PERFORMANCE OF ARTICULATORY- IMPAIRED CHILDREN ON SEVERAL PSYCHOLINGUISTEC MEASURES 1. Yhesis for the flame of Ph. D. MICH3GAN STATE UNIVERSITY PAMELA SEARS RIEDL 1975 Linux)! This is to certify that the thesis entitled Performance of Articulatory-Impaired Children on Several Psycholinguistic Measures presented by Pamela Sears Riedl v has been accepted towards fulfillment of the requirements for 1 21;; D .degree in AM” I and : Speech Sciences 1 Major professor DMe September IQ; 1975 v ‘ ec~x BINDERY L ' L Murry a ABSTRACT PERFORMANCE OF ARTICULATORY-IMPAIRED CHILDREN ON SEVERAL PSYCHOLINGUISTIC MEASURES BY Pamela Sears Riedl Research has suggested that children with moderate to severe articulation problems tend to show inadequate audi- tory discrimination skills, and, when compared to normal speaking children on various language measures, show a less complex grammatical structure, poorer language comprehension. shorter spoken sentences, and so on. Further research was warranted to clarify the auditory discrimination skills of articulatory-impaired children and to eXplore the rela- tionship between articulation, language, and auditory dis- crimination of these children. The purpose of this study was to compare the responses of first grade articulatory-impaired and normal speaking children (1) to an auditory discrimination test and (2) to a sentence imitation task involving the factors of sentence length, sentence type, and word type. Forty first-grade children selected from elementary schools in the Ingham County Intermediate school district of Michigan served as subjects. Twenty, 13 boys and 7 Pamela Riedl girls, mean age 6.3 years, had normal speech, language and hearing. The remaining twenty, also 13 boys and 7 girls, mean age 6.3 years, were selected from speech therapy classes from the same schools as the normal speakers, and had exhibited three or more articulation errors on McDonald's figgeening Qggp Egg; 9; égticulation. Subjects were tested on an individual basis. At the beginning of the test session, the subject was administered the wepman Auditory Discrimination IEEE- This test was chosen since it examines a child's ability to distinguish between word pairs which differ both phonem- ically and linguistically. This test was followed by the primary experimental stimuli of S4 sentences of which 18 were well-formed, 18 were anomalous, and 18 were ill- ordered in nature. Of each set of 18, 6 were 3 words in length, 6 were 5 words in length, and 6 were 7 words in length. Functor and contentive word types were present in each stimulus sentence. The sentences were presented by tape recorder in sound field to each subject. Stand- ardized instructions were read by the eXperimenter re- questing the subjects to repeat exactly what they heard. Errors of omission, substitution, addition, and word re- versal were analyzed by a multifactor ANOVA routine. Results indicated that (l) the auditory discrimination skills of the articulatory-impaired children were inferior to those of the normal speaking children, and (2) the main factors of articulation, sentence type, sentence Pamela Riedl length, and word type and their interactions were signi- ficant at various levels with respect to the error types observed. Articulatory-impaired children generally made more errors than normal speaking children, particularly for five- and seven-word well-formed and anomalous se- quences. Articulatory-impaired and normal speaking children were affected in a similar fashion by the factors of sentence type, sentence length, and word type. Generally, errors increased as sentence length increased. Errors were highest for ill-ordered sequences, followed by anomalous and well-formed sequences, respectively. More errors occurred for the functor word type than for con- tentive word type. These trends paralleled those of earlier studies. Since error rates for seven-word ill- ordered sequences were highest for both groups of subjects, seven items were considered to be past the automatic recall ability of the subjects and, therefore, subjects' short-term memory recall abilities did not appear to differ, regardless of their articulation proficiency. Since normal speaking children did show better recall than articulatory-impaired children for five- and seven- word well-formed and anomalous stimuli, the difference .in performance was attributed to a difference in subjects' language competence rather than a difference in short- term memory recall ability. Normal speaking children appear to be better able to use the semantic and/or Pamela Riedl syntactic structure present in well-formed and anomalous sequences to recall the longer phrases. Based upon the results of this study, suggestions were offered for future research which will more thor- oughly test the recall abilities of misarticulating children and further describe the relationship between the areas of auditory discrimination, phonology, syntax and semantics. PERFORMANCE OF ARTICULATORY-IMPAIRED CHILDREN ON SEVERAL PSYCHOLINGUISTIC MEASURES by Pamela Sears Riedl A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Audiology and Speech Sciences 1975 To My Husband, Helmut, and My Parents 11 ACKNOWLEDGEMENTS I would like to express my gratitude to Dr. D. S. Beasley, Assistant Chairman of the Department of Audiology and Speech Sciences, who provided much time and effort as my thesis advisor. For their time, suggestions, criticisms, and general contributions to this dissertation, my thanks go to my committee members: Drs. D. Beasley, L. Deal, P. Walsh, and L. McIntyre. I also wish to express my thanks to Dr. J. Gill, Professor in the Department of Dairy Science, who was of great help concerning the statistical design and analysis of my dissertation. I am indebted to the following for making it possible to carry out my research in the Ingham Intermediate School District of Michigan: David Haarer, Director of Special Education, and Kenneth Woodring, Special Assistant to the Director in the area of Speedh and Language Impaired. I acknowledge my sincere appreciation to the speech and language pathologists whose c00peration and support were essential to this research projects Diane Wilson, Mary Chester, Betty Ferry, Sue Crino, and Margaret Palmer. Finally, I would like to express my love and grate- fulness to my husband, Helmut, whose support and encour- agement was fundamental to the successful completion of this dissertation. 111 LIST OF LIST OF CHAPTER TABLE OF CONTENTS TABLES O O O O I O O O O O O FIGURES O O O O O O O O O O O I. INTRODUCTION. . . . . . . . . . II. LITERATURE REVIEW . . . . . . . . Language Studies of Misarticulating Chi late" 0 O O O O O O O I O Auditory Discrimination Studies of Misarticulating Children . . . . Short-Term Memory and Imitation Tasks. Statement of the Problem . . . . . III. EXPERIMENTAL PROCEDURES . . . . . . Subjects . . . . . Design and Stimul . . Presentation Procedures Analysis. . . . . . IV C RESULTS 0 O O C O C O O O O O 1. Auditory Discrimination 2. Functor Omissions . . Articulation . . . Sentence Type. . . . . Articulation x Sentence Length Sentence Type x Sentence Length. Sentence Type x Sentence Length X MtiCUlatione e e e e e 3. Contentive Omissions . . . . . Articulation x Sentence Length . Sentence Type x Sentence Length. 4. Punctor Substitutions. . . . . Articulation x Sentence Type. . Articulation x Sentence Length . Sentence Type x Sentence Length. iv Page vi 16 22 32 36 36 37 37 39 40 4O 42 45 45 45 47 55 61 62 64 69 71 76 78 5. 6. 7. 8. Contentive Substitutions. . . . MtiCUlation e e e e e e e Sentence Length . . . . . . Sentence Type x Sentence Length. Rever8a18. e e e e e e e e Sentence Type. . . . . . Sentence Length . . . . Sentence Type x Sentence Length FUHCtor Additions e e e e e Contentive Additions . . . . v. DISCUSSION . C O O O O O O O O Auditory Discrimination . . . . Sentence Length . . . . . . . Sentence Type . . . . . . Recall Patterns for Sentence Types. Word TYpe e e e e e e e e e Articulation . . . . . . . . Implications for Future Research . VI. SUMMARY AND CONCLUSIONS . . . . e . Auditory Discrimination . Punctor Omissions . . . Contentive Omissions . . Functor Substitutions . . Contentive Substitutions . Reversals . . . . Additions- Functor and Contenti Conclusions e e e e e e euousasu«uue ..m.z .ucaxeone adEuocv nodusasUAUHd an «oeuoouo:aue .asoHsEocs .poESOu:HHo3V om>a oucoucom xn A35 .ouo3lco>oe .3m .ou03:o>«u .zm .ouo: :oeuzuv nuchq oucoucom .uouoom nouum ousucouuea SS0: codeaeeo nouocsm .m ensue» 6.363;: 322554 awesomufias eouoeuorfiu 332554 ooeuouados 3h 3m 3M 35 3m 3m 3h 3m 5:” 35 3m 3m 3h 3m 3m 3h 3m 3m _. S . , .fi OWN A 7 .l 5 L a _ a a I e .. omm ,- u 9 3 L o _ , a ll. 3 A v a. O? u a S w a. cm L a OH.“ .mOz .3 s i OH on On 0? Om mm 380883 50 SOVLNHOHHd NVSN 58 .ooEuomuaaozv oaks oucoucom ouoz:co>mw ou03:o>«m ou03:oou£a OH 4 m3 OH 4 .m3 0H 4 he, :3 i], i IL L :owm 3 3 ill 3 e a omm m S L o .l a T 3 . cam _ O 8 3 ..0m .HJN a.mn uuoz:co>em On .A.H.< .oouueofid :>u0ugasu«uus ..m.z .mcexseoa Hassocv caeueasonuu< >A Aouoz:cs>ee .ouoz Io>eu .puOSISSSSuV Samson eocoucem >9 «OH .oouevuo:aaa .4 .u30aeeoce .mz .nouoom Sousa essence cs0: codameeo nouoczh UkO3IO>d h .5 seamen ouoz:eeuna 4 N3 OH 4 hz OH < M3 .m.z L ___A. CA .on flow Em 880883 50 SOVINSOHSd NVSW 59 In the anomalous sentence category, 3W strings again showed the lowest mean percent error for both groups of subjects. For the two articulation groups, a significant increase in functor omissions occurred when length in- creased to 5 words. However, means for SW and 7W anomalous sentences did not significantly differ: that is, a further increase in length from 5 to 7 words did not cause a correSponding increase in functor omissions. The A.I. children made significantly more functor omissions than N.S. children at each sentence length in the anomalous category, a finding which suggests that their use of such syntactic cues was less effective. The IO sentence category provided a different error pattern. Means for each articulation group increased significantly as sentence length increased. The A.I. children made significantly more functor omissions than the N.S. children for 3W- and 5W-IO strings. However, mean error scores for 7W-IO repetitions were the highest for both groups and did not significantly differ from one another. Both means approached or crossed the 50% omission point. Therefore, both articulation groups had equal difficulty recalling functor words when the stimulus was 7 items in length and ill-ordered in sentence type. Figure 7 presents the error means grouped by sentence types within sentence length. Table 7 explains which means were found to be statistically different. In both the N.S. and A.I. groups, no means of sentence types within the 3W length differed significantly. Of the 60 SW strings, the two groups again Showed similar error patterns. Errors increased as the type changed from WF to A to IO, and again, the N.S. group had lower error means than the A;I. group. In the N.S. group, the 5W-WF mean was significantly different only from the SW-IO mean. For the‘A.I. group, the SW-WF mean differed significantly from both the 5W-A and -IO means. For the A.I. group at the SW level, removal of semantic structure caused a shift in errors: removal of syntax had a lesser impact since the means for A and IO strings did not significantly differ in either articulation group. Since the N.S. group's mean for 5W-WF and 5W-A did not significantly differ even though an increase in error scores was observed, it is possible that normal Speakers used the remaining syntactic cues in the anomalous sentences to better advantage than the A.I. group, whose means for 5W-WF and A did differ. The 7W-string error pattern was similar for both articulation groups. Mean percentage of errors increased as sentence type changed from.WF to A to IO. In both groups, the 7W-WF mean did not differ from the 7W-A mean. Finally, the two groups' means were compared for each of the nine sentence type x sentence length cate- gories using the Tukey test. The results are shown on Table 7. The values of these means can be found on Table 4. The means for 3W sentence types did not significantly differ even though the A.I. subjects' means were higher. A11 means for SW sentences of each type were found to differ at p = 0.05. The means for 61 7W-WF and 7W-A of the two groups also differed at p = 0.05. The 7W-IO means, however, were not significantly different. Thus, both groups found 7W-IO strings equally difficult to repeat, a finding which supports the assumption of equal memory Spans among the subjects used in the study, regardless of articulation proficiency. Therefore, whatever recall processing strategy(ies) was used by the normal Speaking children which helped them perform significantly better than the articulatory-impaired children was something other than automatic memory Span ability. The language structure cues present in the other sentence type categories deserve consideration. The fact that N.S. and A.I. children made similar numbers of errors on 3W- and 7w-IO sequences but made significantly different numbers of errors on 5W-IO sequences further indicates a difference in the two groups' processing strategies. This difference in processing of verbal material again reflects, as suggested above, an existing language problem in articulatory- impaired children. 3. Contentive Omissions The analysis of variance table for this dependent variable's relationship to articulation, sentence type and sentence length factors is summarized in Table 8. All main effects were significant, indicating that articulation proficiency, sentence type, and sentence length had an effect on the occurrence of contentive 62 Table 8. Results of ANOVA for Contentive Omission Errors. Source df Mean Square Significance* Articulation (A) 38 130.224 p a 0.004 Sentence Type (ST) 2 4961.229 p¢::0.0005 Sentence Length (SL) 2 8790.389 p¢=:0.0005 A x ST 2 156.342 n.s. A x SL 2 431.012 p = 0.001 or x SL 4 3849.038 p<0.0005 A x ST x SL 4 88.778 n.s. n.s.: not significant omissions. The interactions of Articulation by Sentence Length and Sentence Type by Sentence Length were significant and will be discussed below: Articulation by Sentence Type and Articulation by Sentence Type by Sentence Length were not found to be significant interactions. Table 9 con- tains the mean values for contentive omission errors for the various interactions. Articulatign x Sentence Length The articulation groups differed in their reSponses to the three sentence lengths. Viewing the mean error values from the second section of Table 9, one can see that contentive omissions were less frequent for both articulation groups for 3- and S-word sentences but increased for 7-word sentences. The A.I. group has higher means overall, but their error mean increased to (53 Table 9. Mean Percent Error Scores for Contentive Omissions for the Factors of Sentence Type (well-formed: WF, anomalous: A, ill-ordered: IO), Sentence Length (three-word: 3W, five-word: 5W, seven-word: 7W), and Articulation (normal Speaking: N.S., artic- ulatory-impaired: A.I.). Articulation x Sentence Type: not significant fit A 19 Overall Mean n.s. 2.8 2.4 V 12.3 '5.9 A.I. 3.9 6.8 17.8 9.5 Overall Mean 3.3 4.6 15.1 Articulation x Sentence Length: significant 3W éfl 1y Overall Mean N.S. 2.1 1.9 13.5 5.9 A.I. 3.5 3.4 21.5 9.5 Overall Mean 2.8 2.7 17.5 Sentence Type x Sentence Length: Significant 32_ pg 1W Overall Mgan WP 3.5 1.4 5.1 3.3 A 2.5 1.5 9.8 4.6 IO 2.3 5.1 37.7 15.1 Overall Mean 2.8 2.7 17.5 Articulation x Sentence Type x Sentence Length: not sign. W? A IO Overall Mean Overall 3E é! 1E Mean gig; egg; gig; A.I. N.S. A.I. gig; all; 3.7 3.3 0.8 1.9 3.8 6.3 2.8 3.9 1.7. 3.3 0.0 3.1 5.7 14.0 2.4 6.8 0.8 3.7 5.0 5.3 31.2 44.3 12.3 17.8 2.1 3.5 1.9 3.4 13.5 21.5 64 a greater extent than the N.S. group for the 7w sentences (see Figure 8). Post hoc testing was conducted on these means for within- and between-group mean comparisons. Results are presented in Table 10. Error means for 3W and SW sentences did not signifi- cantly differ within or between articulation groups. Both groups made significantly more contentive omissions in 7W-string repetitions. The 7w-error mean for the A.I. children was significantly higher than the 7W-mean error score of the N.S. children. This error pattern differs from the way in which sentence length affected functor omissions, where both groups' mean scores increased significantly as length increased and the means differed significantly at each sentence length. Both groups recall contentives better than functors at every sentence length. As was the case for functor omissions, most errors occurred for both articulation groups for 7W strings, indicating again that the 7W-item stimulus affected the retention abilities of both articulation groups. Sentence Type x Sentence Length This was significant at p<=:0.0005 and the means are shown on Table 9. Figure 9 displays the subjects' error means for each sentence type within a sentence length, and Figure 10 shows means for sentence lengths within a sentence type. Post hoc testing of the means provided 65 300 m 6 m ---- A.I. a: “1 NOS. ‘5‘ 20“ r---1 m I u I 4: n I E4 z I l m 2 [fl I m ' I an 100 ' I ' I g ' I 2: : I 7 I - 1 L I l Three—Word Five-Word Seven-Word Figure 8. Contentive Omission Mean Percent Error Scores: Sentence Length (three-word, five-word, seven- word) by Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Table 10. 66 Post Hoc Testing Results for Contentive Omission Mean Error Comparisons: Sentence Length (three-word, five-word, seven—word) by Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Mean Pairs Within N S A I Articulation Groups ° ' ° - 3w vs. 5w n.s. n.s. 5W VS. 7W ** we 3w vs. 7w at a. Mean Pairs Across Articulation Groups N's“ vs. A-I- 3w Means n.s. SW Means n.s. 7w Means we neSo' nOt ** p = 0.01 significant 40V w .9 N 4% 10. MEAN PERCENTAGE OF ERRORS _‘I—r-1 [—1x 67 WF A IO WF A IO WF A IO Three-Word Five-Word Seven-Word Figure 9. Contentive Omission Mean Percent Error Scores: Sentence Type (well-formed: WF, anomalous: A, ill-ordered: IO) by Sentence Length (three- word, five-word, seven-word). 68 40 V 1' I 2 2 3° 4. m m m 0 3 £520~b Z Id U m m A Z 5, 10 ,. z 7 L— 3w 5w 7w 3w 5w 7w '3w 'sw 7w Well—Formed Anomalous Ill-Ordered Figure 10. Contentive Omission Mean Percent Error Scores: Sentence Length (three-word: 3W, five-word: 5w, seven-word: 7w) by Sentence Type (well- formed, anomalous, ill-ordered). 69 the results in Table 11. The means of all sentence types within the 3w and SW sentence length categories did not differ significantly from one another: all error means were low. In contrast, the means of all sentence types in the 7w category differed significantly from one another. Therefore, when children had to repeat sentences of 7 words in length, the factor of sentence type became important. Well-formed sentences were easiest to recall, followed by anomalous and ill- ordered sentences. In other words, as language infor- mation was gradually removed, repetitions became less complete for all subjects, regardless of their articulation proficiency. As shown in Table 11 and Figure 10, the length of the stimulus sequence did not have any effect on well-formed sequences. For the anomalous and ill-ordered sentence types, 3W and SW sentences were repeated with relatively few contentive omission errors. However, 7w-A and 7W-IO sequences were repeated with significantly more contentive omissions. Length, therefore, affected the responses for these two sentence types. The combination of the loss of semantic and syntactic structure plus the increase in stimulus length to 7 items brought about a 40% loss of the contentive words in ill-ordered sequences. 4. Functor Substitutions Substitutions occurred less often than omissions. A summary of the analysis of variance table for the Table 11. 70 Mean Error Comparisons: formed: WF, anomalous: A, ill-ordered: IO) by Sentence Length (three-word: seven-word: 7W). Post Hoc Testing Results for Contentive Omission Sentence Type (well- 3W, five-word: 5W, Sentence Type Means 3W 5W 7W WF V8. A “.8. n08. * A vs. 10 n.s. n.s. ** WF vs. IO n.s. n.s. ** Sentence Length Means WF A IO 3W vs. 5W n.s. n.s. n.s. SW vs. 7W n.s. ** ** 3W vs. 7W n.s. ** ** n.s.: not significant * = 0.05 ** p = 0001 7l dependent variable is shown in Table 12. Mean values for this error type are shown in Table 13. Table 12. Results of the ANOVA for Functor Substitutions. Source of Variance df Mean Square Significance* Articulation (A) 38 314. 778 p<0. 0005 Sentence Type (ST) 2 715.542 p-=:0.0005 Sentence Length (SL) 2 1313.705 p¢=:0.0005 A x ST 2 217.779 p = 0.035 A x SL 2 186.588 p a 0.049 ST x SL 4 449.619 p.=:0.0005 A x ST x SL 4 134.718 n.s. n.s.: not significant All main effects and two-way interactions were signi- ficant: the one three—way interaction was not significant. Again, by discussing the interactions which were signifi- cant, it will be possible to determine how the main factors affected the subjects' performance with respect to functor substitutions. Agticulation x Sentence Type The two articulation groups' performance varied with respect to the factor of sentence type. The mean error scores for the subject groups are shown in Table 13 and Figure 11. Table 14 shows the results of the post hoc testing. 72 Table 13. Mean Percent Error Scores for Functor Substi- tutions for the Factors of Sentence Type (well- formed: WF, anomalous: A, ill-ordered: IO), Sentence Length (three-word: 3W, five-word: 5W, seven-word: 7W), and Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Articulation x Sentence Type: significant W: A IQ Overall Mean N.S. 1.7 6.5 5.1 4.4 A.I. 11.5 15.9 10.1 12.5 Overall Mean 6.6 11.2 7.6 Articulation x Sentence Length: significant 3g 5W 1W Overall Mean N.S. 0.8 6.9 5.5 4.4 A.I. 9.9 16.9 10.8 12.5 Overall Mean 5.3 11.9 8.2 Sentence Type x Sentence Length: significant 32 ' 53 15 Overall Mean w? 4.6 8.5 6.7 6.6 A 4.6 18.7 10.4 11.2 10 6.9 8.5 7.5 7.6 Overall Mean 5.3 11.9 8.2 Articulation x Sentence Type x Sentence Length: not sign. .32! 38 Z! LEI £35; 5&1; N.S. A.I. N.Sl, ALIA gig; A;l; WP 0.8 8.3 2.5 14.6 1.7 11.7 1.7 11.5 A 0.0 9.2 12.5 25.0 7.1 13.7 6.5 15.9 10 1.7 12.1 5.8 11.2 7.9 7.1 5.1 10.1 Overall 0.8 9.9 6.9 16.9 5.5 10.8 Mean 73 m m 2 O 232 I ‘5 r"—-1 m u I I 5 I I 210 . F""'I _-- o I ' I m I I I m I I ' I , I E ' I I I I ’ 2 ‘ l I_ I I I I Well-Formed Anomalous Ill-Ordered Figure 11. Functor Substitution Mean Percent Error Scores: Sentence Type (well-formed, anomalous, ill-ordered) by Articulation (normal Speaking: N.S., articulatory- impaired: A.I.). m m 0 85 mZOT m I I 2 ' ' """ A010 5.3 I I I ' I l I I I ' I g I : : I ' I z _""I I I I I I Figure 12. Three-Word Five-Word Seven-Word Functor Substitution Mean Percent Error Scores: Sentence Length (three-word, five-word, seven-word) by Articulation (normal speaking: N.S., articulatory- impaired: A.I.). 74 Table 14. Post Hoc Testing Results for Functor Substi- tution Mean Error Comparisons: Sentence Type (well-formed: WF, anomalous: A, ill- ordered: IO) by Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Sentence Types N.S. A.I.* WF vs. A ** ** A vs. IO n.s. ** WF vs. IO * n.s. N0§0 V8. AOL WF Means ** A Means ** IO Means n.s. * p = 0.05 ** p = 0.01 n.s.: not significant The anomalous sentence type contained the most functor substitutions for both the N.S. and A.I. subjects, as can be seen in Figure 11. The anomalous mean for the A.I. children was significantly higher than the WF and IO means, whereas the WF and I0 means did not differ from one another. In contrast, the N.S. subjects' means for A and IO sentences did not significantly differ. For both groups of subjects, loss of semantic sense caused an increase in functor substitutions. Loss of sentence structure and semantic sense together brought the mean percentage of error back down to the WF level for the A.I. children. Therefore, the A.I. children 75 treated IO sentences differently from the A sentences and, perhaps, processed IO sequences in a manner similar to WF sentences. Since errors were similar for WF and IO sequences, A.I. children demonstrated poorer language processing abilities than the N.S. children. The following are some examples of functor sub- stitutions that were observed: 1) Stimulus: fig; big brother sings his black shoe. Response: My brother wears eggs shoes (and what)? 2) Stimulus: Her bike eats a3 apple. Response: Her bike eats up apple. 3) Stimulus: His sister bakes a book. Response: Her sister bakes a book. Other substitutions included: a/an, a/the, his/the, his/her, and a/her. In 1) above, the repetition changed the type of sentence from anomalous to well-formed. The first possessive pronoun was simply changed to another possessive pronoun. The second substitution, some/his, changed a possessive pronoun to an adjective. The change is not readily eXplained: it may be due, in part, to the change in verb from sings to wears. In 2) above, up/an was possibly due to an effect of coarticulation. The /n/ in 23 might not have been perceived and the vowel in a3 became mixed with the vowel in apple. The p of apple helped form the word up. In 3) above, her/his occurred. This was a common substitution as well as his/her. It is possible that, for these children, such pronouns may be interchangeable. It is also possible 76 that these substitutions were made more often by one sex than another, i.e., more male subjects than female subjects. This should be investigated in future research. Referring again to Table 14, the two subject groups showed significantly different means for WP and A sen- tences but not for IO strings. In other words, sentences in which some language structure was present, either semantic or syntactic, were repeated by the N.S. children with significantly fewer functor substitutions than were observed in the responses of A.I. subjects. Ill-ordered sentences, however, were repeated by all subjects with a similar number of functor substitutions. Articulation x Sentence Length The two subject groups performed differently with respect to the factor of sentence length. Actual mean error values are shown on Table 13 and Figure 12. Post hoc testing was conducted on these scores and results are given in Table 15. Both articulation groups produced the most functor substitutions in SW sentences, followed by 7W sentences. The fewest number of errors occurred in 3W sentences, although for A.I. subjects, the differ- ence between the 3W and 7W means was not statistically different. The N.S. group produced similar numbers of substitutions for SW and 7W sentences. The A.I. subjects produced significantly more functor substitutions than N.S. subjects for 3W and SW sentences. Seven-word sentences were repeated by the subject groups with 77 Table 15. Post Hoc Testing Results for Functor Substitu- tion Mean Error Comparisons: Sentence Length (three-word: 3W, five-word: 5W, seven-word: 7w) by Articulation (normal speaking: N.S., artic- ulatory-impaired: A.I.). Sentence Lengths N.S. A.I.* 3W vs. SW ** ** 5w vs. 7W n.s. ** 3W V80 7W . ** nos. N.S. vs. A.Iy 3w Mean Pairs ** 5W Mean Pairs ** 7W Mean Pairs n.s. n.s.: not significant as p = 0.01 78 relatively equal numbers of errors. Sentenge Type x Sentence Length Figures 13 and 14 group the subjects' error means according to sentence type and sentence length, respectively. Figure 13 shows that most functor substitutions occurred in SW anomalous sentences, followed by 7W anomalous sentences. The other means were fairly close with the fewest number of errors occurring in 3W anomalous sequences. Table 16 shows the post hoc testing results. No error means differed significantly in WF and IO sentences of all lengths. Functor substitutions occurred most often in SW and 7W anomalous sentences. All anomalous means differed significantly from one another. In terms of length as a factor, the means for 3W sentences and 7w sentences of all types were not signifi— cantly different. Five-word sentences had the highest means for every sentence type. Five-word WF and IO means did not differ: both were significantly lower than the SW anomalous mean. 5. Contentive Substitutgons The ANOVA table for this error type is shown in Table 17. The main factors of articulation and sentence length were significant, whereas sentence type was not. There- fore, articulation groups varied with respect to conten- tive substitutions for the 3 sentence lengths. Since the factor of sentence type was not significant, the type of stimulus had little independent effect on the retention of 79 30V m m 0 m m m a. 204h O I“. S a z m i , 8 10.. "—'I (5:. J—‘" . l _I z z 3W 5W 7W 3W 5W 7W 3W 5W 7W Well-Formed Anomalous Ill-Ordered Figure 13. Functor Substitution Mean Percent Error Scores: Sentence Length (three-word: 3W, five-word: 5W, seven-word: 7W) by Sentence Type (well-formed, anomalous, ill-ordered). m m 0 g; 20W m ..___1. m 0 £3 a z 10”- Ia. m F-—-! o .r___3 I . m T—"' ""”' m A , I ‘ 2 I 1 __ WF A IO WF A IO WF A IO Three-Word Five-Word Seven-Word Figure 14. Functor Substitution Mean Percent Error Scores: Sentence Type (well-formed: WF, anomalous: A, ill-ordered: IO) by Sentence Length (three- word, five-word, seven-word). 80 Table 16. Post Hoc Testing Results for Functor Substitution Mean Error Comparisons: Sentence Length (three- word, five-word, seven-word) by Sentence Type (well-formed, anomalous, ill-ordered). Within Sentence Type Means Tested* 3W vs. 5W 5W vs. 7W 3W vs. 7w 1. Well-Formed n.s. n.s. n.s. 2. Anomalous ** ** ** 3s Ill-Ordered {I.S. n.s. “.80 Within Sentence Length Means Tested Wngp. A A vs. IO WF vs. IO 1. Three-Word n.s. n.s. n.s. 20 Five-Word ** ** nose 3. Seven-w ord n.s. n.s. n.s. ".808 “Ct ** p = 0,01 significant 81 Table 17. Results of the ANOVA for Contentive Substitutions. Source of Variance df Mean Square Significance* Articulation (A) 38 16.647 p = 0.05 Sentence Type (ST) 2 3.963 n.s. Sentence Length (SL) 2 95.569 p.=:0.0005 A x ST 2 4.683 n.s. A x SL 2 17.685 n.s. ST x SL 4 90.189 p<0.0005 A x ST x SL 4 17.692 n.s. n.s.: not significant the original contentive words. The significant interaction will be discussed. Mean error scores are provided on Table 18. Artigulation Articulation was a significant main factor, indicating that the two groups of subjects differed in performance with respect to contentive substitutions. The articulation means, 1.7% and 2.6%, were found to be significantly differ- ent, with the articulatory-impaired children making more contentive substitution errors. Sentence Length The main effect means for sentence length were tested for significant differences. The means for 3W and SW sentences, 1.9% and 1.3% respectively, were significantly 82 Table 18. Mean Percent Error Scores for Contentive Substitutions for the Factors of Sentence Type (well-formed: WF, anomalous: A, 111- ordered: 10), Sentence Length (three-word: 3W, five-word: 5W, seven-word: 7W), and Articulation (normal speaking: N.S., artic- ulatory-impaired: A.I.). —‘ Articulation x Sentence Type: not significant g: A_ T9 Overall Mean N.S. 1.6 1.7 1.8 1.7 A.I. 2.5 2.9 2.2 2.6 Overall Mean 2.0 2.3 2.0 )— Articulation x Sentence Length: not significant 3!, pg 13 Overall Megn N.S. 1.9 0.8 2.4 1.7 A.I. 1.9 1.9 3.8 2.6 Overall Mean 1.9 1.3 3.1 Sentence Type x Sentence Length: significant -;gl pg 1!, Overall Mean WP 3.3 0.7 2.1 2.0 A 1.4 0.8 4.7 2.3 IO 1.0 2.5 2.5 2.0 Overall Mean 1.9 1.3 3.1 Articulation x Sentence Type x Sentence Length: not sign. A V 3s 5s 21: “E311 MLALLLieA—LLS—oAiELLA; WP 3.7 2.9 0.0 1.4 0.9 3.2 1.6 2.5 A 1.7 1.2 0.0 1.7 3.5 5.9 1.7 2.9 10 0.4 1.7 2.3 2.8 2.7 2.3 1.8 2.2 Overall new 1.9 1.9 0.8 1.9 2.4 3.8 83 lower than the mean for 7W sequences (3.1%). Percentage error scores for contentive substitutions were low: however, the A.I. group produced more errors on the average, and all subjects tended to produce more contentive sub- stitutions when the stimulus was 7 items in length. Sentgnge Type x Sentenge Length Post hoc testing was performed to compare means of sentence types and sentence lengths. Results of the testing are shown on Table 19. Most contentive substitutions occurred in 7W anom- alous sentences, followed by 3W well-formed and SW anom- alous. As may be recalled, functor substitutions occurred primarily in SW anomalous sentences. It can be stated with some confidence, therefore, that sub- stitutions were brought out by the anomalous sentence type when the stimulus was 5 or more words in length. 6. Reverspgs A reversal is a change of word position during the repetition of a stimulus. For example, if the stimulus sentence was “A girl pulled the pretty toy,” and the subject's repetition was “The pretty girl pulled a toy,“ reversals have occurred. For analysis of reversal errors, two categories were established: 1-2 word reversals (category I) or 3+ (category II), wherein 3 or more words have changed position. An arbitrary weight of 1.5 was given to category I and 3.5 given to category II. The number of times either category was observed in a Table 19. Post Hoc Testing Results for Contentive 84 Substitution Mean Error Comparisons: Sentence Type (well-formed: WF, anomalous: A, ill-ordered: 10) by Sentence Length (three-word: 3W, five-word: 5W, seven- word: 7W). Within Sentence Means Tested* Type 3W vs. 5W 5W vs. 7W 3W vs. 7W 1. Well-Formed ** n.s. n.s. 2. Anomalous n.s. ** ** 3. Ill-Ordered n.s. n.s. n.s. Within Sentence Means Tested Length WF vs. A A vs. IO WF vs. IO 1. Three-Word n.s. n.s. * 20 Five-Word 11.8. “as. has. 3. Seven-Word ** * n.s. n.s.: not significant * p = 0005 85 subject's responses was then multiplied by the apprOpriate weight value: the scores were added to obtain a final reversal error score, or index of error. For example, if a subject made 5 reversals of category I and 3 rever- sals of category II, his score would be: (1.5 x 5) + (3.5 x 3) = 18 Mean scores were then obtained by adding all index of error scores and dividing by the total number of subjects. The ANOVA for this error type is presented in Table 20. Mean index scores for the factors and interactions are presented on Table 21. Table 20. Results of the ANOVA for Reversals. Source of Variance df Mean Square Significance Articulation (A) 38 212.408 n.s. Sentence Type (ST) 2 3810.210 p-=:0.0005 Sentence Length (SL) 2 4608.958 p-:=0.0005 A x ST 2 200.069 n.s. A x SL 2 325.069 n.s. ST x SL 4 2862.291 p—=:0.0005 A x ST x SL 4 116.736 n.s. n.s.: not significant Articulation, in this instance, was not a significant main factor, nor were any interactions in which it was a part. That indicates that all subjects performed with similar error patterns. Main factors of sentence type 136 Table 21. Mean Index of Error Scores for Reversals for the Factors of Sentenée Type (well-formed: WF, anomalous: A, ill-ordered: IO), Sentence Length (three-word: 3W, five-word: 5W, seven- word: 7W) and Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Articulation x Sentence Type: not significant g: A 19 Overall Mean N.S. 1.0 2.2 18.0 7.1 A.I. 2.5 3.2 23.7 9.8 Overall Mean 1.8 2.7 20.8 Articulation x Sentence Length: not significant N.S. 1.3 7.7 12.3 7.1 0.80:. 1.5 1.0.1 14. 908 Overall Mean 1.4 10.9 13.1 Sentence Type x Sentence Length: significant 3T pg, 1!, Overall Mean WF 1.2 1.1 3.0 1.8 A 0.4 1.9 6.0 2.8 10 2.6 29.6 30.2 20.8 Overall Mean 1.4 10.9 13.1 Articulation x Sentence Type x Sentence Length: not sign. as 5m 1m 9%55%*‘ Ee§e .3311 Eeée All; Ea§a ALL; Ei§e -1- WP 1.7 0.7 0.0 2.2 1.5 4.5 1.1 2.5 A 0.0 0.7 0.0 3.7 6.7 5.2 2.2 3.2 10 2.2 3.0 23.0 36.2 28.7 31.7 18.0 23.7 Overall Mean 1.3 1.5 7.7 14.1 12.3 13.8 87 and sentence length were significant, a finding which means both factors were related to the occurrence of reversals. Sentence Typp The sentence type main effect means, shown on Table 21 as the overall means for WF, a, and IO types, were sub- jected to post hoc testing. The overall means for WF and A were not significantly different; however, both were significantly lower than the 10 overall mean. 111- ordered strings, therefore, were highly correlated to the occurrence of reversals. In WF and A sentences, in which word order contributes to semantic and/or syntactic structure, reversals were infrequent. Figure 15 displays mean index scores (overall means) for each sentence type for all subjects and demonstrated the dramatic upturn in reversal errors when the sentence types were ill-ordered. Sentence Len th The main effect means for sentence length (overall means shown on Table 21) were also tested for significant differences. The mean index of error for 3W sentences was found to be significantly lower than the SW and 7W index means: the latter two means were not significantly different. Reversals, therefore, occurred primarily in SW and 7W IO sequences. Figure 16 shows the main effect means for sentence length. 88 30* 20,. 10., MEAN INDEX OF ERROR I 1 v_i—_ WF A 10 Figure 15. Mean Index of Errors for Reversals for Sentence Types (well-formed: WF, anom- alous: A, ill-ordered: IO). 20«- MEAN INDEX OF ERROR H O 3W 5W 7W Figure 16. Mean Index of Errors for Reversals for Sentence Lengths (three-word: 3W, five- word: 5W, seven-word: 7W). 89 Sentenge Type x Sentenge Length This was the only significant interaction. Mean index of error values are shown on Table 21. The highest number of reversals occurred in SW and 7W ill-ordered sequences. Post hoc testing of these means (see Figure 17) showed that the means for IO sentences were affected by the sentence length such that 3W sentences had significantly fewer errors than SW and 7W sentences. 7. Fungtor Adthions Additions of words during the repetition of sentence stimuli were observed and analyzed with the other error types. However, it was a very infrequently observed error type. The means are shown on Table 22. Error scores in this case were a simple total number of additions made by the subject. The highest mean obtained for functor additions occurred in 7W-IO sentences. No information can be derived from such a small error sample. 8. Contentive Additions This error type was the least frequent to occur. Most contentive additions occurred in 7W-IO sequences. Mean values are shown on Table 23. In the chapter concerning experimental procedures, the method of counterbalancing the order of stimulus presentation was mentioned which, hopefully, would control for the effect of fatigue on subjects' error scores. This method was found to be successful since error scores were equivalent for either order of presentation across subject groups. 90 40., I a J INDEX 08 ERROR N o I MEAN 10d» 3W SW 7W 3W 5W 7W 3W 5W 7W Well-Formed Anomalous Ill-Ordered Figure 17. Reversal Error Index Means: Sentence Length (three—word: 3W, five—word: 5W, seven-word: 7W) by Sentence Type (well-formed, anomalous, ill-ordered). 91. Table 22. Functor Addition Mean Error Scores for the Factors of Sentence Type (well-formed: WF, anomalous: A, ill-ordered: IO), Sentence Length (three-word: 3W, five-word: 5W, seven-word: 7W) and Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Articulation x Sentence Type g; A T9 Overal; Mean N.S. 0.07 0.13 0.93 0.38 A.I. 0.20 0.27 1.52 0.66 Overall Mean 0.13 0.20 1.22 Articulation x Sentence Length 3! 2!, 1!, avegg11 Mgap n.s. 0.22 0.35 0.57 0.38 A.I. 0.43 0.53 1.02 0.66 Overall yoga 0,37 0.44 0.79 Sentence Type x Sentence Length fl _5_g __7g Overa 1 1 Mean NP 0.17 0.05 0.17 0.13 A ‘0.07 0.00 0.52 0.20 10 0.72 1.27 1.67 1.22 Overall Mean 0.32 0.44 0.79 Articulation x Sentence Type x Sentence Length as as 2w. e:— WF 0.10 0.25 0.00 0.10 0.10 0.25 0.07 0.20 A 0.05 0.10 0.00 0.00 0.35 0.70 0.13 0.27 10 0.50 0.95 1.05 1.50 1.25 2.10 0.93 1.52 Overall "can 0.22 0043 0.35 0.53 0.57 1.02 Table 23. I 92 Contentive Addition Mean Error Scores for the Factors of Sentence Type (well-formed: WF, anomalous: A, ill-ordered: IO), Sentence Length (three-word: 3W, five-word: SW, seven- word: 7W) and Articulation (normal speaking: N.S., articulatory-impaired: A.I.). Articulation x Sentence Type g: A T9 Overall Mean N.S. 0.05 0.02 0.47 0.18 A.I. 0.08 0.17 1.12 0.45 Overall Mean 0.07 0.09 0.79 Articulation x Sentence Length 9g 9g 1g Overall Mean N.S. 0.00 0.07 0.47 0.18 A.I. 0.03 0.30 1.03 0.45 Overall Mean 0.02 0.18 0.75 Sentence Type x Sentence Length 9g 9g 1g Overall Mean WF 0.00 0.02 0.17 0.07 A 0.02 0.02 0.22 0.09 IO 0.02 0.50 1.85 0.79 Overall Mean 0.02 0.18 0.75 Articulation x Sentence Type x Sentence Length WF A IO Overall Mean 9293111 39 53 121 “can N.S. A.I. N.S. A.I. N.S. A.I. N.S. A.I. 0.00 0.00 0.00 0.05 0.15 0.20 0.05 0.08 0.00 0.05 0.00 0.05 0.05 0.40 0.02 0.17 0.00 0.05 0.20 0.80 1.20 2.50 0.47 1.12 0.00 0.03 0.07 0.30 0.47 1.03 CHAPTER V DISCUSSION Results of the study will be discussed in the following sections with respect to auditory discrimination and the main experiment factors of sentence length, sentence type, word type, and articulation. Suggestions for future re- search conclude this chapter. Auditory Discrimination Twice as many A.I. as N.S. children were found to have inadequate auditory discrimination on the Wepman Auditory DisgriminatTon Tes . Errors on the test ranged from 7 to 15 for the A.I. children and 7 to 8 for N.S. children. These results support previous studies which found a positive relationship between articulation defects and poor auditory discrimination. Weiner (1967) had questioned the reliability of these earlier studies, however, because of the variety of test instruments used and the differences in number of articulatory errors exhibited by the subjects selected. He recommended that replications of earlier studies be performed. The present study did not replicate the study by Prins (1963) entirely, but it did make use of his subject selection 93 94 criteria of 3 or more misarticulations and the same auditory discrimination test. Prins was interested in what types of phonological errors were related to auditory discrimination problems and determined that children whose error sounds were close to the target sounds had the most difficulty with the Wepman test compared to children who exhibited more random articulation errors. The experimental children in the present study exhibited misarticulations which were close to the target sounds, e.g., [:t/s, t/tS, s45 , s/O, t/k:l . Sixty percent of these children had poor perfor- mances on the Wepman test. Therefore, support for Prins' contention was evident. Prins also suggested that auditory discrimination was related to the total language processes. He believed that language processes were involved in the kind of discrim- ination required to distinguish the difference in Wepman's word-pairs wherein acoustic variations are phonemic, sig- nalling a change in linguistic meaning. If a child's lan- guage processing system is intact, he should detect the symbolic differences in the words. If the relationship between language and auditory discrimination is valid, then poor performance on the Wepman test indicates a distur- bance in the language system of the testes. Since N.S. children performed better than A.I. children on the Wepman test, one could then conclude that the language systems of the two groups differ. Also, since the two groups differ only by the factor of articulatory pro- ficiency, one could infer that inadequate articulatory 9S develOpment also relates to poor auditory discrimination and a depressed language processing system. Marquardt and Saxman (1970) had found that kinder- garten children with numerous misarticulations performed poorly on Wepman's test and on the Carrow Auditogy Test TpgrLangupge Comprehenpion. The authors held that lan- guage competence and auditory discrimination were signi- ficantly correlated within the articulatory-impaired group, although not within the normal speaking group. The present study tended to support the theory that articulation pro- ficiency was related to auditory discrimination skills and certain language performance skills. The Wepman test assesses the auditory perception of speech units, linguistic units which are paired but not actually meaningfully related and do not occur in ongoing speech contexts. In order to discriminate the word-pair stimuli, it is necessary to perceive the phonemic and/or semantic differences of the language units. The hearing mechanisms, that is, the peripheral auditory systems, were intact for all subjects. Therefore, poor auditory dis- crimination of such word pairs indicated a failure to perceive and process accurately either the phonemic and/or semantic characteristics of the stimuli. The subjects' test papers were reviewed to investigate these factors. Upon reviewing the test forms, the following obser- vations were made. The control subjects had made errors on 16 of the 30 'different' word pairs on the Wepman test 96 for a total of 75 discrimination errors. The eXperimental subjects had made a total of 148 discrimination errors on 26 of the 30 word pairs. (Since a majority of the 'different' word pairs were missed at one time or another by all subjects tested, it is possible that the words within the pairs were not readily semantically recognized by the subjects. In other words, the test words were foreign to the Spoken and receptive vocabularies of the subjects and did not signal linguistically different information. Further examination of the word pairs more commonly missed supported this hypothesis. The most commonly missed word pairs for both groups were: clothe-clove, sheaf-sheath, fie-thigh, and vow—thou. These words are probably not used nor often heard by the children. They are also phonemically difficult to distinguish since the phonemes involved are [75 , v, f, and 0:]. Voiced and unvoiced fricative sounds of this kind are difficult to detect without the aid of visual cues, i.e., viewing the Speaker. Therefore, these words may be difficult to perceive semantically and phonemically for both subject groups. Other word pairs in the Wepman test are similarly hard to distinguish on a semantic level (lath-lash, shack- sack, shoal-shawl, muff-muss), but some are more readily distinguished phonemically (pork-cork, din-bin, coast-toast). If one can assume that the children from both artic- ulation groups generally hear and use similar vocabulary words and that a variety of the words of the Wepman list were unfamiliar to all of the children, then the factor of phonemic differentiation becomes more important than 97 the semantic variable in distinguishing the performance of the two articulation groups. When the words themselves are not readily distinguished on a semantic basis, the listener must rely on the detection of phonemic variations. On this basis, the normal speakers performed better than the exper- imental subjects. In order to determine whether a rela- tionship existed between the articulation problems of the eXperimental subjects and the errors made on the Wepman test, further analysis of the test papers of the misarticulating children was conducted. Of the 148 errors observed, 52 occurred on words differ- entiated by unvoiced fricatives, 26 on words differentiated by unvoiced StOp plosives, and 22 on words differentiated by voiced fricatives. The remaining errors occurred for words differentiated by vowels, voiced plosives, and nasals. It is interesting to note that these children had articu- lation errors primarily for unvoiced fricative sounds: [:s, S, tS,0, f:]. The distinctive feature theory could eXplain the relationship between their articulation errors and their subsequent discrimination errors by noting that the children did not have the distinctive features in their articulation system necessary for accurate perceptual dis- crimination. The motor theory of Speech perception, on the other hand, would suggest that the discrimination errors occurred because these particular phonemes were referred to incorrect motor patterns of articulation in the perceptual systems of the children. In either case, the discrimination errors observed do correlate with the children's articulation errors. 98 Sentgnge Length The results of this study agreed with those of previous studies (Beasley and Smith, 1972: Aaronson, 1967: Miller, 1956: and Schuckers, Shriner, and Daniloff, 1971) in that error rate tended to increase as stimulus length increased. This tendency was clearly observed for all subjects with respect to functor omission errors. However, the articulatory-impaired children made significantly more functor omissions than normal Speaking children at every sentence length. Since three-word stimuli should be within the short-term memory ability of the subjects, it was inter- esting to obtain significantly different error scores for the two articulation groups at the three-word length. It is possible that the responses reflect more than simply automatic recall of stimuli, rather, that the responses reflect the subjects' level of language competence. Menyuk and Looney (1972) believed that repetitions reflect chil- dren's level of grammatical competence if they are developing language normally. If repetitions also reflect the level of grammatical competence in misarticulating children and this level is below that of normal Speaking children, poorer performance on even three-word stimuli would be eXpected to be observed. Contentive omissions did not generally occur until the stimulus item was seven words in length (see Figure 10). The error rate for seven-word sequences for both articu- lation groups for functor and contentive omissions indicated that the short-term memory Spans of the subjects had been 99 exceeded. Seven items in a stimulus were found to exceed subjects' memory Spans in previous studies (Beasley and Smith, 1972: Miller, 1956: Templin, 1957: Beasley and Acker, 1971: and Schuckers pp_pT., 1973). Functor substitutions were infrequent and, in this case, most errors occurred on five-word anomalous sequences. After reviewing subjects' reSponses to these sequences, a main factor was found to contribute to the high error mean. Of 61 functor substitution errors made by eXperimental subjects and 28 made by control subjects, 28 and 24 of these, reSpectively. were made on the functor word 99 in the five-word anomalous sequences “her bike eats an apple“ and “an airplane chOps the wood“. The word pp was usually re- placed by ppg or‘g or was omitted (contributing to omission error scores). The word pp did not occur in any of the other 52 sentences used as stimuli. It is probable that pp was not used in the expressive vocabulary of the first grade children of this study. Also, the word pp was an unstressed word in these anomalous sequences, a fact which might have interfered with accurate perception. A more exact eXplanation for the effect of the word pp on subjects' performances would have been possible had it occurred in other sentence types in the stimulus sentences. Other functor substitutions which commonly occurred and helped to contribute to the high error mean for 5-word anomalous sequences also occurred in other sentence types. Generally, the most common functor substitutions observed were the following: a/the, her/his, his/her, the/his, or 100 the/her. The substitution a/the was considered an error if the subject had produced the word ppg at other times during the test session. If a subject consistently produced a/the and had shown articulation problems including thetkilsound, then a/the was not an error for that subject. This sub- stitution was common, however, even for normal Speakers. Therefore, the error may be due to their acoustic similarity, particularly when the word 929 is unstressed in a stimulus item. The his/her and her/his substitutions suggested that the subjects may not have 9T9 and 99; clearly distinguished in a linguistic sense. At times, recall accuracy may have been affected by sentence context. For example, her/his frequently occurred in the repetitions of the sequence “his sister bakes a book.“ Two eXplanations appear feasible. The feminine noun, sister, may have brought about the change from a masculine possessive pronoun to a feminine possessive pronoun such that 9T9 was replaced by 99; to agree with a feminine image created by the sequence. The second SXplan- ation is that the coarticulation of ‘his sister' caused the perception of hp; as the /h/ in pr is perceived in close conjunction with the /a‘/ in sistgg. The the/his and the/her substitutions may also have occurred due to the sentence context, although not in every instance. When the/his and the/her occurred in the S-word anomalous sequences “the dog purrs his bone" and “the lady walks her coat," Egg occurred in the sequence, was recalled for its position, and was substituted for the words his 101 and 99;. Therefore, 999 was recalled for both functor positions. However, when the/his occurred in ”his sister bakes a book,“ the change is less readily explained. It is hypothesized that when a stimulus taxes a subject's recall, as an anomalous sentence can, less frequently used possessive pronouns such as pr or pp; may be replaced by a common, useful functor word such as 999. Since the anomalous sentence type may appear to be nonsense to the child, the change from pr to ppg or pp; to pp; may not appear to affect the basic structure of the sentence and, indeed, it does not. If the sentence were meaningful, such as “his dog is chasing her dog,“ in which 9;; and Egg have significant roles to play on a semantic level, substitution of pp; for possessive pronouns would make a major semantic difference. Briefly, pp; can be used to substitute for other functor words and the anomalous sentence remains anomalous. The word pp; appears to be easier to recall than the possessive pronouns, particu- larly when Egg was already present elsewhere in the stimulus sequence. Sentence length affected subjects' performances in approximately the same fashion with respect to the re- maining error types: contentive substitutions, additions, and reversals. Errors were highest on seven-word strings. Senten e e Error rate was highest for ill-ordered strings, followed by anomalous and well-formed sequences, except 102 in the case of substitutions, which has been discussed. This error pattern was observed in previous studies (Beasley and Acker, 1971: Beasley and Smith, 1972: Brown and Bellugi, 1964: and Scholes, 1970). All subjects, regardless of articulation proficiency, made more functor and contentive omissions on ill-ordered sequences, followed by anomalous and well-formed sequences. Experimental subjects made significantly more functor omissions for every sentence type and more contentive omissions on ill-ordered sequences than control subjects. Sentence type appeared to have a greater effect on subject performance than sentence length. For example, for functor omission errors, the mean error for 3-word ill-ordered sequences was higher than the mean error for S-word well-formed sequences in normal Speaking subjects' responses. For experimental subjects, the mean error for functor omissions was higher for S-word anomalous than for 7~word well-formed sequences. In other words, al- though a stimulus may contain more words, the sentence type had more to do with the number of functor omissions which occurred, not the additional words. A longer but well-formed sentence was easier for subjects to recall than a shorter, ill-ordered one. In general, subjects showed better recall of well- formed and anomalous sequences than ill-ordered sequences. This suggests that the language structure (semantic and/or syntactic) present in well-formed and anomalous sentence types contributed toward recall processing, especially 103 when the stimulus length increased to seven words. All subjects showed better recall of seven-word well-formed and anomalous sequences than seven-word ill-ordered sequences. However, the normal Speaking children per- formed with fewer errors than the articulatory-impaired children in their repetitions of well-formed and anomalous sequences. This may mean that language structure was of more assistance to the normal Speaking children than to the articulatory-impaired. Since syntactic structure is present in both well-formed and anomalous sequences and since scores were better for these types than for ill- ordered sequences, it appears that syntax played an im- portant function in the task of recall. At the same time, therefore, syntax was of greater value as a language cue to the normal Speaking children than to the articulatory- impaired. This, in turn, implies that the syntax devel- Opment of the two articulation groups is somehow different, with articulatory-impaired children having less well- developed syntax. This is, of course, speculation based upon the results of the study: and more definite comments with respect to differences in syntax develOpment cannot be made. However, a difference does exist between the recall skills of the articulation groups for the various language stimuli which cannot be eXplained simply by differences in short-term memory skills. The short-term memory skills of the two articulation groups would appear to be comparable. In one instance (functor omissions), normal speaking children performed 104 better than the articulatory-impaired children in the recall of every sentence type and sentence length stimuli except for 7-word ill-ordered sequences. Even the means for S-word ill-ordered sequences were significantly differ- ent, suggesting that articulatory—impaired children have poorer automatic memory recall for S-word stimuli. However, scores for the other error types (substitutions, additions, reversals, contentive omissions) did not Show this differ- ence at the 5-word level. Also, the groups generally showed equal difficulty repeating 7-word ill-ordered sequences across the error types. Therefore, all subjects' memory spans were exceeded by seven-word ill-ordered stimuli: and thus, it can be concluded that memory spans of the two groups were not significantly different. This would support the hypothesis that the performance differ- ences observed on well-formed and anomalous sequences were due to factors other than short-term memory abilities of the separate articulation groups. Language structure, and how it was used by the children as a cue for recall, appeared to differentiate the two groups of subjects. Returning to the factor of sentence type and its relationship to error types observed, reversals provided interesting data. Reversals occurred infrequently in well-formed and anomalous sequence repetitions. However, many reversals occurred for ill-ordered sequences. Word position in well-formed and anomalous sequences is important to the semantic and/or syntactic structure of language units. Therefore, reversals, or changes in word 105 position, would change the meaning and structure of the original stimulus. Ill-ordered stimuli, on the other hand, are semantically and syntactically meaningless. Words in such a sequence cannot easily be chunked according to semantic association or bound together by a perceived grammatical structure. As stimulus length increases, words may be chunked together for memorization, but the order may change as short-term memory loses efficiency. It was interesting to find that, in the repetitions of 5- and 7-word ill-ordered stimuli, reversals actually introduced semantic and/or syntactic structure, such that the sentence type was changed. In other words, ill-ordered stimuli were repeated as anomalous or well-formed sequences. The eXperimental subjects changed the sentence type of 48 of the ill-ordered sequences. Some examples are shown below. Ill-Ordered StTmulus Subjects' ReSponsgs 1. Carrot the bunny eats the 1. The carrots eat the /d/\/ A bunny eats a carrot the 2. Kitten chases my her dog 2. My big kitten chases her big dog A kitty chases a dog My kitten chases her dog 3. Baby his calls the mother 3. His baby calls the mother Baby calls his mother 4. Covers blanket baby the 4. Blanket the baby with yellow the bed , her yellow bed Since the above ill-ordered stimuli were frequently repeated with reversals such as those listed, some factor 106 common to these sequences must have triggered the response behavior. After studying these sequences, it was noted that a portion within each stimulus is actually in sen- tential order: (1) ”the bunny eats the", (2) “kitten chases“, (3) ”calls the mother", (4) “the yellow“ or "the bed“. It is possible that the listeners recognized these particular portions of the ill-ordered sequences as somewhat similar to a real sentence and consequently re- structured the string to attempt to give meaning to the stimulus as a whole. Normal Speaking subjects made 28 reversals which introduced some syntactic and/or semantic structure into the ill-ordered stimuli. A few of the more commonly ob- served responses follow. Ill-Ordered Stimulus 9ubjegts' ReSponses 1. Baby his calls the mother 1. The baby calls his mother Baby calls his mother His baby calls his mother 2. Carrot the bunny eats the 2. The carrot eats the bunny, too The carrot eats the bunny 3. Sister my the Spills water 3. Sister my Spills the water The sister my Spills the water 4. Washes a the man big green 4. Washes the big man's green car car A man washes a car Black man washes big dirty car 107 Ill-Ordered Stimulus (cont.) Subjects' Responses 5. Girl red cow silly draws S. The red cow draws the the a silly the The girl draws the silly cow the 6. Covers blanket baby the 6. Yellow blanket covers yellow the bed the baby yellow bed Again, the stimuli contain a portion which resembles or forms part of a sentence. Normal Speaking children may have reacted to the stimuli in the same fashion as the experimental children, that is, they recognized this mean- ingful portion and reassembled the remaining words to give meaning to the stimulus as a whole. To this point, differences in subjects' performances have been discussed relative to errors observed. ”Better” performance on the part of normal Speaking children can also be shown by tabulating totally correct reSponses of the articulation groups, as shown in Table 24. The table illustrates that the greatest differences between the N.S. and A.I. subjects occurred for 7-word well-formed and anomalous sequences, and for S-word well-formed and anom- alous sequences. Experimental subjects had fewer totally correct repetitions. Seven-word ill-ordered sequences were least often correctly repeated by either articulation group. The fact that normal Speaking children were better able to correctly repeat 7-word and 5-word well-formed and anomalous sequences further supports the hypothesis that a difference does exist in the two groups' abilities to process the linguistic information present in these sentence 108 Table 24. Number of Totally Correct ReSponses by Normal Speaking (N.S.) and Articulatory-Impaired (A.I.) Children for the Nine Sentence Type by Sentence Length Categories. Total Number of Correct Responses Sequence Type Out of 120 Possible §3§a Difference Eel; 3-Word Well-Formed 107 +13 94 S-Word Well-Formed 114 +39 75 7-Word Well-Formed 95 +42 53 3-Word Anomalous 113 +22 91 S-Word Anomalous 88 +45 43 7-Word Anomalous 72 +45 27 3-Word Ill-Ordered 98 +25 73 S—Word Ill-Ordered 53 +34 19 7-Word Ill—Ordered 10 + 7 3 109 types 0 Recall Patterns for Sentence Types Recall patterns were analyzed in order to compare results of the present study to the patterns described by Deese and Kaufman (1957). Deese and Kaufman described the recall patterns of normal adult Speakers for English con- textual material and discrete items. The well-formed sen- tences of the present study were considered to be comparable to the English contextual material. Ill-ordered sequences were considered similar to discrete items. Anomalous sentences were considered to be more like contextual material than discrete items since the words of an anom— alous sequence are bound together by syntactic structure. Results did not follow the predicted patterns, however. Table 25 describes the patterns observed by Deese and Kaufman (1957) and those of the present study. Briefly, the recall pattern described by Deese and Kaufman for English contextual material was observed for ill-ordered sequences of the present study. The recall pattern described by Deese and Kaufman for discrete items was similar to the pattern observed for well-formed and anomalous sequences in the present study. This difference in recall patterns observed may be due to the use of different material, deSpite similarity in linguistic structure or non-structure, or due to the difference in ages of the subjects studied. Table 25. Recall Patterns 110 from the Deese and Kaufman (1957) Study and Present Study Results. Deese and Kaufman Present Study English ContextugT Materia first items recalled best, followed by middle and final items, respectively D;screte_Ttems final items recalled best, followed by first items and middle items, reSpec- tively Well-Formed Sentences final items recalled best, followed by mid- dle and first items, reSpectively Ill-Ordered Spntences first items recalled best, followed by middle and final items, reSpectively Anomplous Sentenpes same pattern as for well-formed sentences 111 Word Type Error means were higher for functor words than for contentive words for both articulation groups with reSpect to omissions and substitutions. Error means were low for both word types with reSpect to additions. This trend for functors to be omitted more than contentives as the task becomes more difficult agreed with reports in earlier studies (Scholes, 1970: Martin, 1968: Brown and Bellugi, 1964: Beasley and Acker, 1971: and Beasley and Smith, 1972). Deleted functors in a repetition produces a “telegraphic“ form of Speech, and this was often observed in the reSponses of severely misarticulating subjects, i.e., for “The old man takes a long walk,” the reSponse was “old man take long walk.“ Art;culation The difference in the performances of the two groups of subjects has been discussed within the other topic areas. Briefly, the articulatory-impaired children demonstrated less adequate auditory discrimination skills than the normal Speaking children based on the Wepman Auditory Disnganation Test results. This finding was felt to indicate auditory perceptual problems caused by or related to their Specific articulation problems. It also may indicate that the articulatory-impaired children do less well than the normal Speaking children in distinguishing the word pairs linguistically. However, the word pairs are likely to be difficult for both subject groups to 112 distinguish on a semantic basis, since the words are probably not familiar to them (e.g., as a part of their receptive vocabulary). The two articulation groups also differed in their ability to recall 5- and 7-word well—formed and anomalous sequences, whereby the articulatory-impaired children had higher error scores. Yet, in the case of 7-word ill- ordered sequences, both groups of subjects made similar numbers of errors. It was assumed in the present study that if a subject failed to repeat 7-word ill-ordered sequences adequately but performed effectively while repeating 7-word well- formed and/or anomalous sequences, the difference could be attributed to the language structure in the latter two sentence types and particularly to the ability of the subjects to extrapolate such information and use it to help them in the recall task. All subjects did show better recall for well-formed and anomalous sequences than for ill-ordered sequences. However, normal Speaking Children had the greater ability to repeat the stimuli, indicating more adequate use of the language structure cues. It is possible, therefore, that the poorer per- formance of the articulatory-impaired children is due to less deve10ped language competence in semantic and/or syntactic areas which may directly relate to their moderate to severe phonological impairments. Since Specific syntactic and semantic structures were not tested in this study, it is not possible to assess the Specific linguistic 113 strengths or weaknesses of the children. One can only say that the recall abilities of the two groups often signi- ficantly differed and that language competence may be a factor contributing to the difference in their performance. Implications fopiFuture Research The present study supported previous studies which demonstrated a positive relationship between articulation proficiency and auditory discrimination ability for children below 9 years of age, Specifically between 5.11 and 6.9 years. Further research Should be done to clarify the role of age in this relationship using subjects from 7-8, 8-9, and 9+ years of age who exhibit 3 or more articulation errors with no associated physical, psy- chological, or learning problems, and testing them with the Wepman Auditory D;scrimination Tes . This would help to fulfill the need for replications of studies of this kind as well as to determine the actual relationship be- tween discrimination, articulation, and language proficiency. The above pOpulations could also be tested using the Wepppn Auditory_Dischmination Test and one or more other discrimination tests, i.e., the Goldman-Fristoe-Woodgogk Auditory DiscriminatTon Test, in order to compare results obtained. Since the Wepman test appears to have some inherent problems, specifically with respect to vocabulary used as stimuli, other tests such as the Goldman-Fristoe Woodcock test may be possible alternatives. 114 An extension of the auditory discrimination portion of this study would also add to the linguistic information needed to describe the differences in language skills of normal speaking and articulatory-impaired children. An in-depth analysis of (1) what word pairs were missed by the subjects: (2) what phonemes and linguistic units, therefore, were not discriminated: (3) what Specific articulation errors were present in each subject tested: and (4) what, then, might be the relationship between undiscriminated Speech sounds and the child's own misartic- ulations. Theories pertaining to the relationship of ar- ticulation proficiency, auditory discrimination, and lan- guage skills would become more clear and perhaps strengthened by such detailed information. The main part of the present study analyzed recall performance with respect to error types of omission, sub- stitution, addition, and reversal of word position on a cursory level to indicate which errors were more frequent and compared the two articulation groups on the basis of error occurrence. A future study could be designed to determine error types in depth, attempting to study which particular words are omitted, substituted, added or re- versed and to what extent linguistic information is added or subtracted. Substitutions such as his/her and her/his could be investigated with respect to the sex of the subject making these substitutions. A brief study of reversals in the present study indicated that children sometimes added syntactic or semantic (or both) meaning 115 to an ill-ordered sequence. Interesting information on the differing approaches to the words of the stimuli could be gleaned from such a descriptive study. With modifications, the stimulus material could be presented to other subject p0pu1ations, i.e., hearing- impaired children or adults, aphasic adults, normal Speaking adults, in order to describe their recall as it is affected by the factors of sentence type, sentence length, word type, and particular handicap, if one is present. The present study could be replicated with the addition of a Speech sample taken from each subject prior to or following the taped portion of the study. The eXperimenter could then determine (1) what kind of sentences are used by the subjects in normal conversation, (2) whether articles such as 999, p, and pp are used differentially by the subject, (3) whether such possessive pronouns as pr and 99; are employed, and other relevant facts which would aid the eXperimenter to more adequately analyze errors observed. The Beasley and Smith study (1972) included the factor of stress to determine how stress affected a normal Speaking child's recall of functor and contentive words. They found that stressing of a word improved the subjects' recall of both functor and contentive word types. In fact, the error rate difference between functor and contentive words was less than the difference between stressed and unstressed words, SSpecially for 7-word sequences and ill- ordered sequences. This meant stress was more important 116 than word type with reSpect to recall. Therefore, they felt that stress may be used, for example, in a speech therapy language habilitation program or as an aid toward strengthening the short-term memory span. Stress may be an equally important factor affecting the recall abil- ities of misarticulating children. This could be deter- mined by replicating the present study, adding stress as a factor, and testing children with moderate to severe misarticulations. Future research, in general, must concentrate on the discrimination problems, recall problems, language problems, and phonological problems of misarticulating children and, insofar as possible, describe how these areas relate to one another. From such information, it should be possible to create effective Speech and language habilitation programs which can make a greater difference in their performance skills at an earlier time in their school or preschool years. CHAPTER VI SUMMARY AND CONCLUSIONS There has been increasing interest in the relation- ship of phonology, syntax, and semantics to each other in the process of normal language deve10pment and the added factor of auditory discrimination to each of them. A review of the research indicated that children with moderate to severe articulation errors have associated auditory discrimination problems and/or delayed gramm- atical deveIOpment. Shriner 99 pl. (1969) suggested that children may suffer from auditory feedback problems which lead to misarticulations which, in turn, induce syntactic deficits. Short-term memory studies, re- synthesis studies, repetition task or sentence-elicited studies have all provided some information on the differential skills of normal speaking and articulatory- impaired children. This SXperimenter believed that a study was needed to fulfill three primary research concerns: (1) replicate an earlier study on auditory discrimination in terms of subject selection criteria and testing material in order to verify earlier findings about auditory discrimination skills of articulatory-impaired children, (2) use a 117 118 repetition task eXperiment in which the stimuli offered language structure (deep and/or surface structure) as cues for recall yet controlled for short-term memory parrotting effects, and (3) compare the performance of normal Speaking and articulatory-impaired children based on an analysis of Specific error types observed. The following questions were investigated: Would there be significant differences in the recall accuracy of first grade articulatory-impaired and normal Speaking children for 3—, 5-, and 7-word sequences? Would there be Significant differences in the recall accuracy of first grade articulatory-impaired and normal speaking children for well-formed, anomalous, and ill-ordered sequences? Would there be significant differences in the recall accuracy of first grade articulatory~impaired and normal Speaking chil- dren for contentive and functor word types? What types of errors would occur and with what frequency on the factors of sentence type, sentence length, and word type for articulatory-impaired and normal speaking children? The eXperimenter hypothesized that normal Speaking children (1) would demonstrate more adequate auditory discrimination than articulatorylimpaired children as demonstrated by the results of the Wepman Authory Dis- crimination Test, (2) normal Speaking children would apply the available language cues in well~formed and anomalous sentences for better recall reSponses than those of articulatory—impaired children, (3) normal Speaking and articulatory-impaired children would demonstrate 119 equivalent Short-term memory Spans as evidenced by similar error scores on ill-ordered sequences, and (4) word type errors would follow results of previous studies: all subjects would find functor words more difficult to retain than contentive words as the task increased in difficulty. Forty first grade children with normal hearing and intelligence served as test subjects. Twenty children with normal articulation, 13 boys and 7 girls, mean age 6.3 years, formed the control group. Twenty children, 13 boys and 7 girls, mean age 6.3 years, who had exhibited 3 or more articulatory errors on McDonald's Screening Depp Test of ArtTculgtion served as the eXperimental group. Each subject was tested individually. Initially, the subject was administered the Wepman Auditory Discrimination T999. The eXperimenter then read a standardized set of instructions directing each subject to repeat exactly what was heard. The subject listened to a tape of 54 randomized stimulus sentences which varied in length and type. Re- Sponses were tape recorded and transcribed on a standard answer sheet. ReSponse errors were classified according to error type: (1) omission—~functor or contentive, (2) sub- stitution-~functor or contentive, (3) addition--functor or contentive, and (4) reversals of word order. The study was a Split-Plot design with repeated measures, and a multifactor analysis of variance was performed using a computer routine available in the Michigan State University Computer Library. Post hoc analysis was 120 carried out using Tukey's “Honestly Significant Difference“ test. Auditory Disgngnggion Results SUpported the findings of Prins (1963) in that articulatory-impaired children had more difficulty with the Wepman test than normal Speaking children. Sixty percent of the eXperimental group failed the test as compared to 30% of the control group. Results were discussed with respect to possible differences in lan- guage skills. Functor Omisgions The occurrence of functor omission errors was directly related to the factors of sentence type, sentence length, and articulation. More functor words were omitted for ill- ordered sequences, followed by anomalous and well-formed sequences, reSpectively. Functor omissions increased as sentence length increased. Articulatory-impaired children made significantly more functor omissions than normal Speaking children for every sentence type and sentence length except for 7-word ill-ordered sequences. This finding indicated that all subjects appeared to demonstrate equivalent short-term memory Spans. Qpnteptive Omissions The factors of sentence type, sentence length, and articulation directly related to the occurrence of con- tentive omissions. Most errors occurred for 7-word 121 ill-ordered sequences: error means were low for the other sentence type and sentence length categories. In this case, articulatory-impaired children made significantly more contentive omissions than normal Speaking children for 7-word ill—ordered sequences. FunctpgySubppTgutions The factors of sentence type, sentence length, and articulation appeared to affect the occurrence of this error type. Functor substitutions occurred primarily in the repetition of S-word anomalous sequences. However, analysis of subjects' reSponses indicated that the reasons for the high error rate were factors outside of the sen- tence type and length of the stimulus. Articulatory- impaired children produced significantly more functor substitutions than normal Speaking children in 3- and 5- word sequence repetitions: 7-word sequences had similar error means. Contentive Substitutions The factors of articulation and sentence length were found to relate to the occurrence of contentive substi- tutions. Most errors occurred for 7-word anomalous sequences, followed by 3-word well-formed and 5-word anomalous sequences. Error means, however, were low for all sentence type-sentence length categories. Reversals Sentence Type and Sentence Length factors related directly to the occurrence of reversals. Errors occurred 122 primarily in 5- and 7-word ill-ordered sequence repetitions. Few reversals occurred for well-formed and anomalous se- quences of any length. Additions - Functor gpg Contentive This error type was least frequent and, although it was analyzed along with the other error types, no meaning- ful information could be derived from the data. When observed, additions occurred in the repetition of 7-word ill-ordered sequences. Conclusions Within the limitations of the present study, the following conclusions seem warranted: 1. Children with moderate to severe articulatory problems reveal less adequate auditory discrimination skills than normal speaking children as determined by the results of the ngppn Auditory Disgrimination Test. 2. First grade children, regardless of their articulation proficiency, make more errors on a recall task as the stimulus length increases. 3. First grade children make most errors in the repetition of ill-ordered sequences, followed by anomalous and well-formed sentences, respectively. 4. Errors are greater for the functor word type than for the contentive word type. 5. Articulatory-impaired first grade children reveal poorer recall than normal Speakers for well- formed and anomalous sentences of all lengths as well 123 as for 3- and 5-word ill-ordered sequences. Seven-word ill-ordered sequences are equally difficult for artic- ulatory-impaired and normal Speaking first grade children. Short-term memory Spans appear to be the same for the subjects tested. 6. Functor omissions occur with greatest frequency of the error types analyzed. Contentive omissions are infrequent until the stimulus is a 7-word ill-ordered sequence. Functor substitutions appear to occur primarily in anomalous sequences for all subjects. Contentive sub- stitutions are infrequent regardless of sentence type or length: error means did not exceed 3%. Reversals occur almost exclusively in 5- and 7-word ill-ordered strings. LIST OF REFERENCES Aaronson, D., Temporal factors in perception and short- term memory. EgychoT, BuTT., 67, 130-144 (1967). Aungst, L. F., and Frick, J. V., Auditory discrimination ability and consistency of articulation of /r/. 9. Speech Hearing Dis., 29, 76-85 (1964). Beasley, D., and Acker, J., The effect of sentence length, sentence type, word type and stress on four and five year olds. Unpublished Master's thesis, Michigan State University (1971). Beasley, D., and Beasley, D. C., Auditory reassembly abilities of black and white first- and third-grade children. J, Speech Hearing Res., 16, 213-221 (1973). Beasley, D., Shriner, T. H., Manning, W. H., and Beasley, D. C., Auditory reassembly of CVCS by children with normal and defective articulation. Unpublished study, Michigan State University (1973). Beasley, D., and Smith, A., Short-term memory and imitation: effects of sentence length, sentence type, word type, stress and grade. Unpublished Master's thesis, Michigan State University (1972). Beatty, J., and Borree, J., Effects of word class on the recall of sentences. J, Verb. Lrng, Verb. Behav., 10, 604-607 (1971). Braine, M. D. 8., Ontogeny of English phrase structure: the first phase. Lgnguage, 39, 1-13 (1963). Broadbent, D., A mechanical model for human attention and immediate memory. Psychol, Rev., 64, 205-215 (1957). Brown, R., and Bellugi, U., Three processes in the child's acquisition of syntax. ngvard Edugatgonal Rev., 34, 133-151 (1964). Brown, R., and Berko, J., Word associations and the acquisition of grammar. Child Develpm., 31, 1-14 (1960). 124 125 Brown R., and Fraser, C., The acquisition of syntax. In Ursula Bellugi and R. Brown, The Acguisition of Language, Monogr. of the Society for Research in Child DevelOpm., 29, 43-79. Yellow Springs: Antioch Press (1964). Carrell, J., The etiology of sound substitution defects. Chomsky, N., Syptactic Structures. Gravenhage, The Netherlands: Mouton (1957). Chomsky, N., and Halle, M., The Sound Pattern of Englgsh. Harper and Row: New York (1968). Cohen, J. H., and Diehl, C. F., Relation of Speech-sound discrimination ability to articulation-type Speech defects. J. Speech Hearing DTp., 28, 187-190 (1963). Daniloff, R., and Moll, K. L., Coarticulation of lip rounding. J, Speech Hearing Res., 11, 707-721 (1968). Davis, E. A., Tag ppveloppent of LinguTstic Skill Tn ngns, §;ngletonsI §ingletons with Siblings, gnd Only Ch;ldren from Five to Ten Years. Minneapolis: Uhi- versity of Minnesota Press (1937). Deese, J.: and Kaufman, R. A., Serial effects in recall of unorganized and sequentially organized verbal material. J, EXp, gsyghol., 54, 180-187 (1957). Farquhar, M. 5., Prognostic value of imitative and auditory discrimination tests. J. Speegh Hearipg Dis., 26, 342-347 (1961). Fontaine, V. E., The auditory memory Span of children for vowel sounds. (M.A. thesis, Univ. Wisconsin, 1940): cited by Van Riper, C. and Irwin, J., Voige and Articulation. Englewood, N. J.: Prentice-Hall, Inc. 1958 . Fraser, C., Bellugi, U., and Brown, R., Control of grammar in imitation, comprehension, and production. J, Vprb, turns. Vgrb, thaV., 2' 121-135 (1963). Freedle, R. 0., Keeney, T. J., and Smith, N. D., Effects of mean depth and grammaticality on children's imi- tations of sentences. J, Vegb, Lrng, Verb, thgv., 9, 149-154 (1970). Hall, M., Auditory factors in functional articulatory defects. J. Exp. Educ., 7, 110-132 (1938). Irwin, R. B., Speech therapy and children's linguistic skills. J. Speech Hearing Res., 5, 377-381 (1962). 126 Johnson, W., Darley, F. L., and Spriestersbach, D. C., Diagnostig Mpthods in Spegph Patholpgy. Harper and Row, Publishers: New York 1963 . Kirk, R. E., figpggimpntg; Qppign, Epogpdppgp for the fiehgviopg; Sgignggs. Wadsworth Publishing Company, Inc.: Belmont, California (1968). Kozhevnikow, V., and Chistovich, L., Spgegh, Aptigulation, gnd ggpgeption. (English transl. fr. Russian) Wash- ington, D. C.: U.S. Dept. Commerce (1965). Kronvall, E. L., and Diehl, C. F., The relationship of auditory discrimination to articulatory defects of children with no known organic impairment. J, Spgech Hgaring Dis., 19, 335-338 (1954). Lackner, J., A developmental study of language behavior in retarded children. Neuro s hol ia, 6, 301-320 (1968). Liberman, A., Harris, K. S., Eimas, P., Lisker, L., and Bastian, J., An effect of learning on speech perception: the discrimination of durations of silence with and without phonemic significance. ‘Lppg;_§pggg_. 4, 175-195 (1961). Liberman, A. M., Harris, K. 8., Hoffman, H. S., and Griffith, B. C., The discrimination of speech sounds within and across phoneme boundaries. J, EXp, 2sy§hol., Lieberman, A. M., Cooper, F. S., Harris, K., and McNeilage, P. M., A motor theory of speech perception. Prggeedipgs of the §peegh Cpppuniggtion Spaingr, Vol. 11. Stock- holm: Speech Transmission Laboratory, Royal Inst. Technol. (1963). Marks, L. E., and Miller, G. A., The role of semantic and syntactic constraints in the memorization of English sentences. J, Verb, Lrng, Verb, Behav., 3, 1-5 (1964). Marquardt. T., and Saxman, J., Language Comprehension and Auditory Discrimination in Articulation Deficient Kindergarten Children. J, §ppech Hearing Res., 15, 382-389 (1972). Martin, J. 6., Temporal word spacing and the perception of ordinary, anomalous, and scrambled strings. J, Vprb, Lrng, Verb, Behav., 7, 154-157 (1968). Martin, E., and Roberts, K. H., Grammatical factors in sentence retention. J, Verb. Lrpg, Verb. Behav., 5, 211-218 (1966). 127 Mange, C. V., Relationships between selected auditory perceptual factors and articulation ability. J, Spgech Hearing Res., 3, 67-74 (1960). Mase, D. J., Etiology of articulatory speech defects. New York: Teachers College, Columbia University, Bureau of Publications, Contributions to Education, No. 921 (1946). McCarthy, D. A., The language DeveIOpment of the Pre- sghool Child. Institute of Child Welfare Monograph Series, No. 4. Minneapolis: Univ. Minn. Press (1930). McNeill, D., Developmental psycholinguistics. In F. Smith and G. Miller, The Genesis of gingupge. Cam- bridge: M.I.T. Press (1966). Menyuk, P., A preliminary evaluation of grammatical capacity in children. J. Verb, Lrng. Verb. Behav., 2, 429-439 (1963). Menyuk, P., Syntactic structures in the language of children. J, Child Dev51., 34, 407-422 (1963). Menyuk, P., Comparison of grammar of children with functionally deviant and normal speech. J, Spgegh prging Res., 7, 109-121 (1964). Metraux, R., Auditory memory span for speech sounds: norms for children. J, Speegh Hearing Di§., 9, 31-38 (1944). Miller, G. A., Free recall of redundant strings of letters. Jg 3X2. PEZShOIQ, 56' 484-491 (1958). Miller, G. A., The magical number seven, plus or minus two: some limits on our capacity for processing information. Ppyghol, Rev., 63, 81-97 (1956). Miller, G. A., and Selfridge, J. A., Verbal context and the recall of meaningful material. Amer, J. Psychol., 63, 176-185 (1950). Morris, H. D., Common skills of children with cleft lips and palates. J, §peegh Hearing Res., 5, 79-90 (1962). Nurse, J. R., and Day, D. E., Imitation, comprehension, and production of grammatical structures. J, Verb, Lrngs verb: Behav., 10’ 68-74 (1971). Ohman, 8., Numerical model of co-articulation. J, Acoust. Soc, Amer., 41, 310-320 (1967). Perozzi, J. A., and Kunze, L. H., Relationship between speech sound discrimination skills and language abilities of kindergarten children. J. Speech Hearing Res., 14, 382-390 (1971). 128 Pollack, 1., Message uncertainty and message reception. J. Acoust, Sop, Amer., 31, 1500-1508 (1959). Powers, M. H., Functional disorders of articulation - symptomatology and etiology. In Travis, L. E., Hand- pggk of Speegh Patholpgy. New York: Appleton- Century-Crofts (1957 . Robbins, D. D., Importance of sensory training in speech therapy. J. Speech Hearing Dis., 7, 183-188 (1942). Schneiderman, N., A study of the relationship between articulatory ability and language ability. J. Speech Hearing Dis., 20, 359-364 (1955). Scholes, R., 0n functors and contentives in children's imitations of word strings. J, Verb, Lrng. Verb, Behav., 9, 167-170 (1970). Scholes, R., The role of grammaticality in the imitation of word strings by children and adults. J, Verb, Lrng, Eggb‘ BQDaVs' 8' 225-228 (1969). Schuckers, G., Shriner, T., and Daniloff, R., Auditory Assembly of segmented sentences by children. (g. Spgegh Hearipg Res., 16, 116-127 (1973). Shriner, T. H., A comparison of selected measures with psychological scale values of language development. J, Sppegh Hearing Res., 10, 828-835 (1967). Shriner, T. H., and Daniloff, R. G., Reassembly of seg- mented CVC syllables by children. J, Spgegh Hgaring Res., 13, 537-547 (1970). Shriner, T. H., Holloway, M. 8.. and Daniloff, R. G., The relationship between articulatory deficits and syntax in speech defective children. J, Speech Hearing Res., 12, 319-325 (1969). Slobin, D., and Welsh, C., Elicited imitation as a re- search tool in developmental psycholinguistics. Readings on Child Language Agguisition, Holt, Rinehart, and Winston, In press: New York (1970). Smith, C. R., Articulation problems and ability to store and process stimuli. J. Speech Hearing Res., 10, 348-353 (1967). Sperling, G., A model for visual memory tasks. Human PaCtOIB, 5, 19-31 (1963). Spriestersbach, D. C., and Curtis, J. F., Misarticulation and discrimination os speech sounds. Quart. Speech, 37, 483-491 (1951). 129 Spriestersbach, D. C., Darley, F. L., and Morris, H. L., Language skills in children with cleft palates. g, Speegh Hearing Reg., 1, 279-285 (1958). Templin, M., Certain Lgngugge §kills in Children. In- stitute of Child Welfare Monograph Series, No. 26, On. of Minneapolis Press: Minneapolis (1957). Templin, M. C., A study of sound discrimination ability of elementary school pupils. J, Speech Dis,, 8, 127- 132 (1943). Vandemark, A. A., and Mann, M. B., Oral language skills of children with defective articulation. J, Speech Hearing Res., 8, 409-414 (1965). Van Riper, C., and Irwin, J., Voice and Articulation. Englewood Cliffs, N. J. : Prentice-Hall (1958). Weiner, P. S., Auditory discrimination and articulation. J, Speegh Hegring Dis., 32, 19-28 (1967). Wepman, J. M., Auditory disgrimination test (genupl of Diregtions ). Chicago: Language Research Associates (1958). Williams, H. M., McFarland, M. L., and Little, M. F., Development of language and vocabulary in young children. Univ. Iowa Stud. Child Welfare, 13(2), (1937). Yedinack, J. G., A study of the linguistic functioning of children with articulation and reading disabilities. J, nget, Psychol., 74, 23-59 (1949). APPENDICES APPENDIX A SENTENCES LISTED ACCORDING TO SENTENCE TYPE APPENDIX A SENTENCES LISTED ACCORDING TO SENTENCE TYPE Well-Formed My mother washes the dishes. My dog runs. The boy plays. Her doll cries. The little boy pulls his red wagon. His brother climbs a tree. My black dog likes the new bone. Her big sister wants a new dress. The pretty teacher reads a funny story. The other girl lost her blue ball. Your father drives a car. Take the book. The boy rides a bike. Wash your face. My cat drinks the milk. Throw the ball. The old man takes a long walk. The girl wears a hat. Ill-Ordered His purrs cat. Eats bear the fresh honey the brown. Man dirty nice cleans shoes the his. Balloon happy the clown carries funny a. Hits car train the a. MY bring bikes Milk spills the. Choose friend a. Door close the. Carrot the bunny eats the. Kitten chases my her dog. Covers blanket baby the yellow the bed. Washes a the man big green car. Spins toy your. Brings a flower lady the. Baby his calls the mother. Sister my the spills water. Girl red cow silly draws the a. 132 133 Anomalous The tall tree washes the dirty dog. The chair flies. My yellow dress cries a large tear. Her big brother sings his black shoe. Read a paint. Drink the table. An airplane chops the wood. The dog purrs his bone. A leaf walks. Comb your teeth. A nice man wears his new horse. Her bike eats an apple. The happy children drink a green spoon. The tree hops. His sister bakes a book. The pretty picutre colors a little boy. The chair plays a drum. The lady walks her coat. APPENDIX B RANDOMIZATION OF SENTENCE TYPE, AND SENTENCE LENGTH APPENDIX B RANDOMIZATION 0F SENTENCE TYPE, AND SENTENCE LENGTH Sentence e No, of the Sentence Well-Formed l, 4, 5, 6, 9, 17, 18, 23, 25, 30, 35, 36, 39, 45, 46, 50, 53, 54. 111-Ordered 2, 10, 11, 14, 15, 16, 19, 20, 21, 22, 26, 27, 28, 29, 31, 37, 38, 52. Anomalous 3, 6, 8, 12, 13, 24, 32, 33, 34, 40, 41, 42, 43, 44, 47, 48, 49, 51. Spntenge Length Three-Word 2, 4, 6, 7, 9, 13, 16, 19, 20, 21 24, 29, 34, 39, 40, 44, 46, S3. Five-Word 1, 5, 15, 18, 22, 26, 31, 33, 36, 37, 38, 42, 45, 47, 49, 50, 51. Seven-Word 3, 8, 10, 11, 12, 14, 17, 23, 25, 27, 28, 30, 35, 41, 43, 48, 52, 54. 135 APPENDIX C INSTRUCTIONS GIVEN TO SUBJECTS APPENDIX C INSTRUCTIONS GIVEN TO SUBJECTS I want you to listen to what the man is saying on the tape recorder and then tell me exactly what you heard him say. If you can't remember everything, tell me as much of it as you can. For example, if the man said, “He likes food,“ what would you say? . Good: Let's try another. If the man said, “Find go him,“ what would you say? . Good! Let's do one more. If the man said, “The door plays,“ what would you say? . Fine, let's begin. 137 APPENDIX D ANSWER FORM USED TO TRANSCRIBE SUBJECTS' RESPONSES .usu cooum own use on» s uofiosz .con or» sod-o» as» anon uoxcuaa auo>oo .000 no: >E mousse convex .ueouo as: a wuss: wounds mun sum .oansu on» scene .ocoa son an» maxed moo xouan a: .oxu auso >cczn on» uouuso .onu moo-o noon .s unseen snoono 139 .uru madamm xaas .oouu s mass-u nonuoun mam .coms3 use as: «Head xon manuaa one .oxan means a: .s on» susuu use sud: Adv .nXHmu know 0:9 AOHV .mN .hm .mN .mN .vm .mm .NN .HN .ON .mH .ma .hH .md .md .s >c69u veauusu cacao on» huge: coo-Hem .ucasa s osom .0one xosan ea: emcee ussuoun mun uom .uwn on» noose aces-u sows muses on: .csoun on» keno: Susan one neon susm .uouuo aaov Hem .usou amus- s seauu ensue sodas» a: .ehsam Son 058 .noaau hence one .usn s sues: Huav 0:9 .usou moo >2 .moo huh-o on» asses: sous Has» 0&8 .uso suede a“: .uonuao on» nonns3 usnuoe at .vH .ma .NH .HH .0a .m .m .h .w .m .v .m .N .H .sne mun pH Amzv .euss sham Dena .sEUDa ovuuosum “0‘ ammodflfi mfl4m0 #m mmmzommmm .mBUmmem mmHmumzaaam 300 can anew .usoo no: exam: homH one .xaas wry mxcaup ago a: .Equ o mhsaa u-sfio one .son mauve” s nuanoo ou:uu«a >uuoua ens .xoon s mmxmn house» max .Vm .mm .mm .Hm .Om .mv .mv .hv .oosu uso>mnus3 .oxan a nooau son one .naon use» 0:8 .coome cosum s Refine couoaucu Seas: 0:9 .oaaas as sues oxen uox .omuOE 30s was mus03.cse so“: 4 .nusou usou £800 .xoon an» axoe .uous3 madden 0:» >6 usuewm .uonuoe any naaso nag knew .usu s ns>uuo wonueu ado» .aaan ones so: uuoa ands uorao one .oxasz used 4 .ocon can pause moo one .0003 on» usage ace-auas at .ozu >psa nuszoau s smeaum .xuouo >cc5u s nosou accuse» Savoua one .MSOS Sou usaam A.BZOUV Q NHGZQAA‘ .mv .mv .vv .mv .Nv .Hv .ov .mm .mm .hm .mm .mm .vn .mm .Nm .Hm .OM .mN APPENDIX E waggag'g AUDITORY DI§CRIMLNATION TE§T FORM I 142 Name: Age: Grade: Date: AUDITORY DISCRIMINATION TEST‘ FORM I X Y X Y 1. tub - tug 21. cat - cap 2. lack - lack 22. din - bin 3. web - wed 23. lath - lash 4. leg - led 24. bum - bomb 5. chap - chap 25. clothe - clov 6. gum - dumb 26. moon - noon 7. bale - gale 27. shack - sack 8. sought - fought 28. sheaf - sheath 9. vow - thou 29. king - king 10. shake - shape 30. badge - badge ll. zest - zest 31. pork - cork 12. wretch - wretch 32. fie - thigh 13. thread - shred 33. shoal - shawl l4. jam - Jam 34. tall - tall 15. bass - bath 35. par - par 16. tin - pin 36. pat - pet _ 17. pet - pack 37. muff - muss 18. dim - din 38. pose - pose 19. coast - toast 39. lease - leash 20. thimble - symbo 40. pen - pin Error Score: x: /30 Y: /10 * Typed copy of original Wepman form in order to accommodate margin size requirements.