844 BELL SYSTEM TECHNICAL JOURNAL 



having an electrical characteristic similar to one of the reference 

 circuits, the syllable index / can be calculated from the observed 

 syllable articulation 5 (as obtained with the new standard lists) by 



means of the equation, 



/ = .5 52/^ + .5 ^. (14) 



This value must now be reduced to the reference condition of practice 

 by the methods which have already been described. In such cases it 

 is thus possible to obtain the syllable index from the observed syllable 

 articulation values, and it is unnecessary to analyze the data for the 

 individual sound articulation values. 



The weighted average, however, is the more logical way of obtaining 

 a single index and should be used when it is suspected that it might 

 give results which are essentially different from the unweighted 

 average. 



It is possible to carry the probability relations a step further and 

 apply them to cases of English words and sentences. In order to do 

 this it is necessary to make assumptions as to how the thought or 

 meaning of the words affects the interpretation of the sounds. These 

 assumptions are not only somewhat uncertain, but owing to psycho- 

 logical factors in testing are difficult to verify experimentally. In 

 general, the meaning associated with words makes them easier to 

 interpret than meaningless words. For single-syllable words, these 

 effects are small. Two-syllable words are easier to interpret than 

 single-syllable words. The interpretation of words containing from 

 three to five syllables, and short sentences, depends almost entirely 

 upon interpreting those parts which are not indicated by the thought 



or meaning. 



Other Testing Methods 



For most articulation studies it has been found desirable to use the 

 standard testing technique which has been described, but it is fre- 

 quently necessary, in special cases, to use other techniques. In such 

 cases it is desirable, if possible to interpret the results in terms of the 

 standard technique. In the course of research work, several different 

 articulation testing methods have been used which give information 

 on the type of correlation between them that may be expected. 



The probability relations have been made use of in constructing 

 two other types of lists which are called vowel-consonant and vowel 

 word-consonant word lists. These lists are designed to give the same 

 values of sound articulation as given by the standard lists. The 

 former lists are shown in Table XVI. The various vowels are com- 

 bined with the same consonant, and the various consonants with the 



