seems clear that the human beings who make the final selection from computer-gen- erated lists are guided by associations in choosing a trade name. For example, even if, as reported (Praninskas 1968: 14), Teflon was a computer- generated name, the last part of it clearly echoes nylon, and the first part is consonant with tough, suggesting a tough, smooth surface. Such considera- tions are very likely to have entered into the choice of the name, which is to that extent not a pure creation. Echoic or onomatopoeic words are a type of creation, for example, burp, bu%% fi^j plop, %ap, %ip. However, they are not pure imitations of sounds, since there are clearly conventions of imitation, and certain sounds, such as /2/ in several of the preceding examples, acquire the value of phones- themes. 2.4 Shifting as a source of new words Shifting may be of shape, grammar, semantics, or pragmatics. 66 Vocabulary 2.4.1 Shift of shapes Shape shifting is illustrated by the back shngjob from boy. It is a minor kind of shifting that involves neither loss nor addition, but alteration of the spelling or pronunciation of a form. 2.4.2 Grammatical shifts English has great freedom of shifting forms from one part of speech to another. Because of the sparse morphological marking for parts of speech, almost any English word can be used as a noun, verb, or adjective-like attributive. Nonce uses are frequent, and so are established shifts. In nonce shifts, for example of nouns to verbs (Clark & Clark 1979), the meaning of the nonce verb derives from that of the underlying noun and the context — both the immediate lexical context and the broad non- linguistic context that we call cultural knowledge. Thus, the meaning of porch in to porch a newspaper 'to deliver by throwing into the porch of a house' depends on the noun sense of porch, the co-occurrence with newspaper, and familiarity with the fact that newspapers are in some locations brought to a private house by deliverers who throw them onto the porch. In one examination of over 8,700 converted forms (Biese 1941) the chronologica l distribution of the forms by percentage was as follows: to 14c 15c 16c 17c 18c 19c .16 .09 .20 .20 .11 .26 Except for a dip in the eighteenth century, which is probably explained by the gap in the OEIJs resources, Modern English has a fairly consis- tent rate of shifted parts of speech, with some increase in more recent times. A type of grammatical shift that has become more important in recent times is the use of a trade name as a generic. Escalatorbegan as a proprietary name, but has long since ceased to be so. The second half of Coca-Cola like- wise has become generic; the company is fighting to prevent its nickname from the first half, coke, from following suit. Ziploc (1970), a brand name for a plastic bag that fastens by sealing two interlocking strips, has become generic under the respelling %iplock (1982). Other trade names that are often used genetically but still maintain legal status as proprietary names are Band-Aid (a US term for an adhesive plaster), Biro (a UK term for a ball- point pen), Cellophane, Filofax (a UK loose-leaf record book), Polaroid, US Scotch tape and its UK counterpart Sellotape. Teflon is likely to win out over the 6? John Aigeo non-proprietary termpolytetrafluoroethylene\ it already has metaphorical use in the political term teflon-coated 'possessing an ability to escape the conse- quences of one's actions'. Hoover"is used genetically only in the UK, even though the trade name was US in origin; it and Xerox have further shifted into verb use. Another highly productive type of shift in modern times is the conver- sion of a verb-particle combination into a noun (Lindelof 1938, from whom the following dated examples are taken): show-off (111 6), cut-up (1782), stand-by (1796), knock-out (ISIS), take-off'(1826), sit-down (1836), turn- back (1847), stick-up (1857), clean-up (1866), pull-over (187'5), go-round (1886), rub-down (\S96), play-off (\906), fly-past (1914), and check-up (1924). The 520 nouns converted from verb—particle combinations examined by Lindelof were chronologically distributed by percentage as follows: to 15c 16c 17c 18c 19c 20c 0. 1 .05 .05 .07 .50 .33 Lindelof's twentieth-century examples were limited mainly to the first third of the century. If we assume that the rate of new forms remains con- stant through the rest of the century, the twentieth century would account for about 60 per cent of the new total, and the nineteenth century for 30 per cent. These figures suggest strongly that this type of conversion has increased strikingly in frequency in recent times. Lindelof (1938: 39) observed that combinations originating in America comprised 6 per cent of the eighteenth-century examples, 17 per cent of those from the first half of the nineteenth century, 33 per cent of those from the second half of the nineteenth century, and about 39 per cent of early twentieth-century ones. He concluded: And there is one thing which has struck me more and more while col- lecting and arranging my examples, namely the very prominent part which the language of America seems to play in the creation of words of our type. This conclusion is in keeping with a widespread but largely unsubstantiated belief that American English is more innovative than British. If we suppose that the number, of innovations in a language may be partly cor- related with the number of persons speaking it, the increasing size of the American population might strike us as suggesting that American innova- tions ought to be more numerous than they have been. Such a comparison might suggest that British English is actually more innovative than American. 68 [...]... German 5.9 5 .4 4.8 5 .4 7 Greek 6.9 4. 8 8.0 1 .2 8 Latin 5 .2 5.1 9 .4 Yiddish 5.7 2. 7 5.0 3.6 10 Arabic 2. 0 3.9 1.7 6.0 11 Chinese 1.7 4. 2 3.6 6.0 12 Portuguese 1.0 2. 7 1.0 1.8 13 Hindi 2. 2 0.9 0 .2 2 .4 14 Hebrew 0.7 1.5 0 .4 1 .2 15 Sanskrit 1.7 1 .2 0.8 Persian 0 .2 1 .2 Afrikaans 0.5 1.5 9 16 1.8 Dutch 0 .2 0.3 Indonesian 0 .2 0.3 0.8 2. 1 19 1 .2 20 0 .2 Malayo- 17 18 bo 0 .4 21 Polynesian Norwegian 0 .2 Swedish... h they record b o r r o w i n g a n d the per­ centage of l o a n w o r d s for each l a n g u a g e (or area) w i t h i n each corpus are as follows T h e ranking is an average of the four corpuses: a) (2) (3) (4) BDNE BDNE2 81W3 BDC French 31 .4 17.5 21 .2 12. 0 1 Spanish 6.6 10.8 6.1 12. 7 2 Rank Russian 3 .4 5 .4 2. 1 24 . 1 3 Japanese 7.9 9.3 6.3 9.0 4 African 6.1 7 .2 6.7 3.0 5 Italian 4. 7 4. 5 10.7 2 .4. .. follows: 84 Vocabulary Barnhart Longman OED2 NEWS BDC Cannon 0.0 0.3 0.0 0.0 0.6 19 .4 23 .4 30.8 9.6 19.7 Shortenings 9.7 10.0 17.5 9.7 17.1 Composites 63.9 54. 3 52. 2 37.6 73.5 53.8 (Compounds) (29 .8) (36.3) (19.8) ( 12. 0) (57.6) (29 .6) (Affixations) ( 34. 1) (18.0) ( 32. 3) (25 .6) (15.9) ( 24 . 2) Shifts bo 0.0 14. 2 Creations Blends 4. 8 9.8 3.3 1.1 0.5 1.0 Loanwords 6.9 4. 3 18.8 6.9 6 .2 7.5 Unknown 0.5 2. 2 0.3... described the first three of the following corpuses, totalling 1 ,26 2 l o a n w o r d s ; the fourth is of the l o a n w o r d s entered in The Barnhart 1 Dictionary Companion, v o l u m e s 1 4: 4 0 7 l o a n w o r d s from The Barnhart Dictionary of New English since 1963 ( B a r n h a r t , S t e i n m e t z & B a r n h a r t 1973) 2 3 3 2 l o a n w o r d s from The Second Barnhart Dictionary of New English. .. Verbs are the part of speech m o s t often b a c k f o r m e d , and the e t y m o n is often an agent n o u n in -er: swindle (17 82) , edit (1793), commentate lift (1 820 ), bushwhack 7^ (18 34) , housekeep (1818), shop­ (1 8 42 ), scavage (1851), sculpt (1S 64) , Vocabulary play-act (18 12) , typewrite (1887), barn-storm (1 923 ),proof-read (19 34) , divebomb ( 1 8 9 6 ) , p a n h a n d l e (19 04) , (1 944 ), name-drop... Malayo- 17 18 bo 0 .4 21 Polynesian Norwegian 0 .2 Swedish 1.0 1.5 0.6 22 0.3 1.0 0.6 0.8 1.0 0.3 0.6 25 1 .2 0.6 26 Bengali 0.5 0.9 0 .2 27 Danish 0.5 Eskimo 0.5 Korean Vietnamese Amerindian 78 23 0.6 1.0 0.3 0 .2 24 28 0.6 29 Vocabulary E a c h of the following l a n g u a g e s , w h i c h share ranks 30—56, represents less than 1 per cent of the total: A m h a r i c , A n n a m e s e , B a s q u e , B h u... n d s in the N E W S corpus is puzzling Part of the explanation for it m a y be that the percent­ age of c o m p o u n d s r e p o r t e d w a s only for n o u n s of three patterns (n+n, a + n , v + n ) a n d adjectives of two patterns ( n + a , a + n ) A g o o d l y propor­ tion of the 6.1 per cent of 'other' w o r d s m a y b e c o m p o u n d s of other kinds O n the other h a n d , the OED enters... n d social medicine studies (1 927 ), socialite (1938), social Darwinism (1 947 ), a n d socialisee (1 928 ), (1939), socio- (19 52) In w h a t is d o u b d e s s o n e of the m a n y accidents of the availability of evi­ dence, the adverb subconsciously (1 823 ) is r e c o r d e d before the adjective sub­ conscious (18 32 4) It w a s later in the century that the n o u n s (18 74) a n d subconscious present-day... s in the recent corpuses contrast strikingly w i t h those in The Shorter OED, as r e p o r t e d b y T h o m a s Finkenstaedt (1973, 1 1 8 - 5 6 ) In the following table, the SOED percentages represent the history of English over approximately 1 ,20 0 years, as recorded in that dictionary T h e OED2 p e r c e n t a g e s are of the s a m p l e from v o l u m e 1 of the s e c o n d edition of the OED... in the a d d e n d a of the 1981 printing of Webster's Third (1961) T h e first three samples are the smallest, but w e r e analysed by the s a m e set of criteria T h e sixth and largest s a m p l e includes all the w o r d s analysed in the first sample, but because of the size of the sixth sample, that duplica­ tion does not seriously affect the results T h e percentages of etymological types in these . follows. The ranking is an average of the four corpuses: a) (2) (3) (4) BDNE BDNE2 81W3 BDC Rank French 31 .4 17.5 21 .2 12. 0 1 Spanish 6.6 10.8 6.1 12. 7 2 Russian 3 .4 5 .4 2. 1 24 . 1. 1.7 4. 2 3.6 6.0 12 Portuguese 1.0 2. 7 1.0 1.8 13 Hindi 2. 2 0.9 0 .2 2 .4 14 Hebrew 0.7 1.5 0 .4 1 .2 15 Sanskrit 1.7 1 .2 0.8 16 Persian 0 .2 1 .2 1.8 17 Afrikaans 0.5 1.5 0 .4 18 Dutch. 9.0 4 African 6.1 7 .2 6.7 3.0 5 Italian 4. 7 4. 5 10.7 2 .4 6 German 5.9 5 .4 4.8 5 .4 7 Greek 6.9 4. 8 8.0 1 .2 8 Latin 5 .2 5.1 9 .4 9 Yiddish 5.7 2. 7 5.0 3.6 10 Arabic 2. 0 3.9