STANDARDIZATION OF HIGHLY AUDIBLE PHRASEOLOGY 



239 



ment the present list of standard submarine 

 phraseology. To assist in the future selec- 

 tion of new terms which are readily intel- 

 ligible in noise, it may be useful to summa- 

 rize the attributes which seem in large part 

 to determine the relative audibility of words 

 (7, pp. 74-77; 18, 19).3 



1. The relative power of the component 

 speech sounds. The most powerful phoneme — 

 "aw" as in all — is 680 times (28 decibels) as 

 intense as the weakest phoneme — "th" as in 

 thin — and the remaining speech sounds are 

 distributed between these extremes (2, pp. 

 70-76). Since the noise of military com- 

 munications raises the threshold of hearing 

 for sounds other than the noise itself, the 

 weaker sounds are lost first. Under such 

 conditions, it is to be expected that words 

 will tend to be audible in proportion to the 

 power of their component speech sounds, and 

 particularly of their component consonants. 

 On the whole, consonants are the weakest 

 of the speech sounds, yet yield the most 

 important clues for identifying a message. 



2. The spectra of the component speech 

 sounds. Each speech sound has a charac- 

 teristic acoustic spectrum, or distribution of 

 energy among the various frequencies (1; 2, 

 pp. 28 ff.; 3). Experiments have shown 

 also that for each sound, certain parts of 

 its characteristic frequencies seem to be the 

 critical regions with respect to its identifia- 

 bihty by a listener (2, pp. 279 ff.). There- 

 fore, the relations of the spectrum of a speech 

 sound both to the spectrum of the interfer- 

 ing noise and to the cutoff point and fre- 

 quency characteristics of a particular inter- 

 communication system will affect its relative 

 identifiability. In addition the identifiabil- 

 ity of a speech sound will vary inversely 



3 These factors are in part theoretical, and in 

 part based on experimentation conducted during 

 the last war. The exigencies of war did not permit 

 complete solution of all the problems involved. 

 There is need for a thoroughgoing experimental 

 program to establish in quantitative detail the 

 basic factors governing the audibility of speech in 

 noise, as a foundation for standardizing military 

 oral codes and phraseology. 



with the number of other speech sounds 

 which it resembles, and for which it is liable 

 to be mistaken. 



3. The number of the component sound ele- 

 ments. The more sound elements there are 

 in a word, the greater the chance that enough 

 of these will be heard above the noise to 

 identify the word in full. And in general, 

 the longer a word is, the smaller is the num- 

 ber of other words in the language, posses- 

 sing similar overall phonetic patterns, for 

 which it is liable to be mistaken. Experi- 

 ment confirms this hypothesis: audibility 

 increases radically with word length, as mea- 

 sured by the number of syllables. For 

 example, in severe noise, disyllables are iden- 

 tified 30 percent more frequently than mono- 

 syllables, and trisyllables are identified 50 

 percent more frequently than monosyllables 

 (7, pp. 62, 76; 18, p. 2). 



4. The competitive context of a word. The 

 "competitive context" is that set of words 

 of which the listener knows in advance that 

 the stimulus word is a member. The chan- 

 ces that a mutilated portion of a word will 

 be rightly identified tend to vary inversely 

 with the number of other words in the com- 

 petitive context that it resembles. For ex- 

 ample, if the word "power" is partially 

 masked, the relatively low-powered voice- 

 less stop, "p," will be lost first, and the 

 listener will hear only "-ower." If there is 

 no prior limitation of context, the listener 

 must choose between such possibilities as 

 "tower," "cower," "hour," "bower," "sour," 

 "dower," etc. The more a military vocab- 

 ulary is standardized, the smaller becomes 

 the competitive context of each word in that 

 vocabulary, and the fewer the chances for 

 mutual confusion. Ideally, no two words 

 in a standardized vocabulary should possess 

 a similar overall phonetic pattern. 



Rough measurements are available of the 

 relative identifiability in noise of various 

 single speech sounds, and also, of the relative 

 intelligibility of dissyllabic words possessing 



these alternative stress-patterns: pro-vide; 



