-. Perceptual Analysis Using a Triadic Similarity Listening Task An Honors Thesis (ID 499) by Tina M. Henry Thesis Director Ball State University Muncie, Indiana August, 1990 Expected date of graduation, August 18, 1990 , j,. perceptual analysis .-~ 1 Perceptual analysis uSlng a triadic similarity listening task INTRODUCTION It has commonplace be(~ome discrimination tests substi tutic·ns deficiencies Research administer for sound gpeech to children exhibiting developmental order in identify to however, the past several years, sound perceptu.al possible underlying production difficulties correlation sound t,o (Locke, has 1980) . questioned between perceptual abilities ag measured by speech discrimination tests and production abilities (Broen, al. 1983). Textbooks written for use in undergraduate courses in articulation auditory et. and phonological disorders discuss the discrimination assessment testing has been given fact a that place in procedures even though empirical data supporting relationship between articulation and sound the discrimination are conflicting (Newman, Creaghead, & Secord, 1985; Weiss, Gordon. Lillywhite, 1987). have Newman, Et. al. discuss the fact that studies suggested that auditory cliscrimination abilities follow developmental progression that current tests do not account Both Newman, Et. al. and Weiss, the provide auditory more no commonly used solution tests. to the a for. Et. al. have suggested that tests measuring internal discrimination would be more useful than of & HOwever, problem of some these textbooks accuratel')r assest:;ing discrimination and still recommend some 8SSeSGment of 8 of auditory discrimination and its use in therapy. - The following study attempts to develop method perceptual analysis 2 assessing speech sound discrimination abilities in children will Before provide useful information for therapy. can he presented, of speech perception current the it is necessary to discuss the major perception and the relationship means of assessing speech sound speech between of discrimination rationale is presented for the use of a method mapping the further Finally, illustrates the need for a more representative method. a study theories In addition. an examination and production. that of perceptual that, will present a true picture of a child· s internal perceptual system. Theorie§.. Q_t ~t;;;..Qh p~.tion Opinions about the nature of speech perception have Infant studies have indicated that, inf:=mts .''is young at: differed. one month of age are able to disciminate sounds which differ by a single et. contrast such as voicing: ba/pa or place: 1971) . al., ind icati,:m that linguistically activated from the infants can al. interpret process speech patterns. According to in perception infancy. is it not linguistically relevant as sounds an in detectors", phonetic Butterfield However, others have (Eilna,,; , fact relevant manlH'!l:' and posseSt:: "feat-ure ability to perceive phontetjc stimuli is Instead, this the mc·t't extreme interpretation of Eimas attained - et by linguistic input, which extract auditory (1974) . Eimas, balga features and Cains is ~.Jorl, thatJ neRrly completely argued that as Eimas invc!l ves a simple d i:::;cr iminat1c)n of sounds this claimed. out \-)f perceptual analysls linguistic context. Some authors have argued that perception is facilitated production. The motor theory of speech perception (Liberman. at. 1967) al. , theory is The major tenet of motor speech perception is based on rules of motor exemplifies that this view. production. This the(H'Y is Menyuk AnderRon and direct contrRst Hith the views jn (1969) who believe productions on perceptual categories. that children A less extreme version theory suggests that speech production can facilitate cttas as proof the fact that chidren who have of producti~n - children other sounds can discriminate them who hand, mutually have not mastered them. suggests Bloom thRt perception and perception the quickly production than on the are two (H17f;). dependent, but different, processes which of Th t r:; mastered more of base motor theory has been proposed by Schvacken (Ingram, 1976). and by interact as language develops. Bas'2!d states The that Feat,llre Phoneme phonemes distinctive features acquisi ticm, Blache producing, product The child. possess used (1978) in Int,erac~tion Theory, interrelated elements the of stated process that, phonological "Whi Ie in comparison to the perceived cultural Cll 1 t,ura 1 representation of norm S()lmc1S that Blache which cal.led a c~h he/she is constantly monitoring his/her own "pe t'ce i ved internal - on l' c· u acoustlc nc,rm refers i ld (p.107) . to that, may be different, in It is this internal representation of sounds that is an each needs perceptual analysis 4 to be represented in order to gain a clear picture of the perceptual system of a young child. Before any study of speech perception can be performed, is first necessary to establi~h a link between speech too. has been an area of much production controver::;'.1 In a study by Eilers and Oller (1975), for researcherr;. it responses on a combination production/perception experiment performed using bet~veen subjects four categori.es in which "+" indicated the The absence of. -perception/+production. responses fe 11 ~~. Here recorded Uf3illg indicated the presence of and categories Much were: +p<~rC(3pt ion./ -perception/-production. +perception/-production. +production. and the agefj of 1: 10 and :2: to their surprise. tht:: '-percept 1.( £11 +product ion catep:ory. int~o possible explanation for this data is that there may be One o~casions Eilers occur 3.f-i .~; · ., t .,1C)[j ' i lsc:rJ.!'nJ.tl."l i. making tat'.;);;: f~P8ec:h ()f load .:.: imposed by the ':.:h 1.1d - auditorallv as childhood - J·ef~lJ.IL ~videnced F"rrOl:f.j r'C:::;:;!)J.t, th,'" 1S by his from p.:;ccfcptual dLffjcu]t.j.e:::~, et. 31 (1981) used the same four Winint;:, pPl'cep"t,inn./rt'odw.:ction {.. ert:~ l.L:;ed in explored further which the -perception/-production indicated categories results Further, exif:~tence link et.. aI. ::>.tudy ,:;f that in~o and either +perception/+production production. precedes perception their study provided of the the or little -perc:eptie,n/+pl'oductic":<D support category et (Fraser, - conclude::! that a"! that, al., 1963; From in agreement with other invpstigators Ingram 1974; Eilers perception does indeed a re8ul t, for suggesting that responses in this category may occur simply by chance. this study, Winitz, et al., for (1931) percept,Ion between They suggested that responses which fell producti'::'!!, . the Hinitz. :.'i categories production task;"'; do and Oller. precede no"t. 1976), pr()c1uction and dem()nf~trate fully underlying linguistic knowledge. In a groups tasks 1983 study by Broen. et al., of 3 year old children on both perception and for /1/, selected /r/, and /w/ were compared. of 81{ ills whi Ie the second gn:mp was demonstrating delays in of was developing chosen articulatory two production The first group on the basis that the children had normally art icu lation basis performances em the development. Results of the production tasl-:indicated that. the majority of the children Children chance in in on bot,}) the groups misarticulated normally developing group perception task testing /1/ Rnd /1'/. scored well above soundf-; t.hey invo~ving percepLual analYS1S misproduced. This was not trup of the children in the articulation There was a much greater range of performance delayed group. this on group with the range including two children with no the level was perception task and one child who on all contrasts. Broen et a 1.. SIJ~':gest, errors at performance of this Overall, considerably poorer than performed the normally developing chance group group. that, thefje findings won Id ind icate that the relationshir between development of perception and production not necessarily a symmetrical one. some may be direct 1 y 1. inked to corref';pond ing perceptual others may not be linked. may be a Likewise, requirement attainments for normal co' ]'. '-' articulatjon errors percepti:::m - Although in one::: , correct of articulatory development, bnt mi::;articulati(i{u=: c<smnnt .''l.l-.7ays be blamed on poor perceptual abilities. Assessme:o...:tL Q.f ~:peech I~f::rcept ion In order to assess speech perception, we typically use sort of speech sound discrimin~tion test (Locke, 1980) . some The:?;e tests, according to Locke, may be placed in two major categories: Type 1 and Type II tests. The Type I test allows for a comparison of the adult surface form and the considered to discrimination consist child's be test internal representation, a necessary quality of a (Locke, 1!:J8Cl). good Typjcally, of pictures or objects, whose names are and t.hlS speec}) Type I If; sound testfj represented lD perceptual analysis 7 the child ";:: receptive vocabulary. live The internally must compare a an represented category of sounds anri decide whether or not a match has been made. Type I category. A prohlem that hac the with the picture identification An he.olrd ~est been fitR identified is that i t is not possible sOlJ.nd,c: have been disc;riminated Gince the .-:;hild ~7hjch discontinue ::;()und The picture irlentification test into t,o know by t.spe or presr~nt.t:,d. with voice child a.nd a. ,.,Jord th,e ;:;earch once an acceptable trJ.3.tch been tVH:: additional problem is that some contrasts are very ITl:=ty made. difficult to represent pictorally. Type Generally, II tests require a child to compare two a child is asked to make a same-different based ()n tl\lC or more ,::;yllablef: heard in close are three task however. judgment First:, on .3 in which two syllables are presented at a a child could conceivably answer "samt:;" to all tester, forms. 3ucc'.':!~·:si(:Il. major problems with the Type II tests. same/different time, adult items. would not know whether this was a result of the child's inability to correclty discriminate aJI of the contrasts or a result undp-rstand criticized of the child's inability to attend the instructions. for the fa(~t Final1y. Type to task thA II that t.hey do not: allow for a or test,s are comparisor: of a child"s internal representation witt) the aduit form and that they do not allow for a dptermination of directionality perceptual error. These are two of the eight of quaJities the that Locke outlines for a good test of speech sound discrilnjnat.ion. 8 speech Go IdTria n - F r t:=; t,oe - Wood coc k tb h The experiment measuring yielded \.Jhat they claim 10 be rnea:=:l1ring and thrlt they may ways (Bountress, Hot 1984). AlternativAs to these standardized tests have been suggested by Hanson (1961) and by Slache (1978). Hanson performed a simple experiment in which vowel sequences were presented to twelve year old children, three vowels at a time. The children were asked to record which of the last two vowels was more similar to the first one in distance vowels then the sequence. bett-leen ea":ch Hanson VCH.. e then ectlmated the 1 pair based on the number The had been classed tc'get;her by subjects. arranged perceptual in a three-dimensional diagram in uf V(lT.Jpi which vowels were placed closer to each other than less similar Hanson claims acoustic that the diagram agrees vowel triangles. r.. i th out. - s were slmllar ones. articuLat,ory However, according to Kalgren, experiment was performed in a very crude manner (Malmberg, In times the 1970's similar experiments involving adults were and t,hls 1968). carried These experiments involved having college studellte 0stimate 9 the similarity thr~e-element sound series The use of a city-block model in which ruulti- between (Blache, two- and dimensional f:trlJctUl'ef; can be geomeLrica lly mapped waf.; snggef::ted as a way of externally representing these internal maps 1978). Based t,he Ha!ll'::on (1961) and Bl achr:: (UJ7 8 ) on ::: tucl ies , modified ven:.ion ( . f t.he triadic fcdmilarity t,e,sl>;. and <':1 frequency of 28 eliciting and response analysis have been selected representing the perceptual "'. :'Oubsequent, system means of of children in the following experiment. Sub j eG.t..Ei The subjects were 19 children ranging in age from 4-1 to 7-] with a mean age of 5-4. facility The children were enrolled at a in Indianapolis, Indiana. from preschool to 1st grade. 5-3) and 12 male were in thp. study. grade le'Jel There were 7 female subjects subject~ given pure-tone hearing They ranged in daycare (mean age, ~::'.creeningE: The screening iuc] !lch-,;;j 5~4.) pried' to bE:1ng f.,f!()B~~,1 (JOClHz, iw:.ludec] ::ClOOHz, 4000Hz tones presented at 20dR. none of l~arning the Cll f~pef'"ch , i Ie! t'!::n had any documdl ted diRabilitleB. ~_li The stimulus combinations of /1/, - items /1'/, con~jBterl of the /w/, and / j / . 24 Eac~h possible triad Ha;~ triadic 3sl=;igned perceptual analysis 10 a random number. and the triade were placed in nurnericttl or'der for presentation in order to assure randomness. The individual triads were recorded in a sound-tr~ated room. Initially, the items were recorded through a Hydrometals Cardioid A Microphone into a portable Marantz tape recorder. was placed using an Beltone This tape in between each triKd. Akai ~S-702D stereo cassette deck and Clinical Audiometer. 2000 was The first f;l ight delay then run played through element of a each triad was recorded using channels A and B of the audiometer, the second element using channel A alone. and the third using channel B alone. the These switches were recorded onto another Marantz using portable tape recorder. tape using when played a stereo tape deck. the first element of the triad played through A.E a reS1J It 1 both speakers. the second through the left speaker. and the thiri through the right speaker. Condi tiOU.llJ,g In order to condition the subjects to the triadic similarity task, two form2: of condit.ioning l.Jere lIspd, visual and The visual conditioning consisted of 3 items. presented white triangle rigbt. 1 ike consisted the triangle square - The first sequence of a white square placed in the placed on the left, and a white The subjects Here told to, eme.i n the midd Ie." auditory. middle, square on "Point, t.-) t.t"le c·ne that The next se\: cons j s ted c!f in the center, a red triangle on the left. and a ,:m the right. The same instructions were given. a the look::; a red white The perceptual analysis 11 final visual sequence consisted of a red triangle in the a red Once square on the left, and a white triangle the same instructions were given. ag.::ti.n. either If justified. trial:=:; , was w~s answer a a were the In this subject successfully completed e.;equence e.f 3 auditory tashs using subject on considered correct since either one then the second training sequence was that the firf;t sound wou.ld, were told to right. case, could each ~:,ollnd "come from the point to the sound (on the right of the f'~ffect,c. The They were midd Ie. " or left) that For each of these triads, one of the sounded like the first one. sound effects corresponded exactly to the first one heard. subject be presented. instructed to listen to three sounds. told center, successfully responded to each of the training If a stimllll, the triadic similarity listening task was initiated. For the purpose of the triadic similarity listening task, Sony The portable tape recorder with detachable speakers two create speakers were placed 1 yard apart in order the left, right, middle distinctions. instruct,:=:d that was to used. clearly The subjects were to, "Listen te, the thl':ee sounds and point to the sounds the m()st 1 ike the first one. a Subjects one were reinforced for listening carefully and responding immediately. Results Due to the fact that this was a relatively small sample, qualitative - analysis of the results was employed. Table a I perceptual analysis 12 (Appendix Al provides qualitative evidence of Blache"s model the semi-vowel system (Newman. et al .. one examines pair 5. replace /wa/ 100% of the time. it can frequency seen BI, If it can be seen that Ira! is chosen over /ja/ to be 1985) (Appendix for that, again. than is /ja/. In the corresponding pair ira/ is Once again. chosen with much 6. more if one examines pair 7, it can be seen that /wa/ is chosen over /la/ to replace Ira/ 78% replace Ira/ 73% of the time. ~ver chosen 18, /ral This Finally. in pair 17, is chosen over Iwa/ to replace Ila/ 47% of indicate that. for children in this age degree free variation. provide in support of this method of assessing the c' ~. pair time. group, distance between these sounds is small with These examples in the perceptual of , I /ra/ /wa/ to replace /la/ 52% of the time while would evidence + In the corresponding pair 8. Iwal js chosen over /IB/ the time. to .-~ L.. ·'L a the great qualitative perceptual systems of children. Discuss i·:ffi. The results of this experiment indicate that the triadic similarity listening task was a successful means of assessing the speech perception of young children. make 6.9JI)!~ or d.i.ff.~n...:t Children were not forced to judgmentt, (.. hich require nothing more that simple discriminations between sound pairs since tl1is is not what actually occurs in the process of perception (Morse, 1974). is believed to occur is actually a much more complicated of - comparing audit.ory input t.o an internal What process representation of perceptual anal ys if~ 13 sounds and making a suitable match or creating a new category a suitable similarity with task simulated this process by presenting the similar The 1978). match cannot be made (Blache, opportunity to hear sounds and to triadic the compare child them The sounds in order to identify suitable matches. 1'i" ... to fact that this experimental procedure simulated the natural process of speech perception would explain its success. This procedure which has been used with adults (Blache. had not been attempted '.lith very yonng childre:u, but 1978) is the group for which this information is the most useful the sound system is still emerging. sequence, simplified facilitated training training respc)nse pointing directions, since the use of this procedure with young children. sequence understanding of allowed the children to gain the nature of the task and the complete 0 importance t.he Bourd tbat is 1 ike the first one," e 1 im ina ted the problem of possible differt;;W.t. 1980). The use of simple direct low:;, of to a respond.ing. The "Point j mmediate The addition of a thi:::: misunderstanding of terminology such for which ot,her met.bods have been crit,icized Finally, the use of a pointing response (Locke, eliminated any intervening variables such as I'ictures which may be unfamiliar to the child (Locke, It study or verbal responses requiring additional processing U~80). was was stated in the introduction that the to develop a method of assessing goal the of this perceptual perceptual analysis 14 informatioll abilities of young chidren that would provide useful for The therapy. listening task as successful use demonstrated this study similiarity perceptual an would task in the triadic this study similarity has definite In the area of assessment. the results implicationE for therapy. of of indicate that the use of the could provide valuable information triadic the ~bout system of a child since it identifies the dimensions individual uses in making decisions about phonemes. In the area of treatment. the use of the triadic similiarity listening task with a successive frequency of response analysis (Appendix A) would provide valuable data for clinicians. The data could be used to determine the need for enrollment into speech therapy and to help identify appropriate therapy goals. child"s from frequency of response analysis deviates th:)se of other children in the same ag/; training may be necessary in order to of For instance. if the deviant perceptual system. considerably group. facilitate However, a perceptual reorganization if child'f.: a frequency response analysis is similar to those of other children in age the contraindicated since group. it perceptua.1 would appear wouLd that perception 1S developing in a normal manner. further Wit~ d i fferen t mapping sound of and standardization classes and larger groups of chi IdreJ:l, the perceptual distance of phonemes as Blache (1978), --- experimentation and proposed a by the triadic similarity task could become a wldely perceptual analysls 15 used and clinical tool. perceptual graph the children. make Standardized frequency of response maps could be developed in order internal perceptual systems of to externally normally developing These charts could he used for comparison in order decisions about the need for perceptual training selection of appropriate therapy goals. This phonemes currently which between - used simply would providf~ more uf..;eful unidimensional methods of assess the pairs of sounds. perceived and method identifying the dimensions used by a child when making about charts similarity or the of decisions info l'n,at ion assessing to than perception difference perceptual analysis 16 References E:. Blache, E.. Baltimore: University Park Press. (1974). Bloom, L. Talking. understanding, In and thinking. ~felblJ.f.i...Qh. Bountress, !\.. G. A second look at tests of speech- (1984). 349-359. Broen, P., Strange, W., Doyle, S., & Heller. J. (1983). Perceptien and production of approximant consonants by normal and art.iculation-delayed prel3:chool children. and llihlLing lieseal.:£::h. z'6_, 601-608. ~b. Butterfield, E.C., & Cairns, G.F. reception research. Edwards, M. Eilers, In (1974). Dlscussion summary - Schjgfe~lilllLGh. PercepticHl and production in chi Ie! (1974). phonology: Child ~lolJ..rnEi.l ()..!' the testing e·f four hypotheses. LftnRJ.li!.g~, 1, Jm.u::na.l Qf 205-220. R. 8[ Oller, D.K. (197!::,). The r(;le of :,,;peech discriminaticHl in develc:.pmental sound subgti tutiow:; . .]olJrn.;U Q.f. Child h;'iIl&.~, Eimas, P.D., Siqueland, (1971). ~L ~-l19-329. E.R., Jusczyk, R., and Vigorito. Speech perception in infants. ~;ci~n.~~, J. 11.1. 303-306. Fraser, C., Bellugi, U., & Brown, R. (1983). Control of grammar in im:L tat-ion, comprehension and production. 1LQJ.ll:rl.§.1. Qf - infant perceptual analysis 17 - . Ys;;Lb.g.L L~.r.u:.rUng. Ingram, D. ami (1974). Yf~.rQal BehgyiQL l. The relationship between comprehension In R.L. Schiefelbusch & L.L. and product,ion. La.llillJ..2£~ ~.:r.t;.~~i.Y.t;;.fL=_a C9 u. i e..i.:t".iQ U->.. int~..r.Y...tl.!.ti.QJ1.. 1 :.~ 1··13[:, . BB 1 timore: Lloyd (Eds.). rf;;.t.:a:rda,:t.;i[.;.I!-,- .?.nd. Oni verf; i ty Park PreFjE:. Liberman, A.M .. Cooper. F.S .• Shankweiler, D., & StuddentKennedy, M. (1967). (19BO). Locke, J.L. Perception of the speech code. The inference of speech perception in the phonological} y disordered child. HeariLW. [:~~der.Q.. JJ·dJJ.l".lli:1.1 ~).i. iiP_e_t).c::h ';;!ll.d 4.5.. 4:31-468. Malmberg. B. Holland Fublishing Company. Menyuk. P .. & Anderson. S. and reproduction of /w/. and. tiearin&:. Morse, P.A. R~flearch, (1972). stimuli. (1969). Children"s identification /1'/. and /1/. J.Q1lnllil Q_f ;:'f'e~.Ql.l 12.. 39-f)2. The discrimination of speech and nonspeech .J ()1J.r.D..al Qi. m'f,; r i IT! e n t.Q.l ~:;tLO 1 qgy ,L4. , 477-432. Newman, P .. Creaghead, N.. & Secord. W. and K'!;::me.i.1iatl.Q1l 9..i. arth;llu.t. .,,!;:..:l 9..llil p-hr,2J1cd.QKif;;;.a.l d..LQ.Qr.d.e.rs.. Columbus. Ohio: rharles E. Merrill Publjghing Cc·mpany. Weiss, C., Gordon, M.S., & Lillywhite, H. !llQ.nQg.=m~D.t Baltimore: ~U. (1987). C..lhUcal £1Lticu latQ..LY GIld. Ph.QI1Qli';2g..1Q;:-1.1 diQ.Q.n::lerl'i... Williams & Wilkins. Winitz, H., Sanders, R., & Kort, J. (1981). Comprehension and perceptual analysls 18 productie,n of the /-ez/ p lu.ral allomorph. ~~hcli.r;guistiQ. ~~.ar.Qh, - ill, 259-271. - APPENDIX A TABLE 1 FREQUENCY OF RESPONSE I - A I B 1- wa-ra-la .737 .263 2. wa-la-ra .579 .421 3. wa-ja-la .316 .684 4. wa-la-ja .526 .474 5. wa-ra-ja .100 0 6. wa-ja-ra .263 .737 7. ra-wa-la .789 .211 8. ra-la-wa .263 .737 9. ra-ja-wa .737 .263 10. ra-wa-ja .421 .579 11- ra-ja-la .474 .526 12. ra-la-ja .632 .368 13. la-ja-ra .421 .579 14. la-ra-ja .526 .474 15. la-ja-wa .368 .632 16. la-wa-ja .526 .474 17. la-ra-wa .526 .474 18. la-wa-ra .526 .474 19. ja-wa-.la .474 .526 20. ja-la-wa .316 .684 21- ja-wa-ra .263 .737 22. ja-ra-wa .632 .368 23. ja-la-ra .158 .842 24. ja-ra-la .684 .316 Appendix B 0 :-Special ,,:5 S E M I V 0 W E L ,0 -PLACE- 2