12.1 Introduction

The sound system of English has undergone considerable change in the 1,500 years or so for which documents of the language exist. So great is this change that the earliest forms of the language are not readily comprehensible to speakers of EnglishFootnote ¹ today. Major sound changes occurred every few centuries, continuously increasing the distance to earlier stages of the language. Furthermore, different types of phonological change occurred during the history of English and it is the consideration of these types and the research on them which form the core of the present chapter.

Some of the changes were motivated by reanalysis on the part of language learners and some by gradual shifts in pronunciation by adult speakers. Both types of change are connected and form trajectories along which the sound system of English has moved for over a millennium and a half. In the history of the language there are further motivations for language change. Contact with other languages had a lasting influence until at least the late Middle Ages (Hickey Reference Hickey, Nevalainen, Pfenninger and Traugott2012b) after which this influence was largely confined to the lexical area which was fed with items not gained through direct contact with speakers of other languages. From the eighteenth century onwards a further factor comes to the fore in language change: the prescriptivism which arose surrounding language use and education which concerned the rising middle classes in the late modern period.

12.2 Research situation

Phonology was a primary focus when the first works on the history of English began to appear at the end of the nineteenth century. A prominent figure of this period is Henry Sweet (1845–1912) who was educated in the Neogrammarian tradition of language analysis and whose approach to English linguistics was largely philological. Sweet, like scholars working in other Indo-European languages, was concerned with documenting and describing the oldest stage of a language, see his An Anglo-Saxon Reader (1876) followed by The Oldest English Texts (1885) and An Anglo-Saxon Primer (1886). The treatment of sound change in such works was largely within the framework of nineteenth-century historical linguistics: the development in Old English of the sound system of Germanic (and further back, of Indo-European) was traced by examining the regularities of change (the application of sound laws) and comparing developments in related Germanic languages. This comparative philology was complemented by considering internal factors in reconstructing the earliest stages of English. By the beginning of the twentieth century enough knowledge of the historical sound system and grammar of English had been amassed for major synthetical studies to be attempted by ambitious scholars in the field. Two monumental overview works of this period deserve mention, the six volume A Modern English Grammar on Historical Principles (1909) by Otto Jespersen (1860–1943) and the two-volume Historische Grammatik der englischen Sprache, ‘Historical Grammar of English’, (1914–40) by Karl Luick (1865–1935). After the Second World War the historical study of English phonology continued with scholars such as Robert Stockwell, working initially within the tradition of American structuralism. With the advent of generative phonology, following the publication of The Sound Pattern of English (Chomsky and Halle Reference Chomsky and Halle1968) the historical phonology of English received new impetus. Scholars such as Richard Hogg, Charles Jones, and especially Roger Lass, began a renewed investigation of the diachronic sound system of English, often concentrating on providing new interpretations of items of change which had already been considered, see the contributions in Lass (Reference Lass1969) and, somewhat later, Lass and Anderson (Reference Lass and Anderson1975), and Lass (Reference Lass1976). By the mid 1970s classical generative phonology had been revised considerably and different versions of non-linear phonology (Durand Reference Durand1990) were the focus of attention and have remained so since. Overview works for the entire history of English or for key periods were produced, see Jones (Reference Jones1972, Reference Jones1989), Hogg (Reference Hogg1992a), and Lass (Reference Lass, Hogg and Denison2006) as representative publications.

Since the 1960s there has been a close link between linguistic theory and phonological studies and this link led to new research into the history of the sound system of English with the advent of two major theoretical models: the first, lexical phonology (Kaisse and Shaw Reference Kaisse and Shaw1985), from the 1980s and the second, optimality theory (McCarthy Reference McCarthy2002), from the early 1990s. The application of lexical phonology to issues in the history of English is associated in particular with April McMahon and culminated in her monograph McMahon (Reference McMahon2000b) while the application of optimality theory to phonological change in English is ongoing and best represented in the work of Ricardo Bermúdez-Otero, see Bermúdez-Otero (Reference Bermúdez-Otero2015) and Bermúdez-Otero and Trousdale (Reference Bermúdez-Otero, Trousdale, Nevalainen and Traugott2012) as representative examples of this research. Bermúdez-Otero is known for his advocacy of stratal optimality theory which assumes the existence of phonological stratification, as does lexical phonology, and which has different domains of application from the phonological word to the higher ‘prosodic’ units of the phrase and utterance. But there remains the essential difference between lexical phonology, which assumes ordered rules for different levels of language, and optimality theory which works with the notion of constraint ranking, this determining the actual sound outputs in a specific language or language period, depending on how this ranking holds. Many variants of optimality theory are closely linked to early language learning and interpret rankings as emerging from fundamental storage and processing mechanisms in first language acquisition.

Apart from this theoretically oriented work, there is also current research on historical English phonology which is theoretically aware but not bound to a particular model, for instance that by Jeremy J. Smith (Reference Smith2007); see also Chapter 25 by Minkova and Zuraw in this volume.

12.3 Expanding the database

Up to the late twentieth century the data examined in historical phonology studies of English has been very largely from England. The Old English database is clearly circumscribed with most texts from the south or midlands with some from the north. With the advent of the Middle English period the situation becomes more diverse with much regional writing available. Nonetheless, this material has been regarded as part of the mainstream database for the history of English. By the Early Modern period, beginning around 1600, the focus moves to forms of English stemming from the south-east. By the Late Modern period, starting around 1800, forms of English in London and the Home Counties represent the core of data used for studies of the language. In addition, non-localizable, supraregional forms of English are given preference over vernacular varieties, either urban or rural.

This situation has had a number of consequences. Forms of English outside the south-east were seen as the domain of dialectology while varieties of English in Scotland (including Scots) and in Ireland were treated as completely separate from English in the south-east. Most of all, varieties overseas were in the main ignored, perhaps with a nodding reference to American English. This situation changed gradually in the 1980s with publications such as Trudgill (Reference Trudgill1986) and overviews of English accents, such as John Wells's (Reference Wells1982) seminal three-volume work.Footnote ² With the increasing scholarly interest in varieties of English worldwide (Kortmann and Schneider Reference Kortmann and Schneider2008), new arenas for the development of English came into view for historical phonologists.

In the following sections several instances of phonological change are presented and discussed with a view to documenting the range of change types available in the history of English and their relevance to phonological change as a whole.

12.4 Examining sound change

It is a truism that details of historical phonetic realization are not recoverable via the orthography and hence attempts at accounting for why change took place are fraught with difficulties. Indeed even with present-day primary data the degree of individual variation and the range of sociolinguistic motivation for particular realizations are too great to say with any certainty why a change is taking or has taken place. However, generalizations can be made over repeatedly attested items of sound change, such as intervocalic voicing, palatalization of stops, the rise of front vowels, to allow scholars to posit general principles of phonological change. It is these principles which form the theoretical backbone of the presentation below.

Like other instances of language change, developments in phonology can in principle be assigned to a motivation internal or external to a language. In many cases dual motivation may be posited. The following two sections consider internal and external arguments, the latter stemming from language contact.

12.5 Changes in syllable structure

Since at least the mid 1970s all phonological analyses appeal, to a greater or lesser extent, to syllable structure when explaining synchronic phonology and diachronic developments (see Hooper Reference Hooper1976). Many changes in the history of English can be best accounted for by reference to syllable structure as the following two instances illustrate.

12.6 Interpreting change: the English long vowel shift

By the Early Middle English period the front vowels of Old English had been unrounded and the retracted /ɑ:/ had largely been replaced by a central variant. Open syllable lengthening (see section 12.5.1) had led to long open mid vowels arising so that the long vowel system looked as follows.

Figure 12.1 Early Middle English long vowels

By the end of the thirteenth century alternative spellings begin to appear in northern England which show e and o as i and u (Wełna Reference Wełna1978: 183–4). This could be seen as the beginning of a series of shifts which represent far-reaching changes in the system of English long vowels.

Both long monophthongs and diphthongs are affected by, and form an outcome of the vowel shift. However, the lexical incidence of long vowels and diphthongs altered radically between late Old English and the beginning of the Early Modern period. The following is a summary of the main movements.

Figure 12.2 Raising of mid vowels; diphthongization of high vowels

There is some disagreement about how the long vowel shift began (Lass Reference Lass1987: 129–31). The view that the mid vowels /e:/ and /o:/ pushed the high vowels towards diphthongization was shared by Karl Luick. Support for this view is forthcoming from the northern dialects of English. Here an unconditioned fronting of /o:/ to a front vowel /ø:/ occurred in the fourteenth century. It is assumed by some scholars, e.g. Roger Lass, that northern ME /u:/ was not diphthongized because /o:/ was not encroaching on its space, having been fronted instead (Lass Reference Lass1987: 226–7).

Figure 12.3 Northern versus Southern Middle English

The notion that the high vowels /i:/ and /u:/ were diphthongized lightly to /ɪi/ and /ʊu/, thus vacating phonological space into which the mid vowels then moved, was supported by Otto Jespersen (Reference Jespersen1909). There has also been dispute about whether the shift occurred for internal structural reasons (an account favoured by Roger Lass) or whether language-external factors, ultimately lying in the social behaviour of speakers, provided the motivation for the shift, see the arguments in Stockwell and Minkova (Reference Stockwell, Minkova, Kastovsky, Bauer and Fisiak1988) contra Lass. These authors also postulate that the lenition of consonants in Old English (Stockwell and Minkova Reference Stockwell, Minkova, Kastovsky, Bauer and Fisiak1988: 356) led to a system of vocalic offglides which provided an impetus for the diphthongization of the long vowel shift.

12.6.1 Problems with mergers and splits

The course of the vowel shift presents problems of phonological interpretation. Various mergers occurred but others did not (see Table 12.2 and discussion above). In particular, the raising of long front vowels has been the subject of discussion. A summary of the movements is given in the following figure.

Figure 12.4 Raising of long front vowels

Table 12.2 Overview of the English long vowel shift

	(1300)	1400	1500	1600	1700	1800	1900
drive	/i:/	/ɪi/	/ei/	/əi/		/ʌi/	/ai/
house	/u:/	/ʊu/	/ou/	/əu/		/ʌu/	/au/
feet	/e:/			/i:/
fool	/o:/			/u:/
beat	/ɛ:/			/e:/		/i:/
foal	/ɔ:/			/o:/		/ou/	/əu/
take	/a:/			/æ:/		/ɛ:/	/e:/
sail	/ai/	/æi/	/ei/	/e:/			/ei/
law	/au/	/ɒu/	/ɒ:u/	/ɒ:/		/ɔ:/

In the sixteenth century ME /a:/ (as in made) merged with ME /ai/ (as in maid) in the London area. But although these may have coalesced with ME /ɛ:/ (as in meat), see previous figure, Modern English shows that the vowels cannot have taken the same path because ME /a:/ and ME /e:/ have different present-day reflexes, namely /ei/ (made /meid/) and /i:/ (mead /mi:d/) respectively. One explanation for the non-merger of these vowels is the assumption that the raising of ME /ɛ:/ (to merge with ME /e:/ as in meet) took place before ME /a:/ reached this value. Another is that different varieties showed different stages of the shift and that pronunciations later arose from more than one source.

1. (5)

   	(A)	(B)	(C)
   meet	e:	i:	i:
   meat	ɛ:	e:	i:
   made	ɛ:	e:	e:
   maid	ɛ:	e:	e:

Samuels (Reference Samuels1972: 147) assumes that the reason for the shift indicated between (B) and (C) was the high functional load of /e:/ due to the homophony of vowels from three sources (meat, made, maid) which was reduced by the shift of /e:/ to /i:/. Evidence for this could be found in traditional dialects of English (Ihalainen Reference Ihalainen and Burchfield1994), e.g. in Scotland and Ireland, in which the vowel in mead has not been raised to /i:/, the last step in the raising of long front vowels.

Some scholars (Labov Reference Labov1994: 349–90, Milroy and Harris Reference Milroy and Harris1980) have pointed out that one may be dealing with near-mergers, or indeed if with mergers, then not perhaps for the entire population speaking a language. There may have been varieties present which did not undergo a merger and which offered a model for the later re-establishment of the distinction between meat and mate after the sixteenth century.

12.7 Change as addition: epenthesis and metathesis

Epenthesis, the addition of sounds (vowels or consonants) to words, has a long history in English. At least from Middle English onwards consonantal epenthesis can be seen in the /-t/ which is added at the ends of words ending in a nasal or fricative as in the following examples (the last two are non-standard forms, Jespersen Reference Jespersen1909: 218ff.).

1. (6a)

   OFr paysan

   →

   late ME peasant

1. (6b)

   OFr cormoran

   →

   late ME cormorant

1. (6c)

   ME agains

   →

   EModE against

1. (6d)

   ME amyddes

   →

   EModE amidst

1. (6e)

   EModE onst

   ∼

   ModE once

1. (6f)

   EModE twist

   ∼

   ModE twice

One possible reason for this epenthesis is that the addition of a stop increases the consonantal nature of the syllable coda, thus optimizing the structure of the syllable (Blevins Reference Blevins and Goldsmith1995). Consonant epenthesis can also occur word-internally but with the same apparent motivation, compare English thunder – with /-nd-/ – and German Donner. A further instance of this type of epenthesis is the insertion of a nasal in an unstressed syllable before /ɡ/, i.e. /ˌV.ɡV/ → /ˌVŋ.ɡV/ as in ME nightigale → nightingale; ME messager→ messenger (Jespersen Reference Jespersen1909: 35ff., Dobson Reference Dobson1968: 1004).

Vowel epenthesis involves the insertion of an unstressed schwa vowel to break up a heavy coda, one consisting of more than one sonorant. This type of epenthesis is not found in standard English today but is attested for historical varieties of English and indicated in spellings like alarum ‘alarm’ in Shakespeare.

Metathesis is the reversal of the linear sequences of sounds in a word. It can involve consonants or a consonant and a vowel. For the latter type one usually has a short vowel and /r/, e.g. ME bridde→ bird; ME wreoc∼ weorc ‘work’. For many varieties there is a further condition that the vowel be unstressed, e.g. Irish English modern [ˈmɒdɹən].

In the history of English, metathesis is common and is attested widely for Old English (Campbell Reference Campbell1959: 184–5). There are many instances involving the clusters /sk/ and /sp/.

1. (7a)

   OE fiscas

   ∼

   fixas

   ‘fish’

1. (7b)

   OE waps

   ∼

   wasp

   ‘wasp’

Modern English ask goes back to a metathesized form of OE acsian which in its turn derived from ascian. The form with /-ks-/ existed at the time when the palatalization of /sk/ to /ʃ/ removed all instances of /sk/ in southern varieties of Old English, cf. OE disc (< Latin discus) /diʃ/ ‘dish’. In Early Middle English the metathesis of /ks/ reintroduced the /sk/ cluster (c.1200, Onions Reference Onions1966: 54) long after the palatalization rule had ceased to be active and so there was no further shift of /sk/ to /ʃ/ (Jespersen Reference Jespersen1909: 25).

12.8 Change as a loss of distinctions

12.9 Long-term processes

Some features of Old English were lost, never to re-occur in any first-language variety of English, e.g. consonant geminates. Other features formed part of a process which has not been concluded yet. The set of word-initial sonorants included the approximant /w/ which could be preceded by /h-/ as in hwic ‘which’ which had a devoicing effect on the approximant. The distinction between /w/ and /hw/ (phonetically [ʍ]) survived longest and was maintained well into the Early Modern period. By the eighteenth century the loss of voicelessness with these segments was clearly advanced. Prescriptivists such as John Walker, author of A Critical Pronouncing Dictionary (1791), commented on this, noting that many speakers did not distinguish between which and witch. The loss of /hw-/ [ʍ] is all but complete for varieties of Present-day English with only vernacular forms of Irish and Scottish English maintaining it, see Hickey (Reference Hickey1984) and Stuart-Smith (Reference Stuart-Smith, Kortmann and Upton2008) respectively.

The voicing of [ʍ] would seem to be linked to the loss of [h] in English: the glottal fricative has been restricted to stressed syllable-initial positions given the loss syllable-final /h/, from an earlier velar fricative, in words like ruh ‘rough’. Today it can be found in words like hat /hæt/ and behave /bɪˈheɪv/ in those varieties which have the glottal fricative. But for virtually all urban varieties in present-day Britain /h-/ does not exist. There is evidence for the early loss of /h/ (Milroy Reference Milroy1992a: 137–45) in phrases like to eat humble pie from umbles ‘entrails’ which shows the hypercorrect insertion of initial /h-/.

The first segment in /hw/ correlates with /h/ word initially, that is, to postulate /h/ + /w/ has additional justification in the fact that initial /h-/ occurs anyway (in all varieties with [ʍ]). Conversely, no variety of English which has /h/-dropping also has [ʍ], that is the lack of /h-/ precludes the cluster /hw-/ [ʍ]. There is a further argument from syllable position. Generally, there is an increase of sonority from edge to centre. Analysing [ʍ] as /hw/ means that one has a fricative /h/, then a glide /w/ (a continuant with open articulation) and a following vowel which is in keeping with the sonority cline for sound segments. Furthermore, English does not have voiceless sonorants or glides, so that to posit /ʍ/ would mean that there would be an unevenness in the distribution of sonority going from syllable edge to centre as can be seen from the following two phonological interpretations of [ʍ].

1. (8a)

   [ʍ] = /hw/
   – – – – – – – #					(sonority)
   voiceless	# – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –
   obstruents	# voiced	voiced	voiced	voiced
   	# obstruents	sonorants	glides	vowels
   [h]			[w]
   Syllable edge – – – – – – – – – – – – – – – – – – – ––> Centre

1. (8b)

   [ʍ] = /ʍ/
   – – – – – – – #			# – – – – – – # (sonority)
   voiceless	# – – – – – – – – – – – – – – –#			# – – – – – – –
   obstruents	# voiced	voiced	voiceless	voiced	voiced
   	# obstruents	sonorants	glides	glides	vowels
   			[ʍ]
   Syllable edge – – – – – – – – – – – – – – – – – – – ––> Centre

12.10 Phonotactics

The remarks made in the previous sections have involved sequences of segments in word-initial position, i.e. they concern English phonotactics, the permissible sequences of sounds in words of the language. Sequences in syllable onsets and codas follow fairly strict patterning and generalizations can be made concerning what type of sounds can occur in what slots. The greatest variation is found in single-element onsets which can have any sound of English bar /ŋ/ and /ʒ/ (if one excludes recent French borrowings like genre [ʒɑ̃ɹə]). For double-element codas there are restrictions: the first sound must be an obstruent and the second a sonorant, except with initial /s-/ which can be followed by a stop. Sequences of two fricatives are prohibited and historically have been changed to a stop + fricative, cf. buhsum /-xs-/ → buxom [bʌksəm]. The only exceptions to this are a few words of Greek origin such as sphere with /sf-/. Three-element onsets must consist of /s/ + stop + sonorant/approximant (Lass Reference Lass1987: 92).

1. (9a)

   /s-/	sit, set, sat, suit, sowed

1. (9b)

   /sp-/

   spit,

   /sl-/

   slit

1. (9c)

   /spl-/

   split,

   /spr-/

   spring, spew /spju:/

The statement just made refers to prototypical phonotactic structures in English. However, there are different sequences attested in loans from other languages. For example, /ʃ/ + sonorant sequences are found in a few German or Yiddish loans, e.g. schnitzel, schmusen, schlock [ʃn-, ʃm-, ʃl-].

Furthermore, certain phonotactic positions are only occupied by words of foreign origin. For instance, /z-/ in initial position only occurs in words which are ultimately of Greek origin, e.g. zoo, zone, zeal, zero, xenophobia, along with a small number of words which have a certain phonoaesthetic value, e.g. zip, zap, ziltch, zombie, and some proper names, mostly of African origin, such as Zulu, Zambia, Zimbabwe. Other sounds do not occur in certain syllable positions at all. The voiced palato-alveolar sibilant /ʒ/ is found word-medially, as in vision, pleasure, fusion, or word-finally as in rouge, prestige, but does not occur word-initially (see remarks above).

Historical changes in phonotactics are documented for English. For instance, the sequence /fn-/ was found in Old English in words like fnēosan ‘sneeze; gasp’, fnær(ett)an ‘snore’ (Lutz Reference Lutz1991: 234) but was later lost either by a shift of /f-/ to /s-/ or due to replacement by another word, e.g. fnēosan ‘gasp’ → Old Norse gasp or Anglo-Norman pant (Lutz Reference Lutz1991: 236).

Simplification of onsets also occurred, e.g. the initial cluster /kn-/ and /gn-/ lost the velar stop, cf. knee and gnaw, both with /n-/ in Modern English (Wełna Reference Wełna1978: 227–8). Foreign words with /kn-/ generally show an epenthetic vowel between the two sounds to break up the cluster, e.g. Knesset (Israeli parliament) [kəˈnesət].

12.11 Relative chronology of change

In the present chapter various changes have been discussed which, for a proper understanding of their outcomes and how they affected the lexicon of the language, must be set in order of occurrence. Ordering changes in a specific time sequence relative to one another is known as ‘relative chronology’. A simple example illustrating how this works concerns two major changes in Old English phonology.

Figure 12.5 Palatalization and i-umlaut in Old English

It is obvious that palatalization preceded umlaut; otherwise the pronunciation of the word for ‘king’ would be [tʃin], that is, the process of palatalization would have appeared to have become inactive before i-umlaut set in so that those words which experienced i-umlaut did not go through palatalization.

A further example is provided by the tendency in the history of English for long /u:/ to be shortened, usually before a single-stop coda. This started in the Early Modern period and continues to the present day. The forms affected by this change differ in their realizations today depending on when they experienced the shortening.

Figure 12.6 Vowel shortenings in the history of English

With these changes one can specify the phonetic environment in which they took place. The earliest shortening affected /u:/ before /d/. It took place before the general lowering of /u/ to /ʌ/ in Southern English in the Early Modern period and hence underwent this latter change. After this shortening came that of /u:/ before /k/. This took place after the lowering of /u/ to /ʌ/ had become inactive, hence the pronunciation /bʊk/ for book and not /bʌk/. The shortening before /m/ has not been completed yet as can be seen from words which have variable realizations in British English: room /rʊm/ or /ru:m/.

A third example concerns the long vowel shift which began in the late Middle English period. By this time most of the French loans (Norman and Central French) had already entered the language and thus underwent the shift, e.g. doubt /daut/ from an earlier /du:t/. However, a significant number of loans were not affected and so one must assume that they were borrowed after the shift had been completed.

Figure 12.7 French loans and the long vowel shift

One must also consider the operation of later analogy. There are a few instances where orthographic ou is realized as /au/ for example with route /raut/ in American English whereas British English still has /ru:t/.

Relative chronology is also apparent when dealing with borrowings, for instance, English wine /wain/ is ultimately a Latin loanword, vinum, borrowed in continental Germanic when Latin v was /w/. The word vine /vain/ is a later borrowing of the same word in the Middle English period from Latin via Old French where the pronunciation of v was /v/.

Figure 12.8 Latin /w/ and /v/

12.12 Conclusion

The history of the English sound system offers a good perspective from which to observe, classify, and interpret types of phonological change. The extensive research carried out on the language ensures that the data and its interpretations are well documented and accessible to scholars wishing to advance the insights of this research themselves. Current phonological research appears to involve two main tracks, one which looks at phonological change from the standpoint of present-day theory, above all optimality theory in its various manifestations, not least stratal optimality theory (Bermúdez-Otero forthcoming). Another track sees the sociolinguistic scenarios which prevailed during periods of change as causally involved in this change. This approach is frequently synchronic at the outset and extends findings backwards in time to developments located in history. With regard to methodology, there is also the burgeoning field of sociophonetics which may come to apply its insights into speaker variation (Docherty and Mendoza-Denton Reference Docherty, Mendoza-Denton, Cohn, Fougeron and Huffman2012) to historical situations, such as the long vowel shift, to cast light on the possible motivation and trajectories for such instances of change. Lastly one could point to the increasing interest in the community of phonologists in corpus-based phonology; see the contributions in Durand et al. (Reference Durand, Gut and Kristoffersen2014) as a heuristic for understanding the complexities of phonological change.

13.1 Introduction

Lexical change is a constant feature of living languages in changing societies, and is one of the most noticeable kinds of language change. Over the years, it has attracted attention from both linguists and non-linguists. Nowadays, for example, newspaper editorials and internet message boards are full of discussions of new words, particularly slang terms or expressions which indicate shifting social situations or current cultural phenomena: for example, the British neologism benefit tourism, reflecting concerns about visitors seeking to exploit the welfare state. Interestingly, similar concerns among speakers of other languages may result in similar linguistic innovation, as in German Sozialtourismus with the same meaning.Footnote ¹

Changes in lexis can be broadly divided into two related strands: innovation and obsolescence. Innovation in the lexicon takes place when new words become established within or across groups of speakers, or when new senses emerge for existing words. For example, two relatively recent new words in English adopted by some speakers are the noun bromance and the related adjective bromantic, which appear to be found from the 1990s onwards in British and American English. The two terms are blends, or portmanteaus, of the noun bro (short for brother) and romance or romantic, and usually describe close non-sexual relationships between two men; they are often used of films or television programmes which centre on close male friendships. New meanings for existing words are also common. Sick is now used by some speakers to mean ‘fantastic, awesome’, showing a similar semantic development to other adjectives like bad and wicked, which have earlier negative meanings but are now also used in positive senses. According to internet discussions, the new meaning of sick emerged in the 1990s, although it is often difficult to establish precise dates for this kind of innovation in language. Comparable here is Finnish sairaan ‘sickly’, which has followed a similar semantic development, possibly under the influence of English.Footnote ²

As all of these examples illustrate, lexical innovation can initially be restricted to particular groups of speakers, such as young people or people in particular professions, and may or may not become more generally established through time. The same is true of the opposite process, obsolescence, whereby words or word meanings gradually become less common and die out. Again, it can be very difficult to be confident about when this has happened, because of lexical variation across groups of speakers. For example, the noun wireless, meaning ‘radio’, is no longer frequently or commonly used, but can still be found in the lexicons of some older speakers. It seems likely that this use will eventually die out entirely as fewer and fewer speakers are aware of it and the concept of ‘wireless technology’ becomes more widespread. A word which is now completely obsolete, and only known by speakers as a historical form, is the verb wray ‘reveal, betray, accuse’ (see OED, wray v1). This is found in various spellings from the Old English (OE) period until the late sixteenth century; it occurs, for example, in ‘The Miller's Tale’ in Chaucer's Canterbury Tales, in the line ‘if thou wreye me, thou shalt be wood’ (‘if you betray me, you shall be mad’) (Benson 2008: 72, line 3,507). The Oxford English Dictionary (OED) records a very late example in the phrase to wree against a person in a northern dialect glossary from 1781, but the word seems to have fallen out of general use by this time, and is not recorded in modern synchronic dictionaries like the Oxford Dictionary of English and the American Heritage Dictionary of the English Language. Wray is a fairly typical example, since many words found in Old English are obsolete in later varieties.

A final point to note is that not all areas of the lexicon are equally susceptible to change. Lexis is often divided into open and closed class words: open class words are those in the ‘major’ word classes, that is nouns, verbs, adjectives, and adverbs, while the closed class includes ‘grammatical’ words such as pronouns, articles, prepositions, and conjunctions. Although there has been some change in all of these groups across the history of English, the open class has changed much more frequently, and is very much more accepting of innovation. Within the open class, we can make a further contrast between core or basic vocabulary, that is between the ‘everyday’ words that are used frequently across all registers, and less common vocabulary that is more restricted to particular registers, such as the technical vocabularies of science, law, linguistics, etc. The distinction between these two layers is not clear-cut, but it is nevertheless useful to think of core vocabulary as a category structured prototypically, with some uncontroversial members but blurry edges. We can recognize a significant difference between words like water and love, which are found in Old English and survive to the present day with relatively stable basic meanings (albeit with spelling variations in different periods), and words like edamame and unimpeachability, which are relatively new to English, much more restricted in their use, and much less commonly used.

13.2 Lexical change in the history of English

One way of tracking how the lexicon has developed is by examining groups of related words arranged in semantic categories, as in the Historical Thesaurus of the Oxford English Dictionary (HTOED, Kay et al. Reference Kay, Jane, Samuels and Wotherspoon2009). While no two categories follow exactly the same path of development, those for concepts with a long history of lexicalization often reveal a good deal about innovation and change in the vocabulary. The combination of HTOED with the more detailed information about individual words available in the OED makes a powerful tool for lexical analysis, especially when reinforced by the increased availability of historical online corpora large enough to enable lexical research by allowing further scrutiny of contexts (see further Chapter 8 by López-Couso in this volume).

While discussing structuralism and its relationship to the historical-philological tradition in semantics, Geeraerts writes:

To some extent, combining the OED and HTOED as sources unites these two approaches. To illustrate this point, we begin by demonstrating what kinds of information can be extracted from words listed in HTOED with the meaning of ‘farmer’, as shown in Figure 13.1. The list consists of a main group of words with the general meaning, followed by subcategories where the meaning is more specific. Within each list, words are arranged in chronological order, starting with Old English, dated simply OE. For later material, the first, and, where relevant, last recorded dates of use are given, as are OED style labels such as (orig. US). Three points should be noted before reading the analysis. First, while ‘word’ is a convenient shorthand term for a lexical unit, it would be more accurate to refer to ‘meaning’ or ‘sense’, since each meaning of a polysemous word appears in the semantically appropriate HTOED category. Second, dates in the print HTOED, from which Figure 13.1 is taken, correspond to those in the second edition of the OED and may differ from those in the online third edition, where revision is underway. Third, the OED as a matter of principle omits all OE words which did not survive into Middle English (ME) while the print HTOED includes them, thus giving a fuller picture of the lexicon overall.

Figure 13.1 HTOED section 01.02.08.03.01 (noun) Farmer.

Copyright University of Glasgow 2009

Lexicalization of the concept of ‘farmer’ dates back to the Anglo-Saxon period, when, from the fifth century ad onwards, settlers from various parts of north-western Europe began to arrive in the British Isles, speaking the Germanic languages which developed into the dialects of Old English. Provision of food for themselves and their families must have been an immediate concern, so it is not surprising that vocabulary in this area is well-attested from an early date, as is vocabulary in related categories such as Hunting and Fishing. It is interesting that the Dictionary of Old English (DOE) glosses both bigenga (see big-genga) and buend as ‘inhabitant…husbandman’, reminding us how the two categories overlapped in early societies, when many inhabitants were presumably also farmers.

During the Anglo-Saxon period, English was influenced by Scandinavian languages as a result of Viking incursions along the coasts. (For further information on English's contact with other languages, see Chapter 19 by Trudgill in this volume.) There is little evidence of Scandinavian influence in the Farmer category except for the rare toftman in subcategory 13, which comes from the Old Norse word for a homestead. Borrowing generally took place during everyday interaction in the spoken language, and many Norse borrowings, such as the better-attested sky and skirt, do not appear in the written record, which was primarily the domain of the educated clerical elite, until Middle English. There is also little evidence here of influence from the indigenous Celtic languages, although a Welsh origin has been suggested for tike/tyke in subcategory 14 (see OED tike/tyke noun 1). Lexical influence from these languages is rare in the recorded OE vocabulary as a whole.

By far the greatest influence on the early vocabulary of English was the Norman Conquest of ad 1066, after which much of the OE lexicon disappeared along with the institutions and cultural practices it described. Only two words from our Farmer group survived this onslaught, neither for very long: OE tilia as ME tilie, and OE æcerman as ME acreman. The words in general use for the concept in Middle English were husband, until the eighteenth century, and its derivative husbandman. Husband was formed in late Old English under Scandinavian influence with the meaning ‘male head of a household’ (OED sense 1) but became polysemous in Middle English through narrowing of meaning to focus on particular functions performed by such a person. It is recorded in c.1220 in the meaning of ‘farmer’ (OED 3a), in c.1290 as a male marriage partner (OED 2a), and from the fifteenth century onwards with reference to various managerial positions such as a housekeeper or steward (OED 4a, 5). Husband can thus be viewed as a network of meanings radiating from a central core, or, in historical terms, as a chain of meanings, each linked to its predecessor by at least one semantic feature. The structure of such a chain may be less clear if intermediate meanings become obsolete in the course of history.Footnote ³

The development of the ‘manager’ meaning of husband may have led to what is sometimes called polysemic clash, when polysemes of close meaning introduce ambiguity into discourse, leading to further semantic development or obsolescence. As the OED citations show, the ‘farmer’ and ‘manager’ meanings were commonly prefaced with adjectives such as good and bad to indicate capability in performing these roles, and it must often have been difficult to know which meaning was intended. Perhaps because of this, neither meaning survives in Present-day English, leaving the field clear for the ‘male marriage partner’ meaning and leading to the introduction of farmer to fill the agricultural gap. The fate of the associated verbs is somewhat different, however, with phrases like ‘husbanding one's resources’ (i.e. managing them well and economically) remaining in use, while ‘husbanding’ or tilling the soil has dropped out and ‘husbanding’ a wife never achieved general currency despite two uses by Shakespeare. The abstract noun husbandry still has some currency, but its use is increasingly restricted to the care of animals, especially in the collocation ‘animal husbandry’ (see, e.g., WebCorp: Synchronic English Web Corpus).

For husbandman, the ‘farmer’ meaning dominates from the outset, although the OED offers a few citations for ‘head of household’ and a single one for ‘marriage partner’. It could have filled the gap left by the obsolescence of husband ‘farmer’, but from the sixteenth century onwards it has gradually been replaced by farmer, the first word from French to appear in our list and now the core member of the group. The obsolescence of husbandman may possibly be attributed to the fact that, in some parts of the country, it referred to a tenant farmer, a person of lower social status than a landowner. A glance at other HTOED terms for agricultural workers, such as tillers and ploughmen, shows a continuing predominance of OE lexis for these lowlier occupations.

Farmer is first recorded in the late fourteenth century in the meaning of ‘tax-collector’ and progresses from there by a series of semantic shifts to mean someone who cultivates land for an owner (OED 4), a tenant farmer (OED 3), and, in 1599, ‘One who “farms” land whether tenant or owner or makes agriculture his occupation’ (OED 5a). It is interesting that the OED definition, written in 1895, puts farm in inverted commas, presumably to identify it as a relative neologism, since the verb in this meaning is not recorded until the eighteenth century, when it is defined as ‘To follow the occupation of a farmer; to till the soil’ (OED verb 2, 5). Till itself would be regarded as an archaism nowadays.

Examining farm and farmer leads to a possible example of homonymic clash, where two words of different origins happen to have the same form and are semantically incompatible. As with polysemic clash, homonymic clash can lead to obsolescence or change of meaning. In Middle English, factors such as loss of OE grammatical endings, lack of standardized spelling, and new words entering from French sometimes combined to create situations where originally distinct forms ended up sharing pronunciation or spelling or both. In the present case OED verb farm¹ is the ME descendant of OE feormian, meaning to ‘cleanse, empty, purge’, often with reference to privies. Farmer¹ occurs with reference to those who carry out this task (OED noun 1, occurring in the compound gong-farmer, from OE gang ‘privy’). Both of these meanings were apparently obsolete by 1600, apart from a single dialect record of the verb, perhaps because of the developing meanings of OED farm² and farmer²: one would not wish one's activities as a tax-collector or agriculturalist to be confused with those of an emptier of privies. A further casualty may have been OE feorm, meaning food or a feast, obsolete after the fourteenth century, whose ME spellings also overlapped with those of farm (OED farm noun 1).

Any attempt to identify homonymic clash is speculative, and often requires considerable linguistic and historical research.Footnote ⁴ As Fischer (Reference Fischer, Christian, Hough and Wotherspoon2004) points out, words meaning ‘toilet’ are affected over the years both by technical developments and by a steady stream of euphemisms intended to divert the mind from the more unpleasant aspects of the topic. In the case of farmer², a glance at the HTOED category Emptying/cleaning of privies shows the gradual disappearance of several compounds in -farmer, such as jakes-farmer and dung-farmer, and their replacement by the innocuous nightman and euphemistic gold-finder.

To return to Farmer, from the seventeenth century onwards we see more words of Latin origin, such as colon, agricole, and agriculturalist, entering English, sometimes mediated by French. This trend reflects both greater familiarity with the classical languages as a result of the Renaissance, and an increasingly scientific approach to agriculture. It is further reflected in subcategory 02 farming scientist, where the Greek prefix agro- ‘field’ is favoured in the proliferation of later coinages. We also find two words with feminine suffixes, farmeress, which probably meant a farmer's wife, and farmerette, cited from 1918 to 1963 and defined by the OED (in 1972) as ‘A woman or girl who farms land; a farmeress’. Such coinages were common enough in the nineteenth and early twentieth centuries, when women began to enter a range of professions, but are increasingly rejected in more egalitarian times. Throughout the rest of the sample (which contains only the first fourteen of twenty-seven subcategories), we find words for different kinds of farmer, including those from world varieties as English took root around the globe, such as cockatoo and wool hat in 09 small farmer.

Although no single HTOED category can come close to representing all the processes which have influenced the development of the English vocabulary, the Farmer category illustrates many of the tendencies which can be found throughout the lexicon, such as the obsolescence of OE words, innovation in response to cultural change, and the development of new meanings through polysemy. Categories of words with more abstract referents are less prone to fine differentiation of types of the main category than categories like Farmer, but may use other processes like metaphorization to fill gaps or supply a range of synonyms (see further Kay and Wotherspoon Reference Kay and Wotherspoon2005). Figure 13.2, the HTOED section Difficult, provides an example.

Figure 13.2 HTOED section 01.05.05.11 (adj.) Difficult.

Copyright University of Glasgow 2009

With the single exception of uneath, all the OE adjectives in this category had disappeared by the early thirteenth century. Uneath survives until the eighteenth century, possibly because it is strongly partnered by its antonym eath ‘easy’ in a range of related meanings (OED eath/eith), and had several adverbial uses. In Middle English, the meaning ‘difficult’ is expressed initially by narrowing of OE words with meanings of general badness or opposition, such as evil, hard, and dear, or the Old Norse ill. French appears with uneasy in 1398 (OED 3), which had entered the language earlier in meanings such as ‘uncomfortable, disagreeable, annoying’ (OED 1, 2). Forms of difficult occur shortly thereafter: difficile from French, difficult and difficul either from the same source or possibly back-formations from the noun difficulty, recorded in English before 1400. However, the category as a whole is less notable for borrowed words than for the metaphorical extension of existing words, especially from concrete to abstract hardness, as with hard itself, craggy, tough, uphill, and so on. Uneasy too had a meaning ‘Difficult to traverse on account of ruggedness, steepness, or other obstacles’ attested between 1566 and 1749 (OED 3c); the Middle English Dictionary (MED, accessed via OED3) offers three earlier citations in this meaning, covering c.1390–a1500 (MED 2a). A subcategory of these metaphors suggests a particularly intrusive kind of hardness, as in spiny and its approximate synonyms, prickly, bristling, and rugged. Overall, these examples show the tendency, much discussed in cognitive linguistics, for metaphorical pathways to become established in a language and attract new members.Footnote ⁵

13.3 New words: borrowing

As section 13.2 shows, the lexicon of English has changed substantially over the years, gaining large numbers of words by a variety of means. One of the major sources of new lexis in English is borrowing, the adoption of words from other languages, often with changes to form, such as spelling or pronunciation, and/or changes in meaning. Present-day English includes words from languages as diverse as French (croissant, garage), Czech (robot, hacek), Japanese (kamikaze, sensai), and Swahili (safari, ultimately from Arabic). Because English has borrowed so widely, semantically related words in Present-day English are often formally unrelated: for example, corresponding to the native Germanic form ear is the borrowed form aural; similar pairs are heart: coronary and mind: mental. Scholars such as Kastovsky (Reference Kastovsky and Hogg1992: 294) describe the lexicon of Present-day English as ‘dissociative’ in this respect.Footnote ⁶ By contrast, Old English borrowed very little and has an ‘associative’ lexicon, in which a high proportion of semantically related words are also formally related by derivation or compounding, processes discussed further in section 13.4. The adjectives cildisc and cildlic ‘childish’, for example, are clearly related to the noun cild ‘child’, as is cildgeong ‘child-young, youthful’, where modern English might prefer infant or juvenile in some contexts, as both childish and childlike have developed more specialized meanings. Hybrid forms, combining elements from more than one source, also occur, as in lovable (OE stem and French suffix), nobleman (French and OE nouns), and ungracious (OE prefix and French stem).

It is often assumed that more words are borrowed nowadays than in any earlier period; while it is certainly true that English today has loanwords from a wider variety of languages than ever before, it may not be true that borrowing is a more frequent process. From the very earliest records there is evidence that English adopted and adapted words from other languages, and Durkin (Reference Durkin2014: 300) presents data suggesting that, as a proportion of all new words in English, loanwords have shown a steady decline since the end of the seventeenth century and a steeper decline since the end of the nineteenth century. It is, however, difficult to compare the kind of borrowing that can be found in different periods in terms of raw figures. The context in which borrowing takes place, the languages from which English has borrowed, and the motivation for borrowing, have all changed from period to period in interesting and varied ways. Our knowledge of borrowing in earlier periods is necessarily affected by the nature of the available evidence. It often seems to be the case that a word is borrowed into the spoken language and later passes into written use, or it may first have appeared in an unrecorded and possibly restricted context, such as the vocabulary of a particular trade. Any generalizations about spoken language are handicapped by lack of evidence, since it is not until Early Modern English that we find significant amounts of data which approximate to speech, such as personal letters, play-texts, pamphlets, and court proceedings.

As we saw in section 13.2, it is important to consider the external history of English and English speakers when examining how and why borrowing has taken place: such considerations often explain patterns of borrowing and show how cultural change can promote lexical change. The picture is particularly complicated in Middle English, where the three main languages of English, French, and Latin interacted with one another, as shown, for example, in macaronic texts of the period. Rothwell writes:

He goes on to argue that Middle English speakers would have regarded the use of any or all of these languages in particular contexts as a natural result of the complex political situation after the Norman Conquest, and that ‘assimilation’ is a better term than borrowing for the process by which French words became part of English (1994: 45–67). (See Chapter 24 by Machan in this volume.)

Words which occur in English texts in the early ME period are clearly associated with the functions of French as the language of the ruling elite. They tended to be of relatively high prestige, particularly in comparison to any native alternatives, and many were concentrated in particular semantic fields, such as royalty (royal, royalty, reign, monarch), the nobility (noble, nobility, court, retinue), and literature (poetry, poet, romance, rhyme). Prestige borrowings, as opposed to those motivated by need or at least usefulness, led to the possibility of selecting one term over another for stylistic purposes. Between the Early Modern and Late Modern periods, the fairly elaborate system of register that exists in Present-day English developed, relying on choices in the lexicon between partially synonymous words which are recognizably different in formality and markedness in particular contexts, as in ask, question, interrogate from English, French, and Latin respectively (see further Minkova and Stockwell Reference Minkova and Stockwell2009).

Borrowing can of course take place without face-to-face contact between native speakers of the source and target languages. The major source of loanwords in the OE period is Latin, not primarily because of contact between English and Latin speakers but because some English speakers were bilingual in English and Latin, in that they used Latin as a lingua franca in writing and (to some extent) speech. Following the spread of Christianity through the Anglo-Saxon territories from the sixth and seventh centuries, Latin was the predominant ecclesiastical language; it was also used in England and elsewhere in Europe as a language of permanent record. For both of these reasons, many people encountered and used the language daily in specific situations and for particular purposes, and this led to some borrowing of Latin words, though on a relatively small scale. The HTOED section 03.07.00.18.03.02 Disciple records two OE nouns meaning ‘disciple’: the native leorningcniht and the Latin loanword disciple, which became the most usual term for this concept and is the one commonly used in Present-day English. Other words borrowed in this period relating to religion and religious practices or to writing and records include Pentecosten, the name for the Jewish and Christian religious festival, from Latin pentecoste; biblioþece, meaning both ‘library’ and ‘scriptures, bible’, from Latin bibliotheca; and carte ‘sheet of writing material’, from Latin charta. Latin loanwords tended not to affect more basic OE vocabulary: for example, although it contains many more entries than Disciple, no borrowing from Latin is found in section 01.01.05.05 Sea/ocean.

As we have seen above, Latin remained in use in the Middle English period. A new development, found to some extent in Chaucer (c.1340–1400) and culminating in the work of later poets such as John Lydgate (c.1370–1449/50?) and William Dunbar (1460?–1513×30), was the development of a literary style marked by ‘aureate diction’ in the form of polysyllabic Latinate words such as melodious, saturnine, tenebrous. This diction had two purposes: first, stylistic enhancement, and, second, and more practically, to supply words which fitted iambic metres at a time when English had become increasingly monosyllabic. Larger-scale importation from Latin occurred in Early Modern English, again related to the functions of the two languages, and the ways in which the classical languages were regarded at the time. Latin was still the major language of scholarship around Western Europe, so that major scientific works (in the broadest sense) were written in Latin. There was also huge interest in classical texts, which would have been familiar as part of the education of literate young men. At the same time, borrowing from Latin, sometimes via French, was encouraged by the many contemporary translations into English, and by a widespread feeling among English speakers that their language lacked both elegance and a full vocabulary. Latin terms were commonly used in a naturalized form in translations rather than English equivalents, and were felt to lend prestige to English texts. Some of the earliest printed texts show the popularity of translations: for example, Caxton in the late fifteenth century published versions of Virgil's Aeneid and Ovid's Metamorphoses, as well as translations of French texts. His work shows the influence that translation can have on lexical development, since in choice of both vocabulary and spelling he has a propensity to follow the language of his source (Blake Reference Blake1969: 136–41, 172–3). In non-translations, too, there are large numbers of new English words borrowed from Latin and French. This sudden influx of innovations led to the publication of monolingual dictionaries, the earliest of which specifically recorded and defined ‘hard words’, i.e. mainly Latinate loanwords. Among those first attested in this period are relatively common survivors such as calculate, education, radiate, and sequence, as well as large numbers of rarer or more specialized words such as ingest, oscillate, and quadrilateral.

A fuller account of borrowing patterns in the history of English, which discusses the kinds of morphosyntactic and semantic changes that loanwords undergo, can be found in Durkin (Reference Durkin2014; see also Chapter 23 by Durkin in this volume). Discussions of borrowing in particular periods are available in chapters on vocabulary in the multi-volume Cambridge History of the English Language: for Old English, Kastovsky (Reference Kastovsky and Hogg1992); for Middle English, Burnley (Reference Burnley and Blake1992); for Early Modern English, Nevalainen (Reference Nevalainen and Lass1999a); and for late Modern English, Algeo (Reference Algeo and Romaine1998).

13.4 Processes of word formation

As the examples in section 13.2 show, the alternative to borrowing words from other languages in order to add new forms to the lexicon of a language is to coin new words from forms which are already established. Across the history of English, a number of different processes of word formation can be observed.

13.5 Tendencies in semantic change

Semantic change, the process whereby words develop different or additional meanings, is a key factor in the history of English. It is also a focus of interest in cognitive semantics (see Chapter 4 by Trousdale and Chapter 15 by Fitzmaurice in this volume).Footnote ⁹ Traditionally, however, it has been considered problematic as a subject of study, because, unlike phonological or morphosyntactic change, it tends not to follow regular patterns. For example, in recent times the adjective gay has acquired a new sense, ‘homosexual’, and the earlier sense ‘happy, cheerful’ has fallen out of use for many speakers. Other words with the same earlier meaning have not undergone the same shift: it is not possible to use either happy or cheerful to mean ‘homosexual’, and we thus cannot make any generalization here about a regular change in the linguistic system of English. The reason for this particular instance of semantic change, and for many others, relates to non-linguistic history, and is difficult to pin down and explain. However, semantic change does not appear to be random, and traditionally a number of fairly broad tendencies have been observed. These can be helpful in making sense of individual word histories, and although they sometimes mask the complexity and messiness that can be involved in semantic change, they still provide a useful starting point for describing the ways in which word meaning can shift over time.

Perhaps one of the most common tendencies is the shift to a more or less restricted meaning, changes known as narrowing (or specialization) and widening (or generalization). Widening is the process whereby the meaning of a word (or one of its senses) becomes more general, so that it can be used to refer to a larger, less specific concept. A well-known and often-cited case is the noun bird, the reflex of OE bridd. In Old English, the usual sense of bridd is ‘young bird, young of a bird’; words for the general meaning ‘bird’ include fugel (modern English fowl) and fleogende ‘flying creature’. For modern speakers, bird refers to all feathered vertebrates, and the OE sense has been lost. The opposite, and more common, process, narrowing, occurs when a meaning becomes more restricted over time and is used to refer to only a subset of the concept described in its earlier meaning. OE mete, to take another well-known example, is the ancestor of Present-day English meat, but is roughly the semantic equivalent of modern food, since it could be used to describe any kind of edible substance, not only flesh from animals. The earlier meaning survives marginally in the phrase ‘One man's meat is another man's poison’.

Another pair of tendencies shows the way in which some word meanings change connotatively, so that their senses become more or less positive over time. Amelioration occurs when a word sense shifts from negative or neutral to positive, or from negative to neutral. A classic example is the adjective nice, a loanword from French (ultimately from Latin). In Middle English, when the word is first attested in the OED, it is used to mean ‘foolish, silly’ and can describe unpleasant entities like sin or strife (OED 1); it is also found with other negative meanings including ‘wanton, lascivious’ (OED 2). By the eighteenth century, after further semantic shifts, it is attested with its Present-day English sense ‘agreeable, pleasant, satisfactory’ (OED 14), which has become the dominant sense for most speakers. The opposite tendency, pejoration, appears to be more common in English, though it is difficult to make any definitive statement about the frequency of either shift. A relatively modern example is the noun cowboy, which is found in English from the eighteenth century. In its earliest uses, it has the neutral meaning ‘boy who tends cows’, and this broadens so that it can be used to refer to adult males as well, including the type of men who feature in Hollywood western films. This sense is still central in Present-day English, but by the twentieth century meanings including ‘reckless (motor vehicle) driver’ and ‘dishonest workman’ had also emerged, with clearly negative connotations. Words for certain categories of people are historically prone to pejoration, thus revealing the negative attitudes of a culture, for example words meaning ‘woman’ (Kleparski Reference Kleparski1997) or ‘foreigner’ (Alexander and Struan Reference Alexander, Struan, Farr and Guégan2013).

The final two tendencies that are often recognized in semantic change are of a somewhat different nature, since they describe conceptual relations that trigger changes rather than actual processes of change. Firstly, changes can be triggered by metaphor, the relationship of perceived similarity between two entities or concepts, which involves the mapping of one concept onto the other. A straightforward example is the polysemous noun mouse (OE mus). Its earliest meaning, which still survives as perhaps its most basic, is ‘rodent’, but, since the advent of personal computers, it has also been used with the meaning ‘computer device’. This sense seems to have emerged because of the similarity in shape and size between the animal and the device, and can be explained by the need to name a new object. Equally important as a trigger for semantic change is metonymy, the conceptual relationship between two entities which are associated in experience. This too involves mapping one concept onto another, and can result in an extension of meaning of a word. For example, the noun crown is found in its Latin form corona in Old English with the sense ‘head ornament worn by a monarch’. Since the sixteenth century, it has also been used to refer to the monarch him- or herself, and although this might be regarded as a figurative use, it is highly conventional, and recorded as a sense of crown in modern dictionaries.

It is important to remember that all of these tendencies may affect a single sense of a word rather than its ‘complete’ meaning: most words are polysemous, and have some senses that are stable over long periods of time and others that have changed or developed alongside existing meanings. The word friend, for example, retains its core meaning of ‘a person with whom one has a close relationship’, but has largely lost its meaning of ‘a relative or kinsman’, and, as noted in section 13.4.3, has developed new uses referring to contacts on the social media site Facebook. Particular meanings are not used equally by all speakers. Friend ‘kinsman’ may still be found in the vocabulary of a few speakers of Scots. Friend ‘Facebook contact’ will be familiar to social media users, be in the passive vocabulary of some other speakers, and probably unknown to many more. Semantic change thus tends not to involve a straightforward change from meaning A to meaning B for all speakers at the same time, but is generally gradual, often partial, and typically happens at different rates and to different extents for different speakers of a language.

13.6 Concluding and looking ahead

In this chapter, we have endeavoured to demonstrate, and if possible explain, some of the processes which have affected developments in meaning and form throughout the history of the English lexicon. Interest in this topic, and significant contributions to it, have increased considerably in recent years. One reason for this is the growing availability of online resources, such as the OED, now linked electronically to HTOED and to period dictionaries such as DOE and the MED as well as to modern dictionaries. Allied to this is the development of text corpora of periods or authors which enable standard corpus analysis techniques to be applied to historical data (once issues of variable spelling and polysemy have been sorted out). Hanks (Reference Hanks2013), for example, shows what can be done with such resources in his analysis of the 2,879 occurrences of enthusiasm in the British National Corpus (2013: 148–60), and more specifically its use by Jane Austen at a time when a change from negative to positive connotations seems to have been in progress. For recent English, the web itself has become a corpus. (On corpus analysis, see also Chapter 23 by Durkin in this volume.)

We have also been at pains to stress the interaction of language and culture as an essential component of lexical analysis. An excellent extended example of this approach is Biggam (Reference Biggam2012), which offers a wide-ranging interdisciplinary examination of the semantic field of Colour, including a good deal of historical information about English. One of the key contributions of cognitive linguistics to semantics has been the restoration to respectability of such encyclopaedic knowledge. As Grondelaers et al. (Reference Grondelaers, Speelman, Geeraerts, Geeraerts and Cuyckens2007: 991–2) say, when discussing the contribution of prototype theory:

In the light of all the above, we can confidently expect to learn a good deal more in future about the English lexicon and the uses to which it has been put.

14.1 Introduction

This chapter is concerned with the process and analysis of morphosyntactic change, and more specifically the description, analysis, and explanation of these changes in the field of English historical linguistics. Naturally, in order to understand what has happened in cases of change, we will need a thorough philological description of the facts, based on corpus data. However, no proper description is possible without a theoretical model (i.e. such frameworks as described in the individual chapters in Part I of this volume) by means of which we can systematize what has happened, thus enabling us to move beyond mere facts. In this chapter we will concentrate on some changes that have occurred and on the frameworks that have been applied to them to find out how useful they have been in furthering our understanding of morphosyntactic change in English.

Before we start, some preliminaries. In contrast to the study of phonetic/phonological change, where it is possible to compare concrete forms (forms representing ‘sounds’) or ‘cognates’, and deduce clear tendencies or even law-like developments from this comparison, the use of a theoretical model or models is an absolute necessity in the analysis of morphosyntactic change, where we compare underlying patterns rather than concrete forms; in other words, in syntax there are no cognates that can be recognized purely from their surface form (see McMahon Reference McMahon1994: 107–8). Thus, two superficially similar concrete forms may have to be given a different analysis over time, because the analysis of their structure depends on the synchronic system in which they function, which itself may have changed over time.Footnote ¹ There exists, however, an almost natural tendency to interpret or analyse an older construction very much from the point of view of the modern system with respect to both its form and its function (see Lightfoot Reference Lightfoot1979: 34ff.). An example of this is provided by the frequent analysis of Middle English infinitival constructions of the type God bad the rede See divide (Gower, CA 5, 1661),Footnote ² read with the rede See as subject and the infinitive as intransitive (i.e. ‘God commanded the Red Sea to divide’) rather than transitive (‘God commanded to divide the Red Sea’), not taking notice of the fact that in Middle English infinitival objects were still regularly placed before the verb, as they were in Old English (see note 1 and Fischer Reference Fischer2007: 19–22, which offers more instances of such misanalyses). In section 14.5 we will discuss a somewhat different case in more detail, where the analysis has been led astray in another way, namely by the assumptions of the theoretical framework that was used to analyse it.

Even though a theoretical framework is essential when analysing morphosyntactic change, we have to be aware at the same time that there is a tendency within linguistic frameworks to concentrate on just one factor, principle, universal, or constraint involved in a change, or a tendency to consider the development of one single construction without taking note of the larger domain of similar constructions in which it functions.Footnote ³ Therefore, in order to achieve more insight in change, the use of one specific theoretical model usually does not suffice. In any case of morphosyntactic change studied in detail, we usually find a multiplicity of factors that has caused the change in question – see the articles in De Smet et al. (Reference De Smet2013). The studies collected in this volume show that there is good reason for historical linguists to be eclectic in their use of theoretical models. Theoretical models used in historical linguistic research (such as e.g. generative grammar or grammaticalization) tend to give priority to some factors and not to others in their efforts to clean up the facts so as to make them more manageable, to provide, as it were, laboratory conditions (see Fischer Reference Fischer2007: chs. 2 and 3). Quite often internal, typological, or universal linguistic factors, being of more interest from the point of view of the model and its theoretical development, are given more weight than external ones. This does not mean, however, that they are therefore more important or more crucial.

Finally, in order to discover historical changes, we have to study the language used in any particular period, that is, first and foremost, the written evidence. It is crucial, however, to keep in mind that it is not the language itself that causes the change but the language users, who, after all, interpret and produce the utterances and are thus in some way responsible for the changes. Quite clearly, since we are dealing with history, these speakers are no longer there. We may assume, however, that people's brains and the way they process language will not have altered much in the course of time (following the ‘uniformitarian principle’, cf., e.g., Labov Reference Labov1972: 161, 274–5). In our search for explanation, it would help, therefore, to take into account how people learn language (since the principles or mechanisms used in language acquisition very likely also play a role in change), and we may also consider ongoing cases of change, where introspection is to some extent possible.

In the following sections, I will first discuss the nature of internal and external factors, how they are usually interlinked in actual cases of change in English, and the difficulty of keeping them clearly separate theoretically (section 14.2). In section 14.3, we will look at the various processes/mechanisms (and the models in which these are distinguished) that may be said to play a role in English morphosyntactic change. Section 14.4 will give a brief overview of the main areas of morphosyntactic change in the history of English, and pay special attention to the relation between the loss of inflections and changes in word order. And finally, in section 14.5, a case of change involving the emergence of the semi-modal HAVE-to will be considered in some detail. It concerns a development that has been studied quite extensively and from different viewpoints in the (recent) past.

14.2 Internal and external factors in morphosyntactic change

A first point that needs to be raised is: to what extent can we make a distinction between internal (i.e. language-systemic) and external (sociohistorical) factors, (i) in theory and (ii) in any individual case of change? It is clear that from a theoretical viewpoint such a dichotomy may be useful because the linguist is first and foremost interested in the linguistic factors that cause a change. In order to understand the internal factors involved, we have to link the changes that we find in language use to the system of grammar that helps generate our language utterances. The problem is, however, that this system of grammar constitutes itself a point of debate. Are we born with a universal (innate) grammar, which together with the PLD (primary linguistic data) shapes our grammar system, or do we deduce our grammar system from current language use with the help of general cognitive abilities that are used for learning of all kinds (see, e.g., Lightfoot Reference Lightfoot1979, Reference Lightfoot1991, Reference Lightfoot1999, Reference Lightfoot and Hickey2003 vs. Pullum and Scholz Reference Pullum and Scholz2002, Gentner Reference Gentner2010)? Whatever the case may be, it is clear that synchronic and diachronic linguists need to work together: an understanding of change and what causes it will help to find the contours of the system of grammar and the way it is acquired in childhood, while theories developed to describe the latter should contribute towards our understanding of change. This was already an accepted idea among some of the Neogrammarians, notably Hermann Paul (Reference Paul1909), and it has been widely followed by linguists in more recent times after Kiparsky's (Reference Kiparsky, Bach and Harms1968: 174) famous statement that change ‘provides a window on the form of linguistic competence’. The danger, however, is that this ‘window’ may be blurred by the ideas (principles, constraints) already present in the particular grammatical model a linguist works in, for which evidence is then sought in change.

Because of the above differences in viewpoint, diachronic linguists differ in what they consider important factors in change. A concentration on the system of grammar, as we see in the formal generative approach, puts the emphasis on internal factors, on grammar change, which mostly involves parameter shifts and rule change. It considers the ways in which the language learner systematizes the variants found in an ongoing change, which may lead to a shift in his grammar compared to that of the previous generation. In contrast, historical linguists working within a usage-based approach, such as grammaticalization theory or construction grammar, which concentrate on language in use, look at the gradual way in which an innovation may spread through the language and its speakers. Here, pragmatic and social (i.e. non-systemic) factors, especially those having to do with the communicational context in the widest possible sense, are considered as important as internal systemic ones, or sometimes even more important. The principle of analogy, which through cognitive science, has regained a position in functional approaches, should also be considered an internal factor. It was ignored in generative studies mainly because it could not be captured in general rules or constraints (see Lightfoot Reference Lightfoot1979: 361–73).

A clear example of a system-based approach is, for instance, the explanation for the development of the category of modal auxiliaries in sixteenth-century English provided in Lightfoot's (e.g. Reference Lightfoot and Hickey2003) and Roberts's (e.g. Reference Roberts2007) work, which is seen as a ‘catastrophic’ change caused by a parameter shift. This involves the loss of the so-called V to I rule, which took away the possibility for finite verbs to move to a position before the subject, as was the rule in Old and Middle English (as in What saw thow there?, [Le Morte Arthur, 3462], cf. the impossibility in Modern English: ^*What saw you there?). Since modal verbs, such as can, may, must, etc. continued to occur in this position (cf. ‘What could you see there?’, they came to be interpreted as instances of a new category, namely Aux.Footnote ⁴

An example of a usage-based approach, where a construction enters the grammar gradually, via ‘intermediate’ steps, is De Smet's (Reference De Smet2012) study on the development of English all but, from a quantifier loosely followed by a prepositional phrase (in sentences such as Go, my boy, and immitate [sic] him in all but his misfortunes, 1766), to an adverbial element. According to De Smet, this happened via its position before a predicative adjective (as in amidst perils from which escape was all but miraculous, 1838); next, affecting positions before attributive adjectives > passive verbs > active perfective verbs > active past tense verbs, and finally reaching a position before a verb in the present tense: We all but apprehend, we dimly forebode the truth (1841). He shows how this takes place by means of small, almost imperceptible analogical steps, with each one neatly and non-saliently paving the way for the next one: all but occurring each time in a position that shares a syntactic feature with the earlier, original position. Thus the original position before NPs shares a feature with predicative adjectives because nouns as well as predicative adjectives may occur as complements after copula verbs (as in She is sick/She is a coward), and in a similar way it spreads from adjectives to adjectival verbs such as passive participles (as in She was sick/She was knocked over by…), perfect participles, etc.

It is evident that in an approach where attention is paid both to the innovation and the spread of a new construction, to its form as well as its communicational function, and where the theoretical framework consists of layers of constructional schemata rather than across-the-board rules, a multiplicity of factors is more likely to emerge in the explanation of a change. In contrast, in a framework that works with rules, principles, and constraints that apply throughout, it is much more likely that the explanation will concentrate on the way the speaker's system of grammar ultimately deals with the variant constructions found, in other words on how the language learner reacts in face of the increased weight and direction exercised by the variants. The two approaches in fact highlight different aspects and different stages of the change. Both are valuable, however, and we need both to acquire a full understanding of why the change happened.

When it comes to making a distinction between external and internal factors in any case of change it is worthwhile to quote the last of the seven general principles drawn up by Weinreich et al. (Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968: 188) in their ‘Empirical foundations for a theory of language change’:

It is therefore important for the historical linguist (representing one side of the linguistic partnership), who wants to describe as well as understand the beginning and the endpoint of the change, to analyse the data as much as possible with a mind open to all frameworks relevant for linguistic processing.

Before closing off this brief discussion about the importance of considering both internal and external factors in cases of change, I would like to draw attention to some further aspects that are relevant here. A first point concerns frequency. It is hard to say whether the factor of frequency should be considered internal or external in nature (it is probably both), but what is clear is that it has come to be seen as more and more crucial in our field. Whereas it is mentioned only in passing in Hopper and Traugott's first (Reference Hopper and Traugott1993) edition of Grammaticalization, it has acquired a separate section in the second (2003: 126ff., cf. Pustet Reference Pustet, Frajzyngier, Hodges and Rood2005: 146). Similarly, in generative work on diachronic syntax, frequency has become much more relevant since the 1980s (contrast e.g. Lightfoot Reference Lightfoot1979 with 1999, and the articles edited by Pintzuk et al. Reference Pintzuk, Tsoulas and Warner2000a and Lightfoot Reference Lightfoot2002). In recent years, we have also witnessed the appearance of many studies related to frequency and the effects it has on language processing (e.g. Bybee and Hopper Reference Bybee and Hopper2001, Pustet Reference Pustet, Frajzyngier, Hodges and Rood2005, Bybee Reference Bybee2007), or volumes devoted to (methods of) quantification (e.g. Köhler Reference Köhler2012) and probabilistic linguistics, in which categorial gradience, frequency, and distributional criteria are central (e.g. Bod et al. Reference Bod, Hay and Jannedy2003). These developments are clearly the result of the use of corpora in linguistic research (and of more and more powerful computers to deal with large amounts of data), which began in the 1960s and took off seriously in the 1980s with the availability of bigger and bigger corpora, synchronic as well as diachronic.

A second feature I would like to draw attention to is the fact that speaker-innovations (which may or may not lead to change) are more often linguistically (i.e. internally) rather than socially motivated in the area of morphosyntactic change compared to that of sound change or lexical change. Internal systemic factors are more crucial here because propagation depends more on an increase in frequency through multiple speaker-innovations and on analogies between the new variant and already existent constructions within the language(-system) than on social needs or attitudes. Thus, a new construction may spread paradigmatically from verb to verb-class (by formal and semantic analogy) or from verb-class to another semantically or formally similar verb-class. For instance, Fischer (Reference Fischer1989) shows how the similarity between physical perception verbs (see, hear) and mental perception verbs (see, hear, perceive, understand, etc.) helps to account for the spread of the so-called accusative and infinitive [a.c.i.] construction in English (in Old English occurring only after physical perception verbs: þurh wundor þe he seah Sebastianum [ACC] don ‘through miracles that he saw Sebastian perform’, Mart5 (Kotzor) 855), so that mental perception verbs could begin to occur with an infinitival complement in late Middle English (where in Old English only a þæt-clause was possible): but they wepte to see and vnderstande soo yonge a knyght to Ieoparde hym self for their righte ‘but they wept when they saw and perceived so young a knight to put himself in danger for their rights’ (Malory 281).Footnote ⁵ A structure may also ‘worm’ its way in by an extension of the type of arguments that it may take. This is how the BE + going to > gonna expression spread from animate subject agents to inanimate subjects, while the nature of the to-infinitive changed from concrete to more abstract. Thus the earliest instances are still ambiguous between ‘movement’ and ‘intention’: we be Frenchmen, pylgrymes, & are goying to offer at y^e holy sepulcre (1534, Danchev and Kytö Reference Danchev, Kytö and Kastovsky1994: 62), while constructions with inanimate subjects or abstract infinitives are quite late. The same new construction may also be ‘helped’ by the occurrence of formally and/or semantically similar constructions formed with other future auxiliaries, such as shall/will, thus enabling going to to develop into an auxiliary that looks like other modal auxiliaries (note that the later form gonna indeed takes a bare infinitive). In Bybee (Reference Bybee, Joseph and Janda2003), it is shown in detail how a similar extension caused the use of can to spread in the history of English, turning the Old English lexical verb cunnan/cann ‘to know’ into a modal auxiliary expressing ability. Again this happened on the basis of formal and semantic analogies with other, already existing, verbal patterns.

For similar reasons, morphosyntactic constructions are also far less likely, compared to for instance lexical items, to be borrowed wholesale from another language. Not all linguists agree on this; there are clearly different schools of thought about morphosyntactic change and the likelihood of ‘borrowing’. These are often closely linked to the theoretical framework used. Generative historical linguists on the whole tend to ignore borrowing even though they do not deny that it occurs (see Lightfoot Reference Lightfoot1979: 374, 381–4). Historical linguists with a more sociolinguistic background believe that, given enough intensity and/or length of contact, anything may be borrowed, including syntactic structures (see Thomason and Kaufman Reference Thomason and Kaufman1988: 14, Harris and Campbell Reference Harris and Campbell1995: 149). Most historical linguists, however, believe that the idea that virtually anything can ultimately be borrowed, puts far too much emphasis on the mere force of social pressures and neglects the interaction with internal developments (e.g. Sankoff Reference Sankoff, Chambers, Trudgill and Schilling-Estes2002). Evidence suggests that only those structures are easily borrowed which fit fairly comfortably into the grammatical system of the borrowing language. Thus Latin a.c.i. constructions such as in…yf it soo be the kynge wylle Iuge her to be brente ‘if it is the case that the king will judge her to be burned’ (Malory 807) easily entered English from the fourteenth century onwards because by then (due to the fact that the general clause pattern had become SVO overall) any NP before a verbal, (non-)finite form (i.e. her before to be brente) could be interpreted as subject of that verbal form rather than object (see Fischer Reference Fischer, Gerritsen and Stein1992a, and the brief discussion in sections 14.1 and 14.4), whereas another Latin-type construction, the so-called parasitic gap construction as illustrated in, A man whom if you know i you must love j (1788 Cowper Letters III 215), is found only in Latinate writers since the Renaissance (see van der Wurff Reference Wurff1989). The reason why this latter construction did not enter the grammar of English is that it is difficult to interpret; firstly because the object man has moved from its base position after love (leaving a gap j), via another gap i (which is thus said to be parasitic on the earlier gap j) to the front position making the relation between object and verb hard to spot, and secondly because patterns like this were not found anywhere else in English, which could otherwise have assisted interpretation (as was the case with the a.c.i. construction).

Finally, there is another, rather elusive ‘external’ factor relevant here, which concerns the effect that the development of a written standard language may have on the linguistic models used to analyse the language or indeed on the way we see the language developing as a whole. The written language is often seen as a yardstick for grammaticality judgements in linguistic theories. An example of too much reliance on the written standard within the functional grammaticalization model is provided by Tagliamonte (Reference Tagliamonte, Fischer, Rosenbach and Stein2000) and Miller (Reference Miller, Fischer, Norde and Perridon2004), who show that the HAVE-perfect in standard English, which is supposed to be the neat unidirectional end point of a grammaticalization chain (with a full lexical verb developing into a perfect marker), is only the rule in standard written English. In non-standard and spoken varieties of English, many different forms, often representing stages on the grammaticalization chain, are (still) very much in use, such as an older word-order type (I have a letter written), or the continuing use of the simple past, the perfect with BE (I'm forgot all these things), or one without any auxiliary (She been married), or what Miller calls the ‘reverse cleft’ construction (That's you finished ‘You have finished’). In other words, the process is made to seem more perfectly unidirectional by concentrating only on the written standard. Grammaticalization is often more messy than that, developing different pathways or not developing at all. These aspects are now paid more attention to. A similar example involving the generative framework is the way the so-called that-trace phenomenon (as in Who do you think that will be buying or selling a house in the coming year) is discussed for English. This phenomenon is linked to pro-drop languages, and hence is said to be absent in English, which is not pro-drop.Footnote ⁶ However, in present-day spoken English it is quite common and it is indeed also found in older written texts, which were often closer to the oral channel (see Fischer Reference Fischer2007: 41–2).

The presence of a written standard may also more directly influence the way a language develops. Feilke et al. (Reference Feilke, Kappest and Knobloch2001) and others (see Schaefer Reference Schaefer, Bergs and Brinton2012) have shown that languages which develop such a standard may undergo a ‘Verschriftlichung der Sprache’, i.e. the spoken language may get ‘moulded’ by the forms of the written standard (a development greatly assisted in the last century by general education). Changes that occur in a language due to this are less directly related to the historical chronological development of the spoken language; they are the result of a channel-shift, of the idea that the written language is primary rather than the spoken variety. The influence of Verschriftlichung is especially strong on syntax and the organization of text. It is notable in this respect that early English texts, written before the present standard developed in the Renaissance period, were often closer to the spoken language than written texts nowadays, so that constructions prevalent in these older texts, and missing in later ones, need not be evidence of any change having taken place in the language overall (see Schaefer Reference Schaefer, Bergs and Brinton2012: 1277–8).Footnote ⁷

14.3 Processes at work in morphosyntactic change

In this section we will briefly review the different mechanisms, principles, parameters, constraints, and, ultimately, explanations that have been put forward for English morphosyntactic change within both the functional (usage-based grammar, grammaticalization, construction grammar) and formal (generative) models. It is clear, as already hinted at in section 14.2, that there is a wide divide between the two approaches in terms of the weight they give to the various factors at work in morphosyntactic change.

In functional models the locus of change is found in language use; there is no innate grammar, no fixed grammatical principles or constraints in the grammar system steering the change, and categories are fuzzy rather than fixed. Instead, the emphasis is on semantic-pragmatic forces at work in communication, on the role played by economy and expressivity in both speakers and hearers, on problem-solving as a way to understand why variation occurs. The grammar consists of an inventory of ‘constructions’ acquired in the course of language acquisition, which range from concrete collocations to more and more abstract schemata deduced from these. Changes are seen as typically gradual, spreading slowly through the language community and in the individual speaker's grammar, often riding ‘piggyback’ on lexical items before they take on a firmer hold. Two brief examples of this were given in section 14.2, referring to perception verbs and all but.

In grammaticalization theory, which deals only with a subset of changes taking place in language, the main factors causing change are functional in nature, related to metaphorical and metonymic thinking. These involve language users’ associations, linked to similarity, on the paradigmatic plane and linked to pragmatic inferencing on the syntagmatic plane, respectively. Analogy and reanalysis are seen as formal mechanisms accompanying metaphorical and metonymic thinking, but typically changes in form are said to follow changes in meaning. Reanalysis is considered primary, and analogy secondary, the idea being that a new interpretation of a construction occurs first, after which it may expand by means of analogy (Hopper and Traugott Reference Hopper and Traugott2003, Traugott Reference Traugott, Narrog and Heine2011). An example of this is the treatment of the BE + going to construction mentioned in section 14.2 and below, and the grammaticalization of HAVE-to discussed in section 14.5.

Construction grammar, which has recently developed a diachronic branch (Noël Reference Noël2007, Gisborne and Patten Reference Gisborne, Patten, Narrog and Heine2011, Trousdale Reference Trousdale, Gisborne and Hollmann2012, Hilpert Reference Hilpert2013; see also Chapter 4 by Trousdale in this volume), provides a more general theory for change, also encompassing changes in constructions that do not involve a grammaticalizing lexical element, such as for instance word-order changes (Noël Reference Noël2007: 182). There are many similarities in the tenets accepted in both construction grammar and grammaticalization theory, but where they differ is in the attention paid to formal matters:Footnote ⁸ constructions in construction grammar are a pairing of form and content, i.e. form and content may and often do change together because they influence one another. In this respect it follows ideas put forward in studies on analogy, such as Itkonen (Reference Itkonen2005) and Wanner (Reference Wanner2006). Not surprisingly, therefore, analogy is here considered to be a much more important factor in change than reanalysis: analogy is not just responsible for what is called ‘host class expansion’ (Himmelmann Reference Himmelmann, Bisang, Himmelmann and Wiemer2004) taking place after reanalysis, but also ‘for the creation of new substantive constructions’ (Noël Reference Noël2007: 183, emphasis added; see also Fischer Reference Fischer2007, De Smet Reference De Smet2012, Fried Reference Fried, Hoffmann and Trousdale2013). For instance, new complex prepositions such as in view of, in terms of need not have arisen by grammaticalization of the core lexical item in the normal way, but may have been formed on analogy of a constructional schema (a Prep–NP–Prep pattern) that itself arose as the result of a grammaticalization process (discussed in Hoffmann Reference Hoffmann, Lindquist and Mair2004a). In addition, the ‘success’ rate of the grammaticalization of a lexical construction may wholly or to a large extent depend on the availability of an already existent, similar grammatical construction making it fit more easily into the language-particular system of grammar as a whole. Thus, I argued in Fischer (Reference Fischer, Narrog and Heine2011) that the success of the BE + going to construction also depended on the frequency of AUX-V constructions already present. It is telling in this respect that to became part of the AUX (going to > gonna) so that the structure resembled other AUX-V constructions – which mostly had a bare infinitive – even more closely.

In the formal, generative model, possible semantic-pragmatic motivations, or phonetic-phonological ones involving the syntactic structures under investigation, are neglected because the syntactic system is deemed to be central and autonomous, and context-free (Schlüter Reference Schlüter2005); the way the forms function in the situational context is thus downgraded. The model constitutes a modular and computational system of grammar, geared towards optimal design, which is closely linked to the LAD (Language Acquisition Device). The principles hypothesized within the model always obtain (unless of course falsified), but there is some room for parametric variation. (This is different from functional models working with heuristic tendencies rather than fixed principles, making falsification difficult.) Parameters will be set on the basis of UG (Universal Grammar) and the PLD that the child is faced with during the period of language acquisition. The resetting of parameters (i.e. reanalysis) forms the basic, major mechanism for change (see the loss of the V to I rule referred to in section 14.2). Next to parameter shifts, there are also minor changes involving recategorization. Lightfoot (Reference Lightfoot1999: 185–97), for instance (quoting Warner Reference Warner1995), discusses the changes the verb BE is undergoing in the late eighteenth century in this way, involving restrictions in the type of ellipsis structures in which it could occur.Footnote ⁹ Variant forms, which provide the staple diet in grammaticalization and construction grammar, are considered possible ‘triggers’, which may or may not lead to a ‘change’ in the grammar (Lightfoot Reference Lightfoot1999: 66ff.), or, if robust, are accounted for by means of a ‘double grammar’. This latter way of dealing with variant forms was first suggested by Pintzuk (Reference Pintzuk1991) in order to explain robust variation in word order in Old English, thus ‘explaining’ the fact that the finite verb in subordinate clauses could appear both in final position (the normal position in SOV languages with a V2 rule) as well as in second position (which indeed became the stable position in late Middle English). In this model, in other words, attention is paid primarily to the endpoint of a change, when it is seen to affect the rules of the grammar system itself. It goes without saying that the explanation for a change is therefore radically different compared to that used in functional models, where ‘triggers’ and ‘changes’ form a continuum, with both in equal measure contributing towards an explanation. Changes within the formal model are typically radical, they involve reanalysis taking place during the acquisition period. Their explanation follows from the system of grammar itself and therefore heavily depends on what are considered parameters or principles/constraints within the system.

A theoretical model in which form and meaning play equal roles, in which context is not ignored, and in which the development of new constructions is not looked at in isolation but as part of the overall system in which the construction functions, is advantageous for the field of historical linguistics because it provides more in the way of explanation. McMahon (Reference McMahon2000a: 115) indeed raises the question ‘whether linguistic change can be modelled, and explained, in formal theories at all’. She continues (p. 125): ‘Progress in terms of external explanation can only be achieved if functional considerations are incorporated into the [formal] model, or if it is accepted that such considerations are relevant.’ It seems quite clear that in order to explain change we must look at what speakers do, we must look at language as it is processed by both speakers (production) and hearers (comprehension). This means that we must not only take the situational context into account but also the apparatus with which speakers and hearers process language (Berg Reference Berg1998); in other words, both the way in which speakers acquire the conventional system and the way in which it continues to shape their speech, play a role.

I will end this section with a brief illustration of two changes in Middle English to show how the different formal and functional ways of dealing with syntactic change affect the outcome and our understanding of what has happened (and through this, how it also affects the way we envisage the contours of the grammar system). Through this comparison, I hope to show not only that it is more likely that changes spread gradually and by analogy, but also that it pays to base an explanation (in as far as any explanation is possible) on careful philological scrutiny of sufficient corpus data. An explanation within a parameter-shift framework does not achieve a similar, satisfactory result; it concentrates too much on the situation when the change is fully completed, ignoring why and how it started, and it is often not really explanatory outside its own model.

The first example concerns the parameter shift invoked by generative linguists to explain the basic word order change from OV to VO, which took place in Middle English. According to Lightfoot (Reference Lightfoot1991: 42–77) such a shift may start off with gradual haphazard changes leading to a threshold. In such a case one would not expect to find patterned variation at this earlier stage. After the parameter has finally shifted, one would expect VO to be the rule for all speakers with a shifted parameter. It is interesting to observe, therefore, that the ‘shifted’ VO speakers still use OV structures at quite a late stage. And more crucially, that these OV structures were found to be patterned (because certain types of direct objects – containing negative and quantifying phrases – retained the earlier word order (Moerenhout and van der Wurff Reference Moerenhout and Wurff2000, Reference Moerenhout and Wurff2005)). Such patterning is difficult to account for within the ‘parameter shift’ model.

The second case concerns changes in constructions with an indirect object. Lightfoot (Reference Lightfoot1991: 124) links the development of a new indirect passive in Middle English (the type He shal be holpen ‘he must be helped’, Chaucer LGW 1984), which replaced the earlier dative passive (Ac ðæm [DAT] mæg bion swiðe hræde geholpen ‘but to-that-one may be very quickly helped’, CP(Cotton)33.224.22), to the simultaneous development of a number of other new constructions that are all said to be caused (according to a rather ‘broadbrush scenario’, Vincent Reference Vincent, Butt and King2001: 29), by the loss of dative case in Middle English. Allen (Reference Allen, Butt and King2001) shows on the basis of detailed corpus work that the change involving the development of the indirect passive took place in separate steps, and that there is no direct relation between the loss of dative case and the emergence of the new indirect passive, because the data show that dative-fronted passives had died out before the new indirect passives arrived (Allen Reference Allen, Butt and King2001: 54). Allen argues, taking into account other, similar (analogical) structures, that there is a direct relation between the loss of the dative-fronted passive (ðæm mæg bion…geholpen) and the fronted dative in active sentences (as in Him geaf ða se cyngc twa hund gildenra pænega ‘Him gave the king two-hundred gold pennies’, ApT 51, 20). Due to this positional loss (i.e. of the fronted dative), the original dative and accusative NPs came to be used side by side immediately after the finite verb (the king gave him 200 pennies/ gave 200 pennies him). It was when the recipient NP came to be fixed in the first position after the verb, that it became reanalysed as an object. This fixed position is what enabled the indirect passive to occur: because of the reanalysis of NPs like him to object, a subject position in passives became possible. The explanation is thus linked to the increasingly fixed word-order patterns (SVO) that became the rule in Middle English. But this is not the whole story. Especially noteworthy is Allen's use of similar-looking active constructions and her suggestion of a ‘fuzzy’ object: an original dative object looking like a direct object because of its position. Such a scenario would not be possible within a parameters framework since it would not allow two types of object. However, this is typical in a ‘fluid’ analogical way of thinking (Hofstadter Reference Hofstadter1995), as we find for instance in construction grammar.

14.4 Brief overview of the main areas of morphosyntactic change in English

The areas most affected in the morphosyntactic development of English involve the nominal, the verbal, and the adjectival inflectional systems – where inflections were lost causing both the rise of new periphrastic constructions (e.g. perfect, progressive, modal auxiliaries, dummy DO, prop word one) and changes in word order, which became increasingly fixed on both the clausal and the phrasal level in the course of time. In the philological tradition, the inflectional attrition taking place during the Old and Middle English periods was seen as the result of a number of internal and external factors such as (i) the fact that in Germanic the word accent came to fall on the first syllable, leaving the final syllables (the inflectional suffixes) unstressed and subject to steady erosion; (ii) the Viking occupation of north-eastern England in the Danelaw and the later French invasion of the British Isles, causing a stronger and swifter attrition in English as compared to its Germanic sisters; (iii) increasingly fixed word order, which can be seen as a result of both (i) and (ii) together (see Weerman Reference Weerman1993, McWhorter Reference McWhorter2002).

We have already seen in section 14.3 that the change in basic word order has been treated differently in the literature, depending on the depth of the philological investigation, and the theoretical model used. Another suggestion put forward is that the loss of variable word order may be due to a certain amount of simplification. In cases of language mix or heavy language contact, it often happens that the more irregular and infrequent forms get levelled out leaving the koiné arising out of the mix with fewer exceptional features. This is the suggestion made in Kerswill (Reference Kerswill, Chambers, Trudgill and Schilling-Estes2002) and Fischer (Reference Fischer, Schreier and Hundt2013). But here again, different models offer different explanations. Kroch and Taylor (Reference Kroch, Taylor, van Kemenade and Vincent1997), who performed a quantitative study of syntactic patterns found in Early Middle English Northern- compared to Southern-Midland texts, note an important syntactic difference between these texts, which they relate to a difference between two types of V2 movement involving two different positions to which the verb may move: either to CP or to a lower IP position. Southern-Midland texts show the same constraints on V2 as Old English texts (involving the inversion of full NP subjects and finite verbs – pronominal subjects, being clitics, do not invert – with exceptions after light adverbs like þa, þonne ‘then’). In contrast, in the earliest Northern prose text, the Rule of St Benet, the complex conditioning of Verb–Subject inversion found in Old English and Southern-Midland texts is virtually absent. Full NP inversion is found in all cases, even after þa/þonne, with pronoun inversion following closely, occurring between 91 per cent and 100 per cent (somewhat lower after adjectival complements).

What could be concluded from this is that a considerable simplification has taken place in the North in that, in the application of the V2 rule, subject pronouns are now treated in exactly the same way as full subject NPs, while the variation after þa/þonne has also disappeared. However, Kroch and Taylor do not accept this ‘simple’ solution; they see this as representing a ‘typological distinction’ (p. 297ff.) involving a parameter shift because they have other evidence showing that the Old English clitic status of pronouns has been preserved even in the Northern text. There is no space to go into details here (for discussion, see Fischer Reference Fischer, Schreier and Hundt2013), but it is clear that the clitic evidence heavily depends on the theoretical model in which the change has been described, and on a very small number of rather exceptional sentences, involving complex and infrequent syntactic constructions. The koinéization scenario offers a more likely explanation: the simplification has been caused by imperfect learning due to language contact with Scandinavian, whereby the most frequent pattern (full NP inversion), by analogy, replaces the less frequent one (i.e. absence of inversion with pronoun clitics). Interestingly, this case shows remarkable resemblance to one of simplification in Modern Dutch, also involving inversion. Imperfect learners, who do not understand the subtlety of Dutch inversion after adjuncts, tend to treat them all as the same pattern (see Fischer Reference Fischer2007: 290ff.).Footnote ¹⁰

The overall loss of verbal, adjectival, and nominal inflections not only led to stricter word order (and stricter word order in turn enabled further losses), but it also set the scene for the development of new periphrastic expressions, new adverbial forms in -ly, and the introduction of the prop word one. In the description of all these cases, we see explanations often firmly tied to theoretical models. We have already seen this in connection with the rise of the modals, and we see it again in the description/explanation of changes in adjectival usage. In Fischer (Reference Fischer, Bermúdez-Otero, Denison, Hogg and McCully2000, Reference Fischer2006, Reference Fischer, Denison, Bermúdez-Otero, McCully and Moore2012), I argued on the basis of extensive corpus evidence, that the loss of the distinction between strong and weak adjectives not only caused the loss of substantive adjectives like se wælhreowa ‘the cruel [man]’ (ÆCHomII, 20 177.89) resulting in the use of the prop word one, but it also had an effect on the development of a new determiner system, it led to a fixed Adjective–Noun order in the NP (postposed adjectives as in OE þin þeowa clæna ‘your servant pure’ (LS (Margaret) 4 23) were gradually lost), and to the consolidation of a new adverbial ending in -ly. A complex of factors having to do with changes elsewhere, and especially with the simplification mentioned above are said to account for these new developments. In contrast, in the discussion of some of these constructions in Haumann (Reference Haumann2003, Reference Haumann2010), who works in a generative framework, developments are seen as the result of more abstract theory-internal elements, such as the loss of pro and the loss of agreement features on the adjective. There is no space to go into the different approaches in any detail, but what is clear is that Haumann's account, even though it throws up interesting questions for other researchers to dig their teeth into, is too schematic and the shift too abrupt, to account for all the data found (see Fischer Reference Fischer, Denison, Bermúdez-Otero, McCully and Moore2012).

Recent research, with investigations going beyond syntax and beginning to include discourse and information structure, has shown that the word order changes in turn affected discourse patterns in English (and vice versa), leading to the increasing frequency of passives overall, to it-clefts and other constructions to express discourse topic and comment (broadly defined as ‘given’ and ‘new’ information), and cohesion (see van Kemenade and Los Reference Kemenade, Los, van Kemenade and Los2006a, Los Reference Los2009, Patten Reference Patten2012, Meurman-Solin et al. Reference Meurman-Solin, López-Couso and Los2012). Briefly speaking, what we see here is that the subject and object positions became increasingly fixed in the clause, losing the ability to express information structure by their variation in position, with the result that the preverbal subject position became restricted to expressing only given information and the post-verbal object only new information (hence the rise of passives and it-clefts to indicate given objects and new subjects respectively). An increase in passive infinitives was also caused by the loss of OV order because constructions in which an NP before the infinitive functioned as its object, as in OE hie gesawan þa deadan men swa þiclice to eorþan beran (‘They saw the dead men so thickly to earth carry’, Orosius 3 10.75.20) clashed with the increasingly common interpretation of any NP before a verb as subject. One way to solve this problem was to make the infinitive passive so that the object NP functioned as subject – compare the above example to its modern version: ‘They saw the dead men be[ing] carried to earth so thickly’ (Fischer Reference Fischer and Kastovsky1991). This change at the same time afforded an opportunity for new, Latin-type a.c.i. constructions to emerge, where the pre-infinitival NP functioned as subject of the infinitive, as in thenne she commaunded the gates to be opened (Malory 642) (Fischer Reference Fischer, Gerritsen and Stein1992a, Los Reference Los2005, De Smet Reference De Smet2013).

What is interesting in all the investigations mentioned here, is that they are the result of a combination of approaches: they rely on studies related to the formal elements of word order, which we owe to generative accounts, on the new interest in discourse patterns arising from functional and grammaticalization models, and on the rise of corpus studies, which was accompanied by the development of increasingly larger corpora and the tools to handle these. The detailed search possible in corpora has also provided us with many more examples, thus making it possible to discover quite subtle differences between constructions (showing the role played by analogy), and it has turned our attention to the important role played by frequency.

14.5 The case of HAVE-to

To show some of the processes discussed above at work, and also the effect that a particular theoretical model may have on the way we view a change, I will briefly discuss one case, involving the grammaticalization/auxiliarization of HAVE-to, where word order, analogical forces, changes in the core-modal system, as well as semantic-pragmatic factors important in grammaticalization can all be seen to play different roles.

In line with similar developments involving a possessive verb like HAVE, where HAVE in combination with an infinitive grammaticalized from a full lexical verb into an auxiliary, it has usually been taken for granted that English HAVE-to represents a regular case of grammaticalization. Thus, van der Gaaf (Reference Gaaf1931), Visser (Reference Visser1969: §1396ff.) – who do not yet use the term – Brinton (Reference Brinton1991) and Łęcki (Reference Brinton, Jucker and Taavitsainen2010) all more or less accept the following three developmental stages for the change from OE ic hæbbe þone mete to etene ‘I have that food to eat’ (Jn (WSCp) 4.32) to modern obligative ‘I have to eat that food’. At the earliest (Old English) stage the construction has the following features: HAVE is used as a full verb, meaning ‘to possess’, the NP (mete) functions as the direct object of HAVE, the to-infinitive is not obligatory serving as an adjunct dependent on the NP, and word order is not relevant – it does not influence meaning. In a subsequent stage of the development, the meaning of HAVE slowly generalizes and acquires obligative colouring in combination with the to-infinitive, which itself becomes obligatory, the infinitive now no longer functioning as an adjunct to the NP but as an object complement of the matrix verb HAVE, and the original object of HAVE (mete) becoming an argument of the infinitive. In the final stage, in which we see the appearance of inanimate subjects (possessive HAVE + to-infinitive had animate subjects), and of intransitive infinitives (i.e. the original ‘possessed’ object can now be dropped altogether), HAVE-to becomes truly an auxiliary. Reanalysis or rebracketing from [I [have [that food [to eat]]]] to [I [have to eat [that food]]] takes place at this stage.

It is made clear in this sketch of the putative development of HAVE-to that the grammaticalization proceeds along a path of pragmatic-semantic change – bleaching of possession first, the development of obligative colouring later – and that the syntactic changes – the word-order change and the rebracketing – are subordinate to it, following hard on the heels of the semantic change. The development is seen as gradual and the various stages are extremely difficult to disentangle. The late and sudden shift in word order is difficult to explain satisfactorily within this account.

For this reason I paid more attention to formal aspects in my discussion of the change (Fischer Reference Fischer1994a), combining a generative sense for the importance of word order with a (then recent) interest in corpora. I considered all the instances in which HAVE is combined with a to-infinitive in the Helsinki Corpus (covering the Old, Middle, and Early Modern periods) distinguishing a number of different syntactic types depending on the presence and the position of the object(s) governed by HAVE and/or the to-infinitive. I came to the conclusion that there is no evidence for a gradual semantic change in HAVE from ‘possess’ via a more general meaning to an obligative sense as envisaged in the grammaticalization account, noting that a bleached HAVE already existed in the earliest recorded period, and that the infinitive could have modal colouring from the beginning (but not necessarily one of obligation; it could also express possibility, likelihood, etc. depending on context). Moreover, this modal sense was only possible in constructions where HAVE and the infinitive syntactically and thematically ‘shared’ an object, as in Ic hæbbe mete to etene, which therefore turned out to be a crucial construction.Footnote ¹¹

I also found that the more firm syntactic evidence for the change (the appearance of inanimate subjects, of intransitive infinitives, clear absence of an object of have, double use of have, etc.) is quite late, occurring only from the Modern period onwards (see also Krug Reference Krug2000: 59, 80). Earlier instances of HAVE-to which Brinton (Reference Brinton1991) considered to be periphrastic – led as she was by the tenets of the grammaticalization scenario and, through it, by a too superficial interpretation of the morphosyntactic forms – were found to be interpreted too much with a ‘modern’ eye. It was not noted that the old, non-obligative interpretation was still fully available for these examples and sometimes the only possible one. New examples put forward by Łęcki (Reference Łęcki2010) as evidence that intransitive infinitives and inanimate objects do occur already in Old English, can also be shown to be highly problematic (see Fischer 2015).

In a nutshell, my argument was that it was due to the SOV > SVO word-order change that the order in the ‘shared object’ construction (where the object occurred between HAVE and the to-infinitive) was fixed to HAVE + to-infinitive + NP because other language patterns no longer tolerated an object before the verb with which it had a thematic/syntactic relation. In other words, the SOV > SVO change caused a regular adjacency of HAVE and the to-infinitive, which ultimately led to the auxiliarization of HAVE-to.

With the advent of larger corpora, and the new interest in analogical processes (not only in historical linguistics but also in language acquisition studies) in both construction and usage-based grammar, I decided to look at this case again, since I was no longer such a firm believer in this simple mono-causal explanation. The question that was not really answered in my 1994a article was: why did it take so long for HAVE-to to develop into an auxiliary (according to Krug Reference Krug2000, it became only truly frequent in the nineteenth century), and why did it develop an obligatory and not some other modal sense. In Fischer (Reference Fischer, Fabienne and Brian2015) I used the much larger online Corpus of Middle English Prose and Verse in the hope to find out more by looking at data from the crucial period.

The evidence suggests a very strong role played by analogy, of both an (i) abstract and a (ii) concrete kind. Concerning (i), it is quite clear that the ‘shared object’ syntactic pattern (the only pattern where obligative meaning was possible) was in strong competition with the pattern with two objects and the pattern with an object thematically tied only to HAVE, where this new obligative sense did not occur. These opposing patterns, as it were, slowed down the development to an auxiliary. Only cases where the shared object was preposed – due to wh-movement or topicalization (as in (a) a prevy case that I have to seye to the ‘a private matter that I have to say to thee’, Gesta Romanorum) and (b) Ye shal my ioly body haue to wedde ‘you shall have my beautiful body to marry’, Chaucer, ShT 1613) – were able to provide an analogical pattern for auxiliarization because here HAVE and the to-infinitive were already consecutive and in some of these cases a modal sense could be pragmatically inferred (e.g. in (a) above but not in (b)). This preposed construction became indeed frequent in Middle English. Concerning (ii), a concrete construction where HAVE combined with the infinitive to do without any objects being involved may have created a formal analogical pattern because this HAVE and infinitive were almost always consecutive, thus enabling them to be seen as bonded. But here only the form could have influenced the development, since the meaning of this construction (e.g. I. wil nowht haue to done of suche Matier ‘I will have no dealings with this matter’, Chaucer ShT 4441) had nothing to do with obligation. What also came out of the data, however, was the very frequent occurrence of HAVE followed by the noun nede and a to-infinitive dependent on nede, as well as the construction mo[s]t[e] nede + infinitive, which both expressed obligation. These two patterns, together showing some formal similarity with HAVE-to, are likely to have aided the semantic development towards obligation.

No doubt further factors were involved, such as developments taking place in the core modal system allowing HAVE-to to begin to function as a gap filler for must, and the fact that to-infinitival complements were becoming more frequent overall, thus enabling the former NP adjunct to be interpreted as a complement of the matrix verb HAVE instead.

What this case teaches us is that new ways of looking at the same morphosyntactic change afforded by new theoretical models may help us gradually gain more insight into the how and why of a particular morphosyntactic change, and through it of change in general.

14.6 Concluding remarks

I hope to have shown that we have come a long way in the explanation of morphosyntactic change. In the early days of English historical linguistics, the emphasis was on a proper description of the morphosyntactic facts with the help of traditional grammar and traditional philological methods. It was only in the early 1970s of the last century that the methodology of synchronic linguistics and the then current theoretical frameworks (especially the ones connected to structural and generative models) came to be applied to English historical linguistics, with the aim of finding more general explanations for change that could be tied to models of language learning. More or less simultaneously, a new and again more philological tradition – because it kept an eye on the data and on the functional aspects of language (unlike the generative model) – began to gain ground, helped also by the development of large diachronic language corpora. The latter, accompanied by the emergence of usage-based models both in language learning and language change and the rise of cognitive science, subsequently led to a new interest in the power of analogy. It is this combination of approaches, where each has furthered and developed our knowledge of morphosyntactic developments in English, that has proved to be quite successful in providing deeper insight into the causes and mechanisms of morphosyntactic change.

15.1 Introduction

The study of meaning in the history of the English language requires the work of semantic reconstruction, which, ‘at its most delicate…ultimately means reconstructing past societies with all their cultural and social ramifications’ (Nevalainen Reference Nevalainen and Lass1999a: 435). As Keller (Reference Keller and Ballmer1985: 234) argues, language is an ‘object of sociocultural evolution’. Accordingly, the English language has changed its semantic range and its communicative functions over time, and it has done so as people adapt to changing external conditions and to understanding and communicating within a changing world. The perception that semantic change is sporadic and unpredictable in its effects, that the sensitive semantic changes that affect individual lexical items are idiosyncratic and fundamentally dependent on historical context, has meant that the systematic study of semantic change has been a major challenge in linguistics. Indeed, until the middle of the twentieth century, the treatment of semantic change tended to be reduced to a taxonomy of mechanisms, including pairs of opposites: specialization – generalization; metaphor – metonymy; amelioration – pejoration. In view of the fact that it was difficult to examine change in terms of notions such as regularity and frequency in the absence of techniques to collect and interrogate large quantities of data before the advent of the electronic corpus, scholars had to rely upon dictionaries for lexical data and on their reading in order to secure corroboration for their findings. The availability of large diachronic corpora and the attendant development of sophisticated search software and quantitative techniques for analysing data for semantic study have changed the field, and so both the objects and the methods of study have changed.

Although the principal object of the study of meaning and meaning change has traditionally been vocabulary (the lexical items of a language), scholars have also examined the role of meaning change in the history of structural items, and more recently, grammatical constructions. In the case of the English language, structural items include modal auxiliary verbs (e.g. must and will), which develop out of Old English preterite present verbs in the course of the Middle English period, and morphosyntactic constructions such as the future construction be going to, which emerges with future reference towards the end of the fifteenth century in English. Section 15.2 reviews some critical assumptions underlying current studies of semantic and pragmatic change in English. Section 15.3 concerns the role of meaning change in grammaticalization – the history of constructional and grammatical meaning – and situates important recent research in terms of the processes and mechanisms that account for those changes. Section 15.4 examines the fields of diachronic cognitive semantics, diachronic corpus pragmatics, and the approach to semantic change from a sociolinguistic perspective. Throughout this treatment, we note the impact of corpus linguistics on advancing the field of historical semantics. Section 15.5 concludes with the discussion of future research areas in the light of major new research trends in this area.

15.2 Assumptions

The first assumption underpinning the study of meaning change in the history of English is the distinction between two kinds of meaning.Footnote ¹ For our purposes, ‘speaker meaning’ is the full meaning that speakers negotiate as they use words or expressions in a particular context (Grice Reference Grice1989). Meaning created through the use of words and sentences in a specific context is variable and flexible; it is negotiated in the course of interaction among language users. In contrast, word or sentence meaning is the meaning a word or sentence evokes in recurrent situations or indeed regardless of context. This kind of meaning is conceptual, denotative, or referential; the relationship of the linguistic expression to its meaning is conventional and arbitrary, and does not require a context of use for interpretation. Meaning is, therefore, both communicative and cognitive in nature.Footnote ² Clark (Reference Clark1996: 126) argues that words and sentences are ‘types of signals, linguistic units abstracted away from any occasion on which they might be used, stripped of all relation to particular speakers, listeners, times, and places’.Footnote ³ The distinction between the conventional meaning of an expression and the meaning an expression has by virtue of its use in the context of utterance reflects the distinction between two approaches to meaning that have been represented in the fields of semantics and pragmatics respectively.Footnote ⁴ We understand semantics as ‘meaning’ and pragmatics as ‘use’.

The second key assumption is that polysemy, which occurs when an expression has more than one distinct but related meaning that is conventionally associated with it, is a basic condition of language. For example, the intransitive verb fly in each of the following quotations drawn from a particular period in the history of English has a distinct yet related meaning:

1. (1a) Birds fly above the trees (OED s.v. fly I.1.a) [self-propelled aerodynamic activity]

2. (1b) The wild birds fly (as the bird-catchers term it)… during the month of October. (1768 T. Pennant Brit. Zool. (new ed.) II. 330. OED s.v. fly I.1.e) [migrate –travel on a journey by means of self-propelled aerodynamic activity]

3. (1c) The faster a given aeroplane flies, the less is the power required to produce its lift. (1959 Chambers's Encycl. I. 112/1 OED s.v. fly I.1.h) [capacity for aerodynamic motion]

4. (1d) The disorder follows loyalist protests at a decision to reduce the number of days the Union Flag flies above Belfast City Hall. 2013 Evening Standard, 8 January/6. (OED s.v. fly 6.a) [suspension of lightweight object]

5. (1e) The week flies by, but by Friday I have an inkling of how gruelling it would be to raise children on my own. (2013 Sunday Times, 31 March/11) (OED s.v. fly 7.b)

6. (1f) Adaptive Mobile, a security firm backed by Enterprise Ireland, Doughty Hanson and Intel Capital, prefers to fly below the radar, fighting international gangs who target mobile phone users with spam and fraud attempts. (2013 Sunday Times, 31 March/6)

Speakers have access to polysemies composed of some or all of these meanings. In examples 1a through f above, fly is used as an intransitive verb with a range of subjects. In 1a and b, the subject birds means that fly may be construed as denoting self-propelled aerodynamic activity, though the linguistic context in 1b requires the reader to adapt the reading so that the verb fly is a kind of shorthand for migration, a periodic journey by flight. In contrast, in 1c, the subject aeroplane renders the verb fly as referring to its subject's capacity for aerodynamic motion. In 1d, the verb refers to the property of suspending a lightweight object, namely a flag. The final two examples, 1e and f, are metaphorical uses of the verb fly; in 1e, rapid movement is attributed to a week, a period of time. This is now a highly conventional figurative use of the verb (cf. time flies). In 1f, the expression fly below the radar, which in this context describes the avoidance of public scrutiny on the part of a mobile phone security company, is adapted from a phrase used to describe the action taken by the pilots of military aircraft to avoid enemy detection by radio waves in the course of World War II. In summary, the fact that the subject of each instantiation of fly has distinct characteristics ensures that the reader has little difficulty in identifying which is the most relevant meaning of the verb required to make sense of the sentence. Polysemy is also a condition of semantic change. The sporadic and non-linear nature of change entails that an expression consists of a family of related meanings whose relationship may alter over time. Over time, one or other of these meanings may become more regularly connected with the expression while others might disappear. Semantic change thus results from the conventionalization of a specific meaning of an expression in consequence of speakers’ persistent use of the expression with that meaning in a range of contexts. However, that change does not prevent polysemy – the continued co-existence of meanings that arise in the context of use. As Ullmann (Reference Ullmann1962: 195) comments: ‘there is nothing final about a semantic change: a word may acquire a new sense or scores of new senses’, giving rise to new polysemies.

The study of semantic change can focus on changes in the function of an expression or on the changing realizations of a particular function. In the case of the first kind of focus, the study of the changes of meaning that a particular lexical expression or construction undergoes over time, the form of the expression remains constant as polysemies develop. The focus on the mapping of form to function over time is labelled semasiological change. The majority of the historical semantics research conducted over the past several decades is semasiological, as exemplified by landmark studies of English cultural keywords such as wit (Barber Reference Barber1976, Nevalainen Reference Nevalainen and Lass1999a). In the second kind of focus, on the ways in which a set of meanings is represented over time by a particular expression or class of expressions, the forms vary while the particular domain of meaning appears constant. This study of the mapping of function to form, for example the development of the coded expression of a particular domain such as true, intellect, or conditional, is labelled onomasiological change (Traugott and Dasher Reference Traugott and Dasher2002: 25).

15.3 The regularity of semantic change

Historical semantics has tended to focus on the type of meaning changes or mechanisms that affect particular lexical expressions. The major mechanisms that generate changes in the meaning of a word, namely, specialization, generalization, metonymy, and metaphor, form what Geeraerts (Reference Geeraerts2010) calls the ‘classic quartet’ of meaning changes. These mechanisms have traditionally been treated as opposites: for example, specialization – generalization, amelioration – pejoration, and metaphor – metonymy. For instance, in traditional accounts of semantic change, specialization and generalization are treated as ‘two socially conditioned tendencies working in opposite directions’ (Ullmann Reference Ullmann1962: 200). They have also tended to be treated in isolation for their roles in individual word histories; accordingly, the changes in the meaning of the lexical noun pilot in English since the fifteenth century can be accounted for by mechanisms of generalization (a pilot is now the navigator, not only of a ship but of any kind of travelling craft) and metaphor (from the sixteenth century a moral or spiritual guide could be called a pilot). Although these mechanisms have traditionally been considered to be central in accounting for the idiosyncratic, historically context-sensitive semantic changes that affect individual lexical items, they were, until quite recently, not understood to function as pairs within an overarching process of semantic change. Traugott and Dasher (Reference Traugott and Dasher2002: 60) make this clear, noting that ‘[f]aced with pairs of changes that seemed to cancel each other out, even if one member of each pair is predominant, and with examples taken out of context, many linguists have tended to see semantic change as essentially unstructured’.

However, work within historical linguistics on the grammaticalization of lexical expressions (following Meillet 1958 [1905–6], Benveniste Reference Benveniste and Meek1966) generated the investigation in English historical linguistics of the emergence of modal auxiliary verbs, intensifiers, discourse markers, and subordinators (e.g. Traugott 1989, Reference Traugott, Adamson, Law, Vincent and Wright1990, Hopper and Traugott Reference Hopper and Traugott1993). This research yielded the important finding that the semantic changes accompanying the syntactic changes involved were regular and frequently unidirectional in nature. For example, the relationship of DEONTIC > EPISTEMIC meaning, as evidenced by the history of the English modal auxiliary verbs, was argued to be a unidirectional derivational one. For example, must ‘I conclude that’ derives from must ‘be obligated to’, not vice versa (Traugott Reference Traugott1989, Traugott and Dasher Reference Traugott and Dasher2002: 120–37). The impact of Traugott's work on meaning change in the history of grammatical categories has provoked considerable debate among scholars working on grammaticalization, particularly around the status of what has been called the ‘unidirectional hypothesis’ as an inherent feature of grammaticalization or as an empirical generalization (Lass Reference Lass, Fischer, Rosenbach and Stein2000b, Campbell Reference Campbell2001: 127, Janda Reference Janda2001: 289ff., Eckardt Reference Eckardt2006: 26, Norde Reference Norde2009: 49–52, 91–2).

15.4 Corpus linguistics and research in historical semantics and pragmatics

The availability of corpora for investigating words and their semantic fields has also meant that it is possible for scholars to conduct diachronic onomasiological studies (for electronic resources, see Chapter 8 by López-Couso in this volume). A recent corpus-based onomasiological study of the domain of ‘anger’ takes as its starting point an earlier claim by Hans-Jürgen Diller (Reference Diller, Fernández, Fuster Márquez and Calvo1994) regarding the history of English anger. Geeraerts et al. (Reference Geeraerts, Gevaert, Speelman, Allan and Robinson2012) analyse the competition in the Middle English period among the lexemes anger, ire, and wrath, applying modern statistical techniques to assess Diller's claim. Thus they demonstrate how contemporary corpus techniques can be used to trace and account for semantic change.

The increasing sophistication of corpus linguistic techniques on the one hand and the development of very large corpora on the other have also prompted the revision of theoretical assumptions regarding the linearity of semantic-pragmatic change. For example, the semantic development of discourse markers such as I mean or well has recently been shown not to be linear. Brinton (Reference Brinton, Fitzmaurice and Taavitsainen2007: 58) illustrates the extent to which apparently distinct categories of meaning or functions can overlap or interact in speakers’ use at the same time. Brinton argues that the semantic development of I mean shows that whereas the expression has a metalinguistic function in all periods, its precise nature differs subtly from period to period. For instance, she argues that the metalinguistic meaning (‘I'm making the previous discourse more precise’) identified in the following Middle English example also has an intersubjective quality to the extent that the speaker attends to the hearer's need for greater explicitness or for illustration (Brinton Reference Brinton, Fitzmaurice and Taavitsainen2007: 47, example 9c, presented here as 3):

1. (3) The claper of his distouned bell…I mene his fals tunge)1450 [?1422] Lydgate, Life of our Lady [Dur-U Cosin V.2.16] 2.922; MED).

   ‘the clapper of his distoned bell…I mean his false tongue’.

In this instance, the writer replaces a metaphorical expression (his distouned bell) with a literal one (his fals tunge) to clarify his meaning. Brinton argues that the pragmatic marker I mean is largely a product of the process of idiomaticization, in which semantic compositionality is lost as the expression gains in opacity as it ‘ceases to express a cognitive state of the speaker’ (2007: 67). Brinton concludes that this process is part of the larger processes of grammaticalization and lexicalization, complementary unidirectional processes that involve freezing, the loss of structural compositionality and what she calls the ‘holistic’ accessing of complex forms (2007: 69). Defour and Simon Vandenbergen (Reference Defour and Simon Vandenbergen2010) also challenge earlier assumptions of the linear semantic–pragmatic development of discourse markers in their corpus-based investigation of the history of well in Middle and Early Modern English. They argue that well, which they term a pragmatic marker, undergoes a more complex development than Jucker (Reference Jucker1997) proposes, exhibiting early interpersonal functions.

15.5 Conclusions

Nevalainen and Traugott (Reference Nevalainen and Traugott2012) insist that we consider familiar concepts and processes of language change afresh, in the light of new evidence, theories, and methods of analysis. There are two recent developments in English historical linguistics in particular that promise to advance the study of semantic and pragmatic change. The first is the development of a range of statistical and visualization techniques for allowing the researcher to apply a ‘bottom-up’ approach to the analysis of linguistic data on the basis of relative frequencies of patterns on the one hand and on the nature of the contexts of the target expressions on the other.Footnote ⁸ For instance, Hilpert (Reference Hilpert, Allan and Robinson2012a) demonstrates the utility of collostructional analysis in identifying collocations that occur in statistically significant frequencies in different periods and how variation in the relative frequency of particular collocates can suggest semantic change. Hilpert's ‘bottom-up’ data-driven approach allows the data to reveal points at which there are important changes in collocational patterns, so determining how the data determines its historical organization into periods. This work indicates the potential for understanding better the points at which coded meanings can be distinguished from pragmatic functions in the change of meaning.

The second development is a shift in the interpretation of the nature of the boundedness of grammatical categories (particularly in constructional grammar approaches to meaning). For instance, Colleman and De Clerck (Reference Colleman and Clerck2011) explore the demise of double object constructions with verbs of banishment and verbs of manner of speaking, as in their examples 5 and 14a, repeated here as 4a and 4b:

1. (4a) And a man that could in so little a space, first love me, then hate, then banish me his house [i.e. banish me from his house]. (Richardson 1740) (Colleman and De Clerck Reference Colleman and Clerck2011: 192)

2. (4b) At her departure she took occasion to whisper me her opinion of the widow, whom she called a pretty idiot. (Fielding 1751) (Colleman and De Clerck Reference Colleman and Clerck2011: 198)

They note that understanding the nature of this grammatical change depends on interrogating the interaction of the semantics of the verbs and the grammatical relations of their arguments. Their study anticipates Broccias's (Reference Broccias, Nevalainen and Traugott2012) argument that the treatment of the relationship of syntax to the lexicon as continuous rather than categorical enables us to examine the possible scope of processes of semantic change (such as specialization) as extending beyond the word to whole predicates. Also, studies such as Hilpert (Reference Hilpert2013) enact the extent to which constructional change incorporates the study of semantic change.

The routine utilization of big corpora, increasingly sophisticated techniques for mining them and capturing changes through frequency in a bottom-up approach, together with theoretical advances in semantics and pragmatics augur well for the development of the field.

16.1 Introduction

The history of English is composed of the histories of its registers and genres (Diller Reference Diller, Diller and Görlach2001: 3), each with its own dynamics and diachronic developments. From this angle, language history is still a somewhat neglected and understudied area of English linguistics, although the importance of genres is well acknowledged and the view that all language use and every text is framed within genres has gained ground. Texts vary a great deal in their linguistic realizations and change in time when sociocultural factors change. There is a great deal of scholarly activity in the field, and the future looks bright. Outlines are already beginning to emerge in various domains of writing, but a great deal remains to be done in the future. The most important result so far is perhaps the insight that instead of one major line of development, there are several, and the picture is more diversified than depicted in language histories and early articles on the lines of development. Diversification begins much earlier than generally assumed, as late medieval texts already show a great deal of variation.

The chapter opens with a short discussion about the revival of genre studies, and an account of various approaches to genres (sections 16.2 and 16.3). Central notions in the field are defined in various ways (section 16.4) and, in addition to literature and philology, historical genre research is informed by modern linguistics (section 16.5). Genre dynamics and genre change are treated from the methodological point of view (section 16.6). The advent of corpus linguistics with large-scale multidimensional studies on historical corpora has opened up a new research paradigm based on solid empirical evidence (section 16.7). Short accounts of various registers of writing follow with focus on what studies on genre dynamics have revealed so far (section 16.8). In the conclusions (section 16.9), I shall discuss the potentials of contextualizing variation and genres as loci of change in the history of English.

16.2 The revival of genre studies and the interface between language and literature

Genres have proved helpful operational tools for classifying language use, but their revival in literary and linguistic research is fairly recent. Genre as a theoretical notion goes back to Aristotelian Poetics and was first recovered in literary theory by scholars paying attention to genre dynamics (e.g. Fowler Reference Fowler1982, Bakhtin Reference Bakhtin1986 [1953], Todorov Reference Todorov1990). Genre studies has been a vibrant field in literary scholarship for some decades now, and it has become one of the most important areas where literature and linguistics interact. An interest in genres in linguistics was prompted by the above-mentioned literary genre revival in the last decades of the twentieth century, and genre studies soon developed to accommodate non-literary texts, especially as ‘each type of discourse usually labeled literary has nonliterary “relatives”’ (Todorov Reference Todorov1990: 11).

But how do genres work in communication and why do genre conventions change? By analysing the sociohistorical and cultural factors, we can relate and perhaps explain the reasons for the linguistic realizations of genres and their changes in a larger frame, taking both parties of communication into account. The relationship between genre and language is one of the fundamental issues in genre studies. In literary studies both text-internal and text-external features are considered, and the same views are shared by some linguistic schools and scholars. Recent large-scale empirical studies on genres without preconceived notions of their linguistic patterns suggest, however, that the issue is even more complex than previously thought. We have come to a new phase from abstract theoretical considerations and qualitative studies of fewer texts to empirically based evidence of genres and their developments based on electronic corpora, as will be demonstrated below (see also Kytö and Smitterberg Reference Kytö, Smitterberg, Biber and Reppen2015).

16.3 Approaches to genre classification

In addition to different definitions, researchers have developed various approaches to how genres are related to each other and how the whole genre map should be viewed. Genres are seen as abstractions in which texts are grouped to form ‘classes’, ‘kinds’, or ‘families’. These metaphorical notions reveal the underlying approaches, whether strict groupings with clear boundaries or loose entities based on resemblances. The basic question is whether genres are rigid and clear-cut categories with obligatory components, or whether they constitute fuzzy sets with overlaps and flexible memberships. Aristotelian classificatory principles distinguished three main ‘kinds’ of literature, expressed with the epithets poetic, dramatic, and fictional; in addition, satire was recognized, and so were essay, biography, dialogue, and history (Fowler Reference Fowler1982: 5). Genres were distinguished on the basis of the object of study, the manner of presentation, the structure of the text, and the effect on the audience (Paltridge Reference Paltridge1997: 18–19). An important aspect of classical genre theory was the tenet that the kinds require certain forms and styles, and thus genres and the language patterns manifested in them were not separated. This guideline for text production served for a long time and can be seen e.g. in medieval literature where descriptions of noble people in aristocratic settings required an elevated style, very different from the more robust language use of rural settings and lower social classes (see 16.4.2). The stylistic requirements held true until modern times e.g. in rules of letter writing (see Palander-Collin Reference Palander-Collin, Jucker and Taavitsainen2010).

The above-mentioned metaphorical terminology of classes, kinds, and families, used for the description of genres reveals how the boundaries between categories are perceived. The view that Fowler (Reference Fowler1982) introduced in Kinds of Literature was quite radical, as he was the first to point out the advantages of the prototype approach to genre studies. According to Fowler (Reference Fowler1982: 3) ‘[l]iterature should not be regarded as a class at all, but as an aggregate’, and the connections between texts should be perceived as ‘family resemblances’ to use Wittgenstein's terminology (see Rosch and Mervis Reference Rosch and Mervis1975). This approach emphasizes blurred edges and fuzzy boundaries. The texts in a genre may exhibit genre features to different extents, so that all features need not be shared, for it is the family resemblance that counts, and multiple membership in more than one family is also possible. In non-literary genres, an early adaptation of the family resemblance approach was carried out in a study on medieval prognostications in the vernacular, where the purpose, e.g. whether defining appropriate times for actions or making forecasts, and the point of departure, whether the phases of the moon or a thunderstorm, served to justify the groupings of texts into subgenres within the larger frame of prognostications that could be found in various forms, either in verse or prose (see Taavitsainen Reference Taavitsainen1988). The same approach was used in a study on herbals (Mäkinen Reference Mäkinen2006), where some of the texts overlapped with recipes, some with health guides as well as exhibiting features in common to various extents.

Another approach to genres takes labels like ‘recipe’, ‘letter’, and ‘will’ as guidelines for genre classification. Görlach (Reference Görlach2004: 24–8) collected some 2,000 such labels that record various text functions, but noted that all are not equally salient.Footnote ¹ These labels serve to indicate the dynamics of writing inherent in the diachrony as some labels are entirely new, some are new applications, and some are old. He points out various processes for English: earlier genres could continue to be written but the components may change over time (this is the case in encyclopaedias), features can be transferred from an existing type to a new type (as in the subgenres of legal writing), and borrowings of genre features from other languages and cultures are common in literature (as seen in Chaucer's work) and religious writing (see section 16.8).

A more recent view considers genre labels from a metacommunicative point of view and states that these lexical determinants reveal aspects of the communicative modes that they encapsulate (Hübler and Busse Reference Hübler, Busse, Busse and Hübler2012: 6).Footnote ² The names of genres employed by contemporary speakers in past periods give us an ethnographic view and can be used as a basis of genre inventories of historical periods. This classification gives an interesting overview of social practices as it emphasizes that genres are cultural products and reflects how people understood them. A diachronic outline of the developments emerges when different synchronic surveys of genre labels are compared.

16.4 Definitions of genre, text type, and register, and the frame model

There is a great deal of terminological variation in the research on genres between individual researchers, scholars, and schools in present-day linguistics. Functional criteria for defining genres are predominant in most recent scholarship, whereas conventional genre definitions emphasize formal criteria, structure, and style. The functional approach to genres emphasizes the link between forms of communication, their users, and the sociocultural contexts that gave rise to them. Users are highlighted in the most recent pragmatic genre considerations (Hübler and Busse Reference Harris2012: 6), and the discursive turn of linguistics can be seen in genre studies as well, as genres, like for instance politeness, are not linked to specific forms but discursively constructed (see Jucker and Taavitsainen 2013: 9). Both genres and text types are abstractions made on the basis of individual texts, and it has proved useful to keep genres and text types apart in historical studies so that genre refers to text-external classification and text type to text-internal linguistic features (Biber 1988: 170; Taavitsainen Reference Taavitsainen, Rissanen, Kytö and Heikkonen1997: 187–90, Reference Taavitsainen2001a). In other words, genres are related to the speaker's or writer's purpose of communication rather than the form. Biber and Conrad (Reference Biber and Conrad2009: 16) emphasize that whole texts are needed for studying genres that can be defined as ‘culturally expected ways of constructing texts belonging to the variety’. In contrast, text types can be defined as groupings according to the linguistic form, irrespective of genre. Language users have an idea of the most salient genres, and their conventionalized features including e.g. the opening lines (see below and 4.2). This calls for another definition of genres as mental frames in people's minds that can be triggered by various conventionalized prompts; these frames become realized in texts for a certain purpose in certain cultural contexts (Taavitsainen Reference Taavitsainen2001a: 140). This is very much related to the schema theory which claims that much of our experience is stored as schemata or scripts that trigger expectations of how various situations develop and help in the meaning-making process (McIntyre and Busse Reference Busse, Busse, Jucker and Taavitsainen2010: 113).

Private letters, for instance, conform to the culturally expected ways of interpersonal communication by appropriate formulaic greetings, taking the recipient into account with polite enquiries about health, and well-wishing at the end. These conventions change in the course of time, and there is a considerable difference between the salutations of early private letters and their equivalents in modern e-mails. Recipes provide another example. Their prototypical opening ‘Take’, or R [for recipe] in Middle English, is enough to trigger expectations of a text belonging to a genre whose function is to instruct in preparing something, be it a culinary dish, medicine, or some other utility like ink. Text type features include imperative forms of verbs, measurements, and an optional efficacy part that may be realized in various ways, with the set phrase probatum est, a vernacular evaluative note like ‘good for headache’, or even a case report to convince the reader (see Taavitsainen Reference Taavitsainen2001b). The advantage of this two-tier model of genres and text types is that it brings out genre dynamics and genre changes.

Register is yet another important term defined as situational language use: language is used in different ways in different fields of writing, and the register perspective characterizes the typical linguistic features of text variation and shows similarities on a more general level than text type; thus it is a superordinate term (Biber and Conrad Reference Biber and Conrad2009: 16, Claridge Reference Claridge, Bergs and Brinton2012: 239). Text type features are distributed throughout the text, and therefore text extracts can be used instead of whole texts, but an extract may not fully represent the linguistic make-up of the whole, as the example below (in section 16.4.1) shows.

16.5 Genre and useful insights in English linguistics, shared by historical linguists

Historical genre studies are also informed by modern linguistics as various ‘schools’ have developed their own ways of dealing with genres. Systemic-functional linguistics focuses on the regularities of unfolding discourse, and the linguistic realization patterns are considered part of the generic description. Genre is defined as a ‘staged, goal-oriented social process’ (Martin and Rose Reference Martin and Rose2003: 7), and the list of examples includes ‘greetings, service encounters, casual conversations, arguments, telephone enquiries, instructions, lectures, debates, plays, jokes, games, and so on’ (ibid.).Footnote ⁵ Attention has been directed to present-day genres and the ways in which they unfold (Eggins and Martin Reference Eggins, Martin and van Dijk1997), but genre change or genre dynamics are not really discussed, though this approach could shed additional light to historical genre development. A pioneer historical study was made by Halliday where he defines scientific English as a ‘semiotic space within which there is a great deal of variability at any one time, as well as continuing diachronic evolution’ (1988: 162), but he does not specify different genres within this space. He takes physical science to illustrate the evolution of scientific writing in a long diachronic perspective with very broad brush strokes. The data from the early periods is restricted to Chaucer's Astrolabe, a handbook for a 10-year-old ‘Little Lewis’, as the only representative of the late medieval period. Since then, early scientific writing has proved much more diversified and dynamic (see section 16.8).

Another modern approach that can enrich historical studies is Critical Discourse Analysis which regards genres as important social practices. Fairclough (Reference Fairclough1995: 13–14) makes a primary distinction between actual texts and the conventions that people draw upon in producing and interpreting them: he sees genre as an ideal type and text types as configurations that have developed and become conventionalized for particular types of social situations. Thus his applications of the terms are different from their present use, but likewise he emphasizes the abstract nature of genres and the fact that some individual texts can vary a great deal within loose genre constraints. Genre dynamics is not central in this line of research either, but applications to earlier periods reveal the potentials, e.g. for studying manipulative styles (McEnery Reference McEnery2006) or newspaper language representations of parties in political conflicts (Prentice and Hardie Reference Prentice and Hardie2009). The modern polarization of ‘us’ and ‘them’ is already present in seventeenth-century texts and the article shows how the approach can shed further light on implicit textual meanings in historical texts.

Applied linguistics has made a great impact on both modern and historical genre studies as many of the notions developed within this field have gained common acceptance. The term ‘discourse community’ (Swales Reference Swales1990) is a case in point. It is widely used in historical studies with some adaptations to the sociohistorical circumstances of the period under scrutiny. In historical pragmatics, for instance, discourse communities and their practices have become central with the insight that authors of past periods adapted their language according to their target audiences as they do now (see Chapter 3 by Mazzon in this volume). Another important contribution to the overall frame is the notion of ‘networks of genres’ in which hierarchies in terms of prestige, chains with chronological sequence, and sets with individual text positions are taken into account (Swales Reference Swales2004: 2, 23). This approach has proved useful for studying genre dynamics in the religious domain (see section 16.8), and there is potential for applying it even more widely.

Rhetorics of science has paid a great deal of attention to genres as communicative units that are important for meaning-making practices. Texts are viewed as dynamic causal entities as discourse communities actively structure and maintain their interests through texts that organize our perceptions of the world (Bazerman and Paradis Reference Bazerman and Paradis1991: 3; see also Gross et al. 2002). Historical studies have paid attention to the creation of new genre conventions and changes of genre features over time in response to changing discourse communities. Authors communicate their ideas to various audiences for various purposes in various ways, and this insight has come to play an important role in generic studies. This is what Bazerman and Paradis (Reference Bazerman and Paradis1991: 8) call the operational mode where the expertise of a discourse community is transformed into ‘socially operative texts that guide human actions’.

16.6 Applying theories of genre change to historical data

In diachronic genre analysis, a central position is occupied by the creation of genre conventions with their developments and mutations, and changes of genre features over time in response to the needs of their users. Inventories of genres at various points of time indicate synchronic stages of genre development, and when such descriptions are compared to one another along the time axis, an outline of developments may be detected (see section 16.4.1). But here as in any diachronic comparisons we have to be sure that the entities we compare are really comparable. Thus the requirement of tertium comparationis, something that remains constant, is methodologically important and applies here as well as in diachronic speech act studies (see Jucker and Taavitsainen Reference Jucker and Taavitsainen2000, Taavitsainen and Jucker Reference Taavitsainen, Jucker, Jucker and Taavitsainen2008: 7). The function of a genre lends itself easily as a reliable point of departure and guarantees that we are comparing like with like, and that our research questions are sensible. Genres show different realizations in different periods, but more prototypical features may remain constant in a long diachronic perspective, and genres differ in their rates of change. Examples of both stable and unstable, short-lived and long-life genres can be found in various domains of both literary and non-literary writing. The seventeenth-century genre of ‘Characters’ was popular for a short period and achieved a polished and conventionalized form, but soon went out of use (Taavitsainen Reference Taavitsainen, Sell and Verdonk1994). A contrary example is provided by autobiography/diary/blog writing as first-person narratives that express a person's inner feelings. Manifestations of the same function range from medieval mystical experiences in Margery Kempe's dictated texts to early modern precursors of the modern diary form, and even modern blogs that reveal a person's emotional states have several features in common with the above-mentioned predecessors. The function of these genres is much the same, to vent private feelings whether intended to be read by others or not. Such texts reflect the fuzzy border between literary and non-literary writing, and the dichotomy of private and public that makes the recent manifestations of self-expression so very different.Footnote ⁶

The prototype approach is particularly well suited to the study of the mechanisms of change in historical genres, as everything need not (and does not) change. Stable features can guarantee membership in a genre, but genre mutations are common, and we can give credit to literary studies for pioneering work in the theories of genre change, as Fowler (Reference Fowler1982) was also the first scholar to deal with the dynamics of genres. He discussed the mechanisms of change from a novel angle, and various models for diachronic studies on genre dynamics have been created since. Other literary scholars have broadened the underlying influences to cultural issues like fashions and prestige and noted that, when styles created for culturally learned audiences spread to less cultured layers of society, the reception of the features changed and what used to be refined conventions became an object of mockery (Margolies Reference Margolies1985: 14–16). In stylistics the mechanisms of change have received attention, and the advantage is that modern methods of corpus linguistics and extensive electronic databases have brought empirical studies on linguistic change in various genres to a completely new level as it is possible to quantify the changing proportions of linguistic features. The level of the text itself is a reliable point of departure, and researchers have learned to rely on qualitative examples to demonstrate what their quantitative findings indicate in concrete terms.

16.7 Large-scale multidimensional studies on genre change

According to the variationist view, language is a continuously developing and changing entity (see, e.g., Milroy Reference Milroy1992a). In genre studies we are dealing with macro structures of language, but we can assume that the same underlying theory applies at least to some extent to both the genre repertoire and the linguistic make-up of genres as to the more micro-level changes of phonological, morphological, and syntactic changes. With corpus linguistic methods we have achieved solid empirical evidence of the development of genres and text types and their dynamics in a long diachronic perspective.

Perhaps the most influential corpus linguistic study of the past decades has been Biber's (1988) Variation across Speech and Writing which pioneered in assessing the styles of contemporary English with an innovative, multidimensional methodology. This study was purely synchronic. The aim was to verify the co-occurrences of 67 linguistic features (whose functions had been identified in previous studies) in data that comprised over 481 text extracts in 23 different genres. The frequency counts of these features across the data were assessed with factor analysis that identified their co-occurrence patterns as underlying dimensions of variation. These patterns were then interpreted according to their communicative functions. The multidimensional method has also been applied to diachronic studies with different research questions.Footnote ⁷ Several continuous scales of variation have been identified, but no single dimension alone is sufficient to account for the whole range of linguistic variation. Fiction, essays, and letters are a fairly coherent group, and a ‘drift’ towards a more oral direction was verified in all, but the rates of change were different (Biber and Finegan Reference Biber and Finegan1989). Drama and medical research writing represent contrary extremes in their situational characteristics, and a long-term diachronic study showed how they developed to different directions, indicating that the patterns of change are not convergent in all domains of writing (Biber and Finegan Reference Biber, Finegan, Nevalainen and Kahlas-Tarkka1997). Further evidence for stylistic developments in opposite directions include newspaper language versus the language of science: the former is becoming increasingly colloquial while the latter has moved towards more specialized, more economical expressions with phrasal noun modifiers (see Biber and Gray Reference Biber, Gray, Nevalainen and Traugott2012). Another statistical computer method, cluster analysis, was used to define linguistically based text types according to linguistic features that cluster together, with core texts of a genre showing a high concentration of particular linguistic features, while more peripheral texts had lower densities of these features (Biber Reference Biber1992b).

16.8 Genre stability versus genre dynamics in different registers of writing

In addition to the genre drifts verified by multidimensional studies, genre changes in various domains of writing have been discussed in the literature. There are several projects that aim at revealing genre dynamics within different registers of writing in the history of English. Common to them all is the fact that they are based on empirical evidence and corpus compilation is part of the project work.Footnote ⁸ A great deal of research activity is going on in various fronts, as the following survey will demonstrate.

Religion is one of the most important domains of language use with a wide influence. Early genre classifications deal with Middle English religious prose in general (Blake Reference Blake and Rowland1974) and as presented in the Helsinki Corpus (Taavitsainen Reference Taavitsainen, Rissanen, Kytö and Palander-Collin1993). A much larger corpus of religious writing is under way at the University of Cologne, and its compilers have already used it for their studies. Kohnen (Reference Kohnen, Pahta, Taavitsainen, Nevalainen, Tyrkkö, Jucker and Taavitsainen2010: 527) provides a frame for further assessments based on the direction of communication, God addressing the Christian community (the Bible) or vice versa (prayer), or members of the community addressing each other (theological discussion and religious instruction). Major genres show fairly static features and do not change much in the history of English, although an assessment of sermons from Old English to the late twentieth century revealed a tendency towards more polite and less rigorous directives reflecting sociohistorical changes, with strengthening of the common ground, secularization, and freedom of thought (Kohnen Reference Kohnen, Pahta, Taavitsainen, Nevalainen, Tyrkkö, Fitzmaurice and Taavitsainen2007b). Rütten (Reference Rütten2011: 219–20) highlights genre dynamics in the religious domain paying attention to the functional profiles and coordination of genres that share an overall discourse function. The network perspective (see section 16.5) shows large-scale shifts in and across genres, and a dynamic picture arises (see below). Changes occurred in the early modern period as the Reformation altered the genre repertoire with vernacular catechisms. The introduction of English to where Latin had prevailed changed the functional profiles of sermons and treatises (Rütten Reference Rütten2011: 220). Further changes in the genre map can be verified e.g. as saints’ lives were replaced by another type of biography.

Legal writing provides another authoritative register where genre developments can be traced in a long diachronic perspective. Lehto (Reference Lehto2010 and Reference Lehto, Jucker, Landert, Seiler and Studer-Joho2013) discusses the formation of genre conventions in Early Modern English in regard to complexity, clausal coordination, textual structure, and formulaic uses of multinominals of three legislative genres (parliamentary acts, royal statutes, and orders by the Privy Council) and proclamations. Clausal coordination is most frequent before 1550, while phrasal coordination is more numerous in the seventeenth century. Lehto's study provides evidence that many of the conventions of modern writing became established in the early sixteenth century with printing and other sociohistorical developments accelerating the changes (Lehto Reference Lehto2015). Earlier studies on legal writing include Hiltunen's diachronic survey from Anglo-Saxon law to the present day (Reference Hiltunen1990), Bhatia's (Reference Bhatia1993) studies on genre and register features, and Biber's (Reference Biber1992a) complexity study. An area of more recent attention is courtroom language use and trial records, which has attracted a great deal of attention in historical pragmatics with new considerations on the authenticity of recorded language (see Doty Reference Doty, Jucker and Taavitsainen2010; Kytö et al. Reference Kytö, Grund and Walker2011b). Another line of research assesses genres as loci of linguistic change. Kilpiö's (Reference Kilpiö, Nevalainen and Kahlas-Tarkka1997) assessment of participial adjectives is a good example of such studies using the Helsinki Corpus as its material. After a meagre beginning in Old English, the frequencies of participial adjectives increased considerably in legal writing, documents, and official letters. A more specific study focuses on the role of the genre of statutory writing in the standardization process of written English in the late medieval and early modern periods (Rissanen 2000). This line of research has great potential for future studies.

Scientific and medical writing is a third prestige register and it has received more attention than any other in the history of English (see Pahta and Taavitsainen 2010). The late medieval period was important for establishing genre conventions, as the register was new and genres had to be created. Early manifestations show a great deal of fluctuation as translators had difficulties in expressing abstract notions, and it took time before learned genres became fully established in English (Voigts Reference Voigts and Edwards1984, Reference Voigts, Griffiths and Pearsall1989; Pahta Reference Pahta2001; Taavitsainen Reference Taavitsainen, Taavitsainen and Pahta2004, Reference Taavitsainen, Alberni, Badia, Cifuentes and Fidora2012a). These processes continued for centuries, as English was used alongside Latin and gained predominance only in c.1700. Vernacularization broadened the genre repertoire, and various multilingual strategies can be detected in late medieval and early modern texts (Voigts Reference Voigts1996; Pahta Reference Pahta, Taavitsainen and Pahta2004, Reference Pahta, Taavitsainen and Pahta2011). In addition to the diachronic changes, there is synchronic variation even in the earliest periods. Texts were adapted to the capacities of their target groups, authors had different educational backgrounds, and styles of writing vary accordingly. The availability of new electronic corpora has inspired a boom of linguistic studies on early modern medical writing focusing on e.g. formulaic language and the development of phrasal constructions (see e.g. Hiltunen Reference Hiltunen2012; Kopaczyk Reference Kopaczyk, Jucker, Landert, Seiler and Studer-Joho2013; Tyrkkö Reference Tyrkkö, Jucker, Landert, Seiler and Studer-Joho2013a, Reference Tyrkkö, Taavitsainen, Jucker and Tuominen2014). The new spearhead genres of the Royal Society (1662–) have been much discussed in the literature as experimental reports were a novel genre created for the new discourse community and published in the first scientific periodical the Philosophical Transactions (Bazerman Reference Bazerman1988; Gotti Reference Gotti2001, Reference Gotti and Brownlees2006, Reference Gotti2008; Moessner Reference Moessner, Dury, Gotti and Marina2008, Reference Moessner2009; Valle Reference Valle1999, Reference Valle and Brownlees2006). At the same time, the old scholastic style continued in the majority of writings, and there were texts for lay readers in the old remedy book tradition. A bird's-eye view reveals a dynamic movement in the styles of writing as features of earlier spearhead genres find their way to writings targeted at heterogeneous audiences and become downgraded; a vacuum at the top is filled in by a new genre and a new way of writing. Some genres persist for centuries with fairly stable and conventionalized genre features, whereas some others, mainly recording cutting-edge science, change more rapidly (Taavitsainen Reference Taavitsainen, Jucker, Schreier and Hundt2009, Reference Taavitsainen, Taavitsainen and Pahta2010). The late modern period is still fairly uncharted, but the situation has improved lately. Moskowich and Crespo (Reference Moskowich and Crespo2012) contains several articles on the scientific English of the eighteenth and nineteenth centuries dealing with both micro-level features and macro-level stylistic developments, inspired by the release of a new electronic corpus. In the specialized discourse perspective, the creation of Present-day English conventions is of special interest. The modern spearhead genre of research articles presenting new knowledge exhibits highly conventionalized patterns with large units of tightly packed noun phrases and the reduction of the role of verbs (see Halliday and Martin Reference Halliday and Martin1996; Gross et al. Reference Gross, Harmon and Reidy2002), but different fields of scientific writing have different traditions with their own developments (see Hiltunen Reference Hiltunen2010). The long diachronic line should also include English as the lingua franca of science, used and developed by non-native speakers for communicative purposes for global discourse communities. These most recent phases have received increasing attention e.g. by Jenkins (Reference Jenkins2007, Reference Jenkins2013), Seidlhofer (Reference Seidlhofer2011) and Mauranen (Reference Mauranen2012), and both written and oral practices have been researched.

News discourse can be traced to the early modern period. Before newspapers, pamphlets provided a medium for news distribution to wide and heterogeneous audiences of topical and polemical issues like witchcraft (Suhr Reference Suhr2011) and tobacco (Ratia Reference Ratia2011). The first real newspaper in England, the Oxford Gazette, was published in 1665 (Claridge Reference Claridge, Jucker and Taavitsainen2010: 600). Letters from foreign correspondents provided much of the material, but other genres such as crimes and accidents, ship news, and advertisements were soon added. Some of them have been studied in a diachronic perspective (see, e.g., Facchinetti et al. 2012). This register shows the influence of technology and sociohistorical developments like improvements in literacy more clearly than perhaps any other domain. The target audience of newspapers widened in the nineteenth century and we can talk about mass communication. At first communication was mostly one way with letters to the editor as an exception, but more interactive modes have been developed recently with the internet, and there is also a change in the editorial side from single to multiple authorship. The broad definition of mass media extends the scope beyond the traditional areas to the linguistic landscape of billboards and signs opening up new genres for study (Jucker Reference Jucker and Jucker2009: 2). How all these novel aspects and rapid developments influence genre dynamics within the register provides a challenge for future studies.

16.9 Conclusions

Genre studies fall into the interface between language and literature to a large extent. The emphasis on the functions of genres and how the co-occurrence patterns of linguistic features contribute to the overall aim of the genre can add to our understanding in many respects, e.g. we can analyse more precisely what triggers reactions in the audience and how genre expectations are created. The aim of several studies has been to identify repertoires of genres and their features in the history of English and, more ambitiously, to probe into the dynamics of genres both between adjacent genres and in the larger overall picture. Different models of analysis have been developed to account both for synchronic variation and diachronic change. Research on the linguistic features of texts in various genres and registers of writing has revealed interesting patterns, but work in the interface could continue with assessments of genres as loci of change (as in section 16.8), or as context for humorous effects and irony (as in section 16.4.2; see also note 4). Electronic corpora have gained an important position in Present-day English linguistics, and material on a wide variety of writings, both historical and present day, is readily available. The multidimensional method has been particularly fruitful in revealing genre variation and change, but qualitative methods and micro-level analyses of linguistic features are also needed to complement the macro-level assessments and anchor them in their wider sociohistorical and cultural context.

The advantage of the two-tier model of keeping genres and text types apart is obvious, as we can achieve a more analytical grid for tracing the developments of genres in a historical perspective. This model can reveal aspects of the composition of texts that belong to the same genre at different times. Texts may differ in their linguistic features and the mechanisms of change and the dynamics of genres can be revealed in a new way by tracing variation in their encoding. For example, assessing the changing proportions of narration, instruction, and argumentation can bring us closer to the variationist methods traditionally applied to micro units of language (see section 16.7), and provide a new way of assessing discourse and genre dynamics at the macro level. What we need is concrete evidence for theoretical claims about the mechanisms of change. This is not an easy task, but research in this field is actively striving for this goal.

17.1 Introduction

In their 1968 study of language in its social context, Weinreich et al. defined the aspects central to the study of the process of change. According to their classification, language change needs to be defined in relation to its timing, its transition from one form to another, its embedding in the prevalent linguistic and social structures, and to its social evaluation. In the last few decades it has been shown that these aspects are not only of importance for studies in present-day sociolinguistics and -pragmatics, but also in historical language studies. Theory is of central concern in the reconstruction of language change. According to Labov (Reference Labov, Lehmann and Malkiel1982: 20), the ways in which we use the past to explain the present and the present to explain the past are based on particular underlying notions. The uniformitarian principle argues that language worked on the same principles in the past as it does now (see, e.g., Romaine Reference Romaine1982: 122). Or, as Lass (1997: 26) puts it in his General Uniformity Principle, ‘nothing that is now impossible in principle was ever the case in the past’ (emphasis original).

Historical sociolinguistics, or sociohistorical linguistics as Romaine's pioneering book from 1982 calls it, is the study of the reproduction and transformation of the relationship between the linguistic and social structure of a language over time (Romaine Reference Romaine, Ammon, Dittmar, Mattheier and Trudgill1988: 1452; see also Chapter 1 by Romaine in this volume). Similar to sociolinguistics which studies present-day language use and its users, it is based on the notion that language variation and change is a social phenomenon, and that its primary goal is to understand language use over time (see, e.g., J. Milroy Reference Milroy1992a: 221; Chambers Reference Chambers1995: 11). If language and society are to be considered inseparable, research in sociolinguistic aspects of earlier stages of language – similar to studies in synchronic sociolinguistics – must include a rich view of the society in which changes occur (also Labov Reference Labov2001: 326–7). Nevalainen and Raumolin-Brunberg (Reference Nevalainen and Raumolin-Brunberg2003: 9–11) also stress the importance of wide knowledge of historical texts of the time covered by the linguistic study itself. In their opinion, reliable reconstructions of how language change works in society are possible only when the social, empirical, and historical requirements for the study are fulfilled.

Historical sociolinguistics could also be said to be the starting point to other historical research on human interaction, namely, historical pragmatics (see Chapter 3 by Mazzon in this volume). In the first article which properly tackles the definition, Stein (Reference Stein1985b: 347–8) identifies two overlapping approaches to historical sociolinguistics. The first studies how the social and stylistic meanings of individual linguistic items affect linguistic change, whereas the other studies the ways in which changes in the outside world influence changes in individual language structure. Historical pragmatics falls under this second type of sociohistorical linguistics and, as Stein (1985b: 348) notes, in its wider sense, comprise such cognitive-related research topics as, for example, the use of address.

Jacobs and Jucker (Reference Jacobs, Jucker and Jucker1995: 11) call those approaches of historical pragmatics which study the describing and understanding of language conventions pragmaphilological. They concentrate on the contextual analysis of historical texts, including the writer (addresser) and the reader (addressee) and the social and interpersonal aspects of their relationship, the overall method and setting of text production and reception, as well as the way in which texts are used to achieve a particular goal. Diachronic pragmatics, in contrast, focuses on different historical stages of the same language and attempts to give pragmatic explanations for either change in language in general or for development of smaller pragmatic units like speech acts and discourse markers, as well as politeness phenomena.

As stated earlier, in historical sociolinguistics and -pragmatics interdisciplinary research efforts include relying, for example, on social history. Both the correlational and the interactional branches of linguistic research can make use of this successfully. The correlational approach concerns the dependence of linguistic phenomena on extralinguistic factors which are traditionally looked for in sociolinguistics, such as social status, gender, and age. The interactional approach, on the other hand, is more concerned with communicative situations and speaker–hearer interaction. Since Romaine's seminal study on historical sociolinguistics, the focus has increasingly been on language-external factors, and historical sociopragmatics has further introduced contextual factors into the analysis of language variation and change.

This chapter will focus on some of these most prominent and well-studied processes of sociolinguistic change in the history of English, such as social embedding and the influence of social networks. It will also concentrate on such issues as social status and hierarchy, as well as social mobility. In studies on the history of English, social hierarchy emerges as one of the most central factors on the basis of sociohistorical (e.g. Wrightson Reference Wrightson2002) and linguistic research. Historical sociolinguists focusing on patterns of language variation and change, for example, have made extensive use of macro-social categories. Nevalainen and Raumolin-Brunberg (Reference Nevalainen and Raumolin-Brunberg2003), who study the diffusion of morphosyntactic features in Early Modern English correspondence, show, for example, that the changes studied can be driven either by status or mobility or both. Network studies as such provide a way of combining structure and agency in language use and ideally allow the researcher to analyse how linguistic innovations spread in a network and how the nature of ties between network members and their varying roles affect linguistic choices (J. Milroy Reference Milroy1992a, Milroy and Milroy Reference Milroy, Milroy, Coupland and Jaworski1997).

The focus will also be on similar, and partly overlapping, processes of sociopragmatic change in English, like the construction of social roles and identities. The social network analysis (SNA) approach has been extended and developed for studies of the way in which not only language innovations spread, but also in which the strength of ties between network members and their different roles vary and change. Diachronic change in the available linguistic and societal resources and the ways in which language, society and individual are connected is an intriguing topic, and research into identities, roles and networks increases our understanding of language as part of social life and opens new ways in which language history could interact with disciplines like sociology and social history (Bucholtz and Hall Reference Bucholtz and Hall2005; Omoniyi and White Reference Omoniyi and White2006).

Because of their social, societal, and interactional basis, both sociolinguistic and -pragmatic processes have been studied in material which involves dialogic communication, such as personal correspondence, drama, and trial proceedings. Both historical sociolinguists and -pragmaticists have widely discussed the validity of such research material. Labov, for one, emphasized the role in historical linguistics of ‘the art of making the best use of bad data’ (1994: 11). This means that not only is the survival of historical texts random, these texts are only indirectly comparable to the spoken idiom. For this reason, the topic of validity also concerns the availability of data. Whereas literary texts have been more easily retrievable, those representing ‘authentic’, non-fictional language have been difficult to find and use. This kind of data is mostly accessible for later periods of the history of English, from late Middle English onwards, and therefore the current chapter will introduce studies which have used texts from the late medieval to the late modern period (1400–1800) as their primary data.

The chapter is organized as follows. It will, first, introduce more general processes of sociolinguistic and -pragmatic change, and then move onto a more detailed discussion of research in certain linguistic phenomena which have been widely studied in the history of English during the past years. Since the sociolinguistic and sociopragmatic approaches often overlap, the chapter will concentrate more on those issues which have been studied from both aspects of diachronic change. These case studies will include address pronouns and modal verbs and expressions, the examples ranging from those adopting a more sociolinguistic approach (e.g. Nevala Reference Nevala, Taavitsainen and Jucker2003) to those with a more distinctively sociopragmatic approach to language change (e.g. Fitzmaurice Reference Fitzmaurice2000).

17.2 Sociolinguistic and -pragmatic processes

In social sciences, any social process can be seen to work on different levels, each of them pertaining to the particular scope of the process. Sztompka (Reference Sztompka1993), for one, proposes the presence of three separate sociocultural process levels: macro processes (i.e. global, national, regional), mezzo processes (i.e. large groups, communities, associations, political parties, etc.), and micro processes (i.e. small groups, families, schools, occupational settings, friendship circles). Language use can also be seen as occurring at varying levels, and within linguistics, sociolinguistic and -pragmatic variation and change are often discussed in terms of macro- and micro-level phenomena. Sociolinguistic processes include, for example, accommodation or standardization, whereas sociopragmatic ones involve identification and identity construction, as well as negotiation of speaker and participant roles (for a more detailed discussion on historical variationist research, see Chapter 1 by Romaine in this volume, and on historical pragmatics, see Chapter 3 by Mazzon in this volume). Both types of processes heavily relate to the concept of power, whether concerning the power relations between people of different social status or those of different social role. The linguistic processes discussed in this chapter will concentrate on the influence of social embedding and social networks in changing the English language, focusing further on social status and mobility as well as on social roles and networks.

17.3 Case studies on sociolinguistic and -pragmatic change

Many linguistic aspects in the history of English from spelling systems to morphosyntactic variation have been studied from the perspective of wider social processes and phenomena. In this section, two particular linguistic features are discussed in more detail: address pronouns and modal expressions. As the discussion will show, both features have been the subject of an increasing number of studies in relation to sociolinguistic and -pragmatic variables during the past few decades.

The studies of forms of address and modal expressions have largely concentrated on the period from Middle English to Early and Late Modern English, mainly due to the availability of data. Research questions have varied both according to the material, i.e. written vs. oral, and according to the type of approach. The sociolinguistic studies mostly focus on the extralinguistic factors of social status, as well as on the implications of power in the relationship between the speaker and the hearer. The more sociopragmatic approach comprises research which concentrates on the interactional side of address use, and which goes deeper into the emotive and situational aspects of interaction. In most cases, however, the analysis employs both the sociolinguistic and -pragmatic approach.

17.4 Concluding remarks

Sociolinguistic change covers a wide range of processes on different levels. But even though such variables as social status, gender, or age, are discussed, the main focus is always on the linguistic behaviour of human beings and language users. An individual speaker is only as influential in language change as the most innovative in his/her group, and our status and role in our own social or language group(s) is constantly strengthened and maintained with the support of other members in it.

Social networks and their role in the spread of innovations or in the death of certain linguistic features have proved to be fruitful ground for both sociolinguistic and -pragmatic research. The historical stages of English are increasingly studied from the point of view of individual variation and change, and the influence of entire networks of people and certain groups or groupings within that offers a much-needed insight into how people in different social roles take on a group identity and how language change as a process works as a tool to convey both diachronic and synchronic shift in how roles and identities are built, maintained, altered, and even discarded. The increasing amount of such qualitative research on historical sociopragmatic processes does not, however, weaken the results gained by quantitative sociolinguistic studies – they complement each other. Quantitative analyses can reveal norms behind particular speaker meanings.

The data suitable for the sociohistorical study of English have been gathered into various electronic corpora in recent years (see Chapter 8 by López-Couso in this volume). These collections of texts have been designed to represent both communicative contexts of genres and registers, such as personal letters, trial proceedings, and fictitious dialogues, and social dimensions, such as those concerning regional, gender, and age variation. As the first historical, multi-genre corpus, the Helsinki Corpus of English Texts (HC), has led the way to various diachronic collections like the Corpora of Early English Correspondence (CEEC-400), which can be used to study not only change in grammatical and lexical features, like -s/-th and you/thou variation (e.g. Nevala Reference Nevala, Raumolin-Brunberg, Nurmi, Nevala and Rissanen2002; Nevalainen and Raumolin-Brunberg Reference Nevalainen and Raumolin-Brunberg2003), but also in pragmatic and discursive patterns, such as reporting and personal reference (e.g. Nevala Reference Nevala, Nurmi, Nevala and Palander-Collin2009; Palander-Collin and Nevala Reference Palander-Collin, Jucker and Taavitsainen2010). Interpersonal communication can also be studied from material that represents speech-based language use as interaction (dialogue) in writing, as in courtroom discourse and play dialogue. For example, the Corpus of English Dialogues 1560–1760 (CED) has been designed to facilitate the study of topics which include politeness phenomena and conversational structure (Culpeper and Kytö Reference Kytö, Jucker and Taavitsainen2010). Regional, as well as register, variation can in turn be explored in historical corpora, such as the Corpus of Scottish Correspondence (CSC) and the more recent Corpus of Oz Early English (COOEE), which both offer data for the studies of linguistic accommodation (Fritz Reference Fritz and Tristram2000) and stylistic literacy (Meurman-Solin and Nurmi Reference Meurman-Solin, Nurmi, Gunnarsson, Bergström, Eklund, Fridell, Hansen, Karstadt, Nordberg, Sundgren and Thelander2004).

Acquiring reliable data will allow further study of both sociolinguistic and -pragmatic historical processes. The situation now is already very different from what it was twenty years ago when building language corpora had only just begun for English historical linguistics; today, we have an increasing amount of material not only to delve into social and societal change in respect to language use, but also to go much deeper into variation and change in individual linguistic repertoires. The answers to questions concerning ‘who’, ‘when’, and ‘where’ will be further supplemented by those relating to the question ‘why’.

18.1 Definitions of standard

Dick Leith defined a standard variety as one which has minimal variation of form and maximal variation of function (Reference Leith1983: 32). However, the word standard has many definitions and connotations, as can be seen from the sheer size of the entry in the OED, where there are three major strands of meaning: a military or naval ensign, e.g. the Royal Standard; an exemplar of measure or weight, e.g. standard deviation; and various senses associated with the verb to stand, e.g. standard lamp. With reference to language, the OED suggests that uses of the word derive from the second strand, with the sub-definition ‘an authoritative or recognized exemplar of correctness, perfection, or some definite degree of any quality’ (s.v. standard n., sense II.10.a). The earliest citation in which standard collocates with language is from 1742:

In attributive use, the OED defines standard with reference to language as ‘applied to that variety of a spoken or written language of a country or other linguistic area which is generally considered the most correct and acceptable form’ (s.v. standard adj., sense I.3.e). Standard thus has connotations of uniformity, correctness, and acceptability. To state that a variety is ‘reckoned’ or ‘generally considered’ to be standard implies a democratic consensus, but, as Jenny Cheshire and James Milroy point out ‘it is those who wield the most power who have most influence in determining the “correct” forms of standard languages’ so that standard varieties are ‘socio-political, as well as linguistic, entities’ (1993: 4).

These definitions relate primarily to the form of a standard variety, which is characterized by uniformity, but standards exist, in language as in weights and measures, to fulfil a function or a set of functions. Cheshire and Milroy use an analogy with objects such as car components to distinguish sub-standard from non-standard uses of language:

Cheshire and Milroy are here arguing against the stigmatization of non-standard dialects, pointing out that non-standard dialects of English are able to perform certain functions (such as differentiating singular you from plural yous) which standard English cannot, just as a handmade tool might be more appropriate for a specialist job than a standard one. This argument is convincing to linguists, but the stigma of non-standard varieties persists because standard varieties are associated with élite groups and with modern nation-states.

It is important at the outset to consider where the processes of standardization sit with regard to the mechanisms of linguistic change set out by Labov (Reference Labov1972: 178–81). Labov divides the stages involved in sound change to those involving change from below, i.e. ‘below the level of social awareness’ (1972: 178) and those concerned with change from above ‘a sporadic and irregular correction of the changed forms towards the model of the highest status group, that is the prestige model’ (1972: 179). While specific variants included in the repertoire of a standard language might have their ultimate origin in change from below, standardization itself always involves change from above, since it requires conscious agency. In terms of the traditional division of linguistic changes into internally and externally motivated changes, changes involving standardization are always externally motivated, even though specific features within the standard repertoire may originally have evolved as a result of system-internal factors. It is also important to distinguish between the processes of standardization and those of supralocalization and focusing, both of which, like standardization, involve a reduction in variation of form, but, unlike standardization, can occur as a result of change from below. Nevalainen and Tieken-Boon van Ostade use the word supralocalization as ‘an umbrella term to refer to the geographical diffusion of linguistic features beyond their region of origin’ and note that while this ‘results in dialect levelling, loss of marked and/or rare elements…many processes of supralocalisation in English…have been induced naturally by dialect contacts without any conscious effort towards producing an official standard language’ (2006: 288). Focusing is a term used by Trudgill (1986: 86) to refer to a stage in new dialect formation where there is ‘a high level of agreement in a language community as to what does and does not constitute “the language” at a given time’ (cited in Nevalainen and Tieken-Boon van Ostade Reference Nevalainen, Tieken-Boon van Ostade, Hogg and Denison2006: 287). This involves minimal variation of form, but focusing occurs in the formation of vernacular as well as standard varieties.

Standard varieties of languages arise when certain functions need to be performed by that language, generally the ‘higher’ or more prestigious functions associated with literacy and education, and/or when the language needs to be used as a means of communication between different groups of users. Historically, the process of language standardization is thus tied in with the introduction of literacy and the use of the language in education, government, religion, and other prestigious registers. Unlike tools, which are standardized for purely functional reasons, the standardization of language often has social and/or political motivations. Consequently, in researching processes of standardization in English historical linguistics, scholars have employed methods which have allowed them to find evidence of: (i) uniformity of form among documents from different locations; (ii) the ‘higher’ functions associated with standard varieties; and (iii) the ideological evaluation of these varieties as superior to others. In this chapter, we first consider the highly influential model of standardization provided by Einar Haugen (Reference Haugen1966) and its later modifications by Lesley and James Milroy (Reference Milroy and Milroy1999), further modified by Terttu Nevalainen and Tieken-Boon van Ostade (Reference Nevalainen, Tieken-Boon van Ostade, Hogg and Denison2006), and by Asif Agha (Reference Agha2003). All versions of this model describe standardization as involving several stages, which have some correspondence with successive periods in the history of English, so sections 18.3–18.7 present accounts of research on standardization of English diachronically, from Old English (section 18.3) to the very recent past (section 18.7). Another reason for this diachronic approach is that the evidence for standardization differs considerably both qualitatively and quantitatively from one period to another. This means that both the research questions posed and the methodological approaches to answering them differ in emphasis from one period under study to the next.

18.2 Processes

The foundational model for the process of standardization in language has been that set out by Einar Haugen (Reference Haugen1966). Haugen argues that, in defining terms such as language and dialect, we need to consider two dimensions, that of structure, involving the formal properties of varieties, and that of function, or its social uses. Haugen notes that, while from a purely structural point of view, all varieties of a language are equal, ‘this dimension of functional superiority and inferiority…is an essential part of the sociolinguist's concern’ (1966: 927). If we apply this dichotomy to varieties of English at various historical stages, we can distinguish between the processes of focusing or levelling, whereby varieties become structurally similar but continue to function as vernaculars with no élite functions or social prestige, and those in which structural uniformity is accompanied by prestige and endorsed by authority.

Haugen (Reference Haugen1966: 933) describes the process of standardization as involving four aspects of development: ‘(1) selection of a norm, (2) codification of form, (3) elaboration of function, and (4) acceptance by the community’. He notes that the first two of these processes relate to the form of the standard variety, and the second two to its functions, and that ‘they form a matrix within which it should be possible to discuss all the major problems of language and dialect within the life of a nation’ (Haugen Reference Haugen1966: 933). It is important to note that Haugen implies no chronological ordering by his numbering of these processes and, as Milroy and Milroy (Reference Milroy and Milroy1999 [1985]) point out, some of the processes may overlap and some are ongoing. They develop Haugen's model further by elaborating on the processes involved in the maintenance of a standard variety:

While Haugen describes the process of standardization in terms of form and function albeit recognizing the importance of prestige with respect to the latter, Milroy and Milroy place more emphasis on the role of ideology in the formation and maintenance of standard varieties. They note that ‘the various stages that are usually involved in the development of a standard language may be described as the consequence of a need for uniformity that is felt by influential portions of society at a given time’ (1999 [1985]: 22). They also consider the processes involved in maintenance of a standard (elaboration of function, codification, and prescription) to be ‘stages of implementation of the standard rather than aspects of standardisation itself’ and state that ‘these additional stages are stages that have been observed to follow from the ideology of standardisation’ (1999 [1985]: 23). Since the ideology of a standard language is still influential, the maintenance of that standard is ongoing.

This ideological approach has been developed further by Asif Agha, who uses the framework of indexicality introduced by Silverstein (Reference Silverstein, Basso and Selby1976) ‘to explain how particular systems of speech valorization come into existence in the first place and, once formed, exist as cultural phenomena over the course of some period for some locatable group of social persons’ (2006: 15–16). Silverstein posits three orders of indexicality whereby linguistic forms are associated with social categories. At the first order of indexicality, there is a correlation between a particular linguistic form and a social category, observable by an outsider such as a linguist, but not noticed by speakers themselves; at the second order, speakers come to rationalize and justify the link between the linguistic form and the social category – they notice that people from a certain place/of a certain social class use the feature. At the third order of indexicality, forms which have been linked with a certain social category become the subject of overt comment. The concept of enregisterment sheds further light on this process. Agha defines this process as involving the identification of a set of linguistic norms as ‘a linguistic repertoire differentiable within a language as a socially recognized register’ which comes to index ‘speaker status linked to a specific scheme of cultural values’ (Agha Reference Agha2003: 231). Within this framework, Agha discusses the establishment of RP as one such ‘linguistic repertoire’ which becomes ‘a socially recognized register’ (2003: 231). I shall return to Agha's account later, but it is worth noting here that, whereas Haugen's and Milroy and Milroy's accounts of standardization refer to written forms of language, and, indeed, Haugen asserts that ‘it is a significant and probably crucial requirement for a standard language that it be written’ (1966: 929), Agha's framework, relying as it does on the association of linguistic features with social characteristics by means of metalinguistic discourse, can be applied to the standardization of spoken varieties, and to any level of language.

In the frameworks provided by Haugen (Reference Haugen1966) and Milroy and Milroy (Reference Milroy and Milroy1999 [1985]), the processes involved in standardization are characterized by the minimization of variation in form, the codification of the forms which emerge as minimally variant, the extension of the standard to a wide range of (prestigious) functions, and the acceptance and subsequent maintenance of this standard. In researching the historical processes of standardization, the research questions which must be asked are: is there uniformity among texts from different locations? Is there evidence of deliberate codification/imposition of norms? Is there metalinguistic evidence that one variety is viewed as superior to others; and is this variety used in prestigious functions? The following sections show how these questions have been approached with regard to different periods in the history of English.

18.3 Old English (c.450–1100)

Whether such a thing as a standard language existed in the Old English period is a matter of debate. If we consider the research questions set out in the previous section, then evidence for such a standard would need to be found in the small number of surviving texts from this period, most of which are written in the West Saxon dialect. This dialect has been given a retrospective and possibly anachronistic status in that, as Richard M. Hogg (Reference Hogg2002: 7) points out, most textbooks use West Saxon as the basis for teaching Old English. Many histories of English take the view expressed by Baugh and Cable that ‘the West Saxon dialect attained something of the position of a literary standard’ but that ‘such a start as it had made towards becoming the standard speech of England was cut short by the Norman conquest which…reduced all dialects to a common level of unimportance’ (2005: 62). While it is the case that, after the Norman Conquest (1066), French took over the functions normally fulfilled by standard varieties, in attributing these functions to West Saxon there is a danger of imposing a modern ideology and, as James Milroy (Reference Milroy2001: 547) suggests, legitimizing the standard language as having an unbroken history as the ‘language of a nation state’.

Research on standardization in Old English has mostly concentrated on evidence from a group of Late West Saxon texts associated with Ælfric in Winchester (963–84). Hogg concedes that these are sufficiently uniform to ‘approach the level of a standard language’ (Reference Hogg2002: 7), but argues that the same could be said about a group of texts in the Mercian dialect written in the Lichfield area from about the beginning of the ninth century. Hogg points out:

The arguments for Late West Saxon being considered a standard variety rest on the evidence of relative uniformity, i.e. minimal variation in form, dissemination over a relatively wide geographical area, and the élite status of manuscripts emanating from religious houses. Helmut Gneuss (Reference Gneuss1972) found evidence of lexical uniformity in the existence of a set of ‘Winchester words’ found in manuscripts emanating from the scriptorium in Winchester under Æthelwold and Ælfric, and, more recently, Mechthild Gretsch (Reference Gretsch2006) has found evidence of the imposition of uniformity of spelling in the revisions made by Ælfric. Thus, with regard to this group of West Saxon texts, some evidence exists for minimal variation of form and for some kind of codification and/or imposition of norms. There is, however, no evidence that Ælfric's West Saxon had been chosen or accepted as a standard beyond the religious community that produced these texts. As Hogg implies by his use of the term focused, it would be more appropriate to view groups of texts such as those emanating from Winchester and Lichfield as evidence of the shared norms of particular communities of practice, in each of which a repertoire of linguistic forms was enregistered as appropriate for writing sacred and/or literary texts.

Some scholars have applied sociolinguistic methodology to obtain insights into the dynamics of these focused varieties of Old English. Ursula Lenker, applying Lesley Milroy's (Reference Milroy1987) model of social networks to the community in Winchester, concludes that ‘the “Winchester School” emerges as a closeknit, localised network cluster functioning as a mechanism of norm enforcement’ (2000: 236) and that ‘the development of standardized, focused linguistic norms, seems…to be a natural, almost inevitable consequence of a process of cultural focusing’ (2000: 238). Even this much more conservative analysis is hampered by lack of evidence: there is none of the metalinguistic commentary which, in Agha's analysis, would indicate enregisterment. It would appear that, while focused varieties had developed in specific communities, the process of standardization had barely begun by the end of the Old English period.

18.4 Middle English (c.1100–1500)

The Early Middle English period is characterized by the dialectal diversity of such texts in English as survive, and by the lack of even an embryonic standard variety. This is because, despite the fact that writing in English continued in some centres into the twelfth century, English ceased to be the language used for the functions typically associated with standard varieties. Haugen points out that ‘whenever an important segment of the population, an élite, is familiar with the language of another nation, it is tempting to make use of this as the medium of government’ (1966: 928). Although he is here referring specifically to post-colonial situations in twentieth-century Africa and Asia, the point is equally valid in the case of post-conquest England, where the majority of the élite spoke the Norman dialect of French as their first language. It is not until the fourteenth century that English gradually replaces French in various high-status functions.

Research on the dialects of Middle English has been dominated by the projects which eventually produced the Linguistic Atlas of Late Mediaeval English (LALME) and the Linguistic Atlas of Early Middle English (LAEME). The methodology developed by Angus McIntosh and used by McIntosh and M. L. Samuels involved what McIntosh (Reference McIntosh1963: 7) refers to as the ‘fit-technique’. By this method, linguistic variants found in texts of unknown provenance are compared with those which can be placed geographically on external grounds. If a manuscript contains a set of features which have been allocated to a specific dialect area on the basis of other locatable texts, then it can be assigned to the same area on the basis of ‘fit’. Subsequently, other features found in this manuscript can be added to the repertoire of the dialect and used to ‘fit’ other texts, so that a fuller picture of ME dialects accrues. William Kretzschmar and Merja Stenroos suggest that this technique ‘is based on the premise that ME linguistic variation forms a regular continuum, into which any dialectally consistent text may be placed’ and that therefore ‘only “dialectally consistent” texts are included on the map’ (2012: 115–16). However, Samuels noted that the survey of Middle English dialects also provided ‘a frame of reference for isolating and classifying those types of language that are less obviously dialectal, and can thus cast light on the possible sources of the written standard English that appears in the fifteenth century’ (1963: 84). On the basis of ‘minimal variation in form’ between manuscripts from different times and places, Samuels (1963: 88) identified four such ‘types of language’: (i) the language of most of the texts associated with the Lollard preacher John Wycliffe; (ii) that of a group of seven fourteenth-century manuscripts including the Auchinleck manuscript; (iii), which Samuels considers ‘representative of London English of 1400’ and which includes the writings of Chaucer; and (iv), which Samuels calls ‘Chancery Standard’, and which ‘consists of that flood of documents that starts in the years following 1430’. Samuels (Reference Samuels1963: 85) considers type (i) to have been ‘a standard literary language based on the dialects of the Central Midland counties’ and ‘the type that has the most claim to the title “literary standard”’ before 1430. He dismisses types (ii) and (iii) as too heterogeneous to constitute even embryonic standards, but concludes that type (iv), the ‘Chancery Standard’ is ‘the basis of modern written English’ (Samuels Reference Samuels1963: 88).

Samuels's claim for Chancery Standard as the predecessor of modern standard English was echoed and reinforced by John H. Fisher (Reference Fisher1977), but has since been challenged. Michael Benskin (Reference Benskin, Kay, Hough and Wotherspoon2004) questions the idea that the documents of type (iv) were produced in Chancery or in any way enforced by administration, and points out that English played a minor role in Chancery compared to French and Latin. Laura Wright (Reference Wright, Fernández, Fuster and Calvo1994: 113), drawing on Haugen's idea of ‘elaboration of function’, questions whether the ‘diversity of function’ characteristic of modern standard English ‘could have arisen solely from Chancery documents’, which serve a very limited range of functions. In terms of form, the documents of Samuels's type (iv) do exhibit ‘minimal variation’, but they lack both the elaboration of function and the codification involved in Haugen's model of standardization.

From the fourteenth century we begin to find evidence of the kind of metalinguistic commentary which suggests that some varieties were more highly evaluated than others, evidence that was lacking from earlier periods. Rather than any specific variety being enregistered as ‘standard’ or ‘prestigious’, we find comments about varieties which are stigmatized and therefore non-standard. Katie Wales points out the North was ‘constructed’ as ‘alien and barbaric’ from the medieval period onwards (2006: 65). She demonstrates how Northern varieties of English in particular attracted negative metalinguistic comments from the late fourteenth century onwards. A key text here is Trevisa's (1385) updating of Higden's (Reference Higden and Babington1865) [1330] Polychronicon in which Trevisa augments with much more evaluative language the earlier author's comment that the language of the Northumbrians, especially at York, is hard for southerners to understand. Translated into modern English spelling, Trevisa's words, as translated by Caxton (1480) are as follows:

The words ‘sharp, slitting and unshaped’ are derogatory in themselves, but the alliteration adds to the impression that Northern English is harsh-sounding. In metalinguistic comments such as Trevisa's we begin to see the other side of the coin of standardization: the indexing of varieties spoken far from the centre of power as ‘alien’ or barbaric, in other words, non-standard. This evidence was to multiply and intensify in the fifteenth and sixteenth centuries, when evidence for the existence of a standard variety of English is more solid. By the end of the Middle English period, the need had arisen for a standard variety of English to replace French as the élite medium, and, following the appearance of several focused varieties, by the end of the fifteenth century, there is evidence that a standard variety of English had been selected and was increasingly being used in a range of official and literary functions.

18.5 Early Modern English (c.1500–1700)

While it is conventional to mark the beginning of the Early Modern English period neatly at 1500, both Manfred Görlach (Reference Görlach1991) and Roger Lass (Reference Lass and Lass1999a) see the introduction of Caxton's printing press to England in 1476 as a watershed in the standardization of English. Although, as we saw in the previous section, there was evidence of incipient standardization earlier in the fifteenth century, the introduction of printing allowed for both the accurate reproduction and the wide dissemination of texts. Görlach notes that from this point on, standard English is the norm for all printed documents and any use of dialect is ‘marked’ and employed for effect. In the preface to his English translation of the Aenead, Caxton complains about the diversity of English, but his very complaint is written in the standard variety. For Caxton, ‘printing his books in London-based English was the only possible decision in his time’ (Görlach Reference Görlach and Lass1999b: 495).

Caxton's remarks constitute an early example of what Milroy and Milroy (Reference Milroy and Milroy1999 [1985]: vii, and throughout) call the ‘complaint tradition’, one of the mechanisms involved in the maintenance of standard varieties. Complaints about English in the sixteenth century often focus on the inadequacy of the vernacular in comparison with Latin, which was in this period ceding ground to English in the ‘higher’ functions of educated and religious usage, and with more codified European vernaculars, notably Italian and French. A standard variety of English had been chosen (selection stage), and the consistency of reproduction facilitated by the printing press allowed for minimal variation in form, at least as far as spelling was concerned (Scragg Reference Scragg1974), but it needed to be further elaborated in order to fulfil all the functions required of a standard language, and, as it had yet to be codified, variation between innovative and conservative forms of morphology and syntax is found in many text types. Early research into standardization in this period was focused on elaboration of function, with an emphasis on the eventual ‘triumph’ of English over Latin in a process of vernacularization (Jones Reference Jones1953), but more recently Paula Blank has re-evaluated this process as a transference of élite status from Latin to standard English: ‘Latin was dead – but long live the King's English’ (1996: 14). The rise of corpus linguistics in the late twentieth century has facilitated quantitative sociohistorical research demonstrating that the variation found in Early Modern texts displays the orderly heterogeneity of linguistic change in progress, with evidence of change both from above and from below (Nevalainen and Raumolin-Brunberg Reference Nevalainen and Raumolin-Brunberg2003, based on a corpus of private letters). Richard F. Jones's extremely comprehensive survey of sixteenth-century attitudes towards English has been highly influential in promoting a historicized triumphalist account of the rise of the vernacular. His chapter headings tell a ‘rags to riches’ story from Chapter I ‘The Uneloquent Language’ (1953: 3–31) to Chapter VI ‘The Eloquent Language’ (1953: 168–213) and focus on what appears to be a dramatic change of opinion in the Elizabethan period. Jones argues that many early sixteenth-century complaints about the inadequacy of English concern the lack of vocabulary suitable for writing about matters which had previously been handled in Latin.

As Jones demonstrates with copious examples of such complaints, English was considered uneloquent because, largely due to the teaching of rhetoric via the medium of Latin, the classical languages embodied eloquence. Jones goes on to demonstrate how, by the end of the sixteenth-century, when the English nation-state was becoming a world power under Elizabeth I, authors were boasting about the copiousness of English rather than bemoaning its barrenness.

This negative evaluation of English as ‘uneloquent’ and ‘inadequate’ was overcome by the wholesale introduction of new words, especially those borrowed from, or modelled on, Latin. Jones based his research on texts including metalinguistic comments on English, but the availability of the OED online has allowed more recent researchers to quantify the contribution of various languages to the augmentation of the English vocabulary in the Early Modern period. Culpeper and Clapham (Reference Culpeper and Clapham1996) thus demonstrate that more than 13,000 new words were borrowed from Latin in the period between 1575 and 1675 and that, of all words whose first citations in the OED are from sixteenth-century sources, around 35 per cent are taken from other languages. Here we can see how modern quantitative studies complement the literary-historical approach taken by Jones to reveal the ways in which elaboration was achieved.

Paula Blank takes it as given that a standard variety existed by the sixteenth century. For her ‘the more pressing question…is not when the centralisation of English got underway, but when individuals began to articulate the idea that one form of English represented a prestige form’ (1996: 14). While her research is based on similar material to that of Jones, Blank is more concerned with matters of language ideology and draws on perspectives from critical theorists such as Bakhtin. Blank's work provides an important corrective to Jones's triumphalism and, while making no reference to Haugen's framework, gives abundant evidence for the acceptance and elaboration of what she calls ‘the King's English’. She also suggests that the process of codification is at work in this period: ‘in Renaissance England, grammarians, lexicographers, and spelling reformers were already concerned with advancing the King's English’ (1996: 14). As we shall see in the next section, codification is usually considered to be the preserve of Late Modern English, but other scholars concerned with specific areas of language have noted the beginnings of this process in the late sixteenth and early seventeenth centuries.

Codification is evident in attempts to standardize the spelling of English in this period. Notwithstanding Scragg's point above concerning the regularizing effect of the printing press, Vivian Salmon notes that although ‘printers of the early sixteenth century demonstrate little obvious interest in working towards a standardized orthography’ (1999: 25), scholars and schoolmasters had been debating issues of spelling and pronunciation from the 1540s. Salmon cites Mulcaster's Elementarie (1582) as ‘the first consistent attempt to codify and promulgate detailed rules for normalising and regularising traditional English spelling’ (1999: 20), but, as is demonstrated in Dobson's comprehensive account of sixteenth- and seventeenth-century orthoepists (1957), there was considerable debate in the sixteenth century about the unphonetic nature of the English alphabet including several proposals for spelling reform, such as John Hart's Orthographie (Reference Hart1569) and William Bullokar's Booke at Large (Reference Bullokar1580). Their ideas for reformed spelling were not widely taken up, but the focus on orthography drew attention to the relationship between spelling and pronunciation, and hence to the choice of a model of pronunciation on which spelling should be based. The comments of sixteenth-century authors intensify the enregisterment of regional accents as non-standard, and identify the pronunciation of educated, upper-class Londoners as ‘the best’. Paula Blank argues that such comments constitute ‘overwhelming evidence that prescriptivism has its sources in the linguistic researches of the sixteenth century’ (1996: 9). This is not to say that the stage of standardization identified by Milroy and Milroy as prescription had yet been reached. What Blank (Reference Blank1996: 9) terms ‘prescriptivism’ in the Early Modern period was ‘diagnostic in its methods and its aims’ and hence, in terms of the framework provided by Haugen (Reference Haugen1966) is more characteristic of the selection phase. Authors such as Hart and George Puttenham (1589) single out Northern and Western dialects as the most outlandish, proscribe the usage of the lower classes, and recommend the speech of London, but as a model rather than as a set of codified rules. Seventeenth-century orthoepists such as Alexander Gill (1619) and Christopher Cooper (Reference Cooper1687) singled out specific pronunciations for condemnation but, as we shall see in the next section, the full codification of pronunciation was a development of the Late Modern English period. Likewise, early grammars and dictionaries of English, as Blank points out, highlight the social and geographical differences between varieties of English and provide models that favour the standard. She cites Robert Cawdrey, whose A Table Alphabetical (Reference Cawdrey1604) is the first extant monolingual dictionary of English:

In contrasting ‘Court talke’ and ‘country-speech’ and favouring the former, Cawdrey is contributing to the selection and elaboration rather than the codification of the standard. Likewise, early grammars of English, such as John Wallis's Grammaticae Linguae Anglicanae (written in Latin, 1652/3), while undoubtedly codifying the language in the sense that rules were set out, lack the prescriptivism of many of their eighteenth-century counterparts.Footnote ¹ Wallis set out his aims as follows:

The use of shall and will became highly prescribed in eighteenth-century grammars, and their ‘misuse’ was a shibboleth of Irish English, but Wallis, whose grammar influenced later authors, sets these out in a descriptive manner:

Here, Wallis is clarifying English usage for foreigners: ‘we’, who ‘do not use them interchangeably’, do not need to be taught how to use them. Of course, in stating that ‘we do not use them interchangeably’, Wallis was idealizing and therefore codifying the use of shall and will. Evidence from the Helsinki Corpus (Kytö Reference Kytö1991) demonstrates that there was no clear preference for shall with first person or will with second-person subjects in sixteenth-century texts. On the other hand, corpus-based studies such as those of Nevalainen and Raumolin-Brunberg (Reference Nevalainen and Raumolin-Brunberg2003) demonstrate that some preferences which were to be codified later, such as the disfavouring of multiple negation, were already apparent in the sociolinguistic patterns of sixteenth-century usage. By the end of the seventeenth century, a standard form of English had been selected and accepted, and the process of elaboration was well under way, but codification was in its infancy.

18.6 Late Modern English (c.1700–1900)

It is a truism of English historical linguistics that the eighteenth century is the age of ‘correctness’ (Leonard Reference Leonard1929), when codification and prescription came to the fore. Leonard's account, based on an analysis of a selection of eighteenth-century grammars, was extremely influential throughout most of the twentieth century, but it has been challenged in recent years. One factor behind the re-evaluation of eighteenth-century ideas on language is the greater availability of a wider range of texts via Eighteenth Century Collections Online (ECCO). Close scrutiny of these texts has revealed a wider range of attitudes, ideas, and motivations among eighteenth-century grammarians (Beal et al. Reference Beal, Hodson, Percy and Steadman-Jones2006, Reference Beal, Beal, Nocera and Sturiale2008; Tieken-Boon van Ostade Reference Tieken-Boon van Ostade2008b). The availability of biographical material, letters, and other writings by the authors of eighteenth-century grammars, along with electronic corpora of texts from this period, has facilitated studies which compare precept and practice (Tieken-Boon van Ostade Reference Tieken-Boon van Ostade1982; Yáñez-Bouza Reference Yáñez-Bouza and van Ostade2008; González-Díaz Reference González-Díaz and Tieken-Boon van Ostade2008; see also Chapter 10 by Yáñez-Bouza in this volume). Another important strand of research into Late Modern English standardization is concerned with language ideology and politics. Tony Crowley (Reference Crowley1991) provides political-ideological commentaries on extracts from a range of texts including several key eighteenth-century sources, and discussion of ideological and political motivations of eighteenth-century authors is a key part of the scholarship challenging Leonard. Since the eighteenth century also saw the publication of authoritative dictionaries such as that of Samuel Johnson (Reference Johnson1755), and guides to pronunciation, scholars in lexicography (Osselton Reference Osselton and Hartmann1983) and phonology (Beal Reference Beal1999) have also discussed codification and prescriptivism. Although Agha's research properly belongs to the anthropology of language, he takes the emergence of RP in the eighteenth and nineteenth centuries as a case study for enregisterment. In doing so, he draws heavily on Mugglestone's (Reference Mugglestone1995) account of attitudes to pronunciation in the Late Modern period.

Late seventeenth- and early eighteenth-century authors express a keen awareness of the need for regulation of the language. Jonathan Swift (Reference Swift1712) attempted to set up an academy on the lines of the Académie française to deal with these problems, but was thwarted (Beal Reference Beal, Branca-Rosoff, Fournier, Grinshpun and Régent-Susini2011). Tieken-Boon van Ostade suggests that ‘it finally became clear, after the death of Queen Anne in 1714, that England would never have an academy’ (Reference Tieken-Boon van Ostade2008b: 4). Instead, a number of individuals took it upon themselves to codify the language by producing their own grammars. Finegan refers to ‘the codification of English usage…not by an official academy, but by a disparate band of independent entrepreneurs’ (Reference Finegan and Romaine1998: 536). These ‘entrepreneurs’ were the numerous authors of grammars, dictionaries, and guides to pronunciation which appeared in the second half of the eighteenth century and increasingly moved from codifying the language to prescribing ‘correct’ usage. Only a handful of extant English grammars were published before 1700, and about 50 between 1700 and 1750, but over 200 appear in the second half of the eighteenth century. Perhaps the most famous of these is Robert Lowth's Short Introduction to English Grammar (Reference Lowth1762).

Leonard (Reference Leonard1929) and many others have suggested that the precepts set out by Lowth and other eighteenth-century grammarians were arbitrarily imposed on the language on such spurious grounds as conformity to Latin, or abstract logic. However, studies of educated usage indicate that, in several cases, the ‘rules’ set out by grammarians were already being followed. A case in point is the proscription of the ‘double negative’: Greenwood in his grammar of Reference Greenwood1711 states that ‘two Negatives or two Adverbs of Denying, do in English affirm’ (1711: 160, emphasis in original). Nevalainen and Raumolin-Brunberg (Reference Nevalainen and Raumolin-Brunberg2003) find that multiple negation was already disappearing from upper-class and educated usage in the Early Modern period, and Tieken-Boon van Ostade (Reference Tieken-Boon van Ostade1982) found this construction to be extremely rare in educated eighteenth-century writing. In this case, as in others such as preposition stranding (Yáñez-Bouza Reference Yáñez-Bouza and van Ostade2008) and double marking of degree (González-Díaz Reference González-Díaz and Tieken-Boon van Ostade2008), eighteenth-century grammarians were codifying, or, to use their word, ascertaining, educated usage rather than simply prescribing arbitrary rules.

While several English dictionaries had been published in the seventeenth and early eighteenth centuries, and they had progressed from providing lists of ‘hard words’ to including everyday words and even slang, N. E. Osselton points out that it was not until the middle of the eighteenth century that dictionaries ‘assumed an authoritarian or normative function’ (Reference Osselton and Hartmann1983: 17). This is not to say that the earlier dictionaries were purely descriptive: Osselton's study on dictionaries published between 1650 and 1750 demonstrates that the proscriptive ‘branding’ or marking of certain words in these dictionaries constitutes a ‘unique authoritarian stage’ (1958: 121) in English lexicography. What changes in the mid-eighteenth century, specifically in the wake of Johnson's (Reference Johnson1755) Dictionary of the English Language, is that readers expect a dictionary to inform them about the ‘correct’ use of words. Although Johnson eventually abandoned the ambition expressed in his Plan of a Dictionary to produce ‘a dictionary by which the pronunciation of our language may be fixed…by which its purity may be preserved, its use ascertained, and its duration lengthened’ (1747: 34–5), he still considered it ‘the duty of the lexicographer to correct or proscribe’ the ‘improprieties and absurdities’ of the language (1755: preface). Johnson also set out to define the boundaries of English: as Tony Crowley (Reference Crowley1991: 43) points out, he uses the language of citizenship in setting out his criteria for including words of foreign origin. Johnson notes how ‘the academicians of France…rejected terms of science in their first essay’ but ‘though they would not naturalize them at once by a single act, permitted them by degrees to settle themselves among the natives’ (1747: 5–6). In comparing his efforts to those of the compilers of the Dictionnaire de l'Académie française, Johnson is effectively setting himself up as a one-man academy.

The last domain of the English language to be codified was pronunciation. We saw in the previous section that there was, by the beginning of the Early Modern period, a consensus that the ‘best’ English was spoken by educated and genteel citizens of London and its immediate hinterland, and seventeenth-century orthoepists singled out certain pronunciations for censure, but it was not until the late eighteenth century that attempts were made to set out ‘rules’ for the correct pronunciation of English. As Börje Holmberg so elegantly states ‘it is in the eighteenth century that the snob value of a good pronunciation began to be recognised’ (Reference Holmberg1964: 20). Likewise, Charles Jones notes ‘a sea-change in the way linguistic usage is perceived to relate to criteria such as social status and place of geographical origin’ between 1750 and 1800 (2006: 117). The main instrument for codification was the pronouncing dictionary, the most successful of which was John Walker's Critical Pronouncing Dictionary first published in 1791 but appearing in more than a hundred editions up to 1904. Walker prefaced the dictionary with more than 500 ‘rules’, some of which provided descriptions of articulation, but many of which set out to codify, prescribe, and proscribe. Terms such as vulgar, which appears ninety-six times in Walker's dictionary, index non-standard variants, while antonyms such as polite, mark standard pronunciations. A good example is Walker's discussion of variant pronunciations of words such as merchant, servant, in which, while conceding that pronunciations with /ɑː/ were formerly acceptable, he asserts that such variants have ‘now become gross and vulgar’ and ‘only to be heard among the lower orders of the people’ (1791: s.v. merchant) (see Beal 2010 for further discussion).

Holmberg's use of the term ‘value’ foreshadows Agha's (Reference Agha2003) discussion of the role of eighteenth-century pronouncing dictionaries and elocution manuals in assigning indexical values of civility and correctness to the variety eventually labelled as Received Pronunciation. When Daniel Jones published his English Pronouncing Dictionary (1917), he explicitly renounced the idea of prescribing pronunciation, stating:

Nevertheless, in using words such as cultivated, Jones is both reflecting and reinforcing the enregisterment of the variety described in his dictionary as optimal, and his Pronouncing Dictionary in its successive editions has had a long life as a model for the teaching of British English pronunciation.

By the end of the Late Modern English period, all the stages in Haugen's model of standardization had been completed. With regard to pronunciation, the pronouncing dictionaries and elocution guides of the eighteenth and nineteenth centuries constitute attempts at codification, but, except where individual words are concerned, it would be difficult, if not impossible, to acquire a consistent pronunciation from books. The role of these printed sources in the codification of pronunciation is thus better understood within Agha's framework of indexicality and enregisterment, whereby individual variants or whole repertoires become associated with the persona of the ‘Public School man’ in the nineteenth century and hence become prestige models.

18.7 Conclusion

This chapter has considered evidence for the processes of standardization from successive periods of the history of British English with reference to the frameworks of Haugen (Reference Haugen1966) and Milroy and Milroy (Reference Milroy and Milroy1999 [1985]), along with the anthropological model of Agha (Reference Agha2003, 2006). Since evidence for what Agha terms enregisterment largely takes the form of metalinguistic commentary, this model can be most usefully invoked with respect to later periods, from which such evidence survives, but when available it provides confirmation for the selection and acceptance of a standard variety. However, it is not only the lack of such evidence that raises doubts about the existence of a standard in the Old English and Early Middle English periods. Although there is evidence of focused varieties with minimal variation of form within communities of practice in these early periods, the very existence of ‘rivals’ to the title of standard English suggests that the selection phase was not yet complete. By the end of the fifteenth century, the relative uniformity of printed documents provides evidence that the selection phase is complete, at least with regard to the written language, and metalinguistic comments suggest its acceptance. During the Early Modern period, elaboration of function is in evidence, along with the beginnings of codification, which intensifies and extends to all linguistic levels in the Late Modern period.

However, the process of standardization is ongoing, because as Milroy and Milroy state ‘once it is well established and has defeated its competitors, a standard language must be maintained’ (Reference Milroy and Milroy1999 [1985]: 20). Despite, or perhaps because of, social mobility in the late twentieth and early twenty-first centuries, prescriptivism, like the poor, is always with us, as demonstrated by the popularity of publications such as Lynn Truss's Eats, Shoots and Leaves (Reference Truss2003), and the resurgence of elocution lessons in the guise of ‘business communication’ or ‘accent reduction’ (see Beal Reference Beal, Beal, Nocera and Sturiale2008, Reference Beal, van Ostade and van der Wurff2009b for further discussion of the ‘new prescriptivism’). For this reason alone, research into at least the prescriptivism stage of standardization will need to keep pace with developments in contemporary society. Research into standardization of earlier periods of English can also benefit from new developments in corpus linguistics, linguistic cartography, and sociolinguistics. In particular, it will be interesting to see how Agha's concept of enregisterment, already applied to historical studies of Northern English by Beal (Reference Beal2009a), Ruano-García (Reference Ruano García, Alegre, Moyer, Pladevall and Tubau2012), and Cooper (Reference Cooper2013) might be further applied to earlier periods of standard English.

19.1 Introduction

There are many factors which have contributed to the linguistic character of modern English, but one of them is undoubtedly contact.Footnote ¹ In this chapter I will be concerned to approach the notion of language contact, and its role in the history of English, from what I hope will be a nuanced, sociolinguistic-typological perspective. By this I mean that sociolinguistics shows us that language contactFootnote ² is not a single, unitary phenomenon, as it sometimes seems from the literature: the linguistic consequences of language contact can vary enormously depending on the particular sociolinguistic conditions in which it takes place (Trudgill Reference Trudgill2011b).

Contact can often, for example, lead to simplification. This is the kind of development which occurs as a result of short-term adult and therefore imperfect language learning – seen at its most extreme in the case of pidgins. But contact can also lead to complexification, as a result of long-term contact – typically co-territorial, and involving child-language bilingualism. In these situations, additional grammatical categories can be transferred to a language from neighbouring languages, as typically seen in Sprachbund type situations, with resulting added complexity.

In its examination of contact from a sociolinguistically informed point of view, this chapter deals specifically with those language-contact events involving English which have had structural consequences for mother-tongue English as a whole (Kastovsky and Mettinger Reference Kastovsky and Mettinger2001; Miller Reference Miller2012, Vennemann Reference Vennemann2011). There have been many other relatively recent, less major contact events which have influenced only specific varieties of the language: I have argued, for example (Trudgill Reference Trudgill2010), that the absence of third-person singular present-tense -s in the dialects of East Anglia is due to the fact that in the 1600s one-third of the population of the English city of Norwich were Dutch and French speaking. And colonial varieties of English have experienced particular contact events in their individual situations – though linguistic influences on mother-tongue colonial English varieties, other than at the lexical level, have mainly been from other European languages, such as Afrikaans in South Africa and German/Yiddish in the USA, rather than from the indigenous languages (Trudgill Reference Trudgill2004a).

But the history of the English language generally is a story which has been intimately connected with language contact from the very beginning. If we are looking for contact-based explanations as to why English as a whole is like it is, then we have to go back a very long way.

19.2 English as North Germanic

The Proto-Germanic language that English descends from was linguistically rather unusual within the Indo-European language family. It had many idiosyncratic features in its lexis, grammar, and pronunciation; and many of these idiosyncrasies – aspects of which are still to be found to this day in modern English – are often thought of as being the result of language contact (Hawkins Reference Hawkins and Comrie1990; Polomé Reference Polomé, Greppin and Markey1990).

The First Germanic Sound Shift, which gives us innovative /f/ in English foot as opposed to the original Indo-European /p/ in French pied, Greek πόδι, Lithuanian péda, has been argued by Wiik (Reference Wiik2003), if controversially, to be the result of a Finnic substrate, Finnic being the parent language of modern Finnish, Estonian, and Sami (Lappish). The hypothesis would be that on their journey westwards from the borderlands of Europe and Asia – if that is where the Indo-European homeland was located, as seems very possible – Indo-European speakers came into contact with Finnic speakers. Some of these Finnic speakers went through a process of language shift, with resulting substratum effects.

It has also been hypothesized that language contact and language shift occurred in the southern Scandinavian homeland of Proto-Germanic itself, after the Indo-European dialect in question had arrived there. Schrijver (Reference Schrijver, Bammesberger and Vennemann2003) provides a detailed phonological discussion of the possibility of a connection between Indo-European/Germanic and Sami, perhaps through some shared, now lost substrate. Vennemann (Reference Vennemann and Hanna2003a, Reference Vennemann, Bammesberger and Vennemann2003b, Reference Vennemann2010a, Reference Vennemann and Hickey2010b) has put forth an adventurous but important hypothesis about what happened. He argues – controversially (see Sheynin Reference Sheynin2004) but with support from e.g. Mailhammer (Reference Mailhammer, Kortmann and van der Auwera2011) – that the people who were already in residence when Indo-European speakers first arrived in southern Scandinavia were speakers of a language which was a member of the language family he calls Vasconic. This is the family which is today represented by Basque as the only modern survivor.

It could perhaps be this episode of contact which would account for the several simplifications which occurred in the transition from Indo-European to Germanic, arguably as a result of pidginizationFootnote ³ brought about by adult-language learning. Braunmüller (Reference Braunmüller2008) cites a number of examples of simplification which he ascribes to contact, all of which remain visible in modern English. Proto-Germanic, for example, had only two inflected tenses (present and past), and it had lost the Indo-European subjunctive (the form called ʻsubjunctiveʼ in Germanic continues the Indo-European optative). In addition, the Indo-European sigmatic aorist tense had gone.

Another possible example of contact-induced simplification is the development in Germanic of predictable word stress. In Indo-European, the location of the main stress in a word was unpredictable; but in Germanic this had changed so that all words had their stress on the first syllable (Lehmann Reference Lehmann1961). This is something that we can still see in modern English, where original Germanic words have this stress pattern unless the first syllable is some form of prefix.

A final candidate for the label of simplification is the growth in Germanic of weak dental preterites (Prokosch Reference Prokosch1939: 194), giving English want–wanted as opposed to the Indo-European-type strong verbs of the type sing–sang–sung. One possible mechanism leading to this development was the grammaticalization of the weak verb dō ‘do’ (e.g. Braunmüller Reference Braunmüller2008, Kiparsky 2009). But in any case, the consequence was that part of the verb system was entirely regular and would have been more readily learnable by non-native adult learners.

But Vennemann's hypothesis does not stop there. Contact with Vasconic, he argues, cannot explain all the peculiarities of Germanic. One of these peculiarities is the fact that Proto-Germanic had a vocabulary which was as much as one third non-Indo-European in origin, as a number of linguists have suggested (Feist Reference Feist1932). Hawkins (Reference Hawkins and Comrie1990) reports that the Germanic words which are not of Indo-European origin belong to the core of the basic vocabulary; they include common words, such as those to do with maritime travel like sea, ship, boat, sail, as well as words for sword, shield, eel, calf, lamb, king, and very many others.

Hawkins has no suggestion as to which language these words might have been borrowed from, but Vennemann (Reference Vennemann and Tristram2000, Reference Vennemann and Brinton2001) does have a proposal which is not without support from others (see Coates Reference Coates, Ahrens, Embleton and Lapierre2009). He claims that the vocabulary is not Vasconic. Rather, the lexical items in question originate in a language he calls ʻSemitidicʼ, which was a member of the Afro-Asiatic (formerly ʻHamito-Semiticʼ) family – and specifically a subgrouping which he labels ʻAtlanticʼ. Vennemann's view – again controversial (Sheynin Reference Sheynin2004) – is that, some time during the 4000s bc, sea-faring Semitidic-speaking people travelled north as colonizers from the western Mediterranean along the Atlantic coastline of Europe, to the islands and along the major rivers, reaching southern Sweden in about 2500 bc. In this Germanic homeland they formed a minority ruling class, and eventually went through a process of language shift – but not without first bequeathing a large amount of vocabulary to the Indo-European dialect which became Germanic (see also Mailhammer Reference Mailhammer, Kortmann and van der Auwera2011).

So we can speculate – and it is speculation – that the ancestor of modern English, Proto-Germanic, was a dialect of the Indo-European language which came originally from the borderlands of Asia and Europe and which was influenced in its phonology by contact with Finnic; in its grammar and phonology by contact with the ancestor of modern Basque; and in its lexis by contact with Afro-Asiatic. The development of Proto-Germanic occurred during and/or after the westward movement of Indo-European speakers across northern Europe into southern Scandinavia, and it seems possible or even probable that language contact played an important role in determining the actual linguistic nature of Germanic (Mailhammer forthcoming).

19.3 West Germanic–Continental Celtic contact

The next important series of contact-induced linguistic events which ultimately influenced the nature of English took place as a result of further migrations. Some time before the first millennium bc, groups of speakers of Germanic moved southwards out of their southern Scandinavian homeland into what is now Germany (Hawkins Reference Hawkins and Comrie1990). There they came into contact with speakers of another Indo-European language which was already in place: Celtic (Roberge Reference Roberge and Hickey2010).

Hawkins's map (Reference Hawkins and Comrie1990: 60), following Hutterer (Reference Hutterer1975), suggests that by 1000 bc Germanic was spoken all along the coastline of Germany, from the mouth of the Ems eastwards to the mouth of the Oder in the region of Stettin; and then inland as far as the regions of Hanover and Berlin. Five centuries later, by about 500 bc, considerable further geographical expansion had taken place. Germanic speakers were now to be found in an area stretching from the North Sea coast, starting well to the south of the Rhine Delta – so in modern Belgium – to as far east as the river Vistula, which runs through today's Warsaw; and then as far south perhaps as Amiens in the west and Cracow in the east.

Map 19.1 Germanic was spoken all along the coastline of Germany

(Hawkins Reference Hawkins and Comrie1990: 60, following Hutterer Reference Hutterer1975)

In all of those areas there was contact with Celtic speakers. Schumacher (Reference Schumacher and Zimmer2009: 247), who provides an in-depth consideration of some of the possible lexical and grammatical linguistic consequences of Germano-Celtic contact at this time, locates the main transition zone as being situated ‘upstream along the Rhine from its mouth, then across the middle of Germany, through the Czech Republic, and into the western part of Slovakia’ (2009: 249).

One much-mentioned example of a possible linguistic contact phenomenon resulting from this meeting of Celtic and Germanic has to do with the West Germanic copula (Keller Reference Keller1925; Schumacher Reference Schumacher and Zimmer2009). Old English had a distinction between two different copulas, wesan and beon. The two paradigms are seen in Table 19.1 (from Wischer Reference Wischer, Lenker, Huber and Mailhammer2010).

Wischer points out that this distinction was a purely West Germanic phenomenon: Bethge (Reference Bethge and Dieter1900) noticed that copula forms derived from an Indo-European ^*b-root were not found in North Germanic or East Germanic. Following on from the work of Tolkien (Reference Tolkien and Lewis1963, Reference Tolkien and Tolkien1983), Lutz (Reference Lutz, Sauer and Story2011) shows that not only does Old English have these two different present-tense conjugations for the copula, but that this was actually also a functional distinction, as in Celtic. According to Ahlqvist (Reference Ahlqvist2010), the Old English eom declension of wesan was used only for non-habitual meanings, while the b- declension of beon expressed the habitual; this system was precisely the one which was found in Celtic.

Laker (Reference Laker, Aertsen and Veldhoen2008) points out that the present-tense paradigms of the verb be actually have two different stems in the other West Germanic languages as well. Old Saxon had the present-tense singular 1, 2, 3 forms bium, bist, is, for example – compare the simpler North Germanic system: Old Norse em, ert, er. This suggests, Laker argues, that there were originally two functionally distinct paradigms of be in continental West Germanic as well as in Old English – with the functionality of the distinction having been lost by the time we get to the historical record. (This same situation is now found in modern English, of course: we have the two separate stems, as in am, be etc., but they are no longer used distinctively – something which in itself might be due to later contact between Old English and North Germanic in Britain.)

But if what Laker suggests is correct, the argument is strengthened that the semantico-grammatical habitual–non-habitual distinction may have been acquired as a result of contact with Continental Celtic even before West Germanic speakers crossed the North Sea to Britain. Vennemann goes one step further than this and argues that the two-copula system of Celtic, which he likens to that of Ibero-Romance (Trudgill Reference Trudgill2011a), was the consequence of contact in continental Europe with Vasconic. That is, Vasconic influenced Celtic as well as Ibero-Romance, and Celtic in turn then influenced Old English. This is an example, Vennemann (Reference Vennemann, Filppula, Klemola and Pitkänen2002b, Reference Vennemann2011) argues, of the transitivity of contact: Old English had linguistic characteristics which were ultimately the result of transfer from Vasconic via Celtic.

19.4 English–Insular Celtic contact: the early period

As is well known, West Germanic tribes eventually crossed the North Sea to Britain from coastal regions of what are now Belgium, the Netherlands, north-western Germany, and western Danish Jutland. This process started in the fourth century ad, but permanent West Germanic settlements in England did not begin appearing in any significant way until the middle of the fifth century.

Now Celtic and Germanic continued or resumed contact in England itself, where the contact was between the West Germanic of the westward-migrating Jutes, Angles, Saxons, and Frisians, and the specifically British form of Celtic, Brittonic. Once again, this contact had some rather major linguistic consequences, something which has not always been acknowledged in the past but is now becoming widely accepted (see for instance volume 13 of English Language and Linguistics 2009).

Initially, the contact took place in the Lowland zone of England (Jackson Reference Jackson1953). Morris (Reference Morris1973), albeit controversially, paints a picture in which, after the withdrawal of Roman imperial government and administration from the island in 410, an independent Romano-Celtic Britain continued to maintain Roman civilization for a number of decades (see further below). Then external threats led to invitations from the British to small groups of Anglo-Saxons to assist in the defence of the nation, probably in the late 420s. By about 442 many more Anglo-Saxons had arrived, and they had become numerous enough to revolt against their British hosts.

But the British fought back, and by about 495 they had subdued the Anglo-Saxons, who then remained for the most part confined, as a majority population, to Norfolk, Kent, and Sussex. There were plenty of ethnic English elsewhere, but they were in a minority. A second Anglo-Saxon revolt against the British began in the 570s, and eventually led to Anglo-Saxon control of most of England, and domination of the Celtic population by the Germanic population – the last British military victory over the English was in 655.

So during the period from approximately 420 to 600, Old English speakers were dominated in most parts of Britain by Brittonic speakers. This led to competent bilingualism on the part of the Germanic speakers, and thus to the borrowing of Celtic grammatical categories into English – a form of complexification typical of long-term co-territorial contact situations involving child bilingualism (see Trudgill Reference Trudgill2011b).

This complexification is attested to by a number of writers, including Ahlqvist (Reference Ahlqvist2010), Filppula et al. (Reference Filppula, Klemola and Pitkänen2002), Filppula (Reference Filppula and Tristram2003), Killie (Reference Killie, Stenroos, Mäkinen and Særheim2012), Laker (Reference Laker, Aertsen and Veldhoen2008), Lutz (Reference Lutz, Sauer and Story2011), Preusler (Reference Preusler1956), and Vennemann (Reference Vennemann and Tristram2000), who have discussed the acquisition by Old English of morphological categories from Brittonic in addition to the two-copula system. One example is the grammaticalization of the progressive aspect (Mittendorf and Poppe Reference Mittendorf, Poppe, Hildegard and Tristram2000; White Reference White, Filppula, Klemola and Pitkänen2002; Filppula Reference Filppula and Tristram2003; Lutz Reference Lutz, Sauer and Story2011, Killie Reference Killie, Stenroos, Mäkinen and Særheim2012): the progressive is generally considered not to be a typical feature of the Germanic languages, and is therefore a good candidate for influence from Brittonic in English, where it remains an important feature to this day (see also Keller Reference Keller1925). Ahlqvist (Reference Ahlqvist2010) gives a very persuasive account, on the basis of his thorough knowledge of Celtic linguistics, of why the argument that the English progressive results from influence from Celtic is a sound one.

19.5 English–Latin contact

But there was another form of language contact going on in England during this early period, too, which also had linguistic consequences. Schrijver (Reference Schrijver, Markku and Juhani2009) believes that, during the course of the Roman occupation, the Lowland zone of England had ceased to be monolingual Brittonic speaking and, that as a consequence of Celtic–Latin bilingualism, a specifically British form of Vulgar Latin had developed.

British Latin, Schrijver suggests, was extremely widely used by Celts, especially the upper classes, as a native language or second-language lingua franca. Schrijver's argument for this is based on a detailed phonological analysis which shows that Lowland Brittonic itself experienced considerable linguistic influence from British Latin, suggesting that Latin was spoken in one way or another by a large proportion of the Lowland population. Adams, too, writes that ‘there is evidence…for the spread of Latin among the Celtic population’ (Reference Adams2007: 622), and he describes ‘a Romanised Celtic population which had not received any sort of literary education’ and speaks of ‘lower social dialects [of Latin] as they were spoken in Britain’.

Crucially, as indicated by the subtitle of his 2002 paper ʻThe rise and fall of British Latin: evidence from English and Brittonicʼ, Schrijver (Reference Schrijver, Markku and Juhani2009: 209) argues that Old English phonology was the product of contact with a Celtic-influenced British Latin as much as with Brittonic itself – Vennemann's transitivity of contact in operation once again. This was because, while

There is no space here to go into the impressive, lengthy, and detailed phonological analyses which Schrijver uses to make his case (see Schrijver Reference Schrijver1995, Reference Schrijver, Filppula, Klemola and Pitkänen2002, Reference Schrijver and Higham2007, Reference Schrijver, Markku and Juhani2009), but the argument for a certain amount of Latin–Old English contact is a strong one.Footnote ⁴

Although Rome withdrew its imperial government from Britain in 410, as just noted, Morris argues that there is no reason to suppose that large numbers of people actually left the island – Roman civilisation ‘lasted some thirty years more’ (1973: 30). And it has been argued that British Vulgar Latin continued to be spoken by Romanized Celts well after the beginning of the fifth century, and even for long after that. Jackson (Reference Jackson1953) cites Pogatscher (Reference Pogatscher1888) as believing that there were Latin speakers in England until about 600. Schrijver goes further and talks of Latin lasting well into the 600s – he writes of ‘large numbers of Latin-speaking refugees from the Lowland Zone entering the Highland Zone…in the period between the fifth and seventh centuries’ (2009: 195).

19.6 English–Insular Celtic contact: the later period

To return to the topic of Celtic–Germanic contact, it turns out that Brittonic speakers were not only responsible for complexification in Old English. There is considerable evidence that they were, subsequently, responsible for simplification as well.

These differential linguistic consequences of contact at different points in time have to do with changes in the sociolinguistic context over the centuries. First, as we have seen, the early period of Celtic dominance involved fully bilingual acquisition of Brittonic by subordinate Old English speakers, which was when complexification occurred. However, this was followed by a later period of increasing Anglo-Saxon dominance. It was during this time that simplification took place as a result of the less than perfect acquisition of Old English by adult Celtic speakers in the Highland parts of southern Britain. The Lowland Britain of the south and east was divided from the northern and western Highland area along a line which can be drawn approximately, starting in the south, from Dorchester via Bath, Gloucester, Wroxeter, Leicester, Lincoln, and York to Corbridge and Carlisle (see Map 19.2).Footnote ⁵

Map 19.2 Late British survival in Highland Britain

(from Trudgill Reference Trudgill2010: 13)

This simplification can be seen from the fact that, while Old English was a highly synthetic fusional language, Middle English was much less fusional, showing a clear move towards a more isolating type of morphology. Many authors have advanced contact-based explanations for this development – Milroy says ‘it seems clear that such a sweeping change is at least to some extent associated with language contact’ (Reference Milroy and Blake1992b: 203). And since simplification is well known to be associated with adult language learning (see above), this would seem to be rather uncontroversial.

However, it is more controversial to argue that it was Celtic speakers who played a role in this contact, since it has been widely accepted that it was contact with Old Norse which was responsible (see below). To argue for the role of Brittonic, linguists who favour this hypothesis have had to counter two widespread misapprehensions, as they would see it.

The first is that Celtic could not have been influential because it died out too early: soon after the first Anglo-Saxon invasions, the Brittonic-speaking population fled, or were killed off, or assimilated. However, there is now much evidence to suggest that this was not at all the case – Celtic actually survived even in the Lowlands of England for many centuries. According to Morris (Reference Morris1973):

Brittonic survived even more strongly, and for much longer, in the Highlands of Britain – and especially in the remotest Highland fringes: in Strathclyde (south-western Scotland) and Cumbria (the neighbouring area of north-western England) where Brittonic Cumbric is believed to have survived until at least the twelfth century; and in Devon, as well as, obviously, Wales and Cornwall. Tristram (Reference Tristram2004: 113) suggests that

Gelling (Reference Gelling1993: 55) allows for more than four hundred years for the shift from Brittonic to Old English to have been completed, and suggests that the process was only complete around 900 ad.

The second anti-Celtic argument is that the chronology does not work: the simplification which occurred in Old English did not take place until centuries after the Brittonic language had died out in England and south-eastern Scotland. Tristram (Reference Tristram2004) counters this by pointing to the persistence of a speech-versus-writing diglossia in Anglo-Saxon England. This was an argument first put forward by Dal (Reference Dal1952), in connection with the very low level of occurrence of progressive verb forms in Anglo-Saxon texts. Wagner had a similar insight about this diglossia (Reference Wagner1959), which was developed further by Vennemann (Reference Vennemann and Brinton2001: 364):

Tristram's specific suggestion is that simplification occurred in Old English very much earlier than has hitherto been thought, at a time when contact with Celtic would have provided a very reasonable explanation. But these fundamental changes occurred in the spoken language without making their way into the written language – which provides the only data that we now have. The written evidence that has survived came from the pens of a small upper-class elite who preserved the knowledge of how Old English was supposed to be written, long after the original morphological complexity had disappeared from everyday speech. Evidence of simplification then appears in the written record only after the social breakdown brought about by the Norman Conquest. This led to the disappearance of the diglossia, producing the first evidence available to us of changes which had actually occurred several centuries earlier.

It makes sense, then, to argue that simplification in Old English was the result of adult language contact, after 600 ad and mainly in the Highland zone of England, between Old English and a substratum of subjugated Late British speakers, with eventual shift on their part to Old English. Crucially, in those Highland areas where Brittonic had survived the West Germanic take-over, its speakers were in a very different type of contact situation with Old English speakers than had been the case before 600 ad. Late British was now in a substratal, not adstratal or superstratal relationship with Old English; and something like a caste system, even involving slavery, was probably in operation (Hickey Reference Hickey1995; Pelteret Reference Pelteret1995; Thomas et al. Reference Thomas, Stumpf and Härke2008).

Demography would also have been very important for these developments – there were relatively few native speakers of Old English around to act as linguistic models. ‘The current discourse [of historians and archaeologists] advocates the theory of an elite take-over of the c. two-million Romano-British population by Anglo-Saxon “fringe barbarians”’ (Härke Reference Härke, Slack and Ward2002: 167).

19.7 English–Old Norse Contact

It thus seems that it was limited contact between a minority of Old English speakers and a majority of socially inferior adult Brittonic speakers in Highland England which set the process of simplification in later Old English going, as the Britons shifted to (their form of) English, and Brittonic was eventually lost. Since it was especially in the Highland zone that this pidginized form of Old English came to dominate, it is not surprising that it is in the north of England that we see the earliest and greatest degree of linguistic simplification in the historical record.

However, this northern locus has more usually been cited as evidence in favour of the argument that simplification in Old English was the result of contact with Old Norse. The Old Norse-speaking population was, after all, geographically concentrated in the north of England. This view concerning the role of Old Norse is inherited from Bradley (Reference Bradley1904: 32), who said that ‘we know for a fact that those districts in which the Danes had settled are precisely those in which English grammar became simplified most rapidly’. Poussa has it that contact with Old Scandinavian was responsible for ‘the fundamental changes which took place between standard literary Old English and Chancery Standard English, such as the loss of grammatical gender and the extreme simplification of inflexions’ (Poussa Reference Poussa1982: 84). Kroch et al. (Reference Kroch, Taylor, Ringe, Herring, van Reenen and Schøsler2000) make the same point. Further extensive and detailed discussions of the nature, extent, and consequences of Old Norse–Old English contact are found in Townend (Reference Townend2002), and in Thomason and Kaufman (Reference Thomason and Kaufman1988: 275–304).

However, a sociolinguistic-typological perspective on linguistic change (see below) suggests that contact with Old Norse is most unlikely to be the explanation for the simplification. Large-scale Viking settlements took place mainly during the ninth and tenth centuries, and led to many areas of eastern and, especially, northern England containing a heavily Scandinavian or Scandinavianized population, as famously witnessed by the hundreds of Norse place-names. The numbers of Scandinavians who actually arrived and settled in Britain is, however, unknown and the subject of much controversy (Härke Reference Härke, Slack and Ward2002; Holman Reference Holman2007).

What we know of the relevant sociolinguistic conditions suggests that contact between Old English and Old Norse speakers in northern England very quickly turned into one of long-term co-territorial cohabitation and intermarriage – very unlike the relationship which existed in the second stage of contact between Anglo-Saxons and subjugated Britons. Townend sees the significance of ‘Anglo-Scandinavian integration and the continued practice of mixed marriage’ (2002: 204), and Laing and Laing (Reference Laing and Laing1979: 185) speak of ‘the cultural fusion of Angle and Dane in the north’.

Moreover, not only did large numbers of words make their way from Old Norse into English (unlike Brittonic words) but, as a very persuasive piece of actual linguistic evidence for the intensity of the long-term contact, the English third-person plural pronominal forms they, them, their, theirs also have their origins in Old Norse, as is well known. The borrowing of pronouns from one language to another is a rather rare type of development which one never sees in short-term adult contact situations but only in long-term co-territorial contact of the type which typifies the growth of linguistic areas or Sprachbünde. According to Nichols and Peterson (Reference Nichols and Peterson1996: 242), the ‘borrowing of pronouns points to unusually close contact’ [my italics].

The fact that the contact was ‘unusually close’, and sufficiently intimate and long-term that pronouns could be borrowed, would seem to render much less tenable the popular thesis that it was contact with Old Norse which led to simplification in late Old English. Old Norse and Old English ‘were roughly adstratal in Viking Age England’ (Townend Reference Townend2002: 204). Contact between Old Norse and Old English was not of the sociolinguistic type that makes for simplification, which typically results from short-term adult second-language learning from limited exposure (Trudgill Reference Trudgill2011b). Rather, the North Germanic people would have had every opportunity to learn this closely related language in a non-pidginized form: this was co-territorial, long-term contact involving child bilingualism and therefore perfect acquisition.

This point of view is strongly supported by work by Emonds and Faarlund (Reference Emonds and Faarlund2014), who cite the Norse character of a number of Middle English syntactic constructions as compared to Old English. For example, van Riemsdijk (Reference Riemsdijk1978) makes the strong claim that the only languages in the world which permit fully developed preposition stranding are members of the North Germanic language family – plus English (example (1)):

1. (1) Reven ble skutt på

   ‘The fox was shot at’

   (Holmberg and Rijkhoff Reference Holmberg, Rijkhoff and Siewierska1998)

According to Emonds and Faarlund, ‘most languages disallow it completely’, and West Germanic Dutch allows it only ‘under very restrictive conditions’ – which was also the situation in Old English. According to Hoekstra (Reference Hoekstra, Haider, Olsen and Vikner1995), it is also disallowed in Frisian. (However, for a discussion of the work of scholars who have suggested possible Brittonic influence on the development of English preposition stranding, see Roma Reference Roma, Ramat and Roma2007.)

A second example is provided by the fact that in North Germanic and English, the original Germanic genitive -s ending has become a phrasal clitic which can be postposed to phrases as well as to head nouns, as seen in these Norwegian and English examples (2) and (3):

1. (2) Sjåførens feil

   ‘The driver's fault’

1. (3) Sjåføren av lastebilens feil

   ‘The driver of the lorry's fault’

In continental West Germanic, -s continues to be simply a genitive case inflection on head nouns only. Emonds and Faarlund mention only Dutch and German, but this is equally true of North Frisian (Löfstedt Reference Löfstedt1968; Hoekstra Reference Hoekstra, Dammel, Kürschner and Nübling2010) and West Frisian (Hoekstra Reference Hoekstra2006).

A third point is that there was a switch between Old English and Middle English from verbal prefixes to post-verbal particles. In modern German and Dutch, directional and aspectual particles still have to precede non-finite and clause-final finite verbs, as in example (4):

1. (4) Sie will den Brief herausnehmen

   ‘She wants the letter out-take’

This was also the situation in Old English. But these grammatical prefixes disappeared in Middle English, and English instead developed a system of post-verbal particles which took over the role of the prefixes. Modern English (example (5))

1. (5) She wants to take the letter out

exactly reflects the Scandinavian structure, as in Danish (example (6)):

1. (6) Hun vil tage brevet ud

   ‘She wants take letter-the out’

Fourthly, the infinitival marker to always occurred immediately adjacent to the verb in Old English, as it still does in Dutch and German. In Scandinavian, on the other hand, the marker is a free morpheme which can be separated from the verb by adverbs and negatives. The English infinitival to operates in exactly the same way (example (7)):

1. (7) Det er viktig å alltid komma i tide.

   ‘It is important to always come on time.’

It seems rather clear from Emonds and Faarlund's work, then, that the character of Modern English has indeed been influenced by Old Norse to some degree in syntax, as well as in lexis.

19.8 English–French Contact

It is equally clear, too, that contact with Norman – and other forms of – French has also had a very considerable impact on the nature of English. As is mentioned in all histories of the English language, this is especially true of English lexis, with 40 per cent of modern English vocabulary often being cited as the proportion of French-based words in the modern language (Rothwell Reference Rothwell1998).

Is it possible, however, that there was also some grammatical influence from French? Dalton-Puffer (Reference Dalton-Puffer and Fisiak1996) has provided us with an extensive and detailed account of the role of French in the development of Modern English derivational morphology. But this was initially a direct consequence of the transfer of massive amounts of French vocabulary to English – as was also, to an extent, the phonemicization of /v/ and /z/.

Otherwise any role for French in the shaping of English grammar would seem, on the face of it, to be unlikely. After all, how would this have happened? The number of native Norman FrenchFootnote ⁶ speakers in England was never very high – Carpenter (Reference Carpenter2004) gives a figure of 8,000 for the year 1086, out of a total population of perhaps 3 million (Hatcher and Bailey Reference Hatcher and Bailey2001). These figures are extremely approximate but, even so, we can be fairly sure that more than 99 per cent of the population of England in 1100 were English (or Norse or Brittonic) speaking. The proportion of French speakers, whatever it was, would have been even lower in Scotland and, especially, Wales; and very many of them were elite aristocratic members of a ruling class who had very little contact with the bulk of the Germanic-speaking population.

Bailey and Maroldt (Reference Bailey, Maroldt and Meisel1977), however, suggest that the influence of French on English was very considerable. Indeed, they go further than this and argue that Middle English was actually a ‘creole’ which developed as a result of contact and interaction between English and Norman French. This is clearly a misuse of the term ‘creole’, as has been powerfully and carefully demonstrated by Görlach (Reference Görlach, Kastovsky and Szwedek1986) and others; and Bailey and Maroldt's arguments have not been widely accepted (e.g. Danchev Reference Danchev and Fisiak1997 – though see Dalton-Puffer Reference Dalton-Puffer and Fisiak1995 and Rothwell Reference Rothwell1998). But their putative examples of French grammatical influence should nevertheless be considered.

The examples of influence Bailey and Maroldt cite include:

1. 1 The analytic gradation of polysyllabic adjectives, as with Modern English more/most beautiful. West Germanic languages, including Old English, use the bound morphemes -er and -est regardless of the length of the adjective (example (8)):

1. (8)

   German:	Sie ist interessanter als ihr Mann.
   	‘She is interesting-er than her husband.’

Middle and Modern English do not do this, and here Bailey and Maroldt see the influence of French, cf. Modern French (la) plus intéressante.

Interestingly, however, their thesis is undermined by Emonds and Faarlund, who argue that analytic comparative and superlative forms of English adjectives are actually due to Old Norse rather than Old French influence. They point to the fact that North Germanic also uses free words meaning ‘more’ and ‘most’ for the gradation of longer adjectives (example (9)):

1. (9)

   Norwegian:	Det var det mest underlege…
   	‘That was the most remarkable…’

The chronological and demographic case for the influence of Old Norse is more powerful than that for French. In particular, the Scandinavian proportion of the population – although, as we have just noted, not known with any certainty – was much higher, especially in the North of England.

1. 2 Bailey and Maroldt also cite as being the result of Norman French influence the use of oblique forms of personal pronouns in all functions in English except for non-conjoined subjects (example (10)):

1. (10) Well done him!

   John and me left

   Who did that? ThemFootnote ⁷

They compare this to the similar usage in Modern French of the oblique forms lui, moi, toi, eux.

Here too, however, Emonds and Faarlund claim Scandinavian influence. They argue that the modern English pattern is reflected in modern Danish, where we have (examples (11)–(13)):

1. (11) Mig og Ole gik i bio

   ‘Me and Ole went to the cinema’

1. (12) Ole er bedre kvalificeret end dem

   ‘Ole is better qualified than them.’

1. (13) Hvem vil have en øl til?

   ‘Who wants another beer?’

   - Os to ‘Us two.’

   - Ikke mig ‘Not me.’

   - Hende derovre ‘Her over there.’

Other examples given by Bailey and Maroldt, however, perhaps deserve a more sympathetic consideration: they suggest, for instance, that the development of English wh-interrogative pronouns into relative pronouns took place under French influence. In the modern languages, the pronouns who and qui both operate as relatives as well as interrogatives, something which is not generally true in North Germanic.

19.9 Conclusion

English is not a creole. But it is the result, in part, of millennia of language contact. At some points, contact has led to the transfer of additional grammatical categories and has thus added complexity to the structure of English. This was the case, it has been argued, with the borrowing of the progressive aspect from Brittonic Celtic. At other points contact has led to simplification as with, it has been argued, the loss of Old English morphology due to adult language learning by native-speakers of Brittonic Cumbric in a substratum situation in Highland England. And at yet other points it has simply led to replacive borrowing, as with the replacement of verbal prefixes by post-verbal particles under Old Norse influence.

In spite of all the different episodes of contact, English is still by general agreement an Indo-European Germanic language. But we can see, in its modern structures, the results of contact between the Germanic languages Old English and Old Norse, as well as contact with other Indo-European languages: Celtic Brittonic, Italic Latin, and Anglo-Norman. Less certainly, it may also show traces of contact with Continental Celtic and, even less certainly, with Finnic, Afro-Asiatic Semitidic, and Vasconic.

20.1 Introduction

The spread of English beyond the British Isles since the early modern period has resulted in a large number of post-colonial varieties of English. While it is difficult to give precise figures on the number of post-colonial Englishes (should acrolectal, i.e. standard Singaporean English, and basilectal, i.e. colloquial Singlish, be counted as one or as two varieties? Does the use of English in Cyprus qualify as a ‘variety’ of English?), it is clear that nowadays more people use English routinely as a second language than there are speakers who have it as their first language (see Crystal Reference Crystal2008). Different varieties of global English have been distinguished on the basis of their historical origin and/or their current function.

The different conditions under which English spread have been described in terms of ‘colonization types’ that, in turn, gave rise to different contact-induced Englishes (Mufwene Reference Mufwene2001, Schneider Reference Schneider, Schreier and Hundt2013): in countries, where English was used as a lingua franca for trade purposes (i.e. so-called ‘trade colonies’) English-based pidgins emerged, varieties of English with a simplified grammar that lacked native speakers. English was also used as a lingua franca among the African slaves and in communication with their owners on the plantations in places like the Caribbean; in such ‘plantation colonies’ the pidgins were later adopted as the first language by the (former) slaves and subsequently developed in grammatical complexity, thus giving rise to English-based creoles. This outcome from language contact has to be distinguished from the development in so-called ‘exploitation colonies’ such as in India or Singapore, where English was transmitted to a relatively small minority of the local population via second-language acquisition in a formal classroom situation, leading to the development of second-language (ESL) varieties of English, which also show traces of the local, indigenous languages, but are much closer, structurally, to first-language varieties of English than English-based pidgins and creoles. The huge number of English speakers who migrated to America, New Zealand, or Australia provided the input for the first-language (ENL) varieties in the ‘settlement colonies’, which are largely the outcome of dialect contact between different regional varieties of British English, but which also show traces of contact with the local indigenous languages (mainly on the level of the vocabulary). Schneider (Reference Schneider, Schreier and Hundt2013) adds the recent spread of English in countries without a colonial past (e.g. English in China, Japan, or South Korea) to the typology and labels it ‘postcolonial attraction’ (also subsuming the ‘grassroots spread’ of English in India as an example of diffusion among speakers of the lower classes under this label).

In terms of current function, a major distinction of post-colonial Englishes is between those which are spoken as a first language (ENL) and those which are institutionalized second language varieties (ESL). Beyond that, English is learnt as a foreign language (EFL) and used as an international lingua franca. Broadly speaking, this tripartite distinction corresponds to the three circles (inner, outer, expanding) in Kachru's (Reference Kachru and Kachru1992) widely known model. This model focuses mostly on standard or standardizing varieties of English, but language contact has also led to the emergence of structurally very different, non-standard Englishes, notably English-based pidgins and creoles. Thus, some models try to capture the result (with a focus on the ‘typology’ of different new Englishes, e.g. Kachru Reference Kachru and Kachru1992) whereas others (notably Mufwene Reference Mufwene2001 and Schneider Reference Schneider2007) focus on the process that led to the development of new Englishes, trying to capture similarities in the development regardless of whether the outcome is an ENL or ESL variety, or an English-based creole. Such models allow researchers to combine insights from fields as diverse as creole studies and corpus-based description.

In Mufwene's approach, the focus is on the factors that influence the selection of competing features from the input feature pool in a restructuring process that is the same in essentials for both dialect and language-contact situations (Mufwene Reference Mufwene2001: 4–6). The founder principle, for instance, is a concept first adopted from biology by Mufwene (Reference Mufwene1996) to explain the genesis of creole varieties. When applied to contact between different regional dialects of English from the British Isles, it explains the characteristics of new ENL varieties as resulting from those used by the speakers who founded the colonies. The emphasis in Schneider's (Reference Schneider2007) dynamic model is on identity as one of the main driving forces in the evolution of new Englishes, i.e. the change from a colonial identity to a post-colonial one that encompasses the various strands (former colonizers, indigenous population, other migrants) under one national (linguistic) identity.Footnote ¹ Biewer (Reference Biewer2012: 126) and Meierkord (Reference Meierkord2012: 64–7) bring together these components and spell out how the process of restructuring is shaped by cognitive principles and processes (i.e. second language acquisition, transfer, typological distance of substrate and target language, etc.), on the one hand, and social factors (identity, accommodation, norm orientation, cultural motivations), on the other hand.

Whenever speakers of English migrated to other parts of the world, they came into contact with speakers of other languages (i.e. the indigenous language(s) and/or other migrant languages). Equally important, migrants came into contact with other English speakers from other parts of the British Isles and Ireland, who tended to use a regional variety such as the Suffolk or Lancashire dialect rather than standard English. The development of new Englishes is therefore the result of both language and dialect contact. Matters are complicated by the fact that a particular feature can often be attributed to language contact with a substrate language or to a vernacular variety that formed part of the input (see Lass Reference Lass and Hickey2004a: 365). Dialect contact has been a major force in the development of ENL varieties in countries such as the US or Australia, but it is also relevant for the development of new Englishes in ESL countries.Footnote ² Similarly, language contact has been key in shaping ESL varieties on all levels, but it also played a role beyond the borrowing of native vocabulary (and occasionally also of structural patterns) in ENL varieties. Varieties most heavily influenced by language contact are the various pidgins and English-based creoles. However, language contact is also an important aspect of structural nativization (i.e. the adoption of local grammatical features)Footnote ³ and thus the development of ESL varieties. The fact that there is overlap in the processes involved in creole genesis and structural nativization in ESL varieties is also apparent in the terminology, notably the lectal continuum which ranges from the basilectal end showing heavy influence of contact with the indigenous language(s), through the less heavily influenced mesolectal forms to the more standard-like acrolectal varieties.

Influence from language contact is not only limited to indigenous languages spoken in the occupied territories; the input of other non-native speakers of English who migrated to countries where English became the majority language or an institutionalized second language add another layer of contact-induced change. While the native-speaker English input is typically referred to as the ‘superstrate’ and the local language(s) as the ‘substrate’, Schneider (Reference Schneider2007: 58–60) introduced the term ‘adstrate’ for immigrants and contract labourers, who acquired English as a second language and thus are agents in the development of a local variety of English.

The aim of this chapter is to outline research on the global spread of English with a focus on the processes of structural change that have shaped the emergence of new Englishes in both dialect and language contact (i.e. koinéization and nativization discussed in sections 20.2 and 20.3, respectively), and those that are the result of isolation (i.e. divergence and (re-) convergence, in section 20.4). Throughout, these processes will be illustrated with examples from standard and non-standard Englishes, and both ENL and ESL varieties. Research on the global spread of English has been conducted within sociolinguistics, it has relied on corpus-linguistic methodology, the transcription of early recordings, or it has used evidence from contemporary varieties as a window on the past. In other words, the corpus evidence used to study the development of new Englishes consists of collections of historical and contemporaryFootnote ⁴ texts, with some varieties being better documented than others. The review of the processes involved in the evolution of new Englishes has been informed by studies from these different fields in linguistics.

20.2 Koinéization and beyond

Koinéization (see Trudgill Reference Trudgill1986: 107ff.) is the term applied to the mixing and subsequent levelling of previous dialect differences as well as the development of interdialect forms. Koinéization happens because speakers of different dialect backgrounds tend to accommodate to each other in their speech behaviour (see Kerswill Reference Kerswill, Chambers, Trudgill and Schilling-Estes2002 for details). The outcome of koinéization would be a relatively homogenous variety with fewer dialect features than found in the input varieties. Indirect evidence of homogeneity comes from contemporary meta-comments. Montgomery (Reference Montgomery, Klemola, Kytö and Rissanen1996) found such remarks on the homogeneity of eighteenth-century AmE in letters and journals:

The relative lack of regional variation in New Zealand (NZE) and Australian English (AusE) (see Bauer Reference Bauer and Burchfield1994) appears to provide further indirect proof of koinéization.Footnote ⁵ But lack of regional variation is not necessarily a ‘given’ in dialect contact situations. A case in point is South Africa, where socially distinct input varieties (mainly working class in the Cape Colony but middle or higher class settlers in Natal) prohibited accommodation and led to the development of two relatively distinct ENL varieties in the course of the nineteenth century (see Lanham Reference Lanham, Bailey and Görlach1982).

According to Trudgill et al. (Reference Trudgill, Gordon, Lewis and Maclagan2000: 303–8), koinéization happens in three distinct stages: the first stage (migrant generation) is characterized by rudimentary levelling of dialect differences; extreme intra- and inter-speaker variability and some additional levelling are typical of the second stage (generation of the first native-born speakers); it is only in the third stage (following generation(s)) that a focused variety emerges through further levelling and reallocation (i.e. previously regional variants acquire social connotations; examples of this would be the current social significance of rhoticity in the US or variation in the trap and start vowels in Australia).

Even though koinéization is a concept frequently mentioned in historical accounts of new Englishes, detailed empirical studies of the process are scarce (see Kerswill Reference Kerswill, Chambers, Trudgill and Schilling-Estes2002: 670). A notable exception is Montgomery, who uses eighteenth-century letters by Irish immigrants since ‘Irish varieties of English can make a strong claim for a prominent role as input to Colonial American English, their speakers comprising perhaps one-quarter of all English-speaking emigrants before the American Revolution’ (1996: 222). He concludes that koinéization was a gradual process that is likely to have continued into the nineteenth century (Montgomery Reference Montgomery, Klemola, Kytö and Rissanen1996: 230ff.);Footnote ⁶ more importantly, the koiné is unlikely to have been widely used:

In the study of koinéization, one methodological problem is that formal, written language tends to be closer to standard English than colloquial spoken English. So ultimately, ‘koinéization of Colonial American English is in actuality the spread of literacy and standards of appropriate usage in writing’ (Montgomery Reference Montgomery, Klemola, Kytö and Rissanen1996: 232). Moreover, recent studies of Late Modern English letters have shown that regionally specific language use tends to be curiously absent even in this text type (see, e.g., Hundt Reference Hundt2015). Spoken language would therefore be a better source for the study of dialect birth.

For NZE, such data became available in the form of recordings made throughout rural New Zealand in the 1940s, including some of the first New-Zealand-born speakers. These enabled the Origins of New Zealand Project (ONZE) to document the embryonic stages of the New Zealand accent. According to Gordon et al. (Reference Gordon, Campbell, Hay, Maclagan, Sudbury and Trudgill2004), the New Zealand accent is not simply the outcome of koinéization. They ultimately argue for multiple factors, namely (1) the founder effect (south-eastern dialect features); (2) input from settlers who came via Australia, and (3) swamping effects from large-scale immigration from the south-east of Great Britain in the 1870s; the NZE accent thus developed as ‘a combination of these, the product of all three working in concert, each later one reinforcing the earlier ones’ (2004: 256).Footnote ⁷

Regarding present-day NZE, Hickey (Reference Hickey and Hickey2003b: 236) claims that the absence of regional variation is not simply due to accommodation and subsequent dialect levelling; instead, he postulates that it arose from supra-regionalization, i.e. ‘the adoption of the focused variety of New Zealand English from areas of high density, varied settlement to areas of lower, less varied settlement’ (ibid.) which resulted in the removal of minority, non-prestige features of local speech in the accents, e.g. of late nineteenth-century Irish immigrants. Additional factors apart from koinéization and supra-regionalization that help explain the apparent homogeneity of New Zealand English are internal migration (see Gordon et al. Reference Gordon, Campbell, Hay, Maclagan, Sudbury and Trudgill2004: 254) and close-knit network ties (see Britain Reference Britain, Bell, Harlow and Starks2005b: 164ff.).

20.3 Nativization

Contact between English and other languages leads to structural nativization, i.e. the adoption of local grammatical patterns.Footnote ⁸ These may be due to borrowing and transfer. In addition to direct influence through transfer and borrowing, more general cognitive processes related to the acquisition or learning of the target language (such as simplification and convergence) also give rise to contact-induced nativization of a new English.Footnote ⁹

More superficial signs of language contact are lexical borrowings which occur early on in language-contact situations and do not necessarily require bilingualism (see the many toponyms, terms for flora and fauna, and cultural concepts that English has borrowed from languages around the globe). Quite a few of these items have become part of global English (e.g. pyjamas or budgie), whereas others (e.g. puja ‘Hindi prayers’ and matai ‘NZ native tree’) are restricted to regional varieties. While borrowing most obviously affects lexis, prolonged language contact and bilingualism may also result in the borrowing of grammatical patterns as part of the structural nativization process. A good example is the kena-passive typical of basilectal Singaporean English (e.g. His tail like like kena caught in the in the ratch hut, ICE-Sing S1A-052), which employs the auxiliary kena and has adversative meaning (see Bao and Wee Reference Bao and Wee1999), i.e. it partially overlaps with the semantics of the English get-passive.

Bilinguals and learners of a second language may transfer structural patterns from their dominant/first language to the target language, especially in a situation of language shift. This happens on all levels of the language system: speakers whose first language is syllable-timed, for instance, often transfer the different rhythmical pattern to English to some degree (see, e.g., Low et al. Reference Low, Esther and Nolan2000). Target-language lexical items may be used with meanings closer to the source language(s), as for instance the wider application of aunty as a respectful term for an older woman (who may or may not be a relative) in different varieties of Indian English (IndE). Examples of transfer in the area of grammar would be the Irish after-perfect (e.g. I am after having my lunch ‘I have just had my lunch’; see, e.g., Hickey Reference Hickey2007: 197–200) or the omission of articles by speakers whose first language does not have articles, as in Her husband is manager of Ø sugar mill; see, e.g., Sand Reference Sand2004: 286). Transfer may also occasionally lead to a reintroduction of grammatical contrasts that were present in historical varieties of English but subsequently lost: an example would be the number distinctions in the pronoun system of pidgins and creoles whose substrate languages distinguish between singular, plural, and dual; such dual pronouns are also found in some basilectal ESL varieties (e.g. Fiji English us-two as a first-person dual inclusive and us-gang as a first-person dual exclusive; see Siegel Reference Siegel and Hickey2010: 832). Early attestations of grammatical transfer pose a problem for the description of new Englishes: on purely linguistic grounds, they are indistinguishable from interference phenomena (i.e. negative transfer in second language acquisition); however, once they are used with a certain text frequency, it can be argued that they are no longer contact-induced errors but have developed into features of the emerging variety.Footnote ¹⁰ In the following example from the Philippine component of the International Corpus of English (ICE-Phil),Footnote ¹¹ the verb teach is used with a different complementation pattern and word order than the one found in ENL varieties:

1. (1) STARTING this year, ‘mobile teachers’ will be deployed by the government in far-flung areas to teach the young with the basics of reading, writing and arithmetic. (ICE-Phil, W2C-020)

It is a pattern that is attested only once in the corpus, but if it were found to be used regularly, e.g. in local newspapers, it would be difficult to dismiss it as a contact-induced error.

In addition to borrowing and transfer, simplification and convergence are processes that lead to nativization in new Englishes. Simplification is a natural phenomenon that occurs in natural and instructional settings of second-language acquisition/learning. Simplification to an extreme degree characterizes the development of English-based pidgins and creoles, but it has also affected the ESL varieties in exploitation colonies. Examples of simplification in grammar that are attested from both contexts would be the omission of the third-person singular present tense ending or plural -s as well as copula omission in sentences such as He sick, She teacher. Reducing irregularity is another aspect of simplification in contact varieties that might lead to, e.g. a more general use of regular past tense endings (particularly in basilectal varieties) or the extension of the regular plural ending to non-count nouns in standard(izing) varieties of ESL, as evidenced in the contemporary material from the International Corpus of English (ICE):Footnote ¹²

1. (2) It has exported over one thousand heavy equipments…(ICE-Ind, S2B-017)

1. (3) Basic equipments for flower arrangement include a pair of scissors to cut branches and flowers, flower holders and vases. (ICE-Phil, W2D-012)

At the constructional level, the regularization of prepositional usage for verbs and related nouns can also be used to illustrate this tendency. In inner circle varieties, one can have a discussion about a topic, but to discuss about it is considered ungrammatical. Again, ICE components provide evidence that prepositional usage is regularized in the outer circle, particularly in spoken data:

1. (4) Now I would like to discuss about Premchands pictorial uses of language. (ICE-Ind, S1A-006)

1. (5) Can we discuss about our presidentiables. (ICE-Phil, S1A-021)

The collocation is occasionally also attested in written language use in contact varieties of English, as in the following example from ICE-Fiji (which, alongside the regularized use of discuss about also features the extended use of would and the progressive, discussed in more detail in Deuber et al. Reference Deuber, Biewer, Hackert, Hilbert, Hundt and Ulrike2012 and Hundt and Vogel Reference Hundt, Mukherjee, Mukherjee and Hundt2011, respectively):

1. (6) First, I would be explaining about the gender inequality, which often leads to the high incidence of poverty amongst women, which is what I would be discussing about in the second part of this essay. (ICE-FJ, W1A-016)

Simplification (and regularization) are a more general process in language acquisition, and it is therefore hardly surprising that discuss about is also a common collocation in learner Englishes (e.g. Nesselhauf Reference Nesselhauf2009: 18).

Convergence occurs if a local grammatical feature is found in the substrate and superstrate languages.Footnote ¹³ In Irish English, the progressive spread in the late modern period just as it did in English English; another example of convergence is the development of group verbs (i.e. phrasal and prepositional verbs) in Irish English and English English (Hickey Reference Hickey2007: 279).

The same processes (borrowing, transfer, simplification, and convergence) apply, in principle, to both substrate and adstrate influence on new Englishes. With more than seven million immigrants, German has had an influence on AmE beyond lexical borrowing: transfer from German has resulted in expressions such as Are you going with? (Wolfram and Schilling-Estes Reference Wolfram and Schilling-Estes2005: 122) and the use of hopefully as a sentence adverbial (meaning ‘it is to be hoped that’) (OED s.v. hopefully adv. 2.). Övergaard (Reference Övergaard1995: 44–6) suggests that the revival of the mandative subjunctive in AmE (e.g. I suggest that he consult a doctor) is likely to have been fostered by language contact with adstrate languages such as German, French, Spanish, and Italian in the Midwest. Adstrate influence is obviously not limited to the US. Clyne (Reference Clyne2003: 153) reports structural influence of German on the English spoken in the Tarrington-Tabor (Western Victoria) area of Australia, where speakers use the definite article with names of languages (the English, the German) while leaving out indefinite articles with occupational terms (e.g. She was teacher from German Sie war Lehrerin). Similarly, in Fiji and South Africa, with sizeable communities of sub-continental Indians (mainly descendants of indentured labourers), influence from Indian adstrate languages (notably Hindi) goes beyond lexical items (e.g. roti, samosa) and includes patterns such as the use of been as a past tense marker, as in You been tell me or affirmative questions without do-support, as in How often she goes to the cinema? (see Mesthrie Reference Mesthrie, Kortmann, Schneider, Burridge, Mesthrie and Upton2004 and Mugler and Tent Reference Mugler, Tent, Burridge and Kortmann2008).

20.4 Divergence vs. convergence – retention and innovation

During earlier periods in the global spread of English, the (relative) isolation of speakers is likely to have led to divergence while the more recent re-establishment of contact, especially via mass media has led to some (re-) convergence. Divergence of new varieties from the input dialects has often been attributed to two competing processes, namely retention (or ‘lag’) and innovation. The terms ‘retention’ and ‘lag’ are at times applied to the alleged wholesale retention of a previous stage of the language. An example would be the claim that, in the Appalachians, people have preserved the language of Shakespeare, i.e. Early Modern English. On closer inspection, wholesale conservation of historical dialects is a myth (e.g. Schneider Reference Schneider and Hickey2004: 262ff.).

One of the most systematic empirical studies of early AmE with respect to retention and innovation is Kytö's work on modals and verb morphology. She finds that modal verbs provide an example of colonial lag (Kytö Reference Kytö1991: 353) whereas verb inflection is an example of colonial innovation (Kytö Reference Kytö1993: 113). Both studies also show that more detailed analyses often reveal more complex historical patterns even in early colonial texts: the use of third-person will is actually a parallel development in both AmE and BrE (see Kytö Reference Kytö1991: 351), whereas AmE is, indeed, more conservative than BrE in retaining -th forms in the specific context of verbs ending in /s/ and /z/ (see Kytö Reference Kytö1993: 132). Likewise, research on southern hemisphere varieties (Trudgill Reference Trudgill1999a and Jeffries Reference Jeffries, Schreier, Trudgill, Schneider and Williams2010) shows that ‘lag’ may have played a role in the development of other (post-)colonial Englishes, but it is typically only one factor responsible for the use of a particular feature.

The main problem with the lag-innovation dichotomy is that it has also been applied to features of post-colonial Englishes (see Hundt Reference Hundt, Rohdenburg and Schlüter2009: 14 for details). However, what at first glance appears to be an archaism may actually be a revival of an old pattern. A case in point is the mandative subjunctive in AmE (see section 20.3), which shows a long-term pattern of decrease followed by a more recent increase in the twentieth century (see Övergaard Reference Övergaard1995; Hundt Reference Hundt, Rohdenburg and Schlüter2009: 31). The (post-colonial) revival of the mandative subjunctive is a change that has been spreading in other varieties in the second half of the twentieth century (see, e.g., Hundt Reference Hundt, Lindquist, Klintborg, Levin and Estling1998; Leech et al. Reference Leech, Hundt, Mair and Smith2009: 53; Peters Reference Peters, Peters, Collins and Smith2009b: 129), i.e. it is a change led by AmE but in which other varieties appear to be following suit. On a global scale, AmE is often found to be the lead variety in ongoing change. This does not only hold for the revival of a conservative feature such as the mandative subjunctive but, as one would expect, also for genuinely innovative constructions. One such example is the spread of pseudo-titles such as lawyer or journalist followed by a proper name (typical of journalistic writing), as in lawyer Peter Russell or journalist Helen Richards: varieties such as NZE or East African English are more advanced in the spread of an originally AmE pattern than BrE (see Bell Reference Bell1988 and Meyer Reference Meyer, Reppen, Fitzmaurice and Biber2002).

Research on one of the most isolated English-speaking communities – Tristan da Cunha – confirms the view that isolation leads to more than simply ‘lag’ or koinéization. Moreover, Schreier's (Reference Schreier2003) sociolinguistic-historical data from Tristan da Cunha provide evidence that dialect formation and change even in small, remote settings need to be attributed to a complex of factors, including the possibility of independent developments (in this case completive done and usta went/had/done).Footnote ¹⁴

Patterns of differential change that go beyond the simple dichotomy of lag/retention vs. innovation can only be studied on the basis of diachronic corpora that sample (national) regional varieties. On the basis of ARCHER (A Representative Corpus of Historical English Registers), which provides evidence of Late Modern BrE and AmE, and databases of narrative prose, Hundt (Reference Hundt, Rohdenburg and Schlüter2009: 32ff.) extends the original dichotomy beyond ‘extraterritorial conservatism’ (a more neutral term for ‘lag’) and ‘innovation’, to include ‘divergence’ (i.e. the development of two varieties in different directions), ‘parallel developments’, ‘resurrection/revival’, and ‘kick-down developments’ (i.e. initial conservatism in one variety which then takes the lead in later stages of the ongoing change). Genuine examples of extraterritorial conservatism in differential change are extremely difficult to find. It is much more common that apparent retentions are actually instances of more recent revival. Currently, differential change is difficult to study in varieties other than BrE and AmE, but diachronic reference corpora are beginning to be compiled for other World Englishes (Fritz Reference Fritz2007 or Hundt Reference Hundt2012), including ESL varieties (Collins et al. Reference Collins, Borlongan and Yao2014). However, the corpora for Early SingE and PhilE only provide evidence for developments in the second half of the twentieth century. Other corpora of (post-)colonial Englishes sample only one or a limited set of text types (e.g. the Corpus of Early Ontario English (Dollinger Reference Dollinger, Renouf and Kehoe2006a), the data used in Rossouw and van Rooy (Reference Rossouw and van Rooy2012) or the Corpus of Irish English Correspondence (McCafferty and Amador-Moreno Reference McCafferty, Amador-Moreno, Migge and Chiosáin2012)). Data collections of emerging new Englishes in exploitation contexts are currently still a lacuna for historical corpus linguistics. Thus, for the majority of new Englishes (both ENL and ESL), diachronic corpora are still unavailable.

20.5 Conclusion and outlook

The overview of processes of language change in global English has shown that there are a lot of similarities across the different settings. Both plantation and exploitation colonies have given rise to varieties characterized by language and dialect contact, typically ranging from more heavily influenced, basilectal varieties (e.g. basilectal Jamaican creole or basilectal Fiji English) to varieties that are close to the standard end of the spectrum (e.g. acrolectal Jamaican and Fiji English). While the feature pool that constitutes the input into new dialect formation will vary somewhat, essentially the same processes have been found to occur in the restructuring of the new varieties.

The second half of the twentieth and the beginning of the twenty-first century are characterized by large-scale migration to ENL countries, not only from the former colonies of Britain and the US but also from non-English-speaking parts of the world. The future of global English therefore needs to take into account the possibility of secondary dialect contact, i.e. the influence of ESL varieties and the use of EFL on ENL varieties. A case in point might be the use of the progressive in NZE (see Hundt and Vogel Reference Hundt, Mukherjee, Mukherjee and Hundt2011): close and frequent contact between ENL and ESL speakers in a small country like New Zealand could foster structural influence from ESL on ENL varieties. Outsourcing of business processes and information technology, notably in the form of call centres located in South and East Asia, but increasingly also in EFL countries like Bulgaria and Brazil (see Lim and Ansaldo Reference Lim, Ansaldo, Nevalainen and Traugott2012: 569) might prove an additional source of secondary dialect contact and resulting influence from non-native varieties on ENL in the long run (see below).

Despite the centrifugal forces that have been at work during the colonial and post-colonial history of the English language, English (unlike Latin) has not separated into different languages. Especially at the acrolectal end of the continuum, English has remained a language that allows for international communication on a global scale. There are no signs that this state of affairs will change in the near future. At the same time, speakers have been making use of the inventory available from the feature pool as well as independent developments to adapt the language to their local needs, at times by developing very localized forms of expression which will occasionally make it difficult for other speakers of ‘English’ to understand what they are saying. Some of these originally highly localized varieties have, in turn, been spreading through popular culture and, recently, through the electronic media. To some extent, these also lead to the adoption of vernacular features from varieties such as African American Vernacular English, Jamaican Creole, or Chicano English, as Mair (Reference Mair, Schreier and Hundt2013a: 317–18) points out:

The two opposing trends that we are likely to see in future developments of global Englishes are globalization and localization. Continued language and dialect contact are expected to lead to the global spread of features (e.g. a possible spread of discuss about to ENL varieties) whereas local identity construction is a force that is likely to result in the development of more localized norms of language use (e.g. the ‘exclusive’ use of basilectal varieties in countries such as Jamaica). In addition to the mass media, another force shaping English(es) is the globalization of the market, which has resulted in the outsourcing of communication services from ENL to ESL countries. As mentioned previously, call centres bring the local varieties back into contact with ENL varieties, but they also bring the ENL varieties into close contact with ESL varieties. In this particular area of globalization, we can expect to see mutual convergence of standard varieties, especially in writing, but they may not always be in the direction of the metropolitan varieties, as the example of the progressive in NZE shows.

Another aspect for future developments concerns norm orientation. In the past, teaching materials for ESL and EFL classrooms tended to be produced in Britain and the US, thus contributing to continued exo-normative influence from the two major reference varieties, BrE and AmE.Footnote ¹⁵ More recently, however, these materials are beginning to be produced more locally and also in ESL countries such as India, thus opening the possibility of external influence from one ESL variety onto another or on geographically closer EFL varieties; this opens up the potential for varieties such as Indian English to develop into regional epicentres, which would be evidenced in convergence of South African Englishes on the IndE model and divergence from the previous, exo-normative British model.Footnote ¹⁶

What is the evidence on which historical studies of new Englishes have been based? The answer to this question depends somewhat on the different structural levels of the varieties under investigation. Lexical borrowings are recorded in sources such as the Oxford English Dictionary allowing us to date, for instance, the first attestation in writing of roti (‘unleavened bread’) from Hindi to 1834; similar historical dictionaries are available for AusE and NZE, and have been used in comparative research on the history of regional lexical vocabulary (see Peters Reference Peters, Hoffmann and Siebers2009a). Recordings of the first locally born speakers of an emerging variety are usually not available (the ONZE project being a notable exception to the rule), and research on the evolution of the local accent therefore usually has to rely on demographic evidence or metalinguistic comments from contemporary sources like school inspectors (on the limitations of such approaches, see Gordon Reference Gordon1998). The evolution of grammatical regional usage can best be traced on the basis of historical corpora, but with the exception of a few varieties from the inner circle (i.e. American, Australian, New Zealand, and Canadian English), we still lack the resources to study grammatical change in real time from the beginning for most global Englishes. The collection of such databases, that should ideally include manuscript material and sociolinguistic background information on the people who produced the texts, is one of the major requirements for future research on the history of global varieties of English. Letters and diaries from immigrants and early newspapers from the colonies that are beginning to be digitized (but not necessarily by linguists) will provide useful resources for future studies of the global spread of English, including the role that adstrate influence may have played.Footnote ¹⁷ It is only on the basis of such corpora that researchers will be able to fill in some of the minutiae of the historical development of these varieties.

21.1 Introduction

21.1.4 Perspectives on ‘frequency’: variationist versus text-linguistic research designs

Empirical research on grammatical change in English has been carried out from two major perspectives: variationist and text-linguistic. These two perspectives approach the quantitative description of language use in fundamentally distinct ways. Simply put, variationist research studies investigate proportional preferences, while text-linguistic studies investigate the rates of occurrence in texts (see Biber Reference Biber2012: 12–17).

The variationist approach was originally developed for sociolinguistic research (see Labov Reference Labov1966, Cedergren and Sankoff Reference Cedergren and Sankoff1974; see also Chapter 1 by Romaine in this volume) but has since been extended to other applications, like research in the Probabilistic Grammar framework (e.g. Bresnan and Hay Reference Bresnan and Hay2008). The variationist method is based on analysis of the variants of a linguistic variable. To be included in the analysis, variants must be interchangeable (i.e. they are both grammatically possible and equivalent in meaning). Tokens of each variant are coded for a range of contextual factors, and then quantitative analysis (often statistical regression analysis) is used to determine the extent to which contextual and possibly language-external constraints (in Labovian parlance, ‘conditioning factors’) favour or disfavour particular variants.

In contrast, in the text-linguistic approach, the quantitative analysis describes the rates of occurrence for linguistic features in texts, typically without extensive annotation of individual occurrences. In this approach, features are not necessarily contrasted with competing features (as they are in the variationist approach). When linguistic variants are contrasted in a text-linguistic design, each variant is treated as a separate linguistic feature.Footnote ²

In terms of their research designs, the primary difference between these two analytical approaches is the unit of analysis (or the ‘observations’):

• In variationist studies, the unit of analysis is each occurrence or non-occurrence of a linguistic feature (‘variant’). Variationists are thus interested in individual linguistic choices, and their constraints.

• In text-linguistic studies, the unit of analysis is each individual text (or each sub-corpus – see below). Text linguists thus analyse linguistic use on a coarser level of granularity.

The units of analysis are the ‘observations’ that are described in a study. For the most part, each observation in a variationist study (i.e. a token of a linguistic feature) has categorical rather than continuous characteristics; the overall patterns can be quantified by counting the frequency of each category across the full set of observations. By contrast, each observation in a text-linguistic study (i.e. each text) is analysed in terms of quantitative characteristics. Variationist studies tell us the proportional preference for one variant over another, but they are typically agnostic about how often we will encounter a grammatical feature in a text. In contrast, text-linguistic studies are designed for this latter purpose.

In the present chapter, we illustrate this methodological difference through two related studies of noun phrases that express genitive relationships, where one noun modifies another noun. The first case study employs a variationist research design, while the second case study employs a text-linguistic research design. As we show in the following sections, these two approaches answer different research questions and lead to different conclusions. Taken together, they provide a more complete description of historical change than either taken on its own.

In the variationist research design, each occurrence of a genitive noun phrase is treated as an observation; the analysis is restricted to those occurrences of genitive noun phrases that are interchangeable with other structural variants (see discussion in sections 21.2 and 21.3). Each of these noun phrases is coded for several linguistic factors, such as the animacy of the modifying noun, the thematic status of the head noun and modifying noun, the length of the head noun phrase and modifying noun phrase. Then, considering the full set of all interchangeable genitive noun phrases, it is possible to determine the factors that favour one linguistic variant over another. For example, we might find that 75 per cent of all interchangeable noun phrases with an s-genitive have an animate modifying noun (e.g., the president's book), while only 20 per cent of the noun phrases with an of-genitive have an animate modifying noun (e.g. the main goal of the president). In this case, we could conclude that an animate modifying noun favours the s-genitive over the of-genitive. (It is also possible to analyse the complete set of interacting predicting factors through a logistic regression; see below.)

In contrast, each text is an observation in the text-linguistic design. In this case, the rate of occurrence is determined for each grammatical feature in each text, and it is subsequently possible to compute means and standard deviations for those rates in different registers. For example, from this perspective of-genitives occur with a mean of 30.2 per 1,000 words in a corpus of 2,005 science research articles (standard deviation = 6.0), and 36.6 per 1,000 words in a corpus of 2,005 history research articles (standard deviation = 8.1) (see Biber and Gray Reference Biber and Gray2013: 122).

These two types of research design can lead to opposite conclusions regarding which linguistic form is more ‘common’. For example, Figure 21.1 presents corpus findings regarding the use of s-genitives versus of-genitives from a variationist perspective (based on Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 302).Footnote ³ At first sight, these findings might lead one to conclude that s-genitives are more common in conversation than in academic writing: c.30 per cent of all genitives in conversation are s-genitives, while only c.5 per cent of all genitives in academic writing are s-genitives. However, because these findings are presented from a variationist perspective, they report proportional preference; they do not actually tell us how often a listener/reader will encounter these structures in texts.

Figure 21.1 Proportional use of s-genitives versus of-genitives modifying a head noun, in late twentieth-century conversation vs. academic writing

(based on Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 302)

In contrast, a text-linguistic design can be used to investigate the rates of occurrence for these different grammatical features. In this case, as Figure 21.2 shows, we would come to exactly the opposite conclusion: s-genitives have a higher rate of occurrence in academic writing (c.2.5 occurrences per 1,000 words) than in conversation (c.0.8 times per 1,000 words).

Figure 21.2 Rate of occurrence for genitive constructions in late twentieth-century conversation vs. academic writing

(based on Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 302)

The apparent contradiction between the two approaches arises because the overall use of genitive constructions (combining all s- and of-genitives) is much higher in academic writing than in conversation: only c.2.5 total genitives per 1,000 words in conversation versus c.34 total genitives per 1,000 words in academic writing. As a result, the proportion of s-genitives is higher in conversation (0.8/2.5 = c.30 per cent; see Figure 21.1), while the actual rate of occurrence for s-genitives is higher in academic writing (see Figure 21.2).

In addition to the nature of the observations (or units of analysis), a second major difference between the two research designs has to do with the population of linguistic instances included in the analysis. Variationist analyses are restricted to a sample of linguistic tokens that are interchangeable variants of the same linguistic variable (Labov's Reference Labov1966: 49 ‘principle of accountability’). The theoretical motivation is that analyses should be restricted only to those linguistic tokens where speakers are genuinely making a choice. Identifying interchangeable tokens of a linguistic variable is a major step in variationist analyses, which often results in a greatly reduced sample of linguistic tokens (see section 21.2). For example, in Hinrichs and Szmrecsanyi (Reference Hinrichs and Szmrecsanyi2007), c.64 per cent of all genitives were considered to be interchangeable; the other 36 per cent were excluded from the analysis.

A related consideration is that the variants of a linguistic variable can differ dramatically in the extent to which they are interchangeable. For example, in the variationist comparison of genitives by Hinrichs and Szmrecsanyi (Reference Hinrichs and Szmrecsanyi2007), c.80 per cent of all s-genitives in their sample were interchangeable and therefore included in the analysis, but only c.56 per cent of all of-phrases (modifying a head noun) were coded as interchangeable.

These differences in inclusion criteria can have major implications for subsequent conclusions. For example, relying on a variationist sample of interchangeable tokens for genitives, Leech et al. (Reference Leech, Hundt, Mair and Smith2009: 225; figure 10.5) show that 58 per cent of these structures were s-genitives in 1991; and based on that finding, they conclude that ‘by 1991, the s-genitive had overtaken the of-genitive in frequency’. However, it is crucially important to be aware of the methodological basis of such claims: this finding is based on the set of interchangeable tokens, which is very different from the total set of occurrences for these features. Thus, Figure 21.2 – based on all occurrences of s-genitives and of-phrases modifying a noun – shows a very different pattern, with of-phrases being much more common than s-genitives (especially in academic writing).Footnote ⁴

In the following sections, we further discuss and illustrate these methodological considerations through a case study of historical change in the use of genitive constructions. We define ‘genitives’ broadly to include any constructions that involve a noun phrase serving as modifier of a head noun. In particular, we investigate the use of three structural variants: s-genitives, of-genitives, and premodifying nouns. This three-way choice can be studied from a variationist perspective, and all three linguistic features can be investigated from a text-linguistic perspective. The following descriptions compare and contrast the kinds of historical patterns that can be discovered through each approach.

21.2 Methods

The study is based on an analysis of a sub-corpus of ARCHER (see Biber et al. Reference Biber, Finegan, Atkinson, Fries, Tottie and Schneider1994a), including all BrE texts from the registers of personal letters, newspaper reportage, and science articles. In total, the corpus used in this study comprises 327 texts and nearly 390,000 words. These three registers were chosen because they differ with respect to their primary communicative purposes, their interpersonal focus, and their intended audience. Taken together, inclusion of these registers allowed us to investigate the ways in which patterns of linguistic variation are mediated by register differences (see also Biber Reference Biber2012).

We coded all texts to identify occurrences of s-genitives, of-genitives, and premodifying nouns, and determine their interchangeability with the other two variants. For s-genitives and of-genitives, we followed the methods used in Szmrecsanyi and Hinrichs (Reference Szmrecsanyi, Hinrichs, Nevalainen, Taavitsainen, Pahta and Korhonen2008) and Wolk et al. (Reference Wolk, Bresnan, Rosenbach and Szmrecsanyi2013), and we then developed a similar set of methods for coding premodifying nouns.

The first step was to automatically identify potential cases of each of the three variants. For the genitives, we searched for of and final ^*'s/^*s’ (as well as final ^*s in seventeenth- and eighteenth-century texts, because apostrophes were often omitted from s-genitives during that period). For the premodifying nouns, we used the Biber TaggerFootnote ⁵ (see Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 35–6) to automatically identify nouns and search for instances of two adjacent nouns. We then manually coded each occurrence, to eliminate cases that were not genitives (e.g. of-phrases as part of prepositional verbs, such as think of, speak of, be composed of), and to then mark the boundaries of the two noun phrases in the remaining cases. In the following discussion, we refer to the two parts of these constructions as the ‘possessor’ and the ‘possessum’, even though most instances of genitives do not actually express the meaning of possession.

For the purposes of the variationist study, we analysed each genitive construction by hand to determine if it was ‘interchangeable’ with one or both of the other two variants. This step was based on our intuitions, deciding whether the structure was functionally equivalent and could be rephrased with another variant to express roughly the same meaning. (There has been considerable debate over the years of the extent to which grammatical variants are truly equivalent or interchangeable; see, e.g., Lavandera Reference Lavandera1978; Dines Reference Dines1980; Weiner and Labov Reference Weiner and Labov1983. Stefanowitsch (Reference Stefanowitsch, Rohdenburg and Mondorf2003) includes a critical discussion of similar issues with respect to genitive constructions.) In general, we required that the rephrasing use the same words (e.g. the county justices versus the county's justices). There were two main modifications to this rule, where we additionally allowed:

1. 1 the optional addition or deletion of a determiner to the possessum for of-genitives (e.g. the government's policy ↔ the policy of the government);

2. 2 the optional pluralization or singularization of the possessor for premodifying nouns (e.g. home prices ↔ prices of homes).

The following special cases were coded as not interchangeable:

1. 1 phrases that have been conventionalized (e.g. Murphy's law, post office);

2. 2 constructions in which an s-genitive is not followed by an explicit possessum phrase (e.g. an associate of John's);

3. 3 titles of books, films, etc. that are premodified with an s-genitive (e.g. Van Gogh's Starry Night);

4. 4 measures expressed as of-genitives (e.g. three gallons of milk);

5. 5 of-genitives where the possessor noun phrase has a post-modifier (e.g. the girlfriend of the man that I met);

6. 6 noun premodifiers that are not definite (since the possessum in s-genitives is always definite; e.g. a London college).

Beyond the guidelines enumerated here, coders were instructed to rely on their best judgement to determine interchangeability. After several rounds of trial coding and subsequent revisions to the coding scheme, two coders rated several texts in order to measure inter-coder reliability. Reliability was calculated for each of the nominal modifiers using simple percent agreement and Cohen's k.Footnote ⁶ The s-genitive (N = 84) coding achieved a simple agreement rate of 95 per cent and a ‘very good’ Cohen's k of 0.91. Reliability for of-genitives (N = 112) achieved a simple agreement of 90 per cent and a ‘very good’ Cohen's k of 0.80. Finally, the reliability analysis for premodifying nouns (N = 91) yielded a lower, yet still acceptable simple percent agreement of 85 per cent, with a ‘good’ Cohen's k of 0.69. Computer programs were developed to automatically count each of the features of interest. The text-linguistic analysis was carried out based on the normalized (per 1,000 words) rates of occurrence for each of the texts in the corpus. The variationist analysis, on the other hand, was based on calculating proportions of the raw counts.

21.3 The variationist analysis of genitive constructions

As described in preceding sections, the first step in the variationist analysis was to consider each linguistic token, to determine if it was interchangeable with one or both of the other variants. One methodological disadvantage of considering three variants is immediately apparent in such an analysis: there are many more alternatives to consider than in a study of a simple dichotomous choice. For example, s-genitives are analysed to determine whether they are interchangeable with of-genitives, interchangeable with nouns as nominal premodifiers, interchangeable with both of-genitives and nouns as nominal premodifiers, or not interchangeable at all. Similarly, of-genitives and nouns as nominal premodifiers are all coded to identify instances that are interchangeable with one, both, or neither of the other two variants.

Tables 21.1–21.3 present the results of this coding. Table 21.1 shows the extent to which s-genitives are interchangeable with the other two variants. Most s-genitives are interchangeable with of-genitives, across registers and across historical periods. In contrast, few s-genitives are interchangeable with premodifying nouns: as low as 5–10 per cent in letters and newspaper articles, and c.20 per cent in science prose. Nearly all s-genitives that are interchangeable with premodifying nouns are also interchangeable with of-genitives (as shown by the last column of Table 21.1).

Of-genitives are more consistently interchangeable with both of the two other variants, but they show greater differences across registers and periods (see Table 21.2). In letters, c.30–35 per cent of the occurrences of of-genitives are interchangeable with s-genitives, and c.25 per cent are interchangeable with premodifying nouns. Those patterns hold across periods. A higher proportion of of-genitives are interchangeable with s-genitives in newspaper articles: c.50 per cent across periods. But newspaper articles are similar to letters in that only c.25 per cent of of-genitives are interchangeable with premodifying nouns. Science writing is interesting in that it shows an apparent historical increase in the proportion of of-genitives that are interchangeable: from c.30 per cent in the eighteenth century to c.50 per cent in the twentieth century. This same pattern holds for both interchangeability with s-genitives and with premodifying nouns. Finally, the last column in Table 21.2 indicates that interchangeability with s-genitives versus interchangeability with premodifying nouns are relatively independent, since the figures for three-way interchangeability are considerably lower than either of the other two columns.

The patterns of interchangeability for nouns as premodifiers (Table 21.3) are similar to those for s-genitives: relatively few nouns as premodifiers are interchangeable with s-genitives (across registers and periods), but c.50 per cent of nouns as premodifiers are interchangeable with of-genitives (with lower proportions in newspaper articles). And here again, we see that if a premodifying noun is interchangeable with an s-genitive, that token will usually also be interchangeable with an of-genitive (as shown by the last column of Table 21.3).

Two general patterns are noteworthy here as background to the interpretation of variationist findings:

1. 1 In general, many tokens of genitive constructions are not interchangeable. In fact, only 10–50 per cent of occurrences are interchangeable for many of the comparisons. The one exception here is for s-genitives, which are usually interchangeable with of-genitives (Table 21.1). But otherwise, fewer than 50 per cent of the occurrences of these constructions are interchangeable with other variants.

2. 2 The extent of interchangeability varies considerably across constructions, across registers, and to some extent, across periods.

While identifying interchangeable tokens is one of the first steps in a variationist perspective, there is usually little consideration of the extent of interchangeability. That is, the analysis is focused on the linguistic variable, operationally defined as the set of interchangeable occurrences. As a result, the extent to which that set of variants represents the total pool of linguistic occurrences has generally been disregarded as theoretically irrelevant. However, a complete historical description of a structural domain must also account for the patterns of variation and change for the non-interchangeable occurrences. We briefly return to this point below and then again in the conclusion.

In the remainder of the present section, though, we adopt the variationist perspective, considering the patterns of variation within the set of interchangeable occurrences of genitive constructions. Figure 21.3 presents our findings for the alternation that has been the focus for most previous work on genitives: the choice between s-genitives versus of-genitives in constructions where the two are interchangeable.Footnote ⁷ The patterns shown in this figure provide some support for earlier claims that the s-genitive has been increasing historically at the expense of the of-genitive (see, e.g., Potter Reference Potter1969; Leech et al. Reference Leech, Hundt, Mair and Smith2009). However, this figure also shows that this historical trend is mediated by register differences. Thus, in personal letters, the s-genitive increased proportionally in use during the nineteenth century, but that pattern then remained relatively stable over the course of the twentieth century. In newspaper reportage, of-genitives became even more strongly preferred in the nineteenth century, followed by a strong shift towards s-genitives during the twentieth century. As a result, newspaper writing and personal letters are relatively similar in showing c.30–35 per cent proportional use of s-genitives in the latter part of the twentieth century. However, the historical trend in science prose contrasts with both letters and newspaper reportage: a small reliance on s-genitives in the eighteenth century, followed by a steady decline in proportional use over the following two centuries. As a result, only c.5 per cent of interchangeable genitive constructions are realized as s-genitives in twentieth-century science prose.

Figure 21.3 Historical change in the proportional use of the s-variant (vs. the of-variant)

The historical trends are less consistent in Figure 21.4, which plots the proportional use of interchangeable s-genitives versus nouns as premodifiers. This is due in part to the fact that these two construction types are in general not interchangeable, and thus the proportions shown in Figure 21.4 are based on very small samples. For example, the nineteenth-century proportion of 45 per cent s-genitives in science prose is based on a sample of only forty tokens (eighteen s-genitives that are interchangeable with noun-premodifiers – see Table 21.1, and twenty-two noun-premodifiers that are interchangeable with s-genitives – see Table 21.3). Despite the fluctuations, the overall historical trends are consistent across registers, with a notable increase in the proportional use of noun-premodifiers (and decline in the proportional use of s-genitives) across the centuries. Science prose shows the strongest increase, with noun-premodifiers being used over 95 per cent of the time in interchangeable constructions from the twentieth century.

Figure 21.4 Historical change in the proportional use of the s-variant (vs. the premodifying noun variant)

The sample of interchangeable occurrences for the of-genitive versus noun-premodifier alternation is much larger (see Tables 21.2 and 21.3), and correspondingly, the historical trends shown in Figure 21.5 are much more consistent across centuries. For all three registers, there is a strong increase in the proportional use of noun-premodifiers at the expense of of-genitives. Letters and newspaper reportage take the lead in this change during the nineteenth century, while science prose shifted strongly towards increased noun-premodifier variants during the twentieth century (see also Biber and Gray Reference Biber and Gray2011a, 2013; Berlage Reference Berlage2014).

Figure 21.5 Historical change in the proportional use of the premodifying noun variant (vs. the of-variant)

Finally, Figure 21.6 plots the proportional use of variants for those occurrences of genitives that can take all three variants. Similar to Figure 21.4, Figure 21.6 is based on small sample sizes for many data points (see the right columns in Tables 21.1–21.3), but one trend stands out from Figure 21.6: the historical increase in the preference for noun-premodifiers in cases where all three variants are possible. This increase is most pronounced in letters and newspaper reportage, but the same trend occurs to a lesser extent in science prose.