No abstract

No abstract

Several indexes to measure contingency of sentences were constructed by considering nouns, repeated nouns, and total number of words. Contingency was operationally defined as reconstructibility in order to test the several indexes against a criterion.The best form of the index was then selected and retested. The contingency ranking, based on the index, of ten sections of text correlated 0-84 with the reconstructibility ranking. It was concluded that the index is a valid initial approximation to a measure of contingency if contingency is defined as reconstructibiIity.Scholars have been concerned with language as a communicative device for many centuries. More recently, part of this interest has been expressed in statistical analyses of language as evidenced in the counting of word frequencies and the calculation of word contingencies. Newman (1951) investigated patterning of vowels and consonants in various languages and found that there were marked differences in the degree to which sequences of vowels and consonants form a restriction on the informational content of the language, He also found the greatest degree of patterning in primitive languages and least patterning in English. Therefore, in information terms, English is noisy and unpredictable. In English there is little structure beyond the 5th or 6th letter (the length of the average word) perhaps, because the statistics of words are on a different level from the statistics of letters. Newman and Gerstman (1952) developed a coefficient of constraint which provides an estimate of the sequential dependencies of letters. They conclude that " neither greater freedom nor greater constraint is discovered when sequences of letters are examined of a length greater than that of one word " ; and they also suggest that " in many respects, analysis at the level of words would be most useful."Miller, Heise and Lichten (1951) found that a word is harder to understand if it is heard in isolation than if it is heard in a sentence because if one hears the first six words of a sentence, the seventh can more easily be determined €or the range of possible continuations is sharply restricted.This sort of formulation, however, has not been applied to the analysis of sentences. Since the number of possible sentences in English is very large, any estimate of the

No abstract

No abstract