812 BELL SYSTEM TECHNICAL JOURNAL 



In forming such syllables only those combinations involving final r 

 and ng that are shown in Table I are included. Much confusion 

 exists concerning the pronunciation of other combinations of these 

 sounds. Syllables that represent slang in English are also omitted. 

 These omissions are made by returning the card upon which the 

 sound in question is written to its box and drawing another card. 

 By combining the sounds at random in this manner any desired 

 number of lists may be made which for practical purposes are all of 

 equal difificulty. 



In addition to containing a certain speech sound content, connected 

 speech is characterized by inflection, accent, a rate of utterance, etc. 

 In the earlier articulation studies the test syllables were called singly 

 at intervals of about three seconds. When considered with respect 

 to connected speech this procedure seems somewhat artificial. Com- 

 parative tests were made in which the syllables were called singly 

 and as parts of introductory sentences. The tests showed the syllable 

 articulation to be somewhat larger when the introductory sentences 

 were used. The increase was due largely to the greater ease in 

 interpreting the initial consonants of the syllables, when they were 

 inserted in the introductory sentences. The effect was most noticeable 

 for the stop and fricative consonants which have relatively short 

 durations. In order to make the technique more nearly like connected 

 speech the syllables are called in the short introductory sentences. 

 The use of such introductory sentences also helps to insure that any 

 element in the transmission system being tested, whose performance 

 depends particularly upon their immediate past history, will be in 

 the condition in which we are interested for determining speech 

 transmission capabilities. 



A list of sentences which is used for this purpose together with a 

 sample record of articulation data is shown in the articulation test 

 record of Table II. For calling purposes, the syllables on the cards 

 are written in the spaces under the columns marked "called" of the 

 test record. These sentences are called uniformly at the rate of 

 15 per minute. When the syllables in the first column are called, 

 the sentences are repeated using the syllables in the second column 

 and then those in the third column. 



The observers are provided with blank articulation test record 

 sheets. They write the sounds which they hear in the corresponding 

 "observed" columns. When the test is completed the observed and 

 called sheets are compared and the various articulations obtained. 



For good results it has been found advisable to use a testing crew 

 of ten people — 5 men and 5 women. Eight people are ordinarily 



