Available online at www.sciencedirect.com
Language Sciences 30 (2008) 31–52 www.elsevier.com/locate/langsci
Deriving verbs in English R.M.W. Dixon
*
Research Centre for Linguistic Typology, La Trobe University, Bundoora, Victoria 3086, Australia Received 13 April 2007; accepted 19 April 2007
Abstract Phonological and semantic principles which underlie the derivation of verbs from nouns and adjectives in English are examined. There is intricate phonological conditioning for suffix -ize and for suffix -(i)fy; a third major process is zero derivation. These derivational processes cover more than a score of semantic relations (some with overlap between processes). It is shown that whether a noun or adjective forms a derived noun (and, if so, how) depends on a combination of (i) the meaning of the noun or adjective; (ii) its phonological forms; and (iii) its etymology, in particular whether it is of Romance or of Germanic origin. The generalizations arrived at are then applied to prediction of how new verbalizations may be created from proper names. 2007 Elsevier Ltd. All rights reserved. Keywords: English; Verbal derivation
1. Introduction In English there are a number of ways of deriving a verb from a noun or adjective. For instance, we find de-nazi-fy, with deverbal suffix -(i)fy, and de-stalin-ize, with -ize. But why these suffixes with these nouns? Why don’t we say *de-nazi-ize and *de-stalin-ify? Why is it the custom to say central-ize, with -ize, but glori-fy, with -(i)fy? The verb to describe making something German in language or character is German-ize, but with respect to French it is French-ify. Once more, why?
*
Tel.: +61 39 479 6401; fax: +61 39 467 3053. E-mail address:
[email protected] 0388-0001/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.langsci.2007.04.002
32
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
There is also ‘zero derivation’, a root which is primarily a noun (or an adjective) having secondary function as a verb. If someone writes a review of a book, they review it (verb with same form as noun). But if they subject a book to scrutiny, they scrutin-ize it (verb derived from noun by means of -ize). When someone puts something into a store (noun) they store it (verb with same form as noun). Yet when a doctor sends a patient to hospital (noun) they hospital-ize them (verb derived from noun by -ize). This paper attempts to put forward principled answers to these and similar questions. We will see that the motivating factors are a combination of phonological form, semantics and etymology. These basic principles are likely to constitute the basis for explanation of word-class-changing derivations cross-linguistically. Before examining in detail derivations through suffixes -(i)fy and -ize, and zero derivation, we should mention that there are a number of minor processes for verb derivation. They include the following: • Suffix -ate (from Latin participial ending -atus). A number of verbalizations – with a fairly wide range of meanings – date from the 19th century, including vaccin-ate, hyphen-ate and orchestr-ate. There are also back-formations from nouns ending in -ation, such as legislate from legislation and aviate from aviation. This derivational process has only a little productivity today, mostly in American English. Suffix -ate continues to be used in joke formations similar to discombobulate ‘confuse, frustrate’ (which was in fact made up, in the USA, in the 1830s). • Prefix en- em- (also of Romance origin) forms verbs from a smallish number of nouns – for example, en-vision, en-slave, en-cage, en-tomb, em-power – and adjectives – en-large, en-rich. It means ‘make into’, ‘put into’, or ‘provide with’ but is no longer productive. • Suffix -en (of Germanic origin) creates verbs from just a few nouns (threat-en, fright-en) and from a fair number of adjectives. There are phonological and semantic factors conditioning which adjectives take -en. (1) The suffix is only added to roots ending in p, t, k, f, s, S , h, and d. (2) It is only used with verbs from the Dimension and Physical Property semantic types (for example, wid-en, deep-en, hard-en, light-en) plus quick-en from the Speed type and three from the Colour type – black-en, whit-en and redd-en. (Dixon, 1977: 32–7; 1982: 21–4.) See Section 4.2 below. • Germanic prefix be- survives – with a variety of meanings – in just a few verbs derived from nouns and adjectives, including be-head, be-devil, be-friend, and be-little. Today it has very limited productivity. • There are a number of pairs where the noun ends in a voiceless and the verb in a voiced fricative. For f/v we have belief/believe, half/halve and proof/prove. For h/ð there are wreath/wreathe plus noun /mauh/, verb /mauð/, both written mouth. And for s/z we have noun /haus/, verb /hauz/, both written as house. There is also a handful of pairs ending in t/d, including noun extent and verb extend, noun ascent and verb ascend. (Proof/prove, extent/extend and ascent/ascend are Romance forms with the remainder being Germanic.) • Nouns in English are more likely than verbs to be stressed on the first syllable. A number of noun/verb pairs of Romance origin involve a stress difference (sometimes with a corresponding shift in vowel quality); for example noun /0insLlt/, verb /in0sLlt/, insult; noun /0sLspekt/, verb /sEs0pekt/, suspect. (For a useful discussion see Ross, 1973.)
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
33
• In a number of instances both noun and verb show initial stress, but there is a difference in vowel quality; for example noun /0estimEt/, verb /0estimeit/, both written estimate. Others include implement and separate. (All are of Romance origin.) The purpose of the present paper is to explore the three productive verbalization processes – adding -(i)fy, adding -ize, and zero derivation – in order to explain and predict which is used when. Forms less rebarbative than de-nazi-fy and de-stalin-ize can be constructed on the names of recent national leaders. One naturally says de-reagan-ize (rather than *de-reagan-ify) and de-bush-ify (not *de-bush-ize). But what about Kennedy? Neither an -ize nor an -(i)fy derivation seems to come naturally here. Before asking why – and then deriving one or more verbs based on Kennedy, in Section 6 – we must examine the phonological and semantic properties of the -ize and -(i)fy derivational suffixes. 2. Phonological conditioning of -(i)fy A word ending in -(i)fy /-fai/ must have (with only very occasional and short-lived exceptions): • the antepenultimate syllable bearing primary stress; • the penultimate syllable, which is unstressed, typically having nucleus /i/, although this can be some other vowel; • the final syllable, /1fai/ bearing secondary stress. For example, beauti-fy /0bju:ti1fai/. There may be one or more unstressed syllables preceding the antepenult as in electri-fy /i0lektri1fai/. Many-(i)fy derivations may accept negative prefix de-, which is unstressed; thus, de-class-ify /di:0kla:si1fai/ and de-electri-fy /di:i0lektri1fai/. We can summarize the types of root to which -(i)fy is added. Note that the great majority are of Romance origin. (i) Disyllabic roots, with initial stress, ending in /i/; the suffix then has form /-1fai/. For example, glori-fy /0glO:ri1fai/ and beauti-fy. Some are most used in past participle form; for example, ladi-fied, gentri-fied, citi-fied, mummi-fied. A fair number have a consonant cluster before the /i/; these include ugli-fy, countri-fy, cockni-fy, dandify, yankee-fy, monki-fy. (ii) Other disyllabic forms with initial stress, ending in a vowel other than /i/; the suffix again has form /-fai/. There is only a small number of established -fy derivations of this form (although more can readily be created). Argu-fy /0a:gju1fai/, dates from 1751. During the time when Negro was a non-taboo word, we had Negro-fy, as in a quotation of 1859 from the OED: A ring in the nose, aided by all the brown unguent in the world, would have been powerless to negrofy the bold Saxon outline of his features. (Interestingly, there is an alternative verbalization, Negrify, which has the more common vowel /i/ before the -fy.) (iii) Added to monosyllables; the suffix then has form /-i1fai/. For example, class-ify /0kla:si1fai/; also gas-ify, jok-ify, speech-ify, fish-ify. The suffix can be added to monosyllables ending in a consonant cluster, as fals-ify, French-ify, sand-ify. There appear to be no established words in which -(i)fy is added to a monosyllabic word ending in
34
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
a vowel. However, these can readily be produced; for example, de-mao-ify /di:0maui1fai/. There is a shift in vowel quality between type /taip/ and typ-ify /0tipi1fai/. (iv) There are some instances of -ify added to a noun or adjective disyllabic root which bears initial stress and ends in a consonant. In the -ify form, primary stress must be shifted to the second syllable of the root so as to fall in the antepenultimate syllable of the word (a well-nigh inconvertible requirement for -(i)fy derivations). For example: acid solid
/0æsid/ /0sOlid/
acid-ify solid-ify
/E0sidi1fai/ /sE0lidi1fai/
Also humid-ify, person-ify, subject-ify and object-ify. Diverse, /dai0vE:s/ has stress on the second syllable, and this is retained in divers-ify, /dai0vE:si0fai/. Similarly for intense/intens-ify. (v) Verbs have been created from a number of adjectives ending in -ic by omission of the final /k/ – electri-fy /i0lektri1fai/ from electric /i0lektrik/, syllab-ify from syllabic, and de-tox-ify from toxic. (vi) English suffix -(i)fy goes back to French -fiar and Latin -ficare. Some -(i)fy forms have been borrowed entire from Romance languages. They include terrify, liquefy, purify, justify, quantify and personify. (Others mentioned under (iv) – acid-ify, solid-ify, object-ify, subject-ify and humid-ify – were created within English.) As already mentioned, many -(i)fy verbs may take negative prefix de-; one can say deglori-fy, de-myth-ify, de-class-ify. Indeed, some -(i)fy derivations are only used with de-; we have mentioned de-toxi-fy and de-nazi-fy. Many (perhaps all) -(i)fy verbs form nominalizations with -ication; for example, glori-fi-cation, class-ifi-cation, pur-ifi-cation, quanti-fi-cation. (Some use -action; for instance rar-ify, with alternative nominalizations rar-if-action and rar-ifi-cation.) Primary stress goes on the -a- /ei/ of -cation, with the original primary stress now becoming secondary stress. Thus glori-fy /0glO:ri1fai/ yields glori-fi-cation /1glO:rifi0keiSEn/. In this derivation there are two unstressed syllables between secondary and primary stresses. And these -cation nominalizations may combine with de-, which is always unstressed, as in de-class-ifi-cation/di:1klæsifi0keiSEn/, and de-nazi-fi-cation. (See Dixon, 2005: 340–1.) 3. Phonological conditioning of -ize The English suffix -ize /-aiz/ relates to French -iser, Latin -izare and Greek -izo. In British English it is often spelled -ise (following French orthography) but American English uses only -ize and the OED standardizes on this spelling. Like those ending in -(i)fy, words ending in -ize have a preferred shape: • the antepenultimate syllable should bear primary stress; • the underlying stem (without -ize) should end in a consonant; • the final syllable, ending in /1aiz/, bears secondary stress. For example, moral-ize /0mOrE1laiz/. As with -(i)fy, there may be an unstressed syllable before the antepenult; for example material-ize /mE0tiriE1laiz/. The antepenultimate stress
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
35
rule is less rigid for -ize than for -(i)fy. Primary stress may go on to the fourth syllable from the end, with two unstressed syllables intervening between this and secondary stress on the final /aiz/, as in institutional-ize /insti0tju:SEnE1laiz/. The types of words to which -ize may be added can be summarized as follows. (As with -(i)fy, the great majority are of Romance origin.) (vii) A polysyllabic noun or adjective ending in a liquid (/l/ or /r/) or in a nasal (/m/, /n/ or /N/). Quite a number of these have final -al (which was in Latin dissimilated to -ar after an l) or -an, showing their Romance origins. For example: • ending in -l: national-ize, spiritual-ize, scandal-ize, capital-ize, legal-ize, specialize, cannibal-ize, symbol-ize, tranquill-ize, and many more; • ending in -r: familiar-ize, circular-ize, popular-ize, character-ize, bowdler-ize, pressur-ize, secular-ize, and many more; • ending in -n: urban-ize, christian-ize, Belgian-ize, union-ize, section-ize, carbonize, patron-ize, revolution-ize, Latin-ize; and many more; • ending in -m: victim-ize, solemn-ize, item-ize, atom-ize, and some others; • ending in -ó: very few commonly occurring words. But this type can readily be formed; for example, Keating-ize (based on the name of a former Australian Prime Minister, Paul Keating). A few words ending in a liquid change vowel quality when -ize is added. For example, satire /0sætaiE/, satir-ize /0sætE1raiz/, and sterile /0sterail/, steril-ize /0steri1laiz/. We noted that adding -(i)fy may induce shift of stress to adhere to the template for this derivation, as in /0sOlid/ to /sE0lidi1fai/. This does not happen with -ize. Although most derivations have penultimate stress, if stress occurs in the underlying stem on a syllable which would be earlier than the antepenult, then it remains. As mentioned before, we then get two unstressed syllables between primary stress and the secondary stress on final /aiz/, as in American-ize, /E0merikE1naiz/ and also in monophthong-ize /0mOnEfhON1gaiz/. (viii) A polysyllabic noun or adjective ending in -ic, /ik/. There are two ways in which a verbalization with -ize is achieved, both producing a word with the canonical stress profile. One involves dropping the -ic, the other retaining it. (viii-a) Omit the -ic syllable, and add -ize. This applies when the original root, to which -ic was added, was disyllabic and had initial stress. Adjective-deriving suffix -ic requires that the preceding syllable be stressed. When -ic is omitted and -ize added, stress reverts to the initial syllable of the root. For example: DISYLLABIC ROOT WITH dogma /0dOgmE/ magnet /0mægnEt/ INITIAL STRESS ADD
-(t)ic,
STRESS
dogma-tic /dOg0mætik/
magnet-ic /mæg0netik/
SECOND SYLLABLE
(THE SYLLABLE IMMEDIATELY PRECEDING REPLACE
-ic
-ic) -ize,
BY
STRESS REVERTS TO INITIAL SYLLABLE (THE ANTEPENULT IN THE WORD)
dogmat-ize /0dOgmE1taiz/ magnet-ize /0mægnE1taiz/
36
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
There are quite a few others of this type, including drama/drama-tic/dramat-ize, system/system-atic/systemat-ize, trauma/trauma-tic/traumat-ize. A similar pattern is found in hypnotic /hip0nOtik/, hypnotize /0hipnE1taiz/ but here there is no simple underlying noun (hypnosis does not fit the pattern). A variant is: democrat /0demEkræt/
democrat-ic /1demE0krætik/
democrat-ize /di0mOkrE1taiz/
The noun democrat bears initial stress. In adjective democrat-ic, primary stress moves to the third syllable, the one immediately preceding -ic (as required by this suffix). In democrat-ize, stress moves to the second syllable, the antepenult, to fit the preferred stress pattern of -ize forms. A similar pattern is found in anaesthetic /1ænis0hetik/, anaesthetize /æ0ni:shi1taiz/ (there is here an irregular relation between noun anaesthesia and adjective anaesthetic). (viii-b) A number of -ic forms have no corresponding noun, and then the -ic syllable is retained, -ize being added after it (with the /k/ of -ic /-ik/ being palatalized to /s/ before /aiz/). For example: romantic italic
/rE0mæntik/ /i0tælik/
romantic-ize italic-ize
/rE0mænti1saiz/ /i0tæli1saiz/
A number of adjectives ending in -ic have no underlying nominal root and consist only of two syllables. The -ic is retained (again becoming /is/ before /aiz/) and stress remains on the initial syllable, thus satisfying the stress requirements for both -ic and -ize. For example: public
/0pLblik/
public-ize
/0pLbli1saiz/
Similarly for critic/critic-ize. The adjective politic /0pOlitic/ does not have stress on the syllable preceding -ic; but it is not an adjective derived from a noun by adding -ic. Suffix -ize is added to politic (retaining the -ic), giving politic-ize /pO0liti1saiz/ with primary stress moving to the antepenult, the preferred stress pattern for -ize derivations. The behaviour of derivations based on poet /0pouit/ is interesting. Since the noun is disyllabic, we get adjective poet-ic with second syllable stressed, /pou0etik/. It falls into set (viii-a) so that the derived verb should be poet-ize /0poue1taiz/. This is attested, from 1581. But from 1804 a competitor arose, poetic-ize /pou0eti1saiz/, an irregular exemplar of set (viii-b). Both verbalizations are in use today. (ix) A noun of at least three syllables ending in /i/. These often refer to a speech act or have abstract reference. The typical profile for this set is a trisyllabic root, with primary stress on the initial syllable (which is the antepenult), ending in orthographic -y (which in almost all cases relates to French orthographic -ie). The final -y /i/ is replaced by -ize /-aiz/, which bears secondary stress. For example:
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
memory harmony agony eulogy
/0memEri/ /0hA:mEni/ /0ægEni/ /0ju:lEdZi/
memor-ize harmon-ize agon-ize eulog-ize
37
/0memE1raiz/ /0hA:mE1naiz/ /0ægE1naiz/ /0ju:lE1dZaiz/
Others of this ilk include energy, elegy, botany, rhapsody, sympathy, synchrony, plagiary, fantasy, jeopardy. There are also some with more concrete meaning, such as subsidy, scrutiny, colony, deputy. Also in this set are stems of four syllables the last of which has nucleus /i/. Stress goes on the second syllable (which is the antepenult) and is retained in the -ize derivation, this again conforming to the preferred stress pattern for -ize forms. For example: economy monopoly
/i0kOnEmi/ /mE0nOpEli/
econom-ize monopol-ize
/i0kOnE1maiz/ /mE0nOpE1laiz/
Also apology, anthology, mythology, philosophy and (with final orthographic e /i/, relating to its ultimate source in Greek) epitome. For category, dictionaries often give pronunciation /0kætigEri/, with four syllables, carrying over into categor-ize /0kætigE1raiz/. This has a non-canonical profile, with stress on the pre-antepenult. Such pronunciations may apply in some circumstances, particularly in the USA. But a very common pronunciation is as /0kætigri/, with three syllables, which then forms categor-ize /0kætig1raiz/, a derivation which does show the canonical stress pattern. Exactly the same remarks apply for military and militar-ize. Theory is a most interesting word. It may be pronounced with three syllables /0hi:Eri/, or with two, /0hi:ri/. From the disyllabic form one would expect the verbalization to be theor-ify /0hi:ri1fai/, in set (i) of the -(i)fy discussion above. From the trisyllabic form we would expect to derive theor-ize /0hi:E1raiz/. This is the occurring form and it is likely that – irrespective of how the underlying noun theory is treated – theor-ize is normally pronounced with three syllables, including unstressed /E/ between the initial stressed and the final unstressed syllables; that is /0hi:E1raiz/ rather than /0hi:1raiz/. When one plays around with -(i)fy derivations, those for most items from set (ix) sound absolutely unacceptable – *botan-ify, *eulog-ify, *jeopard-ify, and so on. In contrast, theor-ify doesn’t sound too bad. This verbalization is in fact attested, but only with marginal occurrence. It appears that the disyllabic pronunciation of theory is gradually taking over. In time, this may lead to theor-ify becoming more common, at the expense of theor-ize. (x) A polysyllabic noun or adjective ending in d, t or s. In contrast to sets (vii–ix), this type of -ize derivation is not at all common. There are a few disyllabic roots, ending in d which take -ize; for example, standard /0stændEd/, standard-ize /0stændE1daiz/. Also with hybrid, bastard, method, liquid, Soviet. Legitimate /li0dZitimit/ may add -ize, giving legitimat-ize (attested from 1791) /li0dZitimE1taiz/. This has major stress on the pre-antepenult and does not conform to the preferred stress pattern for -ize derivations. As a consequence, a shorter form arose
38
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
(from 1848), legitim-ize /li0dZiti1maiz/, which does show the canonical stress pattern. Both are in use today, although legitim-ize is gaining in frequency over its older rival. Nouns synthesis and emphasis drop their final -is when taking -ize, in order to produce a form with the preferred stress pattern – synthesis /0sinhEsis/, synthes-ize /0sinhE1saiz/. There is one totally irregular -ize derivation – propagand-ize /1prOpE0gændaiz/. Here stress is kept on the same syllable as in the noun propaganda /1prOpE0gænda/. The resulting -ize form has stress on the penultimate, rather than on antepenultimate and final syllables. The proper derivation should be something like /prO0pægEn1daiz/. (Try saying this aloud; it is scarcely recognizable as related to propaganda /1prOpE0gænda/.) It was noted that many of the words to which -(i)fy is added end in a consonant cluster – monosyllables such as ugl-ify, countr-ify, fals-ify, sand-ify, and disyllables such as object-ify, intens-ify. In contrast, -ize is added almost exclusively to forms ending in a single consonant. The main exceptions are categ(o)r-ize /0kætig1raiz/ and milit(a)r-ize. Plus final -rd – in bastard-ize, standard-ize, jeopard-ize – for dialects which preserve an /r/ pronunciation here. Nominalizations in -ation may be formed from many -ize verbalizations; for example union-iz-ation /1junjEnai0zeiSEn/. Just like -(i)fi-cation nominalizations, there is primary stress on the penultimate syllable, preceded by two unstressed syllables and then secondary stress on the fifth syllable from the end. And, as with -(i)fy, the negative prefix de-, which is unstressed, can be added to many -ize verbalizations and also to many subsequent -iz-ation nominalizations. For example de-material-ize /di:mE0tiEriE1laiz/ and de-material-iz-ation /di:mE1tiEriElai0zeiSEn/. This last commences with two unstressed syllables, followed by one with secondary stress – the same as that showing primary stress for the underlying noun material /mE0tiEriEl/ – then two unstressed syllables followed by /0zei/ with primary stress and finally unstressed /SEn/. 4. The semantics of verbalization 4.1. Verbs from nouns Looking first at nouns, the great majority of those verbalized through -(i)fy or -ize (or suffix -ate or prefix en-) are of Romance origin. Many more nouns have secondary function as verb without any change in form (this is ‘zero derivation’); they include nouns of Germanic and of Romance origin, in about equal numbers. The relations of meaning between a noun and a verb derived from it cover a fair semantic range. Some of the recurrent meaning relations will be summarized. The great majority of derived verbs are transitive; it is convenient to use A to indicate transitive subject and O for transitive object. A few are intransitive; S indicates intransitive subject. P indicates a peripheral (non-core) argument. X is used to indicate the underlying noun. (a) A behaves towards O as an X would. For example, Mary nursed John indicates that Mary behaved towards John as a nurse would to a patient. Others include bully, butcher, tailor, doctor, boss, host, pilot, referee, shepherd, parrot, dog. A classic pair showing a difference within this type is mother (a child), used to describe someone looking after a child in the way that a mother would and father (a child), which refers to providing (through sexual intercourse) the semen which leads to the conception of a child.
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
39
(b) A treats O as if O were an X – baby, fool. (c) A makes O into an X – cripple, martyr, heap, group, list, knot, cash. Related to these is verb panic ‘make get into a panic’. (d) A puts O in/on X – air (the clothes), table, market, store, can, seat. A subtype of this is where X is used to transport O, as in cart, ship. (e) A puts X in/on O – water, sugar, butter, grease, poison, label, muzzle (dog), man (a ship), bridge (a gap), enamel (a plate). (f) A gets X from O – milk (cow), core (apple), skin (animal), stone (cherry), weed (the garden), tax (citizens). (g) A uses X on O in the way that X is generally used – bayonet, club, knife, axe, iron, ski, brush, comb, mop. Note that there is a second verb based on stone: ‘throw stones at’, as in stone the christians; compare with ‘take stone out of the middle of’ as in stone the cherries. A subtype involves a body part as X, as in elbow (someone out of the way), finger (the precious diamond necklace), and eye (some interesting object). (h) A assigns X to O – price, name, title, date, time, limit, average, total. (i) S (an animal) produces X (a child) – (a cow) calves, (a sheep) lambs, (a mare) foals. (j) S picks X – blackberry, as in she is going blackberrying. This list is by no means exhaustive. There are many minor patterns. A small sample is: • • • • • •
shop for P, ‘look in shops for P’ puzzle over P, ‘think about P, which is a puzzle’ mistake O for P, ‘think O is P but this is a mistake’ chance to meet O, ‘meet O by chance’ fish for compliments, ‘try to attract compliments as an angler does fish’ gesture to someone ‘make a gesture to someone to attract their attention to something’ (motion to someone is similar) • soldier on, ‘continue on with a task despite difficulties, as a soldier might continue on in battle’ (a metaphorical sense)
Turning now to verbs derived from nouns with -ize, we find there are a few which accord with (a–e) above. They include: (a) A behaves towards O as an X would – burglar-ize (a house), patron-ize, poet(ic)-ize, bowdler-ize (a salacious literary work, in the way that Dr Thomas Bowdler expurgated Shakespeare ‘for family reading’). (b) A treats O as if O were an X – a rather metaphorical use of lion-ize (a celebrity) ‘treat as special and unusual (as a lion is among animals)’. (c) A makes O into an X – victim-ize, pauper-ize, liquid-ize, carbon-ize, crystal-ize, magnet-ize, union-ize, colon-ize, dramat-ize, novel-ize, antholog-ize and fantas-ize. (d) A puts O in/on X – hospital-ize (a patient), computer-ize (a set of data), memor-ize (some fact). (e) A puts X in/on O – macadam-ize – in the UK – or bitumen-ize – in Australia (a road). However, most -ize derivations – and also almost all those employing -(i)fy – do not correspond to the recurrent meaning relations just listed for zero derivations. Many -ize and -(i)fy verbs relate to affecting the inherent nature of something. Indeed, this
40
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
(k)
(l)
(m)
(n)
(o)
can be seen by comparing the examples quoted under (e). One can water the plants, sugar the strawberries, man the boats, or bridge a gap, but in each case the added thing remains distinct from that to which it is added. With the -ize derivation macadam-ize, the macadam layers become an indissoluble part of the road. Further semantic relations between a noun and its -ize derivation can now be listed: A provides X to/for O – summar-ize, subsid-ize, eulog-ize. A couple of intransitive verbs also belong here. If S offers an apology to P, then S apolog-izes to P; and if S expresses sympathy for P, then P sympath-izes with P. And eleg-ize may be used either intransitively or transitively. Also in this set are energ-ize, harmon-ize, militar-ize and motor-ize – if an army is motor-ized, it is equipped with motor vehicles. A subjects O to X. If A applies pressure to O, then A pressur-izes O. If A subjects O to criticism/critical remarks, then A critic-izes O. Similarly for satir-ize, scandal-ize, scrutin-ize, jeopard-ize. Traumat-izing O implies doing something such that O goes into a trauma. Agon-ize is an intransitive verb; if S agonizes over P, then S subjects themself to agony concerning P. A represents X of/for O. For example, if A is a symbol of O, then A symbol-izes O, as in The fall of the Berlin Wall symbol-ized (that is, represented a symbol of) the end of the Cold War. Also character-ize, epitom-ize, signal-ize. S achieves X. If someone econom-izes on household expenditure then they achieve economy in household expenditure. Also in this set is: A achieves X over O – if someone monopol-izes the cardboard box market, then they achieve a monopoly over the cardboard box market. S follows the pursuit of X – botan-ize, philosoph-ize, theor-ize, and so on.
Moral-ize ‘indulge in moral judgements’ may also be placed in this set. Note that the verb moral-ize is typically used in a disparaging way, implying that the moralizer is acting in a self-righteous and perhaps hypocritical manner. This is an example of the general principle that every derived form has its own character and meaning, seldom being simply the association of the meaning of the underlying root or stem plus the semantics of the derivational process involved. As with zero derivation, the listing just produced is not exhaustive. For instance, it does not include (among others): • deput-ize for ‘act as deputy for’. • woman-ize ‘pursue women for sexual purposes (in a manner considered socially unacceptable)’. And the verb cannibal-ize (for example, a machine) ‘use parts of one machine to repair another’ appears to have only glancing semantic relation to the underlying noun cannibal. Verbs derived with -(i)fy typically mean ‘(make) take on the character of X’. If someone who comes from the country is said to be citi-fied, this implies that they have assumed some of the habits characteristic of city life. Preach-ify can be used to describe someone who talks as if preaching – their way of talking is reminiscent of the way a preacher delivers a sermon – or else to describe a sermon or series of sermons which are tedious; in both senses the verb has negative connotations.
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
41
There are fewer derivations with -(i)fy than with -ize or zero. Some do fit into the sets presented above. (c) (d) (k) (m)
A makes O into an X – mummi-fy, acid-ify, lique-fy, gas-ify, fish-ify. A puts O in/on X – class-ify, cod-ify. A provides X to/for O – glori-fy, beauti-fy, electri-fy, dirti-fy. A represents X of/for O – exempl-ify, sign-ify. A further set of verbs created by -(i)fy is similar to (c) but has a significant difference. (p) A makes O have (some of) the characteristics of X – Yankee-fy, Cockney-fy, countrify, citi-fy, ladi-fy, gentri-fy, dandi-fy, yuppi-fy, monki-fy (or monkey-fy), object-ify, person-ify. (Some of these are most often used in passive participle form, e.g. ladified.)
In many circumstances, -(i)fy and -ize have the same meaning, their use being conditioned by the phonological nature of the form which is to be verbalized, as set out in Sections 2 and 3. Thus, one can only gas-ify ‘change into gas’, not *gas-ize, and only colon-ize ‘make into a colony’, not *colon-ify. But beyond these instances, there is a tendency for a verb with -(i)fy to have a meaning ‘something like X’ rather than just plain ‘X’. If someone jokes, there are likely to be telling jokes, or making a joke out of something, whereas if they jok-ify they may just be chatting in a jokey sort of way to liven up a conversation. If one says that someone has dirtied an object, there is an implication that it should be kept clean and is now dirty which is not at all a good thing. But if someone is said to have dirtified it, then it is likely that they made it a little bit mucky but this does not really matter at all. A particularly exotic coinage is hungr-ified. Saying I feel a bit hungr-ified can mean that I am just a trifle peckish, as opposed to the pragmatically powerful I’m hungry. Note though that all of these recent coinages have a wide range of meaning. Hungr-ified can also be used for ‘very hungry’, and dirti-fy for ‘do the dirty on, play a dirty trick on’. These are jocular – often termed ‘slang’ or ‘colloquial’ – uses of the -(i)fy derivation. They may be based on nouns or adjectives, and even on verbs. Perhaps the most established of the latter is argu-fy (first documented in 1751) – this can refer to a disagreement about something, less serious than a full-blown argument. Alternatively, argu-fy can be used to describe arguing endlessly and tediously. Similarly, speech-ify (this from a noun) can refer to adopting an arrogant speech-making-like tone in the course of everyday conversation; or else to continuously making formal speeches of a long and boring nature. In its jocular or informal sense, -(i)fy is increasingly productive. One can farm-ify a piece of land, which means undertake a bit of farm-type work on it. One can hom-ify a temporary residence, which might involve setting up some familiar photographs to make it seem more like home. List-ify can be used for doing a bit of listing, perhaps not too systematically or seriously. An object can be gold-ified or silk-ified a bit. And so on and on. 4.2. Verbs from adjectives A verb derived from an adjective, X, will mean either ‘S becomes X’ or ‘A makes O become X’. As mentioned in Section 1, suffix -en derives verbs from simple adjectives in a limited number of semantic types, provided that they end in p, t, k, f, s, S , h, or d. These can generally be used either intransitively or transitively – for example The odds on the favourite shortened or The bookmakers shortened the odds on the favourite. Some of those
42
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
that lack a final consonant to which -en may be added can function as a verb with zero derivation – narrow, slow, cool, dry, smooth, calm, empty and (with a limited sense) thin. Brave only forms a transitive verb with the meaning ‘be brave to do something’, as in He braved the snow to attend that lecture ‘he was brave to go through the snow to attend that lecture’. As verbs, clean, blind, tame, and free are also restricted to transitive function. An unusual feature of English is the opposition of adjectives beauti-ful and ugly, and nouns beauty and ugli-ness. The roots are noun beauty (a loan from Old French) and adjective ugly (from Old Norse). Verbs ending in -(i)fy are typically formed from disyllabic nouns ending in /i/, hence beauti-fy (attested from 1526). By analogy, verb ugli-fy (1576) was created from adjective ugly. Much later, verb pretti-fy (1850) was derived from Germanic adjective pretty. Other verbs ending in -(i)fy which relate to adjectives were borrowed entire from French – fals-ify, pur-ify, simpl-ify and solid-ify. Note that most verbs derived from adjectives through -(i)fy are only used transitively. (Intens-ify is an exception, having both transitivity values.) Tranquill-ize, perhaps the only verb formed by means of -ize from a simple adjective, is also restricted to transitive function. Suffix -ize is extensively used to derive verbs from adjectives which are synchronically polymorphemic; for example fiction-al-ize ‘make become fictional/a work of fiction’, west-ern-ize ‘make become western/have characteristics of the west’; also margin-al-ize, christ-ian-ize, circul-ar-ize, spirit-ual-ize, America-n-ize. In addition, -ize is used with adjectives which, while not being analyzable within the modern language, include an adjectivizing suffix of Romance origin – legal-ize, popular-ize, equal-ize, rational-ize, urban-ize and many more. These -ize verbalizations are like -(i)fy derivations from adjectives (but unlike -en forms) in that they are almost all restricted to transitive use. An exception is materialize, which is typically (but not exclusively) intransitive, as in A lucky opportunity materialized for her. Since the meanings of verbs derived from adjectives by means of -ize are varied, it is useful to classify them as relating to: • • • • • • • • •
• •
– equal-ize, decimal-ize, metric-ize – internal-ize, external-ize, local-ize, global-ize, circular-ize SCOPE – popular-ize, particular-ize, special-ize, general-ize, marginal-ize, personal-ize SOCIAL CONVENTIONS – legal-ize, legitim(at)-ize, criminal-ize, western-ize, urban-ize, democrat-ize, vulgar-ize CONFORMING TO NORM – regular-ize, standard-ize, rational-ize, formal-ize RELIGION – spiritual-ize, pagan-ize, secular-ize, solemn-ize EXPRESSION – fictional-ize, dramat-ize, verbal-ize, material-ize NATION/LANGUAGE – American-ize, German-ize PERSON/IDEOLOGY – de-stalin-ize, Christian-ize We also find -(i)fy verbalizations for the last two headings – French-ify, de-nazi-fy and so on. Whether -ize or -(i)fy is used for a verbalization based on a proper name is basically determined on the basis of phonology. This is discussed in Section 6. There are, as might be expected, a few idiosyncratic -ize verbs derived from adjectives. They include: Familiar-ize. This is typically used reflexively; if one familiar-izes oneself with something, one makes oneself familiar with it. Penal-ize. It could be said that adjective penal means ‘involving punishment’ and verb penal-ize ‘impose punishment’, but the types of punishment differ significantly. A penal QUANTITY LOCATION
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
43
sentence is imprisonment and a penal institution is a prison. But to penal-ize someone is not to send them to prison, rather to impose a penalty, as in The teacher penal-ized her ten marks for not handing in the final assignment.
5. The ‘how’ and the ‘why’ Whether a noun or adjective forms a derived verb – and, if so, whether by zero derivation, by -(i)fy, by -ize, or by one of the other processes described in Section 1 – depends on three factors: • The meaning of the noun or adjective. See the list of meanings associated with zero derivation, -(i)fy, -ize, and -en, enunciated in Section 4. • The phonological form of the noun or adjective. • Its etymology. Some roots were borrowed from Romance languages but their verbal derivations developed within English; they include cod-ify and popular-ize. For others, both root and verbalization were borrowed, as with pure and purify, liquid and liquefy, harmony and harmonize, solemn and solemnize. We first summarize the phonological conditioning factors, described in Sections 2 and 3. And then put forward a number of explanations, while examining contrasting verbalizations.
5.1. Integrated summary of phonological conditioning The examples of zero derivation listed in the first part of Section 4 were almost entirely monosyllabic and disyllabic forms. They included two words which are typically shown as trisyllabic in dictionaries but which are in fact frequently pronounced as disyllables: referee /re0fri:/ and average /0ævridZ/. There is also the verb blackberry, a compound with one monosyllabic and one disyllabic component. Other nominal compounds can also be used as verbs; for example, These hens free-range. Overall, zero derivation verbalizations are typically formed from either Germanic or Romance forms which have one or two syllables, or compounds involving such roots. In Section 2, the phonological conditioning of -(i)fy was summarized, and in Section 3 that of -ize. (Recall that both are used predominantly with Romance roots.) We can now integrate these conditioning environments. It was noted that-(i)fy verbs require primary stress on the antepenultimate syllable and that -ize derivations prefer this. Since -ize/-1aiz/ is a single syllable, this suffix is prototypically attached to a disyllabic form with initial stress – legal /0li:gEl/ and legal-ize /0li:gE1laiz/. When added to a trisyllabic form, with initial stress, ending in -y /i/, -ize replaces the -y – fantasy /0fæntEsi/, fantas-ize /0fæntE1saiz/. The other suffix has two forms. It is disyllabic -ify /i1fai/ when following a monosyllabic root – code /0koud/, cod-ify /0koudi1fai/. And it is monosyllabic -fy /1fai/ when added to a disyllabic root ending in /i/ – beauty /0bju:ti/, beauti-fy /0bju:ti1fai/. Now to summarize the phonological conditioning for -(i)fy and -ize. (Item numbers (i), (ii), . . . from Sections 2 and 3 are cross-referenced here.)
44
R.M.W. Dixon / Language Sciences 30 (2008) 31–52 MONOSYLLABIC ROOT,
(iii).
Can only take -ify, as in gas-ify. POLYSYLLABIC ROOT ENDING IN A VOWEL
– DISYLLABIC ROOT, WITH INITIAL STRESS, ENDING IN /i/, (i). Add -fy /fai/, as in countri-fy /0kLntri1fai/. –
/i/, generally relating to orthographic -ie in French, (ix). Replace final /i/ by -ize /aiz/, as in sympathy /0simpEhi/, sympath-ize /0simpE1haiz/. These can be trisyllabic nouns, with initial stress – such as sympathy – or quadrisyllabic forms with primary stress on the second syllable – as apology /E0pOlEdZi/, apolog-ize /E0pOlE1dZaiz/. Two disyllabic forms which bear initial stress and end in a vowel other than /i/ were mentioned under (ii) in Section 2; suffix -fy is attested in Negro-fy and argu-fy. In fact, verbs formed from nouns or adjectives ending in a vowel other than /i/ are very scarce. How proper names ending in a vowel are treated is discussed in Section 6. ROOT OF THREE OR MORE SYLLABLES, WITH ANTEPENULTIMATE STRESS, ENDING IN
POLYSYLLABIC STEMS ENDING IN A LIQUID
(/l/, /r/)
OR IN A NASAL
(/n/, /m/
OR
/N/), (vii).
Placement of primary stress in the underlying form is maintained. If the underlying form is disyllabic, with initial stress, the preferred antepenultimate stress pattern for an -ize derivation is achieved, as in victim /0viktim/, victim-ize /0vikti1maiz/. The underlying form may have more than two syllables but show penultimate stress; the preferred stress pattern will again eventuate, as in editorial /1edi0tO:riEl/, editorial-ize /1edi0tO:riE1laiz/. Or it may have primary stress on an earlier syllable, as in decimal /0desimEl/, decimal-ize /0desimE1laiz/; this results in a non-preferred pattern, with primary stress on the preantepenult, separated by two unstressed syllables from secondary stress on the final /1laiz/. The corpus assembled for this study includes one disyllabic form ending in /n/ which takes -ify rather than the expected -ize – person /0pE:sn/, person-ify /pE0sOni1fai/, (iv). Note that primary stress has here had to shift to the second syllable to satisfy the penultimate stress condition on -(i)fy verbalizations. This is discussed in Section 5.2. -ic /ik/. -ic) IS DISYLLABIC AND BEARS INITIAL STRESS, (viii-a).
POLYSYLLABIC STEMS ENDING IN
–
ORIGINAL ROOT (WITHOUT
Replace final -ic /ik/ by -ize /aiz/ as in drama /0drA:mE/, dramat-ic /drE0mætik/, dramatize /0dræmE1taiz/. (If the original root had more than two syllables with initial stress, then -ize again replaces -ic and stress goes on the antepenult, different from its position in the underlying root. Under (viii-a) this was illustrated for democrat /0demEkræt/, democrat-ic /1demE0krætik/, democrat-ize /di0mOkrE1taiz/.) –
IF THERE IS NO UNDERLYING ROOT (WITHOUT
-ic), (viii-b).
Add -ize /aiz/ after -ic (which assimilates to /is/ before /aiz/), as in critic /0kritik/, criticize /0kriti1saiz/. The preferred stress pattern pertains.
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
45
There are just a few adjectives ending in -ic, which replace the final -ic with -ify. These were set out under (v) – electri-fy /i0lektri1fai/ from electric /i0lektrik/, syllab-ify from syllabic, and de-tox-ify from toxic. In each instance the -ic form has penultimate stress and so replacing final /ik/ with /i1fai/ – or final /k/ with /1fai/ – results in a verbalization with antepenultimate stress. It seems that all three verbalizations entered English from Mediaeval or Post-Mediaeval Latin, so that the use here of -(i)fy – rather than the expected -ize – has an etymological explanation. POLYSYLLABIC STEM ENDING IN A STOP OR FRICATIVE. There are a fairly small number of nouns and adjectives of this type which may be verbalized.
–
DISYLLABIC STEM, WITH PRIMARY STRESS ON SECOND SYLLABLE,
(iv).
Add -ify. Stress is retained on the second syllable, satisfying the antepenultimate requirement for an -(i)fy verbalization. For example subject /sEb0dZekt/, subject-ify /sEb0dZekti1fai/; also object-ify, intens-ify, divers-ify. Note that these roots could not take -ize /aiz/ unless stress were moved to the initial syllable, something of which there is no instance. Note also that two of the words end in a consonant cluster /kt/. We remarked that -ify is often added after a consonant cluster but that -ize scarcely ever is. –
DISYLLABIC STEM, WITH PRIMARY STRESS ON INITIAL SYLLABLE,
(iv) and (x).
There appear here to be two competing mechanisms: • Add -ify and move primary stress one to the right, as in humid /0hju:mid/, humid-ify /hju:0midi1fai/. Also acid-ify, solid-ify. • Add -ize and retain stress, as in liquid /0likwid/, liquid-ize /0likwi1daiz/. Also standard-ize, hybrid-ize, bastard-ize, method-ize. The question then is: why do some verbs of this type take -ify and others -ize? The answer appears to be etymological. Humid-ify, acid-ify and solid-ify were borrowed entire from French. But for liquid, standard, hybrid, bastard and method, it was just the root which was borrowed from French (or Latin); creation of a verb by adding -ize happened at a later stage within English. This indicates that -ize is the productive verbalization for disyllabic roots ending in a stop and bearing stress on the initial syllable. In summary, we have seen that: • -(i)fy is a productive verbalizer with: – monosyllabic forms; these take -ify /i1fai/; – disyllabic forms, with initial stress, ending in /i/ (and probably also those ending in other vowels); these take -fy /1fai/; – disyllabic forms ending in a stop or fricative, with stress on the second syllable; these take -ify /i1fai/. • -ize, /aiz/ is a productive verbalizer with: – forms ending with a liquid or nasal; – disyllabic forms ending in a stop or fricative, if stress is on the initial syllable;
46
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
– forms ending in -ic /ik/; – forms ending in -y /i/ (corresponding to French orthographic -ie); here -ize /aiz/ replaces final y /i/. There are a number of exceptions to these general rules, some of which will be explained in the next section.
5.2. Some explanations and contrasts Why does the verbalization based on person /0pE:sn/ involve -ify, giving person-ify /pE:0sOni1fai/, which requires stress to shift to the second syllable? Why could not person take suffix -ize, as do patron, human and woman? The answer lies in the semantics of the verbalization. When a verb was first coined from person (the earliest written record is from 1728), it had the meaning ‘attribute human characteristics to’. This belongs to set (p) of Section 4.1, ‘A makes O have (some of) the characteristics of X’. Set (p) relates only to suffix -(i)fy, not to -ize. Thus, the verb created around the 1720s was person-ify, not person-ize. Later, the verb took on a second sense, ‘exemplify, represent’, which falls into set (o) of Section 4.1, that relates to both -(i)fy and -ize. If this second sense had come first, then on semantic grounds either -(i)fy or -ize could have been used, and it is likely that -ize would have been chosen since person has the appropriate phonological form for an -ize derivation. But the ‘characteristics’ sense was prior, and this serves to explain why the verb created was person-ify. (Note that there is another verb based on the derived adjective person-al, and here -ize is the suffix used, giving person-al-ize ‘make become personal’.) There are three major components of music – rhythm, melody and harmony. The verb harmon-ize ‘create harmony, make harmonic’ is of frequent occurrence whereas melod-ize and rhythm-ize (although they have been used) are not in common currency. Why? The likely explanation is pragmatic. Music consists basically of melody (a succession of notes) and rhythm (a pattern of beats or pulses). To these is added harmony (combinations of different notes sounded simultaneously). Given melody and rhythm, one can then harmon-ize. One would less often want to talk of melod-izing or rhythm-izing. Let us now look at a quite different semantic domain, the names of trade and professional people. Some are derived from verbs – bake-r from bake, teach-er from teach. Others are not. Of these, tailor, butcher, doctor, nurse, shepherd and engineer are basically nouns which have a secondary use as zero-derivation verbs. Why cannot milliner and haberdasher also function as verbs? Perhaps because they are too long; it was noted in Section 5.1 that the great majority of nouns which function as verbs through zero derivation have just one or two syllables. But if doctor can be a verb, why can’t dentist? And if butcher can be, why not grocer? (All these nouns are disyllabic.) We can note that the noun burglar (a sort of trade) does form verb burglar-ize. This throws up another question – why do we have burglar-ize when there is a perfectly good monomorphemic verb burgle? Here an answer can be provided. The activity noun burglary (c.1200) and agent noun burglar (c.1268) came from Romance. For more than six hundred years there was no commonly-used verb to describe the activity. Then, more or less simultaneously, two were created – verbalization burglar-ize is first attested (in the USA) in 1871, and the back-formation verb burgle in 1872. As mentioned in Section 4.1, -ize derivations typically describe ‘affecting the inherent nature of something’. In keeping with
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
47
this, burgle generally refers just to the stealing of goods, whereas burglarize is likely to be employed if there is accompanying violence, with the house being ransacked or trashed. Note that burglar has an appropriate form to take -ize – a disyllabic noun with initial stress, ending in r. Dentist, ending in a consonant cluster, does not qualify, which is probably why there is no dentist-ize. But grocer and lawyer have similar form to burglar. There seems no reason why, in the absence of a zero-derivation verb, there should not be verbalizations lawyer-ize and grocer-ize, parallel to burglar-ize – save that language is to some extent an artefact of human whim, rather than a mechanical device. Many of the explanations for ‘why this verbalization?’ are phonological. In Section 1 we enquired why one says central-ize, with -ize, but glorify, with -(i)fy. Simply because a disyllabic form with initial stress, ending in /l/ takes -ize, whereas a disyllabic noun, again with initial stress, ending in /i/ takes -(i)fy. In Section 1 we also brought up the question of why the verb corresponding to scrutiny is scrutin-ize but for review there is the zeroderivation verb review. This is because the form of scrutiny is appropriate for verbalization with -ize whereas that of review /rE0vju:/, with stress on the second syllable and final /u:/, is not appropriate for verbalization by either -ize or -(i)fy. Typically, a Germanic noun will form a verb by zero derivation, while a Romance loan of similar meaning will take -ize. For example: = NOUN shop cripple
VERB
VERB = NOUN PLUS -ize hospital-ize victim-ize
As mentioned before, there are a fair number of examples where a verbalization with (i)fy was borrowed from Romance, together with the underlying noun or adjective. These include: just and justify pure and purify liquid and liquefy
false and falsify simple and simplify terror and terrify
Now just, false and pure are monosyllables. If a verb were to be derived from them within English it could only be by -(i)fy and would have exactly the same form as the borrowed verb. Similarly for simple, which is disyllabic (the second syllable consisting of syllabic l) but could only form verb simpl-ify. However, liquid and terror are disyllabic. They could – and do – form a verb by the addition of -ize: liquid-ize and terror-ize. These have quite different characteristics: • Liquid-ize (1837) has virtually identical meaning to liquefy (1547) and may differ only in terms of context of use. The Cobuild English dictionary states that liquid-ize is used for making something liquid ‘through an electrical appliance’, yet many people use liquefy for this. Current Oxford English dictionaries state that liquefy is mainly used in chemistry, but in fact the usage is much wider. It seems that liquefy and liquid-ize have essentially the same meaning and are generally interchangeable. (There is also the verb liquidate, 1670, ‘put an end to’, which is based on Mediaeval Latin liquidare. Note that liquefy comes from French lique´fier.) • Terrify (1578), based on Latin terrificare, means ‘fill with terror, frighten’. Much later, a new verb terror-ize (1823) was created. At first it meant much the same as terrify (like
48
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
liquefy and liquidize) but by the 1850s a second sense emerged: ‘rule by terror, maintain power through a rule of terror’. There are a number of examples of a noun forming a verb with -ify and an adjective derived from the noun forming a verb with -ize. For example: NOUN
VERB
ADJECTIVE
VERB
myth verb person
myth-ify verb-ify person-ify
myth-ic verb-al person-al
myth-ic-ize verb-al-ize person-al-ize
Myth-ify (1873) and mythic-ize (1840) have very similar meanings ‘create a myth about someone or something, treat as a myth’. There is also the noun mythology (c.1420) from which is derived verb mytholog-ize (1603) ‘follow the pursuit of mythology, interpret with respect to mythology’ (along the lines of philosoph-ize, botan-ize and theor-ize). Verb-ify (1813) means ‘convert (for example, a noun) into a verb’; verbal-ize (1609) also has this sense, and can means besides ‘express in words’ (or ‘express in too many words, be verbose’). Person-ify and personal-ize were discussed before. Some apparent redundancies have an indirect etymological explanation. Why do we have both syllab-ify /si0læbi1fai/ and syllab-ize /0silE1baiz/ from syllable /0silEbl/? Well, syllable (c1384) and syllabize (1656) were taken from Romance, and then syllabization (1926) was created from syllabize. But, before this, syllabification (1838) had been adopted from Mediaeval Latin, and finally syllabify (1926) was back-formed from this. There are also examples of -ize applying to two forms of the same word. Middle English took from Latin colony (1382) and at a later stage derived from this colon-ize (1622) ‘make (some territory) into a colony’. Then came adjective colonial (1776) ‘relating to a colony’ and, based on this, colonial-ize (1864) ‘(make) take on the characteristics of someone living in a colony’. Suffixes -(i)fy and especially -ize are highly productive. What more natural than that they should take over from a suffix which is no longer fully productive, such as -ate? Adoption of domestic-ate (1639) from Mediaeval Latin was closely followed by the creation within English of domestic-ize (1656). Domesticate is still far and away the more popular but domestic-ize continues on. It is only slightly fanciful to predict that domestic-ize is likely to gradually take over in the centuries ahead. 6. Deriving verbs from names Why is it normal and natural to use French-ify (rather than French-ize) to describe translating something into the French language or making someone or something take on characteristics of the culture of France, but German-ize (rather than German-ify) with respect to German language and culture? The answer is phonological. French is a monosyllabic word, which takes -ify; German is a disyllabic word with initial stress ending in -n, naturally taking -ize. In each derivation, primary stress falls on the antepenult. (German-ify, with stress moved to the second syllable, and French-ize have been used, but very sparsely.) One can only readily verbalize a noun or adjective if it has an appropriate phonological form, as set out in Sections 2 and 3 and summarized in Section 5.1. In addition to a name
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
49
for each nation, there is also a noun describing an inhabitant, and an adjective describing the language or culture (the last two may have the same form). As set out in Table 1, a verb may be derived from one of these items, sometimes from more than one, depending on their phonological form. (Note that the table is not exhaustive; there are more examples for almost every set.) In set A of Table 1, adjectives French and Welsh and inhabitant names Dane and Scot are all monosyllabic forms and so take -ify. Danish, Scottish, Wales and Denmark are disyllabic but end in a sibilant or /k/, which are not well attested with -(i)fy or with -ize. Compounds such as Scotland, Frenchman and Welshman sound less felicitous when verbalized (with -ize) than do Scot-ify, French-ify and Welsh-ify. (France is a monosyllable; verbalization Franc-ify is attested but it is many many times less frequent than French-ify.) America, in set B, is trisyllabic, ending in a vowel, with stress on the second syllable. Mexico is trisyllabic with initial stress and final /ou/ while Germany is the same but with final /i/ (not related to French orthographic ie). None of these have an appropriate form to take a verbalizer, but their inhabitant/adjective forms all end in -an and readily verbalize with -ize. Language names in set C are disyllabic with stress on the second syllable, an appropriate form to be verbalized with -ify – Nepal /ni0pO:l/, Nepal-ify /ni0pO:li1fai/; Japan /dZE0pæn/, Japan-ify /dZE0pæni1fai/. A verb can be derived through -(i)fy either from nation name Turkey or from inhabitant name Turk, in set D; the derivations Turki-fy and Turk-ify fall together as /0tE:ki1fai/. Two verbalizations are also possible for sets E and F, but with different forms. In set E, nation name Israel /0izreil/ bears initial stress and forms verb Israel-ize /0izrei1laiz/, while inhabitant/adjective form Israeli /iz0reili/ has stress on the second syllable and gives verb Israeli-fy /iz0reili1fai/. For set F, the use of -(i)fy and -ize are reversed with respect to set E. Inhabitant/adjective form Russian takes -ize to form Russian-ize (first attested in 1831), while nation name Russia /0rLSE/ loses final /E/ and then adds -ify to derive verb Russ-ify /0rLSi1fai/. The OED ascribes the first written instance of this to Queen Victoria,
Table 1 Verbalizations based on nation names NATION
VERBALIZATION
INHABITANT
VERBALIZATION
ADJECTIVE
VERBALIZATION
A1 A2 A3 A4
France Wales Denmark Scotland
– – – –
Frenchman Welshman Dane Scot
– – Dan-ify Scot-ify
French Welsh Danish Scottish
French-ify Welsh-ify – –
B1 B2 B3
America Mexico Germany
– – –
American Mexican German
American-ize Mexican-ize German-ize
C1 C2
Nepal Japan
Nepal-ify Japan-ify
Nepalese Japanese
– –
D
Turkey
Turki-fy
E
Israel
Israel-ize
Israeli
Israel-ify
F
Russia
Russi-fy
Russian
Russian-ize
Turk
Turk-ify
Turkish
–
50
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
in 1865: Good Alice seems quite Russified. Both verbs are in use today, with Russ-ify generally being preferred over Russian-ize. Verbs are also formed from the names of prominent people and ideologies. Bush-ify means ‘make more closely conform to the ideas set forth by President Bush’. We earlier mentioned de-nazi-fy and de-stalin-ize which mean ‘divest of the insidious influence of Nazis/Stalin’. Here there is only one form as a possible candidate for verbalization. Zero derivation is virtually never employed with proper names, so whether there is a felicitoussounding verb depends on the phonological form of the name. For example (note that negative prefix de- may optionally be added to any of these): • •
•
•
•
•
• •
•
take -ify – Bush-ify, Blair-ify, Ford-ify, Burns-ify. /i/ take -fy. These can be disyllabic, as Trotsky /0trÅtski/, Trotski-fy /0trÅtski:1fai/, and (de-)nazi-fy. Or they can be longer, as Disraeli /diz0reili/, Disraeli-fy /diz0reili1fai/, Mussolini /1musE0li:ni/, Mussolini-fy /1musE0li:ni1fai/. DISYLLABIC NAMES, WITH INITIAL STRESS, ENDING IN A LIQUID, A NASAL, /d/ or /t/ take -ize – Carter-ize, (de-)hitler-ize, Reagan-ize, (de-)stalin-ize, Gladston-ize, Whitlam-ize, Howard-ize, (de-)soviet-ize. Names with a different phonological profile are less easily verbalized. For example: From Isaacs and Adams can be formed Isaac-ize and Adam-ize. The final s is here dropped, since -ize does not normally follow a sibilant or a consonant cluster. But note that the monosyllabic name Burns retains the final s in Burns-ify, since -ify may follow a consonant cluster. Consider the name of the long-serving Australian Prime Minister Menzies /0menziz/. One could say Menzi-fy /0menzi1fai/, dropping the s and adding -fy, or Menzies-ize /0menzi1zaiz/. Both appear possible, but both are less felicitous than prototypical verbalizations such as Bush-ify and Clinton-ize. Costello /kE0stelou/ has penultimate stress and ends in a vowel. Costello-fy is possible (similar to Negro-fy) but Costell-ify – with final /ou/ replaced by /-i1fai/ – sounds much better. Roosevelt is a disyllabic form with initial stress ending in a cluster. It is because of the cluster (disliked by-ize) that Roosevelt-ize sounds alright, but only just. Kennedy /0kenEdi:/ is trisyllabic with initial stress, ending in /i/. Neither verbalizing suffix goes well with this. One could use -fy, which would involve shifting stress to the second syllable. Because of the stress difference Kennedi-fy /ke0nidifai/ doesn’t sound all that similar to Kennedy /0kenEdi:/. It is not possible to replace the final -y /i/ by -ize /aiz/ – saying Kennedize, as in subsidy/subsidize – since it does not relate to orthographic ie in French. One could simply add -ize at the end of the whole name, Kennedy-ize /0kenEdi:1aiz/ but this sounds really awkward (it has a non-preferred stress pattern). The fact is that the phonological form of the name Kennedy is such that it cannot very easily be verbalized. The name of the preceding president, Eisenhower /0aizEn1hauE/, fares a little better. An -ize derivation, Eisenhower-ize, is awkward and scarcely acceptable. For an - (i)fy verbalization, primary stress must go on the antepenultimate syllable, giving Eisenhowerify /1aizEn0hauEri1fai/. This is recognizable as related to Eisenhower /0aizEn1hauE/, since there has simply been interchange of primary and secondary stresses between first and third syllables. MONOSYLLABIC NAMES
NAMES WITH PENULTIMATE STRESS ENDING IN
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
51
In conclusion, we saw that the principles for deriving verbs from common nouns and adjectives depend on a combination of etymology, semantics and phonological form. Just the phonological factors carry over into the verbalization of proper names. The study of how verbs may be formed from names confirms these basic principles inferred from investigation of common nouns and adjectives. (There are some exceptions to these principles, but overall a fairly small number.) And the study draws attention to an interesting aspect of speakers’ feelings about their language. People don’t say Kenned-ize (in the way that they do say subsid-ize and energize) because they have an underlying awareness that this derivational process applies only to Romance loans, and that Kennedy does not fall into this set.
Note Phonological representations (including recognition of secondary stress) are based on my own dialect of Educated British English, using the transcriptional system in Jones (1956). Considerable use has been made of the fine analyses and examples provided by Jespersen (1946) and Marchand (1969), and of the definitions and citations in the latest version (2nd edition, 1989, and revisions from M to part-way through P on-line) of the OED, the large Oxford English Dictionary (the publishers’ re-naming of A new English dictionary on historical principles, founded mainly on the materials collected by the Philological Society, edited by James A.H. Murray, 1888–1933, Oxford: Clarendon Press). Plus the Unabridged Random House dictionary (Flexner, 1987) and Cobuild (Sinclair, 2001), which is corpusbased. I have consulted the International Computer Archive of Modern English (1991) collection of English language corpora (including, among others, the London-Lund corpus of Spoken English and the Brown Standard Corpus of Present-day Edited American English) plus google.com. Indications of relative frequency are based on these sources. Since information in sources varies widely it is not feasible to attempt specific statistical statements of relative frequency. Clark and Clark (1979) give an admirable account of zero derivation of verbs from nouns in English, examining the semantic factors involved. Further discussion of this is provided by Aronoff (1980). Recent discussions of suffixes -ize and -(i)fy include Leiber (1998, 2004) and Plag (1999). It would be useful to have a generally-accepted term for ‘word-class-changing derivations’. Sweet (1891, 38–40) did coin the term ‘conversion’ for this. However, in recent years Sweet’s term has been reinterpreted as describing only word-class changing by a derivation which has zero realization. Indeed Bauer and Huddleston (2002, 1640) insist that the term ‘conversion’ should be used instead of ‘zero derivation’ or ‘zero affixation’.
Acknowledgements I am most grateful to Alexandra Aikhenvald who discussed this study with me at every stage, providing perceptive and constructive comments on this essay together with most welcome further exemplification. Further comments, of a most useful nature, were provided by Laurie Bauer, Kate Burridge, Jessica Cleary-Kemp and Nigel Love. And Stephen Morey assisted with harmon-ize, melod-ize and rhythm-ize.
52
R.M.W. Dixon / Language Sciences 30 (2008) 31–52
References Aronoff, Mark, 1980. Contextuals. Language 56, 744–758. Bauer, Laurie, Huddleston, Rodney, 2002. 19. Lexical word-formation. The Cambridge Grammar of the English Language, Huddleston, Rodney, Pullum, Geoffrey K. (Chief authors). Cambridge University Press, Cambridge, pp. 1621–1721. Clark, Eve V., Clark, Herbert H., 1979. When nouns surface as verbs. Language 55, 767–811. Dixon, R.M.W., 1977. Where have all the adjectives gone. Studies in Language 1, 19–80 [Revision published as pp. 1–62 of Dixon (1982)]. Dixon, R.M.W., 1982. Where Have all the Adjectives Gone, and Other Essays in Semantics and Syntax. Mouton, Berlin. Dixon, R.M.W., 2005. A Semantic Approach to English Grammar. Oxford University Press, Oxford. Flexner, Stuart Berg, 1987. Editor in chief of The Random House Dictionary of the English Language, second ed., unabridged. Random House, New York. International Computer Archive of Modern English, 1991. ICAME Collection of English Language Corpora. Norwegian Computing Centre for the Humanities, Bergen. Jespersen, Otto, 1946. A Modern English Grammar, on Historical Principles, Part VI, Morphology. Munksgaard/Allen and Unwin, Copenhagen/London. Jones, Daniel, 1956. Everyman’s English Pronouncing Dictionary, Eleventh ed. Dent/Dutton, London/New York. Leiber, Rochelle, 1998. The suffix -ize in English: implications for morphology. In: Lapointe, Steven G., Brentari, Diane K., Farrell, Patrick M. (Eds.), Morphology and its Relation to Phonology and Syntax. CSLI, Stanford, pp. 12–33. Leiber, Rochelle, 2004. Morphology and Lexical Semantics. Cambridge University Press, Cambridge. Marchand, Hans, 1969. The Categories and Types of Present-Day English Word-Formation, a Synchronic Diachronic Approach, second ed. C.H. Beck, Munich. Plag, Ingo, 1999. Morphological Productivity: Structural Constraints in English Derivation. Mouton de Gruyter, Berlin. Ross, John Robert, 1973. Leftward, ho! In: Anderson, Stephen R., Kiparsky, Paul (Eds.), A Festschrift for Morris Halle. Holt, Rinehart and Winston, New York, pp. 166–173. Sinclair, John, 2001. Founder editor-in-chief of Collins Cobuild English Dictionary for Advanced Learners. HarperCollins, Glasgow. Sweet, Henry, 1891. A New English Grammar, Logical and Historical. Clarendon Press, Oxford.