Where Does One Way of Speaking Stop? A Line on a Map That Nobody Drew

Surveyors visit many locations and record how speakers there answer the same set of questions, asking for the local word for particular objects, the local pronunciation of particular words and the local grammatical form in particular constructions. Plotting the answers for one feature across all the locations shows an area where one form is used and an area where another is, and the boundary between them is drawn as a line. Repeating that for hundreds of features produces hundreds of lines, and where many of them run close together the map shows a bundle, which is what people mean when they speak of a dialect boundary. A single line rarely means much and a bundle usually does.

Boundaries tend to coincide with obstacles and with old divisions:

• •Rivers that were difficult to cross before bridges
• •Ranges of hills and areas of marsh or forest
• •Boundaries of former kingdoms and dioceses, long after they ceased to exist
• •Limits of former settlement by a particular group
• •Market catchments, since people talk to those they trade with
• •Modern administrative boundaries, which mostly do not correspond at all

Large systematic surveys were conducted across Europe and North America from the late nineteenth century and their results reshaped the subject. A German survey begun in 1876 sent questionnaires to tens of thousands of schoolteachers, and the results showed that the neat boundaries assumed by earlier scholarship did not exist, since different features produced different lines that crossed each other rather than coinciding. That finding, that dialects shade into each other by feature rather than dividing cleanly, is among the most important results in the field. The English survey conducted in the 1950s and 1960s interviewed elderly rural speakers specifically, on the reasoning that their speech preserved older distinctions that were disappearing.

Individual words plotted on these maps frequently record specific historical events. Words of Norse origin cluster across northern and eastern England in a distribution matching the area of Scandinavian settlement, and the boundary is visible a thousand years later. Words of French origin concentrate in vocabulary associated with administration, law and cooking, reflecting who used those words after 1066. Terms for objects introduced at a known date spread outwards from the place of introduction in a pattern that can be dated. Words surviving only in remote areas mark the retreat of an older form. And a boundary that matches no physical or political feature frequently turns out to record a former estate, a former parish or a former route.

Traditional local speech has receded substantially and the reasons are the obvious ones, namely mobility, broadcasting, education and the movement of population from countryside to city. That has not produced uniformity, and what has emerged instead is regional levelling, in which very local forms disappear while broader regional varieties strengthen and spread, so the map becomes coarser rather than blank. New features spread outwards from large cities along transport routes. Modern survey work uses telephone interviews, online questionnaires answered by enormous numbers of people, and analysis of social media text tagged by location, which produces maps from millions of data points rather than hundreds and which shows change happening in real time.

Recording how speakers at many locations answer the same questions and plotting each feature produces lines, and where many lines run together the map shows a real boundary. Those bundles follow rivers, hills and the borders of vanished kingdoms rather than modern administrative divisions. A German survey from 1876 established that features produce crossing lines rather than neat divisions, and modern work uses social media text tagged by location.