
- •Contents
- •Figures
- •Tables
- •Contributors
- •Preface
- •Acknowledgements
- •1 Overview
- •1.1 Introduction
- •1.2 The roots of English
- •1.3 Early history: immigration and invasion
- •1.4 Later history: internal migration, emigration, immigration again
- •1.5 The form of historical evidence
- •1.6 The surviving historical texts
- •1.7 Indirect evidence
- •1.8 Why does language change?
- •1.9 Recent and current change
- •2 Phonology and morphology
- •2.1 History, change and variation
- •2.2 The extent of change: ‘vertical’ and ‘horizontal’ history
- •2.3 Tale’s end: a sketch of ModE phonology and morphology
- •2.3.1 Principles
- •2.3.2 ModE vowel inventories
- •2.3.3 ModE consonant inventories
- •2.3.4 Stress
- •2.3.5 Modern English morphology
- •2.4 Old English
- •2.4.1 Time, space and texts
- •2.4.2 The Old English vowels
- •2.4.3 The Old English consonants
- •2.4.4 Stress
- •2.4.5 Old English morphology
- •2.4.5.1 The noun phrase: noun, pronoun and adjective
- •2.4.5.2 The verb
- •2.4.6 Postlude as prelude
- •2.5 The ‘OE/ME transition’ to c.1150
- •2.5.1 The Great Hiatus
- •2.5.2 Phonology: major early changes
- •2.5.2.1 Early quantity adjustments
- •2.5.2.2 The old diphthongs, low vowels and /y( )/
- •2.5.2.3 The new ME diphthongs
- •2.5.2.4 Weak vowel mergers
- •2.5.2.5 The fricative voice contrast
- •2.6.1 The problem of ME spelling
- •2.6.2 Phonology
- •2.6.2.2 ‘Dropping aitches’ and postvocalic /x/
- •2.6.2.4 Stress
- •2.6.3 ME morphology
- •2.6.3.2 The morphology/phonology interaction
- •2.6.3.3 The noun phrase: gender, case and number
- •2.6.3.4 The personal pronoun
- •2.6.3.5 Verb morphology: introduction
- •2.6.3.6 The verb: tense marking
- •2.6.3.7 The verb: person and number
- •2.6.3.8 The verb ‘to be’
- •2.7.1 Introduction
- •2.7.2 Phonology: the Great Vowel Shift
- •2.7.4 English vowel phonology, c.1550–1800
- •2.7.5 English consonant phonology, c.1550–1800
- •2.7.5.1 Loss of postvocalic /r/
- •2.7.5.2 Palatals and palatalisation
- •2.7.5.3 The story of /x/
- •2.7.6 Stress
- •2.7.7 English morphology, c.1550–1800
- •2.7.7.1 Nouns and adjectives
- •2.7.7.2 The personal pronouns
- •2.7.7.3 Pruning luxuriance: ‘anomalous verbs’
- •2.8.1 Preliminary note
- •2.8.2 Progress, regress, stasis and undecidability
- •2.8.2.1 The evolution of Lengthening I
- •2.8.2.2 Lengthening II
- •3 Syntax
- •3.1 Introduction
- •3.2 Internal syntax of the noun phrase
- •3.2.1 The head of the noun phrase
- •3.2.2 Determiners
- •3.3 The verbal group
- •3.3.1 Tense
- •3.3.2 Aspect
- •3.3.3 Mood
- •3.3.4 The story of the modals
- •3.3.5 Voice
- •3.3.6 Rise of do
- •3.3.7 Internal structure of the Aux phrase
- •3.4 Clausal constituents
- •3.4.1 Subjects
- •3.4.2 Objects
- •3.4.3 Impersonal constructions
- •3.4.4 Passive
- •3.4.5 Subordinate clauses
- •3.5 Word order
- •3.5.1 Introduction
- •3.5.2 Developments in the order of subject and verb
- •3.5.3 Developments in the order of object and verb
- •3.5.5 Developments in the position of particles and adverbs
- •3.5.6 Consequences
- •4 Vocabulary
- •4.1 Introduction
- •4.1.1 The function of lexemes
- •4.1.3 Lexical change
- •4.1.4 Lexical structures
- •4.1.5 Principles of word formation
- •4.1.6 Change of meaning
- •4.2 Old English
- •4.2.1 Introduction
- •4.2.4 Word formation
- •4.2.4.1 Noun compounds
- •4.2.4.2 Compound adjectives
- •4.2.4.3 Compound verbs
- •4.2.4.7 Zero derivation
- •4.2.4.8 Nominal derivatives
- •4.2.4.9 Adjectival derivatives
- •4.2.4.10 Verbal derivation
- •4.2.4.11 Adverbs
- •4.2.4.12 The typological status of Old English word formation
- •4.3 Middle English
- •4.3.1 Introduction
- •4.3.2 Borrowing
- •4.3.2.1 Scandinavian
- •4.3.2.2 French
- •4.3.2.3 Latin
- •4.3.3 Word formation
- •4.3.3.1 Compounding
- •4.3.3.4 Zero derivation
- •4.4 Early Modern English
- •4.4.1 Introduction
- •4.4.2 Borrowing
- •4.4.2.1 Latin
- •4.4.2.2 French
- •4.4.2.3 Greek
- •4.4.2.4 Italian
- •4.4.2.5 Spanish
- •4.4.2.6 Other languages
- •4.4.3 Word formation
- •4.4.3.1 Compounding
- •4.5 Modern English
- •4.5.1 Introduction
- •4.5.2 Borrowing
- •4.5.3 Word formation
- •4.6 Conclusion
- •5 Standardisation
- •5.1 Introduction
- •5.2 The rise and development of standard English
- •5.2.1 Selection
- •5.2.2 Acceptance
- •5.2.3 Diffusion
- •5.2.5 Elaboration of function
- •5.2.7 Prescription
- •5.2.8 Conclusion
- •5.3 A general and focussed language?
- •5.3.1 Introduction
- •5.3.2 Spelling
- •5.3.3 Grammar
- •5.3.4 Vocabulary
- •5.3.5 Registers
- •Electric phenomena of Tourmaline
- •5.3.6 Pronunciation
- •5.3.7 Conclusion
- •6 Names
- •6.1 Theoretical preliminaries
- •6.1.1 The status of proper names
- •6.1.2 Namables
- •6.1.3 Properhood and tropes
- •6.2 English onomastics
- •6.2.1 The discipline of English onomastics
- •6.2.2 Source materials for English onomastics
- •6.3 Personal names
- •6.3.1 Preliminaries
- •6.3.2 The earliest English personal names
- •6.3.3 The impact of the Norman Conquest
- •6.3.4 New names of the Renaissance and Reformation
- •6.3.5 The modern period
- •6.3.6 The most recent trends
- •6.3.7 Modern English-language personal names
- •6.4 Surnames
- •6.4.1 The origin of surnames
- •6.4.2 Some problems with surname interpretation
- •6.4.3 Types of surname
- •6.4.4 The linguistic structure of surnames
- •6.4.5 Other languages of English surnames
- •6.4.6 Surnaming since about 1500
- •6.5 Place-names
- •6.5.1 Preliminaries
- •6.5.2 The ethnic and linguistic context of English names
- •6.5.3 The explanation of place-names
- •6.5.4 English-language place-names
- •6.5.5 Place-names and urban history
- •6.5.6 Place-names in languages arriving after English
- •6.6 Conclusion
- •Appendix: abbreviations of English county-names
- •7 English in Britain
- •7.1 Introduction
- •7.2 Old English
- •7.3 Middle English
- •7.4 A Scottish interlude
- •7.5 Early Modern English
- •7.6 Modern English
- •7.7 Other dialects
- •8 English in North America
- •8.1.1 Explorers and settlers meet Native Americans
- •8.1.2 Maintenance and change
- •8.1.3 Waves of immigrant colonists
- •8.1.4 Character of colonial English
- •8.1.5 Regional origins of colonial English
- •8.1.6 Tracing linguistic features to Britain
- •8.2.2 Prescriptivism
- •8.2.3 Lexical borrowings
- •8.3.1 Syntactic patterns in American English and British English
- •8.3.2 Regional patterns in American English
- •8.3.3 Dictionary of American Regional English (DARE)
- •8.3.4 Atlas of North American English (ANAE)
- •8.3.5 Social dialects
- •8.3.5.1 Socioeconomic status
- •8.3.6 Ethnic dialects
- •8.3.6.1 African American English (AAE)
- •8.3.6.2 Latino English
- •8.3.7 English in Canada
- •8.3.8 Social meaning and attitudes
- •8.3.10 The future of North American dialects
- •Appendix: abbreviations of US state-names
- •9 English worldwide
- •9.1 Introduction
- •9.2 The recency of world English
- •9.3 The reasons for the emergence of world English
- •9.3.1 Politics
- •9.3.2 Economics
- •9.3.3 The press
- •9.3.4 Advertising
- •9.3.5 Broadcasting
- •9.3.6 Motion pictures
- •9.3.7 Popular music
- •9.3.8 International travel and safety
- •9.3.9 Education
- •9.3.10 Communications
- •9.4 The future of English as a world language
- •9.5 An English family of languages?
- •Further reading
- •1 Overview
- •2 Phonology and morphology
- •3 Syntax
- •4 Vocabulary
- •5 Standardisation
- •6 Names
- •7 English in Britain
- •8 English in North America
- •9 English worldwide
- •References
- •Index

204 D I E T E R K A S T O V S K Y
(2)gan/gangan¯ ‘go, come, move, proceed; depart; happen’
gang, masc. ‘going, journey; track, footprint; passage, way; privy; steps, platform’; ciri˙c˙-gang ‘churchgoing’, ears-gang ‘excrement’, forlig˙-gang ‘adultery, lit. going to adultery’, hin-gang ‘a going hence, death’, hlaf¯ -gang ‘a going to eat bread = communion’, etc., etc.
gang-ern, gang-pytt, gang-setl, gang-stol¯ , gang-tun¯ ‘privy, toilet, i.e. some place to which one goes’, gang-dæg˙ ‘Rogation Day’, gange-wifre ‘spider, lit. a going weaver’, etc.
genge˙ n. ‘troops, company’
-genge˙, fem. in niht-genge˙ ‘hyena, i.e. an animal that prowls at night’ -genga˙, masc. in an¯ -genga˙ ‘a solitary goer’, æfter-genga˙ ‘one who follows’,
hinder-genga˙ ‘one that goes backwards, a crab’, husel¯ -genga˙ ‘one who goes to the Lord’s supper’, sæ¯-genga˙ ‘sea-goer, sailor; ship’, etc., etc.
genge˙ adj. ‘prevailing, going, effectual, agreeable’
-gengel˙, masc. in æfter-gengel˙ ‘successor’, cf. also æfter-gengan˙, wk vb, to go behind’
gengan˙, wk vb ‘to go’, æfter-gengness˙ ‘succession, posterity’
a¯-gan¯ ‘go, go by, pass, pass into possession, occur, befall, come forth’ be-gan/be¯ -gangan ‘go over, go to, visit; cultivate; surround; honour,
worship’, be-gang/b¯ı-gang ‘practice, worship’, be-geng/b˙¯ı-genga˙ masc. ‘inhabitant, cultivator’ with numerous compounds, be-genge˙, n. ‘practice, worship’, b¯ı-gengere˙ ‘worker, worshipper’, b¯ı-gengestre˙, fem. ‘hand maiden, attendant, worshipper’, be-gangness ‘calendae, celebration’; and many other combinations with the prefixes/prepositions/adverbs fore, forþ, in, niþer, of, ofer, on, oþ-, to¯, þurh, under, up¯ , ut¯ , wiþ, ymb-
Another aspect reflecting the thoroughly Germanic character of OE vocabulary is the preservation of Indo-European ablaut alternations not only as part of the morphology of the strong verbs (e.g. wr¯ıtan : wrat¯ : writon : gewriten˙ ‘write’, cf. ModE write : wrote : written) but also in related nouns and adjectives:
(3)drincan: drinc ‘drink, drinking’, gedrinca one who drinks with another’, drincere ‘drinker’, drenc˙ ‘drink, drinking’, drencan˙ ‘give to drink’, drenc˙-hus¯ ‘drinking-house’, druncen ‘drunkenness’, druncennis ‘drunkenness’, druncnian ‘be, get drunk’, druncning ‘drinking’, drync˙ ‘drink, potion, drinking’
Almost all of these nouns and adjectives were lost in ME, however, apart from a few survivors such as song, writ, breach, drunk, unlike the other Germanic languages, where these nouns and adjectives are still part of the core vocabulary. We will therefore have to tackle the question of why such a substantial part of the core vocabulary was lost, and what it was replaced by.
4.1.4 |
Lexical structures |
|
The traditional view of language is that its grammar is systematic, structured, rule-governed and therefore predictive, i.e. it allows for

Vocabulary 205
generalisations. Its lexicon, on the other hand, is often regarded as unsystematic, not subject to any rules or generalisations; cf. Bloomfield (1935: 274): ‘The lexicon of a language is an appendix of the grammar, a list of basic irregularities’, a quotation which is repeated almost verbatim in Chomsky (1965: 142). But this view is not really appropriate; cf., e.g., Aitchison (1987: 7): ‘Words are not just stacked higgledy-piggledy into our minds – they are organised into an intricate, interlocking system.’ Without understanding this intricate system, we also cannot make sense of the historical development of the vocabulary of a language.
There are basically two types of structure, viz. purely semantic ones based on meaning relations, and formal-morphological ones based on morphosemantic relations. Both intersect with each other, and both are also affected by the borrowing processes mentioned above.
Semantic structures are based on the fact that lexical items sharing a basic meaning are organised in terms of semantic relationships characterising lexical fields (cf. Coseriu, 1973; Coseriu & Geckeler, 1981; Cruse, 1986, 2000; Kastovsky, 1982; Lipka, 2002; Lyons, 1977: 230–335). Typical examples are lexical relations such as hyponymy, antonymy, complementarity and others:
(4)a. hyponymy: flower : tulip, rose, daffodil, crocus, snowdrop, etc.
b.antonymy: big : small, long : short, giant : dwarf, love : hate
c.complementarity: married : single, male : female, man : woman
d.reversativity: open : close, tie : untie, arm : disarm
In hyponymy, there is one general term to which some more specific terms are subordinated. Their number may be open-ended. Antonymy characterises the relationship between two lexical items which refer to the greater or lesser presence of a given property, i.e. the items are gradable and there is a neither-nor area between them (e.g. this box is bigger than that box, this is neither big nor small). Complementaries are not gradable and divide a common semantic area into two mutually exclusive semantic domains (e.g. someone is either married or single, there is nothing in between). Reversatives denote the undoing of a state that may or may not have come about by a previous action.
Assume you have a structure that consists of x lexemes dividing up a lexical space such as ‘come into a place: get out of a place’, as in OE incuman or ingangan and laefan¯, utgangan¯ . Then a number of lexemes such as enter, arrive, depart, exit are added by borrowing. Obviously this changes the structural makeup of the respective lexical fields by introducing additional semantic distinctions (much along the lines of German essen : fressen as against English eat). Thus, as a result of the massive borrowings, lexical fields in English are rather large and complex, with subtle stylistic and register differentiations, and it is unfortunate that not too many synchronic, let alone historical, studies exist of this phenomenon.
Formal-morphological structures in the lexicon are the result of word-formation processes. The delimitation rests on the distinction between simple and complex lexical items; cf.

206 D I E T E R K A S T O V S K Y
(5)author, butcher, carpenter, architect (= simple) : writ-er, build-er, violin-ist, post-man (= complex); empty, full, busy (simple) : air-less, crowd-ed, ston-y (= complex)
Simple lexical items reflect the primary categorisation of extralinguistic reality in a language. Complex lexical items represent a supplementary, derived categorisation, which operates on the basis of this primary categorisation and adds further categorisations according to the needs of the speech community. Both levels interact with each other, i.e. word-formation patterns take up the structure of lexical fields and can be described in terms of the same semantic relations as simple lexical items; cf.
(6)a. hyponymy: tree : oak tree, pear tree, cherry tree, apple tree, etc.
b.antonymy: kind : unkind, friendly : unfriendly
c.complementarity: true : untrue, loyal : disloyal
Complex lexical items therefore fill gaps in lexical fields: a certain notion exists but has no expression in the form of a simple lexical item, and when a communicative need calls for its lexical realisation, an appropriate word-formation process (or borrowing) is activated.
4.1.5 |
Principles of word formation |
|
Besides borrowing, word formation, i.e. the combining of already existing elements to form new lexical items, is thus another way of enriching the vocabulary. Such new, derived lexemes have the advantage that their meanings can be derived from the meanings of the constituents and the patterns that underlie the formations, because these patterns are also present in other complex lexical items. No language would be able to function properly without such patterns. Just think of numbers: it would be impossible to memorise all numbers from 1 to 1,334,566,778 or even higher, if they were arbitrary names. This is why language has means to create transparent names such as one billion three hundred and thirty-four million five hundred and sixty-six thousand seven hundred and seventy-eight.
Borrowing and word formation have co-operated in the history of English, because many Modern English word-formation patterns are the result of borrowing, e.g. arrive : arrival, consult : consultant, employ : employee, post : postal, bake : bakery (here only the suffix is borrowed), history : historic : historicity, militarise : demilitarise, pilot : co-pilot, entangle : disentangle, etc., etc.
As stated above, the basic function of all lexical items is naming, and word-formation processes also create names; cf. instances such as hamburger, cheeseburger, beefburger, alligator-burger, or Watergate, Irangate, nannygate
(referring to President Clinton’s problem in 1992 of finding an attorney general who had not illegally employed an immigrant household help) and, last but not least, Clinton’s zippergate. But word formation is also involved in another equally important function, viz. what might be called syntactic recategorisation, since it

Vocabulary 207
has syntactic properties. Here, a complex lexical item takes up the previous context and repeats it – almost like a pronoun – in nominal, adjectival or verbal form:
(7)a. ‘So you simply can’t change history using stasis?’ . . . ‘No. . . . One small boy . . . is hardly going to be a history-changer because we’ve brought him forward to our era.’ (Isaac Asimov & Robert Silverberg, Child of Time,
1992)
b.A few thought they had noticed someone resembling the man in the picture. I wasted two days tracking one of the supposed resemblers, and found no resemblance at all. (Robert Silverberg, Up the Line, 1969/1975)
c. |
. . . do we assume that the stone-chucker, wire-stretcher, composite |
|
letter-writer, dumper of green lady and telephonist are one and the same |
|
person and that this person is also the murderer of Miss Cost? . . . Miss |
|
Pride . . . is convinced that the ringer-up was Miss Cost. (Ngaio Marsh, |
|
Dead Water, 1964, summing up the previous events of the novel) |
d.If you want shares in Rolls-Royce apply now! . . . A Public Application Form for Rolls Royce shares appears in this newspaper. (Newspaper ad,
The Times)
e.It’s blood on his hands. His hands get covered with blood, not visible to anyone else, and he goes and washes them . . . He wouldn’t give his name and didn’t mention bloody hands. (Rex Stout, Please Pass the Guilt, 1973)
f.Finally, he put the juice in bottles. Bottled juice was easier to store.
Such formations are often used for information condensation, text cohesion, pronominalisation. As a result, the contents of a previous sentence can be easily modified without a clumsy syntactic construction. The most frequent type of recategorisation are nominalisations. These are closely related to relative and complement clauses; cf.
(8)a. I know someone of whom it is alleged that he has discovered how one can travel through time.
b.I know someone allegedly having / alleged to have discovered travel through time.
c.I know an alleged discoverer of time-travel.
These formations are basically interpretable by referring to functions such as Subject, Object, Adverbial (sometimes called ‘argument structure’) or syntacticsemantic functions, such as Agent, Patient (Theme), Locative, etc. (sometimes called ‘thematic roles’, echoing Fillmore’s (1968) deep-structure cases or roles). Thus a history-changer is ‘someone who (Subject/Agent) changes history (Object/Patient = Theme)’ and to bottle means ‘put into bottles’ (Adverbial Complement/Locative). Here we have a close parallelism between syntax and word formation, although the nature of this relationship is controversial.
Such formations provide a problem for the lexicographer, and for someone who wants to describe the history of the vocabulary, too, because it is questionable whether all such formations, whose meaning is fully predictable, should and actually would be listed in a dictionary like the OED, our primary source for the

208 D I E T E R K A S T O V S K Y
history of the English vocabulary. And in fact very often such formations are not included, or included only unsystematically (cf. Kastovsky, 2000). This means that we do not really know what formations actually existed at a certain time, unless we go to large text corpora. This is also the reason why it is impossible to give accurate numbers as to the size of a vocabulary, because with productive patterns such as those involved in (7) it is always possible for a speaker to come up with a new formation at any time. And this makes the vocabulary of a language in principle non-finite. But languages certainly differ in the extent to which they make use of this possibility, and they also change the extent to which they do this over time. Thus OE was more prone to using such processes, whereas Modern English very often relies on borrowing, especially when it comes to the naming function.
The basic principle of word formation is comparison, i.e. ‘to see a thing identical with another already existing and at the same time different from it’ (Marchand, 1969: 1). This principle is the reason why word formations typically exhibit a binary structure, consisting of a determinant (dt, modifier) and a determinatum (dm, head). This kind of structure is usually called ‘syntagma’ (cf. Marchand, 1969, and Kastovsky, 1999). Cf. the following examples:
(9)determinant / determinatum determinant / determinatum
(modifier) |
/ |
(head) |
(modifier) |
/ |
(head) |
rope |
/ |
dancer |
writ |
/ |
er |
cave |
/ |
man |
clear |
/ |
ing/ance |
ex |
/ |
husband |
bak |
/ |
ery |
pre |
/ |
cook |
fl(a)ut |
/ |
ist |
re |
/ |
write |
civil |
/ |
ise |
over |
/ |
zealous |
beauti |
/ |
fy |
un |
/ |
kind |
fear |
/ |
less |
The sequence of dt and dm is language-specific. The Germanic languages are generally characterised by a dt/dm sequence within word formation syntagmas, whereas syntactic structures also allow the reversed sequence. The Romance languages, on the other hand, allow both dt/dm and dm/dt structures in word formation; cf. (10), where the dm is underlined:
(10)a. F chant-eur, Sp. cant-ador ‘singer’, F im-possible, It. im-possibile
b.F chemin de fer ‘railway’, barbe-bleu, It. Barba blu, ‘bluebeard’, F tire-bouchon, Sp. saca-corchos ‘corkscrew’
Constituent order in complex lexical items thus is language-specific, not universal. Assuming that English, like all Germanic languages, has a general dt/dm sequence, an assumption that seems justified on typological and quantitative grounds, we still have to explain certain patterns that seem to violate this principle. This raises questions about the demarcation of word formation from syntax and the incorporation of syntactic constructions, i.e. phrases, into the lexicon, which is also of historical interest.

Vocabulary 209
Compare:
(11)consul(s)´ general´, heir(s)´ apparent´, letters´ patent´, court´ martial(s)´
consisting of a noun and an adjective in that order, which deviates from the regular order adjective/noun both in compounds and syntactic groups, e.g. blackbird, general store, Black Sea, etc., i.e. it has a dm/dt sequence. The explanation in this case is simple: the pattern was borrowed from French, which allows this order (cf. barbe-bleu in (10b) above). The pattern entered English in ME and belongs to legal or semi-legal jargon; only a few instances are part of the general vocabulary, and it has not really become productive in English; see also Section 3.2.4. There are two other loan patterns with the reversed dm/dt order, viz. personal names, e.g. Fitzherbert (<Anglo-Norman fitz ‘son’ de Herbert), MacArthur ‘son of Arthur’ (Gaelic-Scottish). Again, these patterns have not become productive.
Instances such as (12) also pose problems for the posited general dt/dm sequence in English word formation:
(12) a. father-in-law, jack-in-the-box, jack-of-all-trades, man-in-the-street
b.come in, go out
They illustrate a further means of extending the vocabulary not mentioned so far. From a lexicological point of view, we would like to treat these cases as lexical items, and since they clearly are complex, they should be regarded as results of word-formation processes. But they exhibit certain anomalies not found in the typical word-formation patterns. Thus they do not conform to the regular dt/dm order but exhibit a dm/dt order. This order reflects the order NP + PP, typical of syntactic constructions, which have to be kept separate from word formations proper. Additionally, the formations in (12a) contain syntactic markers
like prepositions or articles – very uncommon in word formations:
(13) |
church-going |
vs going to church, |
brain-surgery vs |
surgery of the brain |
|
wf |
syntax |
wf |
syntax |
And also compare (12b) with formations like income, incoming, outgoing (mail), where the particle (preposition/adverb) is preposed rather than postposed.
On the basis of these structural properties, it would seem to be justified to treat formations like (12) as syntactic constructions rather than as word formations in the strict sense. Nevertheless, such formations are usually listed in dictionaries, i.e. they have been accorded the status of lexical items. This means that complex lexical items not only arise from word-formation processes but also through the lexicalisation of syntactic phrases, which adopt a special meaning and are then treated as lexical wholes, although they preserve their original syntactic properties. Moreover, it should be added that with nominal compounds the delimitation between word formations and syntactic constructions is also problematic, because they also consist of independent lexemes. Thus fore-stressed instances such as house´-door,` steam´-boat`, etc. are usually recognised as compounds, whereas the status of level-stressed formations (with an end-stressed variant), such as gold´

210 D I E T E R K A S T O V S K Y
watch´ gold` watch,´ glass´ case´ glass` case´, etc. is controversial. Marchand (1969: 20ff.) treats them as syntactic groups, Bauer (1983: 102–12) as compounds just like the fore-stressed combinations. On the other hand, everybody seems to agree that level-stressed dancing´ g´ırl dancing` g´ırl ‘girl who is dancing’, or black´ b´ırd black` b´ırd are syntactic groups, whereas fore-stressed dancing´ g`ırl, blackb´`ırd are always treated as compounds, probably because there always seems to be a semantic difference between the two types of construction, which is not always true of N + N-compounds, cf. ´ıce cream´ and ´ıce cream`, which have the same meaning. A similar problem arises with formations such as beg´ınner’s luck´, which look like regular syntactic groups consisting of a preposed genitive + N and also have the stress pattern associated with syntactic constructions. But in this case the compound interpretation is obvious: incredible beginner’s luck cannot be paraphrased as ‘luck of an incredible beginner’; the adjective clearly modifies the determinatum luck, unlike beautiful singer’s voice, which allows both ‘voice of a beautiful singer’ and ‘beautiful voice of a singer’ (cf. Marchand, 1969: 27– 8, 65–9). This is why the -s in beginner’s luck and similar instances should not be treated as a genitive morpheme, but rather as a so-called ‘linking element’ (Fugenelement), as in G Ankunftszeit ‘time of arrival’, Frauenkirche ‘church of Our Lady’.
Lexicalisation of such syntactic phrases is thus a further means of extending the vocabulary, which may in fact go very far and extend to fully idiomatised phrases such as kick the bucket, red herring, pull one’s leg, etc. This kind of lexicalisation and idiomatisation is also relevant to word formations. Once a formation has been accepted into the general vocabulary, it is subject to changes of meaning just like any simple lexeme. Thus OE hali¯ gdæ˙g˙ was indeed a ‘holy day’, but holiday no longer is, and even though blackboards where originally and legitimately called blackboards, the change in colour of the referents (now usually green) has created some degree of idiomatisation, and the most recent exemplars are therefore justifiably called whiteboards. All these processes have of course been at work throughout the history of English.
The morphological status of the constituents dt and dm as either words or bound morphemes (affixes) results in a subdivision of word formation into two basic subcategories. If both constituents are actual or potential lexemes, we speak of compounds, as in bird/cage, girl/friend, letter/writer, writing/table, racehorse/owner, colour/blind, home/sick, home/made, etc. If one of the two constituents is a bound morpheme which is not the representation of a lexeme, we speak of affixation or derivation, with a subdivision into prefixation (the bound morpheme occurs before the lexeme representation), e.g. counter/propaganda, dis/believe, in/justice, re/write, and suffixation (the bound morpheme occurs to the right of the lexeme representation), e.g. arriv/al, king/dom, champion/ship; feather/ed, hope/ful, atom/ic, whit/ish; dark/en, legal/ise.
A further possibility of creating new lexemes is converting a lexeme belonging to one word class into a member of a different word class, e.g. bridge sb > bridge vb, cheat vb > cheat sb (BrE), or clean adj > clean vb. For this, the term

Vocabulary 211
‘conversion’ is widely used, but it is problematic, because it obscures the fact that the difference in word-class affiliation is regularly accompanied by a meaning difference, i.e. that this involves more than just a shift of word-class affiliation. Moreover, this meaning difference matches the type of semantic contrasts we find with suffixation; cf.:
(14)a. cheat vb : cheat sb = write vb : writ/er sb (‘someone who vb’)
b.stop vb : stop sb = land vb : land/ing sb (‘place where one vb’)
c.cash sb : cash vb = atom sb : atom/ise vb (‘to convert into sb’)
d.clean adj : clean vb = legal adj : legal/ise vb (‘cause to become adj’)
Obviously, the same semantic content, i.e. ‘someone who V-s (habitually, professionally)’ in (14a), ‘place where one V-s’ in (14b), ‘convert into N’ in (14c), and ‘cause to become Adj’ in (14d) can be expressed both by suffixes and conversions. From this it follows that conversions are regular derivatives like normal suffixations, except that they exhibit a morphological peculiarity: the deriving suffix is not expressed overtly.
In order to express this parallelism between suffixal derivatives and conversions, the latter have been interpreted as ‘zero derivatives’ by some linguists, e.g. Marchand (1969: 359–89), i.e. as word formation syntagmas where the determinatum is a zero morpheme, in order to preserve the binary structure of word formations; zero morphemes indicate a functional position which is not filled by a formal exponent. This analysis is somewhat controversial, and whether one accepts zero derivation or not is a matter of theoretical orientation (from a historical point of view it makes sense, however, since the zero element always replaces an originally overt morpheme, usually a stem formative; cf. Kastovsky, 1980, 1996). Alternatively, one might also call this process ‘affixless derivation’. The main point is that this process, in view of its semantic properties, should be recognised as a sub-category of derivation, and not merely as a shift of word class.
It is sometimes claimed that this type of word formation is restricted to Modern English or is particularly characteristic of Modern English, but in fact it also exists in OE with its fully-fledged inflectional system:
(15)a. cum-an ‘to come’: cum-Ø-a ‘one who comes, a guest’; gief˙-an ‘to give’: gief˙-Ø-u ‘what is given, gift’; hunt-an ‘to hunt’: hunt-Ø ‘hunting’: hunt-Ø-a ‘hunter’
b.beorht ‘bright’: beorht-Ø-ian ‘to make bright’; ar¯ ‘honour’: ar¯ -Ø-ian ‘to honour’; munuc ‘monk’: munuc-Ø-ian ‘to make into a monk’
Assuming that there is a strict functional delimitation between inflection (the derivation of word forms of lexemes) and word formation (the creation of new lexemes), the endings -a, -u in (15a), and -ian in (15b) have to be regarded as part of the inflectional system (see Section 2.4.5.1) and thus do not have any derivational function but mark grammatical functions such as case/number or infinitive/person/number. Formations such as cuma, giefu,˙ beorhtian thus lack

212 D I E T E R K A S T O V S K Y
a derivational affix just like their Modern English equivalents, except that this is somewhat obscured by the pervasive presence of inflectional endings with grammatical functions.
In view of the important typological changes in the history of English word formation, a few remarks as to typological properties of morphology might be appropriate, which primarily relate to the structure of inflection. Here we can distinguish two types, viz. word-based and stem-based morphology. In the first type, there is one unmarked form in an inflectional paradigm which can function as a word in a sentence without any inflectional ending, e.g. Modern English cat, cat-s; kiss, kiss-es, kiss-ed, kiss-ing. The form without inflectional ending acts as input to the inflectional and derivational processes, which is why we can speak of word-based morphology. In the second case, the inflectional paradigm does not contain any form without an inflectional ending which can function as base form, as, e.g., in OE verbal paradigms such as luf-ian, luf-ie, luf-iaþ, luf-od-e, ge˙-luf-od, or nominal paradigms such as cum-a, cum-an, cum-ena, cumum; tal-u, tal-e, tal-a, tal-um. Thus when we cut off the inflectional endings we are left with a lexeme representation like luf- or tal-, which cannot occur on its own as a word. Such a lexeme representation is called a stem, which is why we speak of stem-based morphology in this case. This notion can be extended to instances such as Modern English scient-ist, dramat-ist, astro-naut, tele-gram, where scient- (cf. science), dramat- (cf. drama), astro-, -naut, tele-, -gram do not occur as independent words, but only in combination with other morphological elements. They therefore have to be regarded as stems. This distinction is historically very important, since in OE we had a mixture of stemand word-based morphology, which gradually developed towards a purely word-based morphology in ME. At the same time, borrowing from languages with stem-based morphology, such as French, Latin or Greek, introduced a new type of stem-based morphology into word formation, which, however, is restricted to the non-native part of the vocabulary and has again created a typologically heterogeneous system, although of a different nature: now the distinction marks the delimitation of a native vs a non-native word-formation stratum. The same is true of morphophonemic/allomorphic alternations. These were pervasive in OE, but were gradually lost in ME and were replaced by a non-alternating morphological system. Alternations in inflections such as keep kept, drink drank drunk, mouse mice, foot feet characterise these forms as belonging to the class of irregular verbs and nouns. Similarly, native or nativised word-formation patterns do not exhibit alternations of the base, so that alternating formations such as sane sanity, h´ıstory historic´ histor´ıcity, Japan´ Japan`ese´ are marked as belonging to the non-native stratum.
The synchronic delimitation between compounding and affixation is not always clear-cut because of change: constituents of compounds may gradually change to affixal status, e.g. Modern English out- in outbid ‘bid higher than an opponent’; under- in underestimate ‘estimate below a fixed norm’ (see Marchand, 1969: 96– 100), where the particle has lost its original locative meaning; or -like in manlike,

Vocabulary 213
boldlike; -monger (< OE mang-ere ‘merchant’ < mangian ‘to sell’) in fishmonger, warmonger; -wise in anti-clockwise, moneywise, weatherwise, which Marchand (1969: 356) calls ‘semi-suffixes’. This is due to the fact that a constituent of a compound may be used for forming a whole series of combinations, which may result in a semantic bleaching of its meaning. Eventually, even a change of grammatical category might happen, as is the case with -wise, where from a purely morphological point of view the formations look like nominal compounds, but they function as adverbs. This kind of phenomenon can be regarded as an instance of grammaticalisation. Thus the Modern English adverbial suffix -ly continues OE -l¯ıc˙-e as in blind-l¯ıc˙-e ‘blindly’, which in turn contains the adjective-forming suffix -l¯ıc˙, but this in fact was originally a noun, viz. l¯ıc˙ ‘body, form’. The origin of this type of formation thus is some nominal construction ‘in x’s form’, just as with the Modern English -wise-derivatives. Similarly, some Modern English suffixes like -dom, -hood can be traced back to full words in OE, whose status as second members in complex lexemes was ambivalent. The noun dom¯ originally meant ‘judgement’, but in derivatives such as martyrdom¯ ‘martyrdom’, freod¯ om¯ ‘freedom’ it adopted the meaning ‘state, condition, fact of being’, which is also the meaning of Modern English -dom, e.g. dukedom, stardom, freedom. A similar development can be assumed for had¯ ‘state, condition, rank, order’, as in abbudhad¯ ‘rank of an abbot’, cild˙-had¯ ‘childhood’.
One more domain of word formation has to be mentioned as contributing to the extension and modification of the vocabulary, less rule-governed than the processes discussed so far, although some of the categories involved have become very important since the early twentieth century: phonetic symbolism (onomatopoeia), clipping, blending and word manufacturing; see Marchand (1969: 397–454).
Phonetic symbolism uses the possibility of language to imitate sounds either directly or indirectly, sometimes associating some metaphorical interpretation with the sounds in question, e.g. extending sound to the representation of movement. Examples are puff, pop, splash, swish, chitchat, zig-zag, pitter-patter, hocuspocus, mumbo-jumbo. Many of these formations are primarily part of spoken language, which is why a historical account of the development of these formations is rather difficult, but at least some formations must already be of OE origin, and many are attested from late ME and early Modern English.
Clipping consists in the reduction of a word to one of its syllabic parts, e.g. ad < advertisement, exam < examination, flu < influenza, lab < laboratory, plane < airplane. This process seems to be particularly popular with designations of persons, especially first names (see also Section 6.3.8), often in combination with the emotive suffix -ie, e.g. Al, Alfie < Alfred, Andy < Andrew, Archie < Archibald, Aussie < Australian, commie < communist, Debby < Deborah, Fred(die) < Frederick, granny < grandmother, looney < lunatic.
Clipping is primarily a modern phenomenon, but examples can already be found in early Modern English, e.g. coz < cousin (1559), gent < gentleman (1564); seventeenth-century examples are van < vanguard, brandy < brandywine

214 D I E T E R K A S T O V S K Y
(a Dutch loan), wig < periwig; eighteenth-century formations are brig < brigantine, gin < geneva, spec < speculation. That century was especially fond of this process; cf. Swift’s remark in his Introduction to Polite Conversation (quoted in Jespersen, 1942: 29.41):
The only Invention of late Years, which hath any way contributed towards Politeness in Discourse, is that of abbreviating or reducing Words of many Syllables into one, by lopping off the rest . . . Pozz for Positive, Mobb for Mobile, Phizz for Physiognomy, Rep for Reputation, Plenipo for Plenipotentiary, Incog for Incognito, Hyppo or Hippo for Hypocondriacks, Bam for Bamboozle, and Bamboozle for God knows what.
And it would seem that this fashion has been revived again today.
Blending is a combination of clipping and compounding, i.e. two lexemes are clipped and then combined into one unit, e.g. smoke + fog > smog, motor + hotel > motel, breakfast + lunch > brunch. This process is moderately productive today, though the results more often than not are rather short-lived.
Rather important today is word manufacturing, which consists of two subcategories, acronyms (‘letter words’) and fully arbitrary creations. Both play an important role especially in the naming of technical products and their marketing in terms of brand names. Acronyms are usually based on initial letters of names for organisations, processes, gadgets, etc. The letters are either pronounced separately, as in EU < European Union, UN < United Nations, or the combination is pronounced like a regular word, as in Aids < acquired immune deficiency syndrome, laser (1960) < light amplification by the stimulated emission of radiation, radar (1941) < radio detection and ranging, snafu < situation normal, all fouled/fucked up, and many others. Arbitrary creations play an important role in brand names, e.g. Kodak, Viagra, Xerox, etc.
In the preceding section, word formations were classified in terms of the morphological properties of their constituents. But there is another possibility, which is based on the function of the constituents. This again results in two basic categories, viz. expansions and derivations, which intersect with the above classification (cf. Marchand, 1969: 11). This classification uses the criterion of whether the head can stand for the whole combination or not, which can be represented by the two formulas AB = B (expansion) and AB =B (derivation). With compounds like house/door, rattle/snake, colour/blind, the head can stand for the whole combination, i.e. a house-door is a door, a rattlesnake is a snake, etc. The same is true of prefixations: a co-author is a kind of author, hyper-active is a kind of active, and re-write is a kind of write. They thus satisfy the formula AB = B and are expansions. Formations such as sleep-er, pott-er, ston-y, atom-ise do not satisfy this formula, since the head cannot stand for the whole combination, because it is a bound morpheme, which transposes the determinant into another word class or semantic category. They therefore qualify as derivatives. This criterion has consequences for the classification of formations which look like compounds but do not satisfy the AB = B criterion: a paleface is not a face,