Centrum för kognitiv semiotik (CCS)

Humanistiska och teologiska fakulteterna | Lunds universitet


Normativity in a Nutshell

Esa Itkonen (Department of Linguistics, Turku University)

Rule-sentence A = “In English the definite article [i.e. the] precedes the noun [e.g. man]” (or “In English tree and three mean a plant and a number, respectively, not vice versa) vs. empirical hypothesis B = “All ravens are black” (or “All pieces of metal expand when heated”).

My thesis (presented in Linguistics and Metascience, in 1974): A and B are fundamentally different. B is falsified by the occurrence of a non-black raven (or of a heated and non-expanding piece of metal). By contrast, A is unfalsifiable, because it is falsified neither by the occurrence of utterances like *Man the came in or *Two and tree make five (because they are incorrect) nor by occurrence of utterances like The man came in or Two and three make five (because it is correct). Why are A and B different? Their referents are normative and non-normative, respectively.

The standard objections against my thesis (some of which overlap to a higher or lower degree) are as follows.

(i) “The truth of A cannot be accepted as long as it has not been experimentally established.”

(ii) “If English were different, A would be falsified.”

(iii) “In English (as it is today) A is verified whereas the opposite rule-sentence “The definite article follows the noun” (or “Tree means a number”) is falsified.

(iv) “It is not the case that the definite article always follows the noun; in a list like ‘the man, the woman, the child ...’ man and woman are followed, not preceded, by the.” (Or: “If I decide that tree means a number, this will be so.”)

(v) “Maybe A is not falsified by simple observations, but neither are scientific theories (as has been taught by Kuhn and Lakatos).”

(vi) “Since A contains theoretical terms like definite article and noun, it is not comparable to B which is expressed in ordinary language.” (This,as it stands, is OK; but notice the mention of the and man.)

(vii) “A and B are formulated in dissimilar ways, i.e. B contains the universal quantifier all while A does not; and this makes an exact comparison between the two impossible.”

(viii) “There are rules different in kind from the one described by A.”

(ix) “Language is not like a game.”

(x) “It is just preposterous to claim that grammars are unfalsifiable.”

(xi) “Only conceptual truths are unfalsifiable; conceptual truths, being necessary, entail the necessary existence of what they are about; but the existence of the rule described by A is contingent.”

(xii) “Only analytical truths are unfalsifiable; but A is not analytical.”

(xiii) “Unfalsifiable truths are known with certainty; but if language is known with certainty, there is nothing left for the grammarian to do.”

(xiv) “Some data is not known with certainty.”

Non-standard objection (xv): “A and B are too different for any comparison to be viable because A, being an (implicit) imperative, does not admit of any truth-value.”

(Common mistake: I say “There is an area of certainty of which X is a part”. The opponent responds: “But Y is not known with certainty.” No connection! Don’t change the topic!) Why should my thesis have aroused so much opposition? My thesis just explicates the meaning of normativity. If the objections – apart from (xv) – were justified, there would be no normativity. Most people do not accept this conclusion. So what do they want? An alternative explication of normativity? But there is no such thing.

Consider the following compromise (due to Robert Nozick 2001): “Hume was right that an ought cannot be derived from an is; normativity cannot be derived from descriptiveness. Yet the descriptive fact that we have a normativity operator within our cognitive apparatus can be given a thoroughly naturalistic and non-normative evolutionary explanation.” Maybe so. But today at least (as opposed to what was the case 300'000 years ago) the difference between A and B does exist!