Introduction

Abstract

The occasional statistical treatment of language data is by no means new. In fact, it may be truly said that linguistics is not possible without some degree of statistical classification. How else than by patiently recording and classifying linguistic forms according to their function can we imagine that the grammar of a language was extracted from the actual speech occurrences? And the same applies to the lexicon of a particular language, which quite obviously presents a major statistical effort. Moreover, a regularity in the historic development of language, like Grimm’s law, say, of the consonant shift, could have revealed itself to the investigator’s mind only after a long and patient collection of data, and thus by a method which we usually describe as statistical. Just to give another instance, this time from comparative philology, let us think of the gene-tical relationship known as the Indogermanic family of languages, which was established by careful comparison of a great mass of linguistic data, and in which the element of frequency of occurrence of certain linguistic forms, lexical and grammatical, is used as the criterion for the degree of nearness of relationship.