3. The relationship between synchronic and diachronic

3.1 Synchronic versus diachronic

The existence of linguistic change is somewhat difficult to assimilate if the intention is, as Labov states (1994), to arrive at a general theory of language; it becomes difficult to accept - even in the context of the focus on language that the theory of variation has, that is to say, as an instrument of communication used by a speech community based, as with all other theories of language in existence, on associations between arbitrary forms and their signifiés. It was accordingly based on the Saussurian concept of opposition and distinctive differences.

Linguistic change is an awkward fact, it disturbs the form/meaning relationship such that the speakers who are affected by the change no longer signal the meaning in the same way as the speakers who have yet to be affected by the change – that is: the old people in the community, or speakers of the same age belonging to other communities. And this circumstance brings with it linguistic instability.

Nonetheless, the instability of linguistic systems is not the only difficulty facing linguists, especially variationists, since the nature of linguistic change also poses substantial methodological difficulties. As Labov says, "if linguistic change were a constant linguistic factor it would be easy to analyse"; however, linguistic change is sporadic, it is disseminated rapidly through the several different component parts of the linguistic structure, until it loses strength or is distorted and is not recognised for more than a century or two. It may come to a halt so suddenly that the rules on a given linguistic phenomenon, which seemed normal and inevitable, become inconceivable and unnatural in the course of a decade, and may then disappear for millennia, fostering the idea of stability.

Most of the topics treated by historical linguistics have to do with phenomena rather than principles. The existence of disagreements on the question of linguistic phenomena is known as the evidence paradox. And the procedures employed by historical linguistics to overcome this paradox were firstly a) to re-examine the internal evidence, or b) bring in external evidence from other fields: history of settlement, literature, demography.

The occurrence of unexpected phenomena is an indication that there are phenomena searching for a principle, and so are in an intermediate state between phenomena and principles. For example, according to Labov (1994: 15), many phenomena studied by historical linguistics violate principles. This is so in the case of the convergence, observed during the 18th century, of the English diphthong /ay/ in the word "vice" and the diphthong /oy/ in the word "voice", which came to be pronounced identically and which then separated again in the 19th and 20th centuries. This is a violation of Garde's principle, Garde (1961), according to which convergence phenomena are linguistically irreversible.

On the other hand, the practice and procedures of historical linguistics are predicated on the idea that phonological change is regular. In contrast, the data used by dialect geography lend support to the opposite idea, that is, almost every word has its own story. In fact, the lexical isoglosses of dialect geography do not coincide with the predictions of regularity.

In view of these considerations, it became clear that overcoming the contradictions and paradoxes of historical linguistics could not be based simply on the reanalysis of linguistic phenomena known and studied by other theories, but rather would specifically need to be based on the processing of a different type of data, and it is this proposal which informs the New Paradigm. Such data would reflect and embody changes going forward now, and throw light on the linguistic past, and on features connected with unfinished business of historical linguistics. 

3.2 Diachronic versus synchronic

The article by Labov (1989), "The child as linguistic historian", which in part reproduces the research by Houston (1986), situates us in the opposite context: that is to say, the utilisation of historical explanations to understand the present, specifically in the form of the synchronic behaviour of the English variable (ING),. The latter is a morpheme studied from a purely synchronic point of view in a number of different speech communities (Labov 1966; Trudgill 1974).

The synchronic variable rule formulated, implies that the variable (ING) - - may be pronounced /in/ in unaccentuated syllables/:

The findings of synchronic research carried out during the eighties showed conclusively that the /in/ variant is produced most frequently in progressive verbs, less frequently in adjectives, and seldom or never in gerunds and nouns.

The diachronic explanation of the research carried out by Houston and reviewed by Labov can be summed up in the following way: /in/ derives from the old English ending for the participle –inde and was the result of simplification of the /-nd-/ cluster in unaccentuated syllables; // was arguably the reflex of the verb noun ending spelt -inge or -ynge. Thus, this study showed an opposition in Old English had become a social and stylistic variable – in other words a marker.

Furthermore, Labov (1989) shows that some linguistic variables, which are not synchronically motivated, show historical continuity with few changes over long periods of time, indicating that the separation into synchronic and diachronic linguistics is no longer viable, since children are perfect linguistic historians. Thus, findings on the English variable (ING)–and they are not alone in this –contradict the principle stating that historical linguistics is irrelevant to the study of synchronic linguistics.

4. Internal and external variation

The separation into internal and external (linguistic) factors and external (social, stylistic) factors which constrain the occurrence of variable linguistic phenomena, seems less than practical for linguists who think, as I do, that language is a whole where "tout se tient". In fact, these two forms of variation cannot be completely separated:

a) When it is looked at as internal, the social distribution of the linguistic variable is also considered. The most important source of data becomes the spontaneous production of members of specific speech communities, and since such data are identified or characterised by participants, time and place, they do not lose relevance for the community where they were collected. They are data which represent the processes of real change and variation produced in a community, and as such they are "sociolinguistic". In fact, most variation studies have a sociolinguistic basis. Nevertheless, either because there are variables that are constrained only by internal factors, or because certain research contexts entail a purely linguistic approach, it is sometimes only possible to refer to linguistic systems.

b) In contrast, consideration of internal and external factors together brings us to the field of micro-sociolinguistics, or sociolinguistic variation, strictly defined, that is to say, that which has as its object the study of language, which analyses the social construction of the speech and speech variety of a given community, and which establishes who are the innovators of the linguistic variation and change based on six independent social variables considered most relevant: sex, age, social class, ethnicity, race and size of community.

The analysis of these social factors in turn brings us to consideration of the status of the linguistic variables and speaker variables within the community and the relationship the latter has with other communities, together with patterns of communication and the homogenising and intensifying effects of social networks. Additionally, the route taken by the transmission of variable elements from generation to generation across different historical periods has to be considered.

According to Labov, what we are discussing here has to do with the notion of integration, a key concept in understanding the New Paradigm. Integration here is meant in two senses: firstly, there is integration of a particular linguistic form and structure into the structural matrix alongside other linguistic forms – given that linguistic variation and change are constrained, redirected and accelerated through their relationship with other linguistic forms in the system. Secondly, there is integration into the structure of a speech community, where to understand the causes of variation and change, we need to know where they were first produced within the social structure, which groups lead the innovation, how it has extended to other groups, and lastly which have been the most resistant groups. (See Turell 1995a and 1995b on a number of Catalan-speaking communities). Consequently, the integration of an explanation of the variation and change into a larger structure brings in consideration of multiple causation; and this is the meaning of the term multivariant analysis for the variable treatment of linguistic phenomena.

Variationist inquiry carried out over the last twenty years has made it possible to establish a whole series of principles related to the nature of internal and external factors responsible for the behaviour of linguistic variables under study. These principles can be resumed as follows:

a) Internal factors are independent from external factors; if an internal factor is removed or undergoes change, the change in behaviour may be reflected in the other internal factors, but external factors do not change; if an external factor is removed or undergoes change, the other external factors change, but internal factors remain the same.

b) Internal factors are independent among themselves, while external factors are interactive (See Turell 1995b for a definition of interaction).

As the study of variation and linguistic change has advanced, other questions have emerged, such as the cognitive consequences of linguistic change, the evaluation made of the variation and change by individuals in the community, and the status of the variable rules in synchronic grammar. These questions have been posed, based on the study of comprehension across dialects as well as on observations on the acquisition by young children of the patterns of variation in their community. Similarly, support has come from longitudinal studies of the same consultants over time at a series of different periods. Which of course brings us directly up against the notion of real time and the study of syntactic change over long periods of time, and for example the study of progressive syntactic change observed in creole languages that are developing.

5. Change in apparent time and change in real time

5.1 Change in apparent time

The simplest way to study linguistic change is to study it in apparent time, based on the analysis of the distribution of linguistic variables across different age groups. This distribution across age groups should not be confused with the regular linguistic behaviour of age grading, repeated in every generation, which has to do rather with differences resulting from the language development found in all individuals.

One of the methodological difficulties most frequently met with in studies of change in apparent time, is the issue of selecting the age of the consultants that will best serve to obtain samples of spontaneous speech of sufficient quality. This methodological difficulty has to do with a series of factors: the attention paid to speech and the lack of a stable language variety in the case of young people, and the possible physical deterioration (loss of teeth, voice, lax articulation) or possible mental deterioration ( loss of memory, interest and attention) in the case of the very elderly.

These methodological questions are important since most studies show that adolescents particularly, between 11 and 18 years, but also pre-adolescents aged between 8 and 10, are the leaders of ongoing language change. At other times, for relevant theoretical reasons, such as the effect of a specific of the social factor in the characterisation of linguistic change, it may be decided that the setting up of age groups cannot done according to biological –chronological age, but rather has to be based on the external social factors in question. This was the case with the study of linguistic change in the Ribagorçà (on the northern Catalonia-Aragon border) (Alturo and Turell (1990), Alturo (1995)). Here, the age groups were determined by the relationship of the speakers with the social upheavals undergone by members of the speech community in the village of El Pont de Suert, between 1930 and 1970. For example, there were those who were adolescents during the Spanish Civil War, aged around 70 at the time the data were collected, secondly those who were adolescents between 1940 and 1950, who experienced the industrial development of the village and who were aged around 50 and 60 years at the time of data collection; and lastly those who were adolescents of the sixties, between 1960 and 1969, who saw stabilisation come to the village, aged between 30 and 40 years old at the time of data collection.

5.2 Change in real time versus change in apparent time

The distribution and occurrence of a given variable by age groups (involving research into linguistic change in apparent time) does not indicate definitively that such a linguistic change really is under way in the speech community in question. Instead it may represent a characteristic age grading pattern, a pattern which develops during the lifetime of individuals and which repeats itself generation after generation.

Thus, beginning with a distribution across age groups in apparent time, the research question would be: do these results really show the existence of linguistic change in progress? According to Labov (1994), the only way to solve the problems posed by studies in apparent time is by providing support for the research findings based on linguistic observation in real time, that is, observing a speech community at two discrete points in time.