William Labov (1994), Principles of Linguistic Change, discusses the nature of facts and theory and illustrates 2 contrasting approaches to theory construction.

(The text below closely paraphrases Labov, but has been edited to make brief points.)

• We build a model that corresponds point-for-point to each element of language structure,
• then we state the rules for relating parts of the model to each other and to the empirical facts.
• Using this model, we attempt to provide an answer to the most general questions, on the basis of whatever information we currently have.
• From the model, make deductions about empirical facts which will confirm or disconfirm it.
• This procedure moves from the unknown to the known. It is rapid and productive, though it does not necessarily have cumulative results. It is deductive, moving from a few examples to the statement of an unrestricted principle, from which we predict further confirmatory facts.

• Definitions: Theories as a form of general knowledge are to be highly valued. Ultimately, however, the chief value of a theory is in establishing the most important matters of fact.
• The term ‘fact’ designates a true synthetic predication about a particular singular object. Facts form the data of a field of inquiry. There is no such thing as a false fact. Every piece of data thus represents a fact, though we may be mistaken about exactly what that fact is.
• Predications made about a plurality of objects are ‘generalizations’. These may be disproved.
• A ‘principle’ is a generalization that is unrestricted in its application in time or space.
• Many generalizations are thus in limbo, struggling for the status (the homogeneity) of facts. Unexpected regularities are also facts, facts in search of a generalization or principle.
• The Procedure: We begin with general questions about the routes, mechanisms and causes of linguistic change (questions derived most often from the work of earlier investigators).
• Select communities showing changes in progress that promise to be illuminating, and
• Make observations of a representative sample of speakers from each community.
• Proceed from observation to inferences about what’s happening in the community as a whole
• Select other communities which are best suited to test the generality of these inferences.
• Combine all the data we have to confirm, correct or reject those inferences.
• This expansion of knowledge leads to a small number of generalizations of wide scope, or principles, which we have good reason to think are true.
• The principles must be logically related to each other.
• As the principles grow stronger, it is increasingly possible to make deductions about what we can expect to find in new communities undergoing change.
• These deductions are strategies to locate sites for renewed testing & refinement of principles.
• As a major goal, these linguistic generalizations or principles should interlock in such a way that they may be combined into simpler and more general statements.
• Such a set of logically related general linguistic principles may deserve to be called a theory.
• But since the fact of language change is not consistent with our fundamental conception of what language is, the explanation of language change will find its causes outside linguistics (in domains such as physiology, acoustic phonetics, social relations, perception, cognition).
• A set of propositions that relate general findings about language change to general properties of human beings or of human societies will deserve to be called a theory of language change.
• This procedure moves steadily from the known to the unknown, enlarging the sphere of knowledge on the foundation of observation and experiment in a cumulative manner. It is inductive, moving step by step as the data base grows to statements of increasing generality.

Back to LG457 Language Variation & Change page