Published in a slightly different format in The European English Messenger 9:9-15 (2000)




Salikoko S. Mufwene

University of Chicago


İİİİİ This essay is a combination of provocative observations on the evolution of English. Questioning some established positions and working assumptions, I raise some issues on the role of contact in language evolution and on the nature of language transmission. These factors are presented as critical to understanding how English started and has speciated into so many varieties today. The discussion is an appetizer to my book The ecology of language evolution (to appear in 2001 at Cambridge University Press) which spells out the contribution that research on the development of creoles can make to genetic linguistics, thus also to scholarship on the history of the English language (HEL). What follows is a reconstruction of a hitherto unwritten plenary presentation at the MAVEN II Conference at the University of Lincoln and Humberside, England, in September 1999. Although space limitations make it difficult to use the illustrations that complemented the oral presentation, every effort has been made to keep this written version on the manifold disputation as straightforward as possible.

İİİİİ Today the word English applied to language can no longer be defined without qualifications as ëthe language of the English peopleíócontrary to how one can identify Japanese, Somali, or Kinyarwanda in relation to the specific peoples that are their primary speakers and bear the same names. Such a characterization would be historical, for instance, ëthe language that over one thousand years ago was the vernacular spoken primarily by the English peopleí.[1]

İİİİİ An alternative diachronic characterization is: ëname associated with language varieties that can be traced genetically to Old Englishí. An important reason for considering it is the following: since the moment the construct English was associated with England and its inhabitants, the English, the language has been appropriated by various other populations. Some of these inhabit territories so many thousands of miles away from its birth place that we cannot be surprised why some of the member varieties have been disfranchised with labels such as pidgin English, English Creole, new English, and indigenized/nativized English.[2]

İİİİİ Most of the above terms reflect a social bias that is inherent in genetic linguistics. They suggest that pidgin, creole, and indigenized/nativized varieties are by-products of language contact, whereas other varieties such as spoken in England and by Whites in North America are not. This assumption is at best disputable. As I focus on contact below,İ let us now consider the label new English. It most saliently subsumes indigenized Englishes, though sometimes it also includes other non-pidgin/creole varieties spoken outside the United Kingdom, if not outside of England (and Scotland) alone. Whichever way the label applies, it is inaccurate. There are no 17th nor 18th-century English varieties still spoken today in England or the United Kingdom.

İİİİİ To be sure, there are some linguists who like to oppose ìold(er)î languages to ìyoung(erî ones. Paradoxically, the most common examples in the first category include English and French (since classical languages such as Latin and Old Norse are presumably dead) and those in the second category include creoles. As made more obvious in the following paragraphs, advocates of this distinction ignore the fact that English creoles are as legitimate evolutions from colonial forms of English as other varieties which developed in the same territories and during the same time period but have been kept within the English franchise. They also fail to consider some important unresolved issues to which I return below.

İİİİİ Since every spoken language is adapted by its speakers to current communicative needs and contexts, there cannot be any ìold(er)î languages. Such varieties would be maladaptive óthe way Latin would if it had not developed into the Romance languages. So, the term new English should apply to all varieties identifiable as English todayóin the same way that Romance languages can be claimed to be new forms of the Latin vernaculars spread by Roman soldiers. Conversely, to the extent that English pidgins and creoles, as well as indigenized Englishes, can ultimately be traced back to Old English, they all have a long history.

İİİİİ Genetic linguists should re-examine their reasons for disfranchising especially pidgins and creoles as not genetically related to their lexifiers. Their typical arguments are that these varieties are contact-based and not mutually intelligible with presumably the more legitimate offspring that, say, colonial English vernaculars are. Letís think about it. Doesnít modern linguistics also profess that dialects of the same language need not be mutually intelligible? Isnít Cockney unintelligible to speakers of some other English dialects? Is New Zealand vernacular English more intelligible to those who are not familiar with it than Jamaican Creole is? (Yes, familiarity is with any variety or language is critical to intelligibility!) So why do genetic linguists capitalize on mutual intelligibility?

İİİİİ Regarding contact-based evolution, how accurate is the history of English sketched above? Does it really start with a full-fledged Old English in England by, say, the 7th century? Did the Germanic invaders of England in the 5th century speak Old English? Did the Jutes, Angles, Saxons, and, later, Frisians speak mutually intelligible Germanic languages or dialects? Even if these were mutually intelligible (which is doubtful if we consider, for instance, Dutch and German today), were they typologically identical? Letís assume, like Crystal (1995), that the Germanic settlers drove the Native Celts to the frontiers of their colonies (Northumbria, East Anglia, Mercia, Essex, Sussex, Wessex, and Kent)óin a way similar to what Europeans did to Native Americans when they colonized the New World. Isnít Old English likely to have originated in the contacts that the newcomers had among themselves across tribal and/or dialectal lines? History also suggests that the Germanic invaders settled Englandİ according to their origins, so that East Anglia was primarily a colony of the Angles and Essex, Sussex, and Wessex were colonized primarily by the Saxons, etc. The different Old English communities that developed then seem to reflect interaction patterns that correspond to these settlement trends, making allowance also for contacts with the Native Celts, however minimal their influences may have been. These considerations should account for the dialectal variation in Old English noted by, for instance, Hogg (1992) and Milroy (1992).

İİİİİ Recent history may shed light on what happened in that distant past. Kurathís (1928) position that dialectal regions in North America reflect immigration patterns of the early colonists has found support in socio-economic histories such as Bailyn (1986) and Fischer (1989). Even indentured servants had a choice over the particular parts of the American colonies where they wanted to immigrate. Variable as they were from colony to colony, new patterns of population and dialect contacts produced new, colonial dialects. By the Founder Principle, structures of todayís dialects were largely predetermined by what was spoken by the founder populations and what emerged from their interactions as they accommodated each other. Of course, stochastic events,[3] such as massive immigrations from other parts of the British Isles or from continental Europe, bore on those initial evolutionary trends and modified some of them, producing todayís American English varieties.

İİİİİ If the Germanic invaders of England were as normal post-Neanderthal humans as the English colonists in North America, Australia, and New Zealand, something similar to the latterís settlement experience must have happened. The birth of Old English is an England-based linguistic development, a colonial product like American, Australian, and New Zealand Englishes. According to the OED, the name England is about as old as the name English. With their earliest attestations situated in the late 9th century, they are etymologically related to Old English terms for the Angles, the former used for their land and the latter for their language.[4] One may conjecture that after the term England was extended to the wider territory of which the older Anglian colony is only a part and Old English as the new colonial language was supplanting the Germanic vernaculars brought from the European continent, English could be identified as the vernacular of the new, wider England. In this respect, the development of Old English is reminiscent of that of creole vernaculars and their other colonial kin, such as American and Australian Englishes, which are all local evolutions. Etymologically, creole vernaculars are varieties typical of certain settlement colonies, like some plants and animals that were/are also identified with the same adjective, e.g., creole cow. Regarding Old English, one important question that remains open is: What role did the Celts play in its development?

İİİİİ The above observations seem so obvious to those of us not doing HEL. It is, however, not clear why experts have generally not considered contact an unmarked factor in the evolution of English in England and among Whites in, say, North America. Exceptions to this observation apply to areas such as the lexicon, where external influence is too obvious to ignore. The high percentage of French lexical items in the English vocabulary actually makes it difficult to deny the role of French in the actuation of the Great Vowel Shift. That there is no comparable influence from Native American languages on American English can be explained by differences between the kinds of interactions the Norman invaders had with the then English populations and those which the English colonizers had with the Native Americansóa fact on which I need not elaborate here.

İİİİİ What is shocking is that Thomason & Kaufman (1988), for instance, dismiss the possible influence of Old Norse and Norman French on some of the grammatical changes that have affected English. They argue that the same changes would have happened even if English had not come in contact with these other languages. One of the issues may be whether external influence should be limited to only those cases where structural features external to a language are adopted. Can it not be extended to those where xenolectal characteristics favor some native variants at the expense of others? I favor the extended meaning of external influence, assuming that every specific ecology of language use determines its local evolutionary trajectory. (I return to this dimension of language evolution below.)

İİİİİ To be fair to them, Thomason & Kaufman (1988) are influenced by the then seemingly sound position in creolistics that creolesí grammatical features are largely contact-induced innovations. This assumption also leads them to stipulate that creoles are not genetically related to their lexifiers, a position that I also dispute. It was based on several mistaken assumptions, some of which make genetic linguistics itself circular. For instance, it was assumed that restructuring qua system-reorganization occurred mostly in the development of pidgins and creoles but not in other cases, or at least not that extensively. However, if restructuring did not happen in other cases, there would be no new dialects of English. If it is a question of degree of restructuring, then where does one draw the line, even if this must remain a fuzzy boundary? In the case of North America, note that since the early stages of its colonization, Englishİ among Whites came in contact with other European languages, whose speakers were no more apt to acquire English than the African slaves were. Documentation on runaway slaves also suggests that many African slaves who lived and interacted regularly with Europeans spoke the same colonial English that their masters or white indentured homologs did. This observation questions the argument that creoles developed in part because there were too few European models to transmit the lexifier faithfully. I show below that the argument is flawed.

İİİİİ The association of transmission of the lexifier with Europeans led to another mistaken assumption, viz., break in the transmission of the lexifier. If there really was a break in transmission, nothing would have been passed onóperiod. To suggest that one needed Europeans to pass the lexifier on to African slaves in the New World betrays ignorance of the history of the colonies. In the early stages, on the small farms that preceded the large plantations (which never developed overnight), there were no African majorities, and the Africansóespecially their childrenóspoke like the Whites with whom they lived fairly intimately. By the time the plantations developed and segregation was instituted, transmission of the English language did not depend on the proportion of Whites any more than it does today in Africa or Asia, where English is propagated by non-Whites. No break in transmission could be posited in such cases.

İİİİİ Besides, to insist so much on the need for more Europeans to transmit the lexifier is overlooking the fact that several European indentured servants didnít speak the lexifier natively and contributed their share to the restructuring of the lexifier among the European colonists and their descendants. In English colonies, both among Europeans and among non-Europeans, it simply was restructured colonial nonstandard English, identified sometimes as a koinÈ, that was transmitted from one group to another and from generation to generation. In every group, with every transmission, the lexifier was being restructured. The difference lies in degrees of deviation from the original and the sources of external influences. It just doesnít make much sense to suggest that in those settings external influence on colonial English was relevant only among Africans and their descendants, but not among Europeans and their descendants.

İİİİİ The explanation for differences among the varieties spoken by descendants of Europeans and non-Europeans lies not in the absence of contact for Europeans and the role of contact for non-Europeans, but rather in how the ecology of the contacts among the colonists (free, indentured, and enslaved) varied. Ecology includes, inter alia, the kind of English to which those appropriating it were exposed, the kinds of languages that came in contact with English (how different they were structurally from it), the patterns of interaction between speakers of the target language and those appropriating it, whether or not segregation was instituted in the colonial population and at what stage, etc.

İİİİİ An important issue has been what role languages other than the target played in the restructuring that took place. Research by, for instance, Shana Poplack and Sali Tagliamonte over the past 15 years has shown that overall the features typically associated with African-American vernacular English (AAVE) can be traced back to colonial English; they are not particularly African in origin. While I basically agree with them, one must also consider the following: because those features are not uniformly, nor universally, used in all English dialects (despite the strong parallelisms that Poplackís and Tagliamonteís studies reveal in the strengths of conditioning factors of the variants), the ecologies of contact with African languages favored the peculiarities that distinguish AAVE from other English vernaculars. It does not matter at all that most of these peculiarities are statistical. Basically the same explanation applies also to English creoles, though restructuring was obviously more extensive in this case. The origins of the features are undoubtedly English, given the fact that the Africans were shifting to, and targeted, it. However, the restructuring that produced the new vernaculars could not be independent of the influence of substrate languages. The indirect evidence for this position lies in all cases in which a language has been appropriated by a foreign group.

İİİİİ The latter observations find support in the development of indigenized Englishes. Despite the guided mode of transmissionóthrough the scholastic mediumólocal ecologies have favored particular features, including downright structural importations from the substrate languages. The endogenous ecologies of indigenized Englishes, with often more structural homogeneity among the indigenous languages, were certainly more favorable to substrate influence.[5]

İİİİİ Can such an ecological explanation not be extended to the evolution of English since Old English, factoring in influences from the Celtic languages, Old Norse, and Norman French? To answer this question, we must compare varieties of Old English and these other languages, so that we can determine whether in the relevant respects they did not indeed influence the selections that were made in favor of what we now identify as changes. One has to factor in the proportion of indigenous to non-indigenous populations and the patterns of interaction between them as groups. Recall also that changes are not necessarily replacive. They can be additive (introducing new units or rules to the current system) or subtractive (involving, for instance, loss of variants). But they can also involve weakening or strengthening of current variants. The latter two are probably the most significant ones when a particular language is clearly targeted by an expatriate group.

İİİİİ To be sure, the endogenous contacts with the Scottish and Irish Celts which produced Scots and Irish Englishes were ecologically not of the same kind as the contacts with the expatriate Norse and Normans. However, as Kroch et al. (in press) show, this does not prove that there was no xenolectal influence on the evolution of English in England. It is indeed interesting that in these particular cases it is the powerful invaders who lost their languages. However, it seems misguided to dismiss contact as a factor providing an ecological explanation to the evolution of the language appropriated by such numerically important and prestigious groups of immigrants. The Norse and Normans cannot have been more perfect learners of a foreign vernacular than modern people, especially in settings where they were not fully integrated, since they interacted most directly with their servants and administrative auxiliaries but remained generally removed from the masses of the English population.[6]

İİİİİ What does all this lead to? The evolution of English has so far been ecology-specific, depending on the natures of the specific idiolects and dialects that were spoken in a setting, how cross-idiolectal or dialectal differences were negotiated, whether or not it was also being appropriated by speakers of other languages, what influence these languages exerted on it, and overall how it adapted to new communicative needs of its speakers. We should remember that a language is transmitted piecemealónot wholesaleóand that hybridization is fundamental to language transmission, starting with the mutual accommodations native speakers make to each otherís idiolects. In settings involving large proportions of non-native speakersóespecially if the latter are not fully integrated with the native speakersóthey too contribute to the pool of structural features and pragmatic constraints from which are selected the set of features that become characteristic of future generations of speakers. The essence of this is that all language evolution is contact-induced, except that contact is situated at the level of idiolects.

İİİİİ To understand all this, one must start with the premise that, like biological species (not organisms!), languages exist only as extrapolations from idiolectsóthe real counterparts of biological organisms. Some of these differ only minimally from each other but others more extensivelyófor instance, idiolects of speakers of mutually unintelligible dialects.İ In the latter case, the differences may be as significant as in the coexistence of native and non-native idiolects. Quite significant in all this is also the fact that languages change not because speakers want to change them but rather through exaptations that take place in the communicative acts of their speakers, as these accommodate each other, fail to meet some target sounds, forms, or constructions, or adapt forms and constructions to new communicative demands. All of these are processes that have been associated with language contact. The innovative part of the argument lies in situating contact at the level of idiolects, making this part of every day linguistic behavior, making individual speakers the primary unwitting agents of change, and embedding language evolution in ethnographic ecologies situated in specific socio-economic historical contexts. Thus, languages do not change on their own nor in abstraction, and the same is true of English.

İİİİİ Consequently, the distinction between internally and externally-motivated change becomes purely sociological, having to do with determining whether a change occurred due the influence from another language or simply due to the dynamics of the interactions of native idiolects. From a structural point of view, the distinction is useless, since the only contact that matters is that of idiolects, regardless of whether or not they contain xenolectal features. A fundamental question here is why idiolects are different. A simple answer to the question is: because, cases of Siamese twins set aside, no two idiolects develop in identical ecologies. Every speaker evolves in a network of communication that only overlaps with those of their kin and friends. Under the conception of ëidiolectí assumed above, there are no restructuring processes that in kind are specific to changes induced by contacts of languages but cannot be induced by contacts of native varieties. This all suggests that it is difficult to imagine how English could have evolved in England, as among Whites in North America, without the causation of contact.

İİİİİ Evolutionary processes are thus speaker-based phenomena which spread within a population through the accommodating acts of speakers. Linguistic features and changes spread like viruses, depending very much on contacts of interacting individuals, allowing the coexistence of competing influences from different sources in the same speaker. Since changes are not necessarily replacive, contact accounts for intra-speaker variation in linguistic behavior. Every speakerís feature pool is an arena of competition and selection that bears on the evolutionary trajectory of a language qua species. Variation in social interaction patterns accounts for the regional and social speciation of English into varieties such as British versus American Englishes, or Southwestern British versus Scottish English, or African-American versus White American varieties of English. Some evolutions are strictly contained in specific social classes. For instance, most of the English said to have been heavily influenced by French vocabulary is educated, middle-class English, not folk, nonstandard speech. In North America, the English said to have been influenced by African linguistic features is African-American English, because African languages also contributed to the new feature pool from which features of African-American English would be selected.

İİİİİ A re-examination of the HEL based on population contacts suggests that Old English must have developed through more or less the same restructuring processes that have yielded its more recent descendants. It helps to start from the competitions and selections that took place among the Germanic idiolectal features brought over from continental Europe in the 5th century (with or without the contribution of Celtic features), and then to factor in layers of the contributions of other languages to the ever-changing feature pools, as English was appropriated by more and more non-English populations. Different selections have produced different varieties, all of them new.

İİİİİ Although the approach to the HEL sketched here has been very much inspired by research on the development of creoles, it does not at all mean that we must revive the hypothesis that Middle English developed by creolization. There is no such global restructuring process that must be identified as creolization. The term is only a disfranchising label that has failed to recognize in the development of creoles processes of the same kind as have taken place in all cases of language evolution. The validity of the distinction between creole and non-creole languages is sociohistorical, not structural (Mufwene 2000). From the point of view of language evolution the distinction can be ignored. No specific terms other than change, evolution, drift, speciation/diversification, and the like already in usage in genetic linguistics need be coined, except for structural processes hitherto unidentified. Likewise, indigenized English is a sociohistorically-grounded term, alluding to the scholastic kind of English that was transmitted to non-native speakersóI dare observe, more or less like a dead language to be revived by non-native speakers for new communicative functions. Indigenization does not denote a structural process and the development of indigenized Englishes is as natural as those which marked the evolution of English in England from Old English to todayís late Modern English, bearing in mind that the ecology of each restructuring process is different in some ways.

İİİİİ In conclusion, we should take advantage of questions addressed in research areas hitherto subsumed by language contact and reopen the books on some established positions and unanswered questions on the evolution of English and other languages. We cannot take it for granted that the working assumptions of older scholarship are all justified. I have just scraped the tip of the iceberg.


Bailyn, Bernard. 1986. The peopling of British North America: An introduction. New York: Random House.

Crystal, David. 1995. The Cambridge encyclopedia of the English language. Cambridge: Cambridge University Press.

Fischer, David Hackett. 1989. Albionís seed: Four British folkways in America. New York and Oxford: Oxford University Press.

Hogg, Richard M., ed. 1992. The Cambridge history of the English language. Vol. 1: The beginnings to 1066. Cambridge: Cambridge University Press.

Kroch, Anthony, Ann Taylor, & Donald Ringe. in press. The Middle-English verb-second constraint: A case study in language contact and language change. In Textual parameters in older language, ed. by Susan Herring et al. Amsterdam: John Benjamins.

Kurath, Hans. 1928. The origin of dialectal differences in spoken American English. Modern Philology 25.385-95.

Milroy, James. 1992. Linguistic variation and change: On the historical sociolinguistics of English. Oxford: Blackwell.

Mufwene, Salikoko S. 2000. Creolization is a social, not a structural, process. In Degrees of restructuring in creole languages, ed. by Edgar Schneider and Ingrid Neumann-Holzschuh, 65-84. Amsterdam: John Benjamins.

Poplack, Shana & Sali Tagliamonte. 2001. African American English in the diaspora: tense and aspect. Oxford: Blackwell.

Thomason, Sarah G. & Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press.


[1] This suggests accurately that the Celts, who then were the only Natives of todayís England, gradually shifted to English as their vernacular during their colonization by Germanic populations. This shift is reminiscent of the case of Native Americans in the United States and Canada, who have been abandoning their ancestral languages in favor of the ìsameî imported vernacular over the past couple of centuries.

[2] As has often been observed, the vast majority ofİ (native) English speakers live outside England. Statistically, speakers of British English are a minority compared to those of American English alone, bearing in mind that such national names are oversimplified constructs. I have included English creole among the labels for two reasons: 1) speakers of such creole vernaculars claim they speak (nonstandard) English; and 2) as noted by Crystal (1995), statistics of speakers of major international languages include pidgin and creole speakers. On account of the arguments presented below, one can certainly question why linguists disfranchise them as separate languages.

[3] For the sake of efficiency and to encourage more reflections, I use some technical terms here, some of them taken from other disciplines. A stochastic event in chaos theory is one that changes the trajectory of a process by redirecting it in another direction. A lexifier is the language from which a pidgin or creole has inherited most of its vocabulary. Assuming that contact has something to do with the emergence of any ìnewî language (variety), then every language (variety) has a lexifier. In biology, the term exaptation is used for unplanned or spontaneous adaptations that typically have nothing to do with the original function of an organ, such as the use of the tongue or epiglottis for speech. Many adaptations that constitute linguistic changes are accidental developments of the same kind.

[4] This is reminiscent of how terms such as Uganda and Luganda are etymologically related, the former for the land and the latter for the language of the Ganda people.

[5] Recall that most creoles are exogenous evolutions, an environmental factor that may have reduced, or simply made less obvious, the significance of straightforward substrate influence.

[6] Better studies may prove that this observation about social interaction should be qualifiedóthe regular soldiers undoubtedly mingled with the regular Native populations.