Rated acoustic (articulatory) similarity for word pairs varying in number and ordinal position of common letters

LiiJ not dirfer in rCc'o,"nit illn whcn huth targe't :ItHj Llistractor worLis wcrc presentcLi togcthe'r during stuLly, Since Ss haLl an opportunit) to c01l1parc thc target anLi Llist rador activcl) LllIring study, they could have attcnLleLi to tllOse word features (selllantic, phonetic, associative, visual) thaI scrved 10 makc synonyms and homophones as discriminable as Iwo unrelalcd words, RLlIRFNCES .-\LJA\IS, j, A, 111Imoll memory, Ne" York: \k(;r""-lilIL 1967, A'>ISIII j), \L & K"APP, \L Assoeiation, ,ynol1~ mity. and dirl'l'lionality in falsl' reeognition, lourn"1 01' Experimental P\)'chology, 1968, 77, 171-179, Bl'SCHKI, H" & LINON, R_ Eneoding hOIllOphol1l'\ anu 'ynonym\ for verbal Ji'l'Timination anti TCl'ognition. P'Yl'honomic SeienL'l', 1969, 14, 269-270_ \!ARSIiAlL l, I" ROl'SI,:, R, 0" lR" & TARPY, R, \1. AL'OU\tk vcrsU\ a\\odatiVl' III od eh ur ,hort-term memory L'oding. P\}ehonomk SeienL'c, 1969, 14,54-55, NOR\!A", D. A, :\femur\' onJ O((ellt;OIl ..,111 ;lItroJ1Ict;OIl to humoll ;;'formot;oll procen;lIg Nell York: Wilcy, 1969, ehap, 7, IHOR'>LJIKI, I, L & LORt;1 , L /'l,,' feGcher', hOIlJhook oJ JU,UIJIJ \\'OrJ.\. ,"eil York: Bureau 01' Publkatioll\, leadle" College, Columbia L niVL'f\ity, 1944, l!NLJIRWOOLJ, Il, L & IRIT:'>iD, 1. S, Irrors in fccol!nition learnine and retention. Journal uf Lxp~rimen tal Psyehology, 1968, 78,55-63, NOTFS L This invcstigation was supported by USPHS Research Grant MH-08556 and USPHS Research Seicntist Award K3-MH-23,796 to Herman Busehke from the National Institute of Mental Health, The assistanee 01' Kirk Gibson in programming and the commcnts of Howard Ranken are gratefully aeknowledged, 2, USPHS Postdoetoral I'ellow, Fellowship No,I-102-HD-29,231-01. 3, Now at the Saul R, Korey Department of Neurology, Albert Einstein College of Medieine, Bronx, New York 10461. 4, Thc eorrection formula employed was a standard one: [P(C) - g] /(1 - g), where P(C) is thc obscrved proportion eorrect and g is the probability 01' a eorrect guess, 0,5 in this case, 5,1t is possible that the words comprising many S pairs were not complctely synonymous, so that the ~emantic content of a study word was slightly different from that of its distractor. In that case, reeognition of S pairs would be expeeted to be ~omcwhat better than that obtained when both words of an S pair are perfeet synonyms, Rated acoustic (articulatory) similarity for word pairs varying in number and ordinal position of com mon letters 1 DOUGI,AS L NEI.SON and LOu/SE D, NELSON, University o[ South Florida, Tall/pa, Fla, 33620 Ratings o[ "similari~v-in-sound" [or pairs o[ words sharing letters in various ordinal positions wcrc obtaincd using a 7-ealegory graphie scalc on whieh degrees o[ similarily were specified by verbal labels, Judgeu acoustic (arlieulatory) similarily increaseu as //l/lllber o[ shared letters illcreaseu [rom zero to onc, {Wo, alld Ihrce letters, Pairs overlapping in first letters wac rateu significantlv more similar thall pairs sharing midule ;" last letters, For pairs sharing letters within two oruinal positions, juugeu similaritv increaseu in the order of jirst-anu-milldle, jirst-and-Iast, m iddle-allu-last, Implicatiolls lor the hypo thesis thaI verbal stimuli are ('(Jueu bv pronullciation were discussed . Manipulations of ordinal position of identical letters within sets of three-Ietter words have produced relatively consistent findings thaI have dcpended UPOIl characteristics of the learning lask, When Psyehon, Sei.. 1970. VoL 19 (2) discriminations between stimuli are required, as in paired-associate (Nelson & Rowe, 1969) and serial-reeall learning (Nelson, 1969), diffieulty of acquisition increased in the order middle (M), last (L), first (F) for stimulus sets sharing letters within single locations and in the order middle-and-Iast (M + L), first-and-middle, (F + M), first-and-Iast (F + L) for sets in which leiters were shared within two ordinal positions, A similar ordering for the dual-Iocus conditions also has been reported when paired-associate stimuli were nonsense syllables (Richardsoll & Chisholm, 1969; Runquist, 1968a), When stimuli could be grouped together as in free reeall (Nelson, 1969) and when paired-associate stimuli shared leiters with their responses (Nelson & Garland, 1969), orders of difficulty by identity locus were reversed, One explanation of these results assumes t hat t he stimuli were coded by pronunciation, rendering the codes subject to acoustic (or articulatory) interference or facilitation depending lIpon the task requiremcnts (Runquist, 1968a), Accordingly, to account for the cffects of variations in locus or identicalletters in the various Icarning tasks, this hypothesis predicts Ihat rated acoustic similarity should increase in the order M, L, and F for pairs of words sharing letters within a single locus and in the order M + L, F + M, and F + L for word pairs sharing letters within two ordinal positions. The only available data '-lave been inconsistent with this prediction, Runquist (l968b) found that word pairs sharing first letters were rated as more similar than were pairs overlapping in middle or last letters, which do not diffeT. No apparent differences were found between pairs sharing letters within two positions, However, the method used for obtaining ratings required estimations of the percentage (0-100) of similarity shared between stimuli of the pair, with the similarity attribute unspecified, To the in d eterminate extent that similarity estimates were made on the basis of counting the number of common letters, as is suggested by ratings around 66% for all dual-Iocus pairs, differences as a function of locus would be attentuated, The purpose of the present study was to sc ale pairs of words overlapping in various ordinal positions with the similarity attribute specified as "sound similarity." Specification of the acoustic dimension was expected to reduce the Iikelihood of rating on the basis of letter counting and, perhaps, reveal greater differences within the overlap conditions, Moreover, ratings made on the basis of acoustic similarity should provide evidence relevant to the adequacy of the pronunciation hypothesis as an explanation for the ordinal-position effect. MATERIALS The items to be rated consisted of 108 pairs of words, Eighty-four of these pairs were generated by taking all possible pairwise comparisons within each of the following three lists: PAN, PAT, PIN, PIT, FAN, FAT, FIN, FIT; BAD, BAG, BED, BEG, LAD, LAG, LED, LEG; and HAM, HAT, HUM, HUT, RAM, RAT, RUM, RUT, This procedure produced four pairs of items wi thin each list that shared letters in all possible positions, including zero, Thus, there were 12 pairs of words representing each of the seven overlap conditions, The 12 pairs representing complete identity were generated by selecting four single words from each list and by pairing each word with itself. These words were selected so that each different initial, medial, and termin (...truncated)