Sei sulla pagina 1di 6

Semantic Approach to the Perception of Complex Sounds

Lawrence N. Solomon

Citation: 30, 421 (1958); doi: 10.1121/1.1909632


View online: http://dx.doi.org/10.1121/1.1909632
View Table of Contents: http://asa.scitation.org/toc/jas/30/5
Published by the Acoustical Society of America
THE JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA VOLUME 30, NUMBER 5 MAY, 1958

Semantic Approach to the Perception of Complex Sounds


LAWRENCE N. SOLOMON
U.S. Navy tlectronicsLaboratory,San Diego52, California
(ReceivedDecember18, 1957)

In an attempt to derive a limited number of descriptiveadjectives with which to characterizepassive


sonarsounds,50 U.S. Navy sonarmen(median sonar experienceof one year) rated 20 different passive
sonarsoundson 50 seven-pointscalesdefinedby polar-oppositeadjectives (e.g., heavy-light). The inter-
correlationsbetweenscales(calculatedover both subjectsand sounds)werefactor analyzedby the Thurstone
completecentroidmethod.Seveninterpretableorthogonalfactors (accountingfor approximately40% of
the variance of judgments) were extracted from the data.
The generality of the methodologyfor psychologicalresearchon auditory perceptionis consideredand
suggestions for researchalong theselinesare presented.

INTRODUCTION
commonlydenotedby the term timbre. The unsystem-
atic terminologyused by musiciansand acousticiansto
HE problem
tion of
has long the
beendimensionsofauditory
one of considerable percep-
interest to describe their tonal experiencesled Lichte to test
researchers in the field of audition. Pitch and loudness experimentally whether or not complex tones have
emergedrelativelyearly aspromisingunitary functions identifiable attributes other than pitch and loudness.
in this domain, but other dimensions,such as volume, He concludedthat complextonesdo possess, in addition
brightness,and density evadedclear-cutspecification. to pitch and loudness,at least three attributes' bright-
Today, Harris and others have thrown somesuspicion ness,roughness,and fullness.
upon the notion that even loudnessand pitch dis- Two years later, in his factorial study of auditory
crimination are unitary in nature and, instead, are function,Karlin9identifiedpitch-quality discrimination
expandingthe study of primary auditory abilitiesinto as one of the dimensionsin the auditory domain, and
a search for subfunctionsamong what were formerly Hanley, in 1956, isolated "unpleasantness"as an
thought to be unitary realms. identifiablefactor in the perceptionof speech.
Working with pure tones over thirty years ago, Other than these studies, however, little has been
Rich ' and Halverson a were able to determine the done to exploresystematicallythe qualitative "mean-
differential thresholds for tonal volume as functions of ing" or connotativeaspectof complexsounds.Timbre
frequencyand intensity. Although there was some continuesto be used as a loose,nonspecificreference
difficultyin replicatingtheseresults,Stevens,
4and later to various qualitative attributes of tonal experiences.
Thomas,* were able to obtain consistentequal-volume Mosier,n in 1941, carried out an experiment, along
contours. From these results, volume was established somewhat different lines, closely allied to Lichte's
as a separateattribute of tonal experience,distinct interest. Mosier was one of the first to attempt a
from pitch and loudness. systematic,psychometricstudy of the meaningof cer-
Stevens6wasalsoable (still workingwith pure tones) tain sounds.He had subjectsrate 296 words and word
to determineequal-densityand equal-brightnesscon- combinationson an l 1-point scale of favorableness-
tours,thus seeminglyestablishingtheseexperimentally unfavorableness.Mosier's technique, however, lacked
as functionsin auditory perception.In 1936, however, the multidimensionalityrequired to do justice to the
Boringand Stevens 7questioned whetherbrightnessand polyvariate nature of meaning.
densitymight not be differentwordswhichwerebeing Much the same problem that prompted Lichte's
usedto apply to the sameattribute. With this concern, researchpresenteditself to the presentauthor in 1953.
they helped focus attention upon the languagethe It was decided to utilize something like Mosier's
listenerusedto describehis tonal experiences,and they techniqueas the methodologymost suited to the task.
set the stage for investigationinto the sematicsof The soundsunder study werepassivesonarrecordings,
auditory perception. and the looseand variable terminologywas the "sonar
Then Lichtea did a seriesof studiesin an attempt to vocabulary" that sonarmen had spontaneouslyde-
quantifythe characteristics
of complextoneswhichare veloped to aid the communicationof their auditory
experiences.
J. D. Harris, Medical ResearchLaboratory Memorandum The submarinesonarman,faced with the realities of
Report 57-4 (1957).
'G. J. Rich, Am. J. Psychol.30, 121 (1919). underseawarfare, has had to develop his own means
aH. M. Halverson,Am. J. Psychol.35, 360 (1924). of analyzing sonar signalsand of describingthem to
4 S.S. Stevens,Am. J. Psychol.46, 397 (1934).
5 G. J. Thomas,Am. J. Psychol.62, 182 (1949). others.In an informal way, a "sonarvocabulary" has
6S.S. Stevens,Proc.Natl. Acad.Sci.U.S. 20, 475 (1934).
?E.G. Boringand S.S. Stevens,Proc. Natl. Acad. Sci. U.S. j.E. Karlin, Psychometrika7, 251 (1942).
22. 514 (1936). Clair N. Hanley, J. SpeechHearing Disorders21, 76 (1956).
8William H. Lichte, J. Exptl. Psychol.28, 455 (1940). C. I. Mosier, J. Soc.Psychol.13, 123 (1941).
421
422 LAWRENCE N. SOLOMON

developedamong sonarmen,similar to the looseand indicating,with a pencilcheckmark,whichof the polar


variable vocabulary used by musiciansand psycholo- termsappliesto the stimulus.The subjectindicatesthe
gists to describetheir tonal experiences.Accordingto intensity of his associationby the extremenessof his
the sonarman, some soundsare "heavy," others are checkingon a seven-pointscale.Thus, on the heavy-
"light"; some are "bright," others are "dull"; some light scale,a judgmentof ! indicatesextremelyheavy,
are "hard," others are "soft." The use of such terms and 7 extremelylight. A judgmentof 4 indicateseither
as "heavy," "bright," and "hard" to characterize "neutrality" or equal associationto both endsof the
sounds seems to be an example of synesthetic or scale.
metaphoricalthinking. By suchmeans,it is possibleto expressthe relation-
In most studies dealing with the recognition and shipbetweenitemsand scales.Any two stimulihaving
detectionof passivesonarsounds,subjectivereactions exactlythe samecheckingon all scalesin the Differen-
to the soundshave beengenerallyignoredas a possible tial may be saidto havethe sameconnotativemeaning.
means of differentiation. However, one must consider The presentresearchwascarriedout in threephases'
the "frameof reference"of the personor, in otherwords, (1) the compilationof a seriesof descriptivescales
take into accountwhat the stimulusmeansto the person. uponwhichsonarsoundsmay be assigned quantitative
It is here proposedthat the experiencedsonarman values; (2) the determinationof the meaningfulness of
operateson the basisof a semanticframe of reference this task by an analysisof the consistency
with which
which is incompletelyverbalized.Each sonarstimulus an experimental populationof observers
assigns certain
is evaluated within this frame of referenceand judged sonarsoundsto certain positionson thesescales;and
(or discriminatedas) warship, submarine,etc. (For (3) a factor analysisof the intercorrelationsbetween
example, since it has a "heavy" sound, it is a cargo scalesto derive a limited numberof generaldimensions
ship.) It seemsfeasible to structure this frame of of meaninghaving maximaldifferentiatingpower for
reference explicitly, thus providing a standardized the soundsamplesstudied.
"ruler" which may aid in the making of sonar judg- The compilationof scalesand the evaluationof the
ments. Actually, such a processwould involve the consistency of the connotativeassociations
to passive
developmentof severalrulers, and theserulers would sonar soundshave been reported elsewhere23 Briefly
be the dimensions of the frame of reference characteris- summarized,a seriesof 50 seven-pointscaleswas com-
tic of the experiencedsonarman. plied, each scalebeing definedby a pair of polar-
The term "dimension" as used here bears a relation- opposite adjectives.The adjectives
wereobtainedfrom
ship to the Cartesian coordinatesof physical space. the followingsources:(1) a preliminarystudy run on
The term "semantic space," as differentiated from collegestudentswhereinthe subjectswrote as many
physical space, is used to denote the hyperspace descriptivewordsas they felt were applicableto de-
(multidimensionalrealm) which includes the sonar- scribevarioussonarrecordings presentedto them; (2)
man's perceptual reactions to sounds as they are a list of scalesbeingusedby Osgoodand SuciTM in their
verbalizedin terms of descriptiveadjectives. factorialstudyof meaning;(3) rationalanalysisof the
sensory
inputsof the humanorganism;
and (4) the list
METHODOLOGY of high-ranking,helpful sonar recognitioncues de-
The task of structuring the sonarman'sframe of velopedby BlackandHarsh5for usein ship-classifica-
tion teachingmethods.The 50 scalesare listed in
referencein terms of descriptiveadjectiveshas been Table I.
approachedin this study through the use of multi-
Sixteenmale subjectsthen rated four passivesonar
dimensionalscaling techniques.By means of these
soundson the 50 scales.Test-retestresults,usinga new
techniques,it is possibleto determinewhat psychologi-
statisticaltest of consistency(the inter-test interval
cal dimensionsthe sonarmanrespondsto when placed
beingfive days)showedthat subjectsare consistent in
in a complexstimulussituation.Suchstimulussitua-
their ratings.The null hypothesis(subjectsmaketheir
tions may be scaledalong multidimensionallines by
first assumingthat certain dimensionsare operativein
ratingsby chance)wasrejectedat or beyondthe 1%
level of confidence for all 50 scales.
that particularsituation,moreor less"blanketing"the
possibilities.Then, through factor analysisone may At thispoint,the factor-analysis
portionof the study
reducethe numberof dimensionsto a minimal quantity wasundertaken.The subjectsemployedwere50 Navy
having maximal differentiatingpower. The method- sonarmenwith mediansonarexperienceof oneyear.
ologyemployedutilized the SemanticDifferential de- The stimuli usedwere 20 recordingsof passivesonar
velopedby Osgood? soundstaken from records utilized by the Navy to
The graphicform of the SemanticDifferential is a train sonaroperators.Each soundwas recordedon a
paper-and-pencil test whereinthe subjectis presented
L. N. Solomon,Navy ElectronicsLaboratory Report 443
with a pair of descriptivepolar terms,suchas heavy- (1953).
light, and a stimulusto be judged.The task consists
of 4C. E. Osgood
andG. J. Suci,J. Exptl.Psychol.50,325 (1955).
5 E. L. Black and C. M. Harsh, Navy ElectronicsLaboratory
CharlesE. Osgood,Psychol.Bull. 49, 197 (1952). Report 342 (1953).
SEMANTIC APPROACH TO PERCEPTION OF SOUNDS 423

TABLEI. Final rotatedfactormatrix. (Decimalpointshavebeenomitted.)

I II III IV V VI VII VIII h

* 1. pleasant-unpleasant 16 60 07 23 07 25 16 --01 5365


2. repeated-varied 01 05 41 16 --06 12 06 05 2204
3. smooth-rough -24 54 12 17 06 18 11 20 4806
4. active-passive -01 - 15 27 -- 11 --06 30 14 -- 15 2433
* 5. beautiful-ugly -11 67 13 17 07 22 12 --12 5889
6. definite-uncertain 03 04 52 06 --06 39 04 - 10 4438
* 7. low-high 71 05 - 16 14 02 18 -02 12 5994
8. powerful-weak 60 -09 35 --22 --08 06 -01 -- 17 5780
9. steady-fluttering 13 15 40 --07 --04 30 -- 16 01 3216
10. soft-loud -19 32 --28 33 12 25 16 21 4724
11.full-empty 53 13 27 -09 - 13 03 22 -- 12 4594
12.good-bad 09 59 28 16 05 02 18 -- 14 5151
'13. rumbling-whining 76 -02 --04 12 05 21 09 09 6568
14. solid-hollow 30 20 21 -07 -09 22 09 - 12 2580
'15. clear-hazy 06 21 58 -08 -01 -03 06 -01 3952
'16.calming-exciting 18 20 --06 48 05 - 12 --07 10 3382
17.pleasing-annoying 19 58 17 15 27 25 12 --06 5773
'18.large-small 79 --07 12 - 11 -01 01 -04 - 13 6742
19.clean-dirty -27 40 25 07 06 -06 21 --16 3772
20.resting-busy -02 16 -15 29 28 03 -20 13 2688
21.dull-sharp 67 -09 --20 19 13 --03 00 14 5705
22.deep'-shallow 71 04 09 -04 06 -05 08 --01 5280
23.gliding-scraping -02 29 15 01 30 29 04 22 3312
24.familiar-strange 14 14 37 18 05 35 15 --03 3569
25. soft-hard -07 24 - 12 28 36 04 21 13 3475
*26.heavy-light 84 - 13 04 --06 --06 00 -08 -- 11 7498
*27.wet-dry 21 -07 --03 --04 21 34 18 02 2440
28.safe-dangerous -06 27 12 48 20 --17 14 --17 4387
29. concentrated-diffuse 11 12 43 --07 --21 --06 03 16 2905
30.pushing-pulling 10 02 -02 21 --08 --09 09 03 0784
31.labored-easy 50 --19 --04 --05 --34 -17 --13 01 4517
32.dark-bright 57 -29 --16 -14 -04 14 --17 17 5332
*33. even-uneven -08 17 51 12 -06 17 14 16 3875
*34.loose-tight 19 13 -- 11 02 57 12 --04 --08 4128
*35. relaxed-tense 14 26 --02 20 64 -03 03 --06 5426
*36. colorful-colorless 06 28 17 - 13 06 03 41 00 3004
37. hot-cold -08 24 22 -- 13 --09 -26 13 08 2283
38. rich-thin 45 13 16 -08 17 --07 34 13 4177
39. obvious-subtle 09 06 43 10 --16 --16 12 07 2771
*40. wide-narrow 73 -07 10 -02 15 -01 05 05 5758
41. deliberate-careless 25 08 41 --03 -21 -01 30 16 3977
42.happy-sad -26 33 20 18 06 --06 31 --16 3778
43.gentle-violent -- 15 25 --05 51 34 20 14 --05 5253
44. mild-intense --02 25 -05 53 33 --06 08 03 4661
45.rounded-angular 36 07 04 02 26 23 04 17 2875
46. slow-fast 60 --09 -06 23 11 -05 --21 --04 4849
47.rugged-delicate 65 -05 15 -22 -12 --08 -02 01 5172
*48.simple-complex -04 02 15 48 11 --14 -01 00 2867
49.green-red 07 --13 - 14 20 18 16 11 -08 1579
50. masculine-feminine 70 -03 11 -- 18 -09 11 --01 15 5782
Mean squaredfactor loading 0.148 0.064 0.058 0.045 0.039 0.029 0.022 0.013

loopof magnetic anauditorystimulus duringthe test sessions


tapeto produce in SanDiegoBay alongside
the submarinefrom whichthe subjectswere obtained.
of a continuousnature. Care wastaken to avoid adding
anychance characteristics
to the sounds throughthe Each soundwaspresentedto the subjectsat a con-
recordingtechnique. stantpowerlevelandallowedto playcontinually until
The 20 soundsconsisted from the last subjecthad rated it on the last scale.This
of five representatives
eachof the fourmajorcategories of soundsastheyare procedure wascontinued untilall 50subjects hadrated
differentiated
by the Navy (i.e., warships,light-craft, each of the 20 sounds on all 50 scales. The raw data
cargoships,and submarines). Thesewerepresentedwerepunched into1000IBM cardssothat eachsubject
to thesubjectsovera Western ElectricNo. 775speaker had a separatecardfor eachsound.Each cardhad a
mountedin the centerof the ceilingof a soundproof numberfrom ! to 7 punchedin each of 50 columns,
room.The subjectssat aroundthe perimeterof the one column for each scale. Pearson product-moment
roomapproximately equidistantfromthesound source. correlations were calculated between each scale and
The soundproof roomwasmounted ona convertedeveryotherscaleusingthe data of bothsubjects and
craft(LCM),andthiscraftwasmoored sounds.
Navylanding
424 LAWRENCE N. SOLOMON

The 50X 50 correlationalmatrix thus generatedwas termsasbeautiful,pleasant,good,pleasing,and smooth.


factor analyzed using Thurstone'scomplete centroid Factor III has beenlabeled"clarity," and is identified
method. Eight factors were extracted from the data by such terms as clear, definite, even, concentrated
beforeMcNemar's criterion7for completeness of factor and obvious. Factor IV seems to represent the
extraction was met. The highest coefficientin each "security" aspectof the sonarsounds,and is associated
column of the correlation matrix was used as the with such descriptiveterms as mild, gentle, calming,
estimateof communalityof the corresponding variable. safe and simple. Factor V was labeled "relaxation"
The eight referencesaxes were rotated by means of and is characterizedby the terms relaxed, loose,soft,
the Zimmerman graphic, orthogonalsystem of rota- gentle, and mild. Factor VI seemsto reflect the degree
tion.a The rotations were guided by the criteria' of "familiarity" the sonarmanhas with the sounds,
positive manifold, simple structure, and psychological and it is expressedin suchterms as definite,familiar,
meaningfulness.Thirty-three rotations yielded an wet, active, and steady. Factor VII is the "mood"
orthogonal solution with seven interpretable factors dimensionto auditory perceptionand is characterized
and one residual. by suchterms as colorful,rich, happy, deliberate,and
full.
RESULTS
Of the 50 adjective-pairsused, 15 can be chosento
The final rotated factor matrix is presentedin Table provide a "factorially pure" instrument with which
I. From the mean squared factor loadings at the to quantify a subject's reactions to complex sounds
bottom of the columnsin Table I it may be seenthat (passivesonar sounds)along the seven dimensions
the eight factors extracted accountfor approximately isolated in this study. They are: heavy-light; large-
42% of the total variancein judgments. small;rumbling-whining;wide-narrow;low-high;beau-
The task of interpretingthe extractedfactors (i.e., tiful-ugly; pleasant-unpleasant;clear-hazy; even-un-
identifying them by name) is somewhatfacilitated even; calming-exciting;simple-complex;relaxed-tense;
because of the nature of the variables involved. A loose-tight; wet-dry (or, as an alternate' familiar-
cluster of semantic scaleshighly loaded on a given strange); and colorful-colorless.These adjective-pairs
are marked with an asterisk in Table I.
factor usually leaveslittle doubt as to the dimensionof
semantic space which that factor represents. For The number of scalesin the above list representing
example,the adjectives"heavy," "large," and "rum- a given factor is proportionalto the percentageof the
bling" heavily loaded on factor I aids in identifying varianceaccountedfor by that factor.
that factor as a "magnitude" dimension of auditory Many of these attributes, undoubtedly,reflect the
perception.The genericterm which will embracethe specificdenotativemeaningthat passivesonarsounds
majority of the scalenamesfound to be highly loaded come to have for the trained sonarmanand may not
on a given factor is usually taken to identify that be fully representativeof the characteristicswhich
particular factor. However, when interpretingfactors, would emergefrom the testing of a sample of the
it is necessaryto go beyond the given quantitative generalpopulation.For that reason,the interpretation
loadingsand to conceiverationally of a propersemantic of the factors isolated is only minimally presented
label for the mathematical dimensions. Most factor and a fuller discussionmay be found in Solomon?
analysts recognize that subjective interpretation is
involved in the identification of factors. DISCUSSION

The meaningof the first sevenfactorsmay be briefly Although the present results may be restricted to
interpreted from the factor loadings presented in the population employedand by the nature of the
Table I as follows:
stimuli used, the methodologyis general and may be
Factor I may be called a "magnitude" dimension, appliedto any set of auditory stimuli, judgedby any
and is characterizedat one extreme by such terms as sampleof the population.The generalapplicationof
the presentmethodologyto problemsin auditory per-
heavy, large, rumbling, wide, and low. The other end
ceptionis immediatelyapparent.The sametechnique
of the dimensionis identified by the polar opposite usedwith sonarsoundscouldbe employedwith simple
terms,--light, small, whining, narrow, and high. Only tones,shapedwhite noise,systematicallyvaried com-
one end of the dimensionwill be presentedhere, since
plextonesandmusicalselections. The multidimensional
the other end is always characterizedby the opposites scaling technique offers a systematicand objective
of the terms given.
approachto a problem which has traditionally been
Factor II is the aesthetic-evaluative dimension of
handledin an unsystemat.ic and introspectivemanner.
auditory perception, and is characterizedby such
19A "factorially pure" scaleis one which is highly loaded on
16L. L. Thurstone, Multiple-factorAnalysis (University of one,and only onefactor, thus representingonly one dimensionof
ChicagoPress,Chicago,1947), Chap.VIII. the domain tested.
17Q. McNemar, Psychometrika7, 9 (1942). s0L. N. Solomon,Navy ElectronicsLaboratory Report 563
18W. S. Zimmerman, Psychometrika11, 51 (1946). (1954) (Confidential).
SEMANTIC APPROACH TO PERCEPTION OF SOUNDS 425

Once a set of sounds has been rated on a series of having the "sameconnotativemeaning"may be thus
scales(such scalesrepresentingthe dimensionsof a isolated and analyzed physically to determine the
"semanticspace"within which judgmentsare made) physicalcorrelatesof the psychological judgmentsof
the relationsbetweensoundscanbe quantifiedby means "similarity" or "togetherness." The presentauthor is
of a statistic called the D measure. 2t Clusters of sounds currentlyundertakingthis phaseof the researchwith
'tC. E. Osgoodand G. J. Suci, Psychol.Bull. 49, 251 (1952). passivesonarsounds.

THE JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA VOLUME 30, NUMBER 5 MAY, 1958

Information Transmission with Elementary Auditory Displays*


WLLArH. SurBY,DAVIS CItAalBLISS,iANDIRWIN POLLACK
Operational
Applications
Laboratory,
Air ForceCambridge
Research
Center,BollingAir ForceBase25, D.C.
(ReceivedJanuary 22, 1958)

A procedure for transmittingthe lettersof the alphabetby tone-codedsignalswasexamined


in quietand
againsta noisebackground. The procedure employedsuccessive selections,
eachfromamonga smallnumber
of alternatives,in order to transmit a target vocabularyof 25 letters. Four stimulusvariables:tonal
frequency,soundlevel, location,and durationwere examined,one at a time. Successive selectionswere
madeamongtwo, three,and five alternativesper variable.The highestreceptionrate wasobtainedwith a
three-alternative,frequency-coded display.Receptionof tone-codedsignalsin noisewasnearly equivalent
to that in the quiet,whenthe tonal signalswereabout3 db abovemaskedthreshold.

INTRODUCTION
Our approach was to examine information trans-
missionwith eachof four tonal variables:the frequency,
NDER
extremely
unfavorable
speech-to-noise
(S/N) ratios, specialproceduresfor speechcom-
soundlevel, location, and duration of a tone. While the
municationmay becomeimportant. For example,in major part of our work was carried out in the absence
tests of communicatinga single word from a target- of noise, sample tests carried out against noise back-
vocabularyof 64 wordsunderextremelylow S/N ratios, grounds will alsobe presented.
it was found that a network sequenceof six successive PROCEDURE
two-alternativeselectionsyielded communications per-
formancesubstantiallysuperiorto the six-foldrepetition Tonal patternswereencodedto representlettersof the alphabet.
Within a given test, the tones varied with respect to a single
of the target-word.t It was alsosuggested that, in the stimulus continuum, e.g., the frequency of a tone. The extremes
low-informationnetwork, words could be replaced by of the range of the stimulus continuum furnished two steps for
tonal signals.
2 Furthermore,well-developed techniques the purposesof informationcoding.With the addition of another
of noise rejection for narrow-bandsignalsmight be intermediatestep, the continuumfurnishedthree stepsfor coding;
and with the addition of still another two intermediate steps,the
expectedto provide an extensionof communication continuum furnishedfive stepsfor coding.These three codeswill
rangewith tone-codedsignals. be referred to as binary-, trinary-, and quinary-codeddisplays.
The presentstudyexaminesthe feasibilityof employ- For example,with respectto frequency,the binary-codeddisplay
ing tonal signalsfor low rates of informationtrans- employedthe two frequencies200 and 3500 cps;the trinary-coded
missionwith naive observers.We did not attempt to displayemployedthe three frequencies 200, 750, and 3500 cps;
and, the quinary-codeddisplayemployedthe five frequencies200,
achieve the rate of transmissionpossiblewith Morse 420, 750, 1810,and 3500 cps.
codeby extremelywell-trainedoperators.Rather, we Successiveindependent selectionsof tones yield an expanded
attempted to develop a procedurewhich could be set of alternatives. For example, five successiveindependent
employedby naive observersfor very modestrates of binary selectionsyield 25 or 32 alternatives; three successive
trinary selectionsyield 33 or 27 alternatives; two successive
transmission.This aim is not trivial. There are many quinary selectionsyield 5 or 35 alternatives.
military operationalsituationsin which a modestrate The procedureof successive selectionsto yield letters of the
of informational transmission with naive observers is alphabet is demonstratedby the networks of Fig. 1. Figure !
representsthe three setsof tracing boardsemployedby the sub-
extremelyuseful,e.g.,long-rangeemergency procedures. jects in the present experiment. These boards permitted the
subjectsto trace the tonal pattern through a network to arrive
* This is TechnicalReport AFCRC TR 57-12 of the Air Force at one of 25 letters. Tracing was accomplishedby moving the
CambridgeResearchCenter, ASTIA DocumentNo. AD 146779. finger along solderstrips to successive choicepoints. The choice
This researchsupportsProject 7682 of the Air Force Research points are indicated by the large dots in Fig. 1.
and DevelopmentCommandProgram in Human Engineering.
Now at the Massachusetts Institute of Technology, Cam- An example of the subject's task follows: with five tones
bridge, Massachusetts. differingin frequency--(beep)(bip) (hap) (bop) (boop), the A
I. Pollack,J. Acoust. Soc.Am. 30, 196-201 (1958). was encoded:inthe binary tests,as (beep) (beep) (beep) (beep)
Reference1, footnote 11. (beep); in the trinary tests,as (beep) (beep) (beep); and in the

Potrebbero piacerti anche