418 MILDRED WEST LORING 



respectively, the decrease in reaction time for the backward 

 association is small compared to the average increase for other 

 cases. It amounts only to 162 sigma and 126 sigma. A larger 

 number of cases would probably throw these results in the same 

 direction as the others. 



Conclusion for experiment IV 



1. The reaction tune is longer for "backward" than for "for- 

 ward" double associations, using these terms to refer to the 

 normal language order. The reaction tune is longer in associat- 

 ing from intransitive verbs back through noun subjects to mod- 

 ifying adjectives than from adjectives forward through noun 

 subjects to verbs. 



SUMMARY 



One of the most important results of this investigation has 

 been the preparation of lists of stimulus words suitable for use 

 in the word association reaction. These lists total 10,888 words 

 and are classified into groups of adjectives, nouns, transitive 

 verbs, and intransitive verbs, each list arranged in chance order 

 alphabetically and further subdivided into groups of one, two, 

 and three syllables. Furthermore each of these groups has been 

 once tested for the suitability of its words to the word associa- 

 tion reaction and separated into selected words, rejected words, 

 homonyms. It is insisted that this evaluation of the stimulus 

 words can by no means be considered final or of universal fitness, 

 but will surely serve greatly in a proper selection of words for 

 a particular purpose in view. 



The complete lists without eliminations were used once with 

 four subjects to study the adjective-noun, noun-adjective, verb- 

 object and verb-subject controlled associations, and to obtain 

 data for the selection and rejection of words. 



The adjective-noun and noun-adjective controlled association 

 was more intensively studied on eight new subjects using 1800 

 words from the revised lists, keeping the number of stimulus 

 nouns and adjectives equal, and constant for all subjects at each 



