Coding in time
In this paper, I argue that linguistics is a historical science in more than one sense: Not only is the object, language, embedded in time, but so is the study of it. This has consequences for our conception of language change. A central result of previous sociolinguistic analyses of spoken Copenhagen Danish, starting with Brink & Lund 1975, is that during the latter half of the 19th century the common European low back vowel (a) was differentiated in the Copenhagen speech community into at least four different vowel qualities all of them bearing both linguistic and sociolinguistic information. I present evidence from an unbroken chain of Copenhagen informants ranging from birth years 1905 until 1962–71. Various sections of this sample have been studied by different researchers using auditory classification of variants, and the total sample has been coded once more by the LANCHART centre. The analysis shows that auditory coding reveals the same patterns of differences between sociologically characterized groups but the relative figures classified as belonging to the various variants diverge quite dramatically and seem to be dependent on the age of the coder and the point in time at which the coding takes place. I suggest explanations for these facts and discuss whether this is a problem for the validity of sociolinguistic research or perhaps an inescapable condition for research within the language sciences.