- Email: [email protected]
The effects of semantic and phonemic prestimulation cues on picture naming in aphasia
BRAIN
AND
LANGUAGE
41, 496-509 (1991)
The Effects of Semantic and Phonemic Prestimulation Cues on Picture Naming in Aphasia MARK Department
of Communication
A. STIMLEY Disorders,
Indiana State University
AND
J. DOUGLAS NOLL Department
of Audiology
and Speech Sciences, Purdue University
In this study of auditory prestimulation cues, picture naming performances under phonemic and semantic conditions were compared to picture naming performance under a neutral condition. Twenty aphasic subjects named 324 pictures (108 pictures x 3 conditions) each. Responses were scored using a coding system adapted by the investigator from classification systems used by Williams and Canter (1982) and Kohn and Goodglass (1985). Results indicated that naming accuracy was facilitated by phonemic and semantic cues. An examination of the distribution of errors under the three conditions revealed systematic effects of phonemic and semantic cues on the frequency of occurrence of specific error types. Increases in semantic paraphasia proportion scores and decreases in unrelated word error proportion scores were associated with the semantic condition, while increases in phonemic paraphasia proportion scores were associated with the phonemic condition. The finding that naming performance of aphasic adults varies as a function of the type of information provided by the cue is discussed in relation to cascade visual confrontation naming models. 0 1991 Academic Press, Inc.
Several studies have examined the effects of phonemic and semantic cues on visual confrontation naming in aphasia (Li & Canter, 1983; Li & Williams, 1989; Love & Webb, 1977; Pease & Goodglass, 1978; Podraza & Darley, 1977; Rochford & Williams, 1962). Although repeatedly demonstrating that phonemic and semantic cues facilitate naming accuracy, This paper is based on a dissertation completed by Mark A. Stimley under the direction of J. Douglas No11and submitted to Purdue University in partial fulfillment of the requirement for the Ph.D. degree in May 1990. Requests for reprints should be addressed to Mark A. Stimley, Ph.D., Department of Communication Disorders, Indiana State University, Terre Haute, IN, 47809. 496 0093-934x/91 $3.00 Copyright 0 1991 by Academic Press. Inc. All rights of reproduction in any form reserved.
PRESTIMULATION
CUES AND NAMING
497
previous studies have generated only a minimal amount of data on which to base explanations for the effects of cues. One source of information that might prove useful in understanding the effects of cues is the analysis of errors made by aphasic subjects under different cued conditions. An examination of changes in the relative frequency of occurrence of specific error types that aphasic subjects make in response to various types of cues has the potential of providing a better understanding of the naming process, its disruption, and the effects of cues. Models of visual confrontation naming (Ellis, 1985; Forster, 1976; Morton, 1985; Riddoch & Humphreys, 1987; Stemburger, 1985) predict that aphasic naming performance will vary as a function of the type of information provided by different types of cues. Within these models, cues have their effects by temporarily reducing the consequences of breakdowns occurring at or between stages in the naming process involving (1) visual analysis of the stimulus to extract visual features, (2) recognition and/or structural categorization of the stimulus, (3) accessof semantic representations, and (4) access of corresponding phonological word forms in the phonological output lexicon. In addition to influencing the number of correct responses made, cues might also influence the proportions of specific error types made. Using these models, Howard and Orchard-Lisle (1984) hypothesized that phonemic cues operate at the level of accessing phonological word forms, because it is at this stage that phonologically coded information becomes available. If an aphasic adult has difficulty accessing a particular phonological word form, the combination of partial activation by the phonemic cue and activation from the semantic system may be sufficient to access the correct name. Semantic cues might be expected to have similar effects via improved access of semantic representations. If an aphasic adult has difficulty specifying complete semantic representations, the results of increased activation levels of semantic representations produced by the semantic cues would be evident in subsequent stages in the naming process. The purpose of the present study is to determine if error analysis reveals qualitative differences in picture naming attempts under phonemic cuing, semantic cuing, and neutral cuing conditions. If error distributions under phonemic and semantic cuing conditions are identical, one may infer that a single mechanism is responsible for the effects of cues and that phonemic and semantic cues do not differentially influence the visual confrontation naming process. However, if error distributions under the phonemic and semantic cuing conditions differ, separate mechanisms may be inferred. METHODOLOGY Subjects. The experimental sample consisted of 20 adult aphasic subjects (9 males and 11 females). As shown in Table 1, subjects ranged in age from 30 to 83 years (Mean = 65.5 years). Postonset times ranged from 1 to 156 months (Mean = 15.0 months). Aphasia
498
STIMLEY AND NOLL TABLE 1 IDENTIFKATION Age
Subject 1 2 3 4 5 6
7 8 9
10 11 12 13 14 15
16 17 18 19 20
Sex
(Years)
M F M F M F M F F F F M F F F M M F M M
67 81 30 65 56 76 60 75 58 74 68 72 68 64 74 69 80 39 51 83
INFORMATION
Months Postonset 9 5 56 4 12 156 4 3 4 2 12 7 4 6 5
11
FOR EACH SUBJECT
Aphasia type
Severity (Aphasia quotient)
Conduction Broca’s Broca’s Anemic Anemic Transcortical Motor Broca’s Anemic Broca’s Anemic Anemic Transcortical Motor Anemic Wernicke’s Conduction Anemic Anemic Anemic Wernicke’s Anemic
70.0 46.8 65.0 76.4 87.2 66.0 63.4 68.4 61.4 83.4 77.0 74.2 78.4 56.5 73.6 84.2 70.6 90.4 56.0 70.4
quotients, as determined by performance on the Western Aphasia Battery (WAB) (Kertesz, 1982), ranged from 46.8 to 90.4 (Mean = 71.0). There were 4 Broca’s, 2 transcortical motor, 2 conduction, 2 Wernicke’s, and 10 anemic aphasic subjects in the sample as determined by performance on the WAB. To be included in the study, each subject was required to pass three screening tests. In the first screening test, subjects were required to point correctly to at least seven of the first eight items on the “Discriminating between Paired Words” subtest of the Minnesota Test for Differential Diagnosis of Aphasia (Schuell, 1965). In this subtest, subjects were presented with a series of test plates that contained two pictures of common objects having similar sounding names and were asked to point to one of the pictures when given its name. The second screening test involved the same eight test plates. To pass this test, subjects needed to point correctly to at least seven of the eight target pictures when given their oral descriptions. To avoid having subjects make correct responses based on memory of performance on the first screening test, four of the objects that served as foils in the first test served as targets in this second test. In the third screening test, subjects were asked to name 15 pictures (cake, car, chair, bear, bird, bread, brush, key, lock, pan, piano, spider, telephone, train, and wagon). Each picture was presented one at a time simultaneously with a sentence completion cue (“It’s a -“). Subjects were given 15 set to respond to each stimulus item. To be included in this study, subjects needed to name correctly at least 3 but no more than 12, of the 15 pictures. After passing all three screening tests, severity and type of aphasia were determined using the WAB (Kertesz, 1982). Criterion for inclusion in this study was a score below 93.8, the aphasia cutoff score established for the WAB (Kertesz, 1982). Aphasia testing was completed
PRESTIMULATION
CUES AND NAMING
499
no more than 3 days before the initial data collection session. On the basis of screening and WAB criteria, 5 of 25 referred subjects were excluded. All subjects were native speakers of American English and were alert, cooperative, and medically stable at the time of testing. No subject demonstrated evidence of dementia, mental retardation, or psychiatric disorder as determined by medical history, neurological examination, and/or speech and language evaluation. No subject had uncorrected peripheral visual and/or peripheral hearing defects as determined by report. Whereas Kohn and Goodglass (1985) reported considerable similarity in the pattern of errors made by aphasic subjects from different diagnostic categories in response to picture naming tasks, and whereas previous cuing studies (Pease & Goodglass, 1978; Li & Williams, 1989) revealed considerable overlap in the diagnostic groups’ quantitative response to phonemic and semantic cues, no attempt was made in this study to group subjects according to aphasia type for data analysis purposes. Test stimuli. To elicit visual confrontation naming responses, 108 black and white line drawings having 90% or higher naming agreement were selected from a standard set of drawings (Snodgrass & Vanderwart, 1980) and were copied onto 10.2 x 12.7 cm cards. Pictures having names that began with vowels, that had compound words as names, or that did not allow the development of short simple semantic cues to be associated with them were excluded from use in this study. Frequency of occurrence of each target word as a noun was determined based on the word sample reported in Frequency Analysis of English Usage: Lexicon and Grammar by Francis and Kucera (1982). The target words ranged in frequency from 717 to less than one occurrences per million. Five words had frequencies of “less than one per million” and were assigned the value “one per million” so that the mean frequency of occurrence could be calculated. The mean frequency of occurrence for the 108 words was 64.1 per million. Cue assignment. Stimulus items were created by pairing the 108 pictures with phonemic, semantic, and neutral auditory cues. During the testing procedure, subjects were presented with the auditory cue before the picture to be named was presented. Immediately after the cue, the picture was presented simultaneously with a short carrier phrase to be completed by the subject with the name of the object in the picture. The carrier phrase (i.e., “It’s a -.“) served to indicate when subjects were to respond. Under the phonemic prestimulation condition, subjects were told what the first sound of the target word was before the picture stimulus was presented. The phonemic cue was presented to complete the carrier phrase “This is something that starts with (phonemic cue).” If the target word began with a single fricative, nasal, or liquid consonant, the phonemic cue consisted of the initial sound of the word in isolation. If the word began with a single stop or affricate consonant, the phonemic cue consisted of the initial consonant plus the neutral vowel /A/. If the target word began with a consonant cluster, the cue included all the initial sounds followed by the neutral vowel. Under the semantic prestimulation condition, subjects were told something about the object in the picture before the picture stimulus was presented. Semantic cues took the form of descriptive attributes (e.g., for kangaroo “. has a pouch,“) functional associates (e.g., for bell, “. . you can ring,“) or functional contexts (e.g., for tiger, “. lives in a jungle”). All of the semantic cues were designed to complete the carrier phrase, “This is something that. .” Under the neutral prestimulation condition, subjects were told “This is something that you will name” before the picture stimulus was presented. The presentation of the neutral prestimulation cue alerted subjects that a picture stimulus was about to be presented and gave them an auditory processing task to perform immediately before each naming attempt. Procedure. The 324 stimulus items (108 pictures X 3 conditions) were presented in a pseudorandom order determined by word frequency, cue order, and cue type. Before
500
STIMLEY AND NOLL
beginning the test, all subjects were informed of the nature of the task and were trained to perform the cued visual confrontation naming task. For each stimulus item within the experimental task, the examiner looked at the picture and provided the appropriate prestimulation cue. After providing the cue, the examiner placed the picture in front of the subject and gave the sentence completion cue to indicate to the subject to respond. The subject then had 15 set to respond. If the subject made a specific response in less than 15 set or failed to make a response, the picture was removed from the subject’s sight and the next stimulus item was presented. If the subject made a nonspecific response (described in the Appendix), the picture was not removed until the original 15 set response period elapsed, the subject made a specific response, or the subject indicated that he was not going to make another attempt to name the picture. Subjects were monitored for signs of fatigue and were provided breaks whenever necessary. Data collection with each subject was completed over a minimum of two sessions. The training procedure was presented at the beginning of each data collection session. The maximum length of any single test session was 45 min. Each subject was tested individually in a quiet room available at the different test sites. The subject’s responses were recorded on a Technics RSB29R tape recorder using a Sony ECM-155 microphone attached to the collar of the subject’s clothing and a Sony ECM-150 microphone placed on a stand on the table in front of the subject. Scoringprocedure. Responses were scored by listening to the tape recordings, determining what the first verbal response to each stimulus item was, and assigning a code to that response. Each response was assigned a single code using a classification system that was adapted by the investigators from classification systems used by Williams and Canter (1982) and by Kohn and Goodglass (1985). Responses were coded as being corrects, phonemic paraphasias, phonemically and semanticahy related errors, semantic paraphasias, phonemically flawed semantic paraphasias, morphologically and semantically related errors, partwhole errors, circumlocutions, unrelated word errors, neologisms, perseverations, nonspecific responses, and no responses. Descriptions used to classify each response are listed in the Appendix. Reliability. Interjudge and intrajudge reliabilities were determined to be 94 and 96%, respectively. Interjudge and intrajudge disagreements did not appear to be systematically related to the coding system or any error category within the system.
RESULTS Correct Responses
Results of a one-factor analysis of variance (ANOVA) with repeated measures for experimental conditions (Winer, 1971) on correct responses were significant (F(2, 38) = 5.25, p < .Ol). A Newman-Keuls test revealed that subjects named more pictures correctly under the phonemic (52.8) and semantic conditions (51.8) than under the neutral condition (46.5). No significant difference was observed in number of correct responses between the phonemic and the semantic conditions. Nonspecific and No Responses
Results of a series of one-factor analyses of variance (ANOVA) with repeated measures for experimental conditions (Winer, 1971) on arcsintransformed proportions of nonspecific responses, no responses, and combined nonspecific and no responses are summarized in Table 2. Of the three measures, the resulting F ratio for the condition on the proportion
PRESTIMULATION
501
CUES AND NAMING
TABLE 2 SUMMARY OF ANALYSES OF VARIANCE WITH REPEATED MEASURESFOR EXPERIMENTAL CONDITIONS ON NONSPECIFICRESPONSES,No RESPONSES,AND THEIR TOTALS
Sum of squares
df
MS
F
PR > F
Nonspecific errors Subjects Condition Error
7.103 0.289 0.819
19 2 38
0.1405 0.0216
6.70*
0.0032
No responses Subjects Condition Error
7.372 0.067 0.442
19 2 38
0.0335 0.0116
2.87
0.0688
Combined nonspecific errors and no responses 19 11.575 Subjects 2 0.075 Condition 38 0.890 Error
0.0375 0.0234
1.61
0.2141
Response type
Note. Criterion measure is arcsin-transformed proportion of error type.
* p < .Ol.
of nonspecific responses was the only one that was significant (F(2, 38) = 6.70, p < .Ol). A Newman-Keuls test revealed that subjects produced significantly greater proportions of nonspecific responses in the semantic condition (14.0%) than in the neutral (10.8%) and phonemic conditions (9.0%). Differences between the phonemic and the neutral conditions were not significant. Specific Errors Subjects produced a total of 2686 specific error responses. The proportions of specific error types made by the group under the three experimental conditions are reported as percentages of specific errors in Table 3. Semantic paraphasias, unrelated word errors, phonemic paraphasias, and perseverations were prominent in all three conditions comprising 26.77, 26.36, 16.90, and 14.67% of all specific errors made, respectively. To determine whether specific error hierarchies between conditions were similar, Spearman rank-order correlations based on the ranked order of these percentages were computed between the three conditions, with the following results: Semantic-Neutral, 0.93; Phonemic-Neutral, 0.93; Semantic-Phonemic, 0.98. Each of these is statistically signficant (‘JI < .Ol) and indicated that each experimental condition tends to evoke similar patterns of specific errors, even though the absolute amounts may differ across conditions.
502
STIMLEY AND NOLL TABLE 3 PERCENTAGE OF
Response type Phonemic paraphasias Phonemically and semantically related errors Semantic paraphasias Phonemically flawed semantic paraphasias Morphologically and semantically related errors Part-whole errors Circumlocutions Unrelated words Neologisms Perseverations Total number of specific errors
SPECIFIC ERROR RESPONSES IN EACH CONDITION
Semantic condition
Phonemic condition
Neutral condition
15.34% 0.23%
19.23% 0.47%
16.24% 0.41%
31.50% 3.75%
23.33% 2.81%
25.64% 3.98%
2.46%
2.23%
2.45%
0.82% 2.22% 22.60% 6.91% 14.17%
0.23% 2.11% 29.31% 6.33% 13.95%
0.31% 2.55% 27.07% 5.62% 15.73%
854
853
979
To examine the distributions of specific errors further, a series of onefactor analyses of variance (ANOVA) with repeated measures for experimental conditions (Winer, 1971) on arcsin-transformed specific error proportion scores was completed. Of the ANOVAs completed on the four prominent error categories, only three (semantic paraphasias, unrelated words, and phonemic paraphasias) indicated that significant differences existed among experimental conditions. For semantic paraphasias (F(2, 38) = 8.69, p < .OOl), a NewmanKeuls test revealed that subjects produced significantly greater proportions of semantic paraphasias in the semantic condition (36.2%) than in the neutral (29.4%) and phonemic conditions (27.2%). No significant difference was observed in the proportions of semantic paraphasias between the phonemic and the neutral conditions. For unrelated words (F(2,38) = 7.39, p < .Ol), results of the NewmanKeuls test indicated that subjects produced significantly smaller proportions of unrelated words in the semantic condition (19.3%) than in the phonemic (26.0%) and neutral (24.6%) conditions. No significant difference was observed in the proportions of unrelated words between the phonemic and the neutral conditions. For phonemic paraphasias (F(2, 38) = 4.80, p < .OS), the NewmanKeuls test revealed that subjects produced significantly greater proportions
PRESTIMULATION
CUES AND NAMING
503
of phonemic paraphasias in the phonemic condition (22.5%) than in the neutral (19.6%) and semantic (18.6%) conditions. No significant difference was observed in the proportions of phonemic paraphasias between the semantic and the neutral conditions. Of all the other specific error types, the only error type that appeared to be affected by experimental condition was the part-whole error. However, whereas part-whole errors were produced by only eight subjects and comprised less that one-half of a percent of all of the specific errors made in this study, they appear to contribute little to the understanding of the effects of phonemic and semantic cues and therefore will not be discussed further in this article. DISCUSSION Results indicate that aphasic naming performance varies as a function of the type of information provided by phonemic and semantic prestimulation cues. Naming performance under the phonemic condition was characterized by increased naming accuracy and an error distribution characterized by increased proportions of phonemic paraphasias. By contrast, naming performance under the semantic condition was characterized by increased naming accuracy and an error distribution characterized by increased proportions of semantic paraphasias and decreased proportions of unrelated word errors. The Effects of Phonemic Prestimulation
Cues
Results of this study indicate that subjects produced more correct responses under the phonemic than the neutral condition and thus confirm the general findings of previous studies on phonemic cues (Li & Canter, 1983; Li & Williams, 1989; Love & Webb, 1977; Pease & Goodglass, 1978; Podraza & Darley, 1977; Rochford & Williams, 1962). Nonsignificant differences in the proportions of combined nonspecific and no responses between the phonemic and the neutral conditions indicate that increases in the number of correct responses were not the result of an increased number of naming attempts under the phonemic condition. Subjects produced 126 more correct responses under the phonemic condition despite producing an identical number of specific responses under the phonemic and neutral conditions. Increased proportions of phonemic paraphasias in response to phonemic cues do not appear to be simply quantitatively related to increased successful word-finding attempts, but instead may reflect qualitative differences in word-finding and word-production processes. By comparison, increases in the number of correct responses under the semantic prestimulation condition occurred without concomitant increases in the proportion of phonemic paraphasias. Results of the current study support the hypothesis that phonemic cues
504
STIMLEY AND NOLL
influence that stage of the naming process involving phonological aspects of selecting and producing word level responses. Interpreted within the framework of cascade models of visual confrontation naming, activation at the phoneme level by the phonemic cue facilitates activation by the semantic system of phonological word forms that share the phoneme presented in the cue. Activation at the phoneme level by the phonemic cue also inhibits activation of phonological word forms that do not share the phoneme presented in the cue. “Noise” within the visual confrontation naming system is reduced not only by further visual processing of the target item (Riddoch & Humphreys, 1987), but also by inhibition of inappropriate phonological forms. As a result, the disordered naming system is more likely to make a response based on the appropriate phonological form. The observed increases in phonemic paraphasias also support a specific hypothesis made by Stemburger (1985) that activation from the semantic system is channeled through a speech output lexicon to phonemic level representations and that verbal responses are generated from these phonemic level representations rather than directly from the phonological forms located in the speech output lexicon. While reflecting increased accuracy in selecting and/or activating correct phonological word forms, increases in phonemic paraphasias also reflected the more generalized consequences of increased activation levels of particular phonemic level representations. Phonemic paraphasias may result from subjects producing verbalizations that have not been completely and accurately specified. The increases in phonemic paraphasias seen in this study suggest a failure to inhibit inappropriate but sufficiently activated phonemic level representations that may exist either before or after the phonological word form has been correctly selected. Howard and Orchard-Lisle (1984) note that phonemic cues could result in increased naming accuracy whether naming deficits result from failure to initiate verbal responses, raised thresholds in the output lexicon, and/or deficiencies in the verbal semantic system. Results of this study suggest that phonemic cues facilitate accessof phonological representations without specifying how or why access would have been insufficient without the cue. The Effects of Semantic Prestimulation Cues Results of this study indicate that subjects produced more correct responses under the semantic than under the neutral condition and thus confirm the general findings of previous studies on semantic cues (Pease & Goodglass, 1978; Li & Williams, 1989; Rochford & Williams, 1962). As under the phonemic condition, increases in naming accuracy are not explained by differences in the number of naming attempts made under the semantic condition. Subjects produced 106 more correct responses
PRESTIMULATION
CUES AND
NAMING
505
under the semantic condition despite producing 19 fewer specific responses. Findings of the current study also partially confirm those of the Podraza and Darley (1977) investigation of semantic prestimulation cues. Semantic cues in both studies resulted in increased semantic paraphasias even though the forms of the semantic cues were considerably different. In contrast to the findings of Podraza and Darley’s study, the error distribution under the semantic condition in the current study was characterized further by decreased proportions of unrelated paraphasias. An examination of data sets revealed that only one of Podraza and Darley’s subjects produced a sufficient number of unrelated word errors under the neutral condition to avoid a floor effect under the semantic condition. This single subject demonstrated the effect observed in the current study by producing 14 fewer unrelated paraphasias under the semantic condition than under the neutral prestimulation condition. Perhaps by coincidence, this subject was the only subject in the previous study to demonstrate a significant increase in naming accuracy in response to semantic cues. Results of this study support the hypothesis that semantic cues influence the access of information in the semantic system. Interpreted within the framework of cascade models of visual confrontation naming, activation of semantically related representations by the semantic cue facilitates the activation of semantically related representations by the visual recognition/categorization system and increases the amount of relevant information available on which to base the selection of the phonological word forms. “Noise” within the system is reduced not only by further visual processing of the target item (Riddoch & Humphreys, 1987), but also by the inhibition of semantically unrelated representations within the semantic system. As a result, subjects are more likely to make correct responses and semantic paraphasias and less likely to make unrelated word errors. Once a picture has been recognized, word-finding difficulties can arise either at the level of the semantic system or at the level of the speech output lexicon (Ellis & Young, 1988). One consequence of a breakdown at either of these levels is that the correct phonological representations of the target words are not sufficiently activated to allow production of correct verbal responses. Although semantic cues ultimately affect access of phonological word forms, semantic cues would not have resulted in the pattern of responses observed unless they had their initial influence within a defective semantic system. If the semantic system were already working properly, increased activation of semantic representations by semantic cues would not have significantly facilitated the access of phonological forms of target words and their close semantic associates (as indicated by increases in frequency of occurrence of correct responses and semantic paraphasias) and would not have significantly inhibited the access
506
STIMLEY AND NOLL
of phonological forms of unrelated words (as indicated by decreases in frequency of occurrence of unrelated word errors). Results of this study suggest that semantic cues facilitate accessof phonological representations by increasing the accuracy and completeness of semantic representations used in the access process. Implications
Regardless of the type of cue used in this study, cues ultimately increased the efficiency and the accuracy of accessing correct phonological word forms. Interpreted within the framework of interactive visual confrontation naming models, the results of this study suggest that phonological word form access is facilitated from the top-down by phonemic cues and from the bottom-up by semantic cues. In the current study, not all of the subjects demonstrated the effects that were reported for the group. The use of error analysis in studies based on matched groups of subjects from different aphasia diagnostic categories may facilitate identification of subject characteristics associated with different error distributions. Such studies may lead to a better understanding of naming disorders as they exist in the different types of aphasia. Although previous studies (e.g., Pease & Goodglass, 1978; Li & Williams, 1989) revealed considerable overlap between aphasia groups in their quantitative response to phonemic and semantic cues, similar studies using error analysis may be able to detect qualitative response differences between groups. APPENDIX: RESPONSE TYPES AND DESCRIPTIONS Correct Responses
Responses in which the subject correctly named the picture (minor dysarthric distortions were permitted). Phonemic Paraphasias
Responses that were approximations of the target word, with one or more of the phonemes in error (“taw” for saw). To be considered a phonemic paraphasia of the target word, a response needed to contain more than 50% of the target word’s sounds. Phonemically
and Semantically Related Errors
Responses that resulted from the deletion of specific sounds from the target word and that were also real words semantically related to the target (“toast” for toaster). Semantic Paraphasias
Responses that were superordinates (“plant” for tree), in-class coordinates (“spoon” for fork), contextual associates (“kitchen” for pan) of
PRESTIMULATION
CUES AND NAMING
507
the pictured object, or names of the material from which the object was made (“wax” for candle). Phonemically
Flawed Semantic Paraphasias
Responses that were approximations of semantic paraphasias, with one or more phonemes in error (“miger” for lion). Morphologically
and Semantically Related Errors
Responses that could be considered to be a semantic paraphasia of the target consisting of the root morpheme (the target word) plus other morphemes that substantially changed the meaning of the response (“screwdriver” for screw). Pluralized target words were not considered to be morphologically and semantically related errors. Part- Whole Errors
Responses that named a part of the picture rather than the whole picture (“door” for truck). Circumlocutions
Responses that were related to the target but did not involve an attempt to name the object in the picture consisting of an adjective (“heavy” for hammer), an adverb (“loudly” for drum), an -ing form of the verb in isolation (“drinking” for glass), or a long phrase of other multiword responses related to the target (“you wear it on your head” for crown). Unrelated Words
Recognizable word responses that showed no obvious phonological or semantic resemblance to the target words or object. Neologisms
Responses that were phonologically flawed to the point of not being clearly recognized as an attempt to produce an English word (“gloket” for hat). Perseverations
Whole word responses that had been used to name one of the previous five pictures. To be scored as a perseveration, responses that could have been scored as phonemic paraphasias, phonemically and semantically related errors, semantic paraphasias, phonemically flawed semantic paraphasias, morphologically and semantically related errors, part-whole errors, and circumlocutions had to have occurred previously as part of an established pattern of perseveration. To be considered part of an established pattern
508
STIMLEY AND NOLL
of perseveration, the response had to have been scored as a perseveration at least once within the last five responses. Correct responses were always scored as correct whether they appeared as part of an established pattern of perseveration or not. For unrelated words and neologisms, logic does not suggest that the repetition of a response in either of these error types was anything other than a perseveration. Repetition of unrelated words or neologisms within the five stimulus item limit were coded as perseverations. “No responses” and nonspecific responses were never coded as perseverations. Nonspecific Responses Responses in which the subject responded with a general comment or question (“I have one of those”), repeated part of the prestimulation cue, substituted a vague (general) word for the target word (“thing” for bed), produced a broken-off noun phrase with a modifier (It’s a big . . .), negated a specific response before it was made (It’s not a dog), attempted to spell the word, or produced open-syllable responses which did not contain more than 50% of the sounds in the target word. If a subject produced one or more nonspecific responses followed by a specific response within the first 15 set following the presentation of the stimulus, the nonspecific responses were discounted and the first specific response was scored. No Responses Responses in which the subject either failed to make a response (said nothing) or stated an inability to respond. REFERENCES Ellis, A. W. 1985. The production of spoken words: A cognitive neuropsychological approach. In A. W. Ellis (Ed.), Progress in the psychology of language. Hillsdale, NJ: Erlbaum. Vol. 2. Ellis, A. W., & Young, A. W. 1988. Human cognitive neuropsychology. Hillsdale, NJ: Erlbaum. Forster, K. I. 1976. Accessing the mental lexicon. In R. J. Walker (Ed.), Explorations in the biology of language. Montgomery, VT: Bradford Books. Francis, W. N., & Kucera, H. 1982. Frequency analysis of English usage: Lexicon and grammar. Boston: Houghton Mifflin. Howard, D., & Orchard-Lisle, V. 1984. On the origin of semantic errors in naming: Evidence from the case of a global aphasic. Cognitive Neuropsychology, 1, 163-190. Kertesz, A. 1982. Western Aphasia Battery. New York: Grune & Stratton. Kohn, S. E., & Goodglass, H. 1985. Picture-naming in aphasia. Brain and Language, 24, 266-283. Li, E. C., & Canter, G. J. 1983. Phonemic cueing: An investigation of subject variables. In R. Brookshire (Ed.), Clinical aphasiology conference proceedings. Minneapolis: BRK Pub.
PRESTIMULATION
CUES AND NAMING
509
Li, E. C., & Williams, S. E. 1989. The efficacy of two types of cues in aphasic patients. Aphasiology, 3, 619-626. Love, R., and Webb, W. 1977. The efficacy of cueing techniques in Broca’s aphasia. Journal of Speech and Hearing Disorders, 42, 170-178. Morton, J. 1985. Naming. In S. Newman and R. Epstein (Eds.), Current perspectives in dysphasia. New York: Churchill Livingston. Pease, D. M., & Goodglass, H. 1978. The effects of cuing on picture naming in aphasia. Cortex, 14, 178-189. Podraza, B. L., & Darley, F. L. 1977. Effect of auditory prestimulation on naming in aphasia. Journal of Speech and Hearing Research, 20, 669-683. Riddoch, M. J., & Humphreys, G. W. 1987. Picture Naming. In M. J. Riddoch & G. W. Humphreys (Eds.), Visual object processing: A cognitive neuropsychological approach. Hillsdale, NJ: Erlbaum. Rochford, G., & Williams, M. 1962. Studies in the development and breakdown on the use of names. I. The relationship between nominal dysphasia and the acquisition of vocabulary in childhood. Journal of Neurology, Neurosurgery, and Psychiatry, 25, 222227. Schuell, H. 1965. Minnesota Test for Differentia! Diagnosis of Aphasia. Minneapolis: Univ. of Minnesota Press. Snodgrass, J. G., & Vanderwart, M. 1980. A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174-215. Stemburger, J. P. 1985. An interactive activation model of language production. In A. W. Ellis (Ed.), Progress in the psychology of language. Hillsdale, NJ: Erlbaum. Vol. 2. Williams, S. E., & Canter, G. J. 1982. The influence of situational context on naming performance in aphasic syndromes. Brain and Language, 17, 92-106. Winer, B. J. 1971. Statisfical principles in experimental design. New York: McGraw-Hill.
AND
LANGUAGE
41, 496-509 (1991)
The Effects of Semantic and Phonemic Prestimulation Cues on Picture Naming in Aphasia MARK Department
of Communication
A. STIMLEY Disorders,
Indiana State University
AND
J. DOUGLAS NOLL Department
of Audiology
and Speech Sciences, Purdue University
In this study of auditory prestimulation cues, picture naming performances under phonemic and semantic conditions were compared to picture naming performance under a neutral condition. Twenty aphasic subjects named 324 pictures (108 pictures x 3 conditions) each. Responses were scored using a coding system adapted by the investigator from classification systems used by Williams and Canter (1982) and Kohn and Goodglass (1985). Results indicated that naming accuracy was facilitated by phonemic and semantic cues. An examination of the distribution of errors under the three conditions revealed systematic effects of phonemic and semantic cues on the frequency of occurrence of specific error types. Increases in semantic paraphasia proportion scores and decreases in unrelated word error proportion scores were associated with the semantic condition, while increases in phonemic paraphasia proportion scores were associated with the phonemic condition. The finding that naming performance of aphasic adults varies as a function of the type of information provided by the cue is discussed in relation to cascade visual confrontation naming models. 0 1991 Academic Press, Inc.
Several studies have examined the effects of phonemic and semantic cues on visual confrontation naming in aphasia (Li & Canter, 1983; Li & Williams, 1989; Love & Webb, 1977; Pease & Goodglass, 1978; Podraza & Darley, 1977; Rochford & Williams, 1962). Although repeatedly demonstrating that phonemic and semantic cues facilitate naming accuracy, This paper is based on a dissertation completed by Mark A. Stimley under the direction of J. Douglas No11and submitted to Purdue University in partial fulfillment of the requirement for the Ph.D. degree in May 1990. Requests for reprints should be addressed to Mark A. Stimley, Ph.D., Department of Communication Disorders, Indiana State University, Terre Haute, IN, 47809. 496 0093-934x/91 $3.00 Copyright 0 1991 by Academic Press. Inc. All rights of reproduction in any form reserved.
PRESTIMULATION
CUES AND NAMING
497
previous studies have generated only a minimal amount of data on which to base explanations for the effects of cues. One source of information that might prove useful in understanding the effects of cues is the analysis of errors made by aphasic subjects under different cued conditions. An examination of changes in the relative frequency of occurrence of specific error types that aphasic subjects make in response to various types of cues has the potential of providing a better understanding of the naming process, its disruption, and the effects of cues. Models of visual confrontation naming (Ellis, 1985; Forster, 1976; Morton, 1985; Riddoch & Humphreys, 1987; Stemburger, 1985) predict that aphasic naming performance will vary as a function of the type of information provided by different types of cues. Within these models, cues have their effects by temporarily reducing the consequences of breakdowns occurring at or between stages in the naming process involving (1) visual analysis of the stimulus to extract visual features, (2) recognition and/or structural categorization of the stimulus, (3) accessof semantic representations, and (4) access of corresponding phonological word forms in the phonological output lexicon. In addition to influencing the number of correct responses made, cues might also influence the proportions of specific error types made. Using these models, Howard and Orchard-Lisle (1984) hypothesized that phonemic cues operate at the level of accessing phonological word forms, because it is at this stage that phonologically coded information becomes available. If an aphasic adult has difficulty accessing a particular phonological word form, the combination of partial activation by the phonemic cue and activation from the semantic system may be sufficient to access the correct name. Semantic cues might be expected to have similar effects via improved access of semantic representations. If an aphasic adult has difficulty specifying complete semantic representations, the results of increased activation levels of semantic representations produced by the semantic cues would be evident in subsequent stages in the naming process. The purpose of the present study is to determine if error analysis reveals qualitative differences in picture naming attempts under phonemic cuing, semantic cuing, and neutral cuing conditions. If error distributions under phonemic and semantic cuing conditions are identical, one may infer that a single mechanism is responsible for the effects of cues and that phonemic and semantic cues do not differentially influence the visual confrontation naming process. However, if error distributions under the phonemic and semantic cuing conditions differ, separate mechanisms may be inferred. METHODOLOGY Subjects. The experimental sample consisted of 20 adult aphasic subjects (9 males and 11 females). As shown in Table 1, subjects ranged in age from 30 to 83 years (Mean = 65.5 years). Postonset times ranged from 1 to 156 months (Mean = 15.0 months). Aphasia
498
STIMLEY AND NOLL TABLE 1 IDENTIFKATION Age
Subject 1 2 3 4 5 6
7 8 9
10 11 12 13 14 15
16 17 18 19 20
Sex
(Years)
M F M F M F M F F F F M F F F M M F M M
67 81 30 65 56 76 60 75 58 74 68 72 68 64 74 69 80 39 51 83
INFORMATION
Months Postonset 9 5 56 4 12 156 4 3 4 2 12 7 4 6 5
11
FOR EACH SUBJECT
Aphasia type
Severity (Aphasia quotient)
Conduction Broca’s Broca’s Anemic Anemic Transcortical Motor Broca’s Anemic Broca’s Anemic Anemic Transcortical Motor Anemic Wernicke’s Conduction Anemic Anemic Anemic Wernicke’s Anemic
70.0 46.8 65.0 76.4 87.2 66.0 63.4 68.4 61.4 83.4 77.0 74.2 78.4 56.5 73.6 84.2 70.6 90.4 56.0 70.4
quotients, as determined by performance on the Western Aphasia Battery (WAB) (Kertesz, 1982), ranged from 46.8 to 90.4 (Mean = 71.0). There were 4 Broca’s, 2 transcortical motor, 2 conduction, 2 Wernicke’s, and 10 anemic aphasic subjects in the sample as determined by performance on the WAB. To be included in the study, each subject was required to pass three screening tests. In the first screening test, subjects were required to point correctly to at least seven of the first eight items on the “Discriminating between Paired Words” subtest of the Minnesota Test for Differential Diagnosis of Aphasia (Schuell, 1965). In this subtest, subjects were presented with a series of test plates that contained two pictures of common objects having similar sounding names and were asked to point to one of the pictures when given its name. The second screening test involved the same eight test plates. To pass this test, subjects needed to point correctly to at least seven of the eight target pictures when given their oral descriptions. To avoid having subjects make correct responses based on memory of performance on the first screening test, four of the objects that served as foils in the first test served as targets in this second test. In the third screening test, subjects were asked to name 15 pictures (cake, car, chair, bear, bird, bread, brush, key, lock, pan, piano, spider, telephone, train, and wagon). Each picture was presented one at a time simultaneously with a sentence completion cue (“It’s a -“). Subjects were given 15 set to respond to each stimulus item. To be included in this study, subjects needed to name correctly at least 3 but no more than 12, of the 15 pictures. After passing all three screening tests, severity and type of aphasia were determined using the WAB (Kertesz, 1982). Criterion for inclusion in this study was a score below 93.8, the aphasia cutoff score established for the WAB (Kertesz, 1982). Aphasia testing was completed
PRESTIMULATION
CUES AND NAMING
499
no more than 3 days before the initial data collection session. On the basis of screening and WAB criteria, 5 of 25 referred subjects were excluded. All subjects were native speakers of American English and were alert, cooperative, and medically stable at the time of testing. No subject demonstrated evidence of dementia, mental retardation, or psychiatric disorder as determined by medical history, neurological examination, and/or speech and language evaluation. No subject had uncorrected peripheral visual and/or peripheral hearing defects as determined by report. Whereas Kohn and Goodglass (1985) reported considerable similarity in the pattern of errors made by aphasic subjects from different diagnostic categories in response to picture naming tasks, and whereas previous cuing studies (Pease & Goodglass, 1978; Li & Williams, 1989) revealed considerable overlap in the diagnostic groups’ quantitative response to phonemic and semantic cues, no attempt was made in this study to group subjects according to aphasia type for data analysis purposes. Test stimuli. To elicit visual confrontation naming responses, 108 black and white line drawings having 90% or higher naming agreement were selected from a standard set of drawings (Snodgrass & Vanderwart, 1980) and were copied onto 10.2 x 12.7 cm cards. Pictures having names that began with vowels, that had compound words as names, or that did not allow the development of short simple semantic cues to be associated with them were excluded from use in this study. Frequency of occurrence of each target word as a noun was determined based on the word sample reported in Frequency Analysis of English Usage: Lexicon and Grammar by Francis and Kucera (1982). The target words ranged in frequency from 717 to less than one occurrences per million. Five words had frequencies of “less than one per million” and were assigned the value “one per million” so that the mean frequency of occurrence could be calculated. The mean frequency of occurrence for the 108 words was 64.1 per million. Cue assignment. Stimulus items were created by pairing the 108 pictures with phonemic, semantic, and neutral auditory cues. During the testing procedure, subjects were presented with the auditory cue before the picture to be named was presented. Immediately after the cue, the picture was presented simultaneously with a short carrier phrase to be completed by the subject with the name of the object in the picture. The carrier phrase (i.e., “It’s a -.“) served to indicate when subjects were to respond. Under the phonemic prestimulation condition, subjects were told what the first sound of the target word was before the picture stimulus was presented. The phonemic cue was presented to complete the carrier phrase “This is something that starts with (phonemic cue).” If the target word began with a single fricative, nasal, or liquid consonant, the phonemic cue consisted of the initial sound of the word in isolation. If the word began with a single stop or affricate consonant, the phonemic cue consisted of the initial consonant plus the neutral vowel /A/. If the target word began with a consonant cluster, the cue included all the initial sounds followed by the neutral vowel. Under the semantic prestimulation condition, subjects were told something about the object in the picture before the picture stimulus was presented. Semantic cues took the form of descriptive attributes (e.g., for kangaroo “. has a pouch,“) functional associates (e.g., for bell, “. . you can ring,“) or functional contexts (e.g., for tiger, “. lives in a jungle”). All of the semantic cues were designed to complete the carrier phrase, “This is something that. .” Under the neutral prestimulation condition, subjects were told “This is something that you will name” before the picture stimulus was presented. The presentation of the neutral prestimulation cue alerted subjects that a picture stimulus was about to be presented and gave them an auditory processing task to perform immediately before each naming attempt. Procedure. The 324 stimulus items (108 pictures X 3 conditions) were presented in a pseudorandom order determined by word frequency, cue order, and cue type. Before
500
STIMLEY AND NOLL
beginning the test, all subjects were informed of the nature of the task and were trained to perform the cued visual confrontation naming task. For each stimulus item within the experimental task, the examiner looked at the picture and provided the appropriate prestimulation cue. After providing the cue, the examiner placed the picture in front of the subject and gave the sentence completion cue to indicate to the subject to respond. The subject then had 15 set to respond. If the subject made a specific response in less than 15 set or failed to make a response, the picture was removed from the subject’s sight and the next stimulus item was presented. If the subject made a nonspecific response (described in the Appendix), the picture was not removed until the original 15 set response period elapsed, the subject made a specific response, or the subject indicated that he was not going to make another attempt to name the picture. Subjects were monitored for signs of fatigue and were provided breaks whenever necessary. Data collection with each subject was completed over a minimum of two sessions. The training procedure was presented at the beginning of each data collection session. The maximum length of any single test session was 45 min. Each subject was tested individually in a quiet room available at the different test sites. The subject’s responses were recorded on a Technics RSB29R tape recorder using a Sony ECM-155 microphone attached to the collar of the subject’s clothing and a Sony ECM-150 microphone placed on a stand on the table in front of the subject. Scoringprocedure. Responses were scored by listening to the tape recordings, determining what the first verbal response to each stimulus item was, and assigning a code to that response. Each response was assigned a single code using a classification system that was adapted by the investigators from classification systems used by Williams and Canter (1982) and by Kohn and Goodglass (1985). Responses were coded as being corrects, phonemic paraphasias, phonemically and semanticahy related errors, semantic paraphasias, phonemically flawed semantic paraphasias, morphologically and semantically related errors, partwhole errors, circumlocutions, unrelated word errors, neologisms, perseverations, nonspecific responses, and no responses. Descriptions used to classify each response are listed in the Appendix. Reliability. Interjudge and intrajudge reliabilities were determined to be 94 and 96%, respectively. Interjudge and intrajudge disagreements did not appear to be systematically related to the coding system or any error category within the system.
RESULTS Correct Responses
Results of a one-factor analysis of variance (ANOVA) with repeated measures for experimental conditions (Winer, 1971) on correct responses were significant (F(2, 38) = 5.25, p < .Ol). A Newman-Keuls test revealed that subjects named more pictures correctly under the phonemic (52.8) and semantic conditions (51.8) than under the neutral condition (46.5). No significant difference was observed in number of correct responses between the phonemic and the semantic conditions. Nonspecific and No Responses
Results of a series of one-factor analyses of variance (ANOVA) with repeated measures for experimental conditions (Winer, 1971) on arcsintransformed proportions of nonspecific responses, no responses, and combined nonspecific and no responses are summarized in Table 2. Of the three measures, the resulting F ratio for the condition on the proportion
PRESTIMULATION
501
CUES AND NAMING
TABLE 2 SUMMARY OF ANALYSES OF VARIANCE WITH REPEATED MEASURESFOR EXPERIMENTAL CONDITIONS ON NONSPECIFICRESPONSES,No RESPONSES,AND THEIR TOTALS
Sum of squares
df
MS
F
PR > F
Nonspecific errors Subjects Condition Error
7.103 0.289 0.819
19 2 38
0.1405 0.0216
6.70*
0.0032
No responses Subjects Condition Error
7.372 0.067 0.442
19 2 38
0.0335 0.0116
2.87
0.0688
Combined nonspecific errors and no responses 19 11.575 Subjects 2 0.075 Condition 38 0.890 Error
0.0375 0.0234
1.61
0.2141
Response type
Note. Criterion measure is arcsin-transformed proportion of error type.
* p < .Ol.
of nonspecific responses was the only one that was significant (F(2, 38) = 6.70, p < .Ol). A Newman-Keuls test revealed that subjects produced significantly greater proportions of nonspecific responses in the semantic condition (14.0%) than in the neutral (10.8%) and phonemic conditions (9.0%). Differences between the phonemic and the neutral conditions were not significant. Specific Errors Subjects produced a total of 2686 specific error responses. The proportions of specific error types made by the group under the three experimental conditions are reported as percentages of specific errors in Table 3. Semantic paraphasias, unrelated word errors, phonemic paraphasias, and perseverations were prominent in all three conditions comprising 26.77, 26.36, 16.90, and 14.67% of all specific errors made, respectively. To determine whether specific error hierarchies between conditions were similar, Spearman rank-order correlations based on the ranked order of these percentages were computed between the three conditions, with the following results: Semantic-Neutral, 0.93; Phonemic-Neutral, 0.93; Semantic-Phonemic, 0.98. Each of these is statistically signficant (‘JI < .Ol) and indicated that each experimental condition tends to evoke similar patterns of specific errors, even though the absolute amounts may differ across conditions.
502
STIMLEY AND NOLL TABLE 3 PERCENTAGE OF
Response type Phonemic paraphasias Phonemically and semantically related errors Semantic paraphasias Phonemically flawed semantic paraphasias Morphologically and semantically related errors Part-whole errors Circumlocutions Unrelated words Neologisms Perseverations Total number of specific errors
SPECIFIC ERROR RESPONSES IN EACH CONDITION
Semantic condition
Phonemic condition
Neutral condition
15.34% 0.23%
19.23% 0.47%
16.24% 0.41%
31.50% 3.75%
23.33% 2.81%
25.64% 3.98%
2.46%
2.23%
2.45%
0.82% 2.22% 22.60% 6.91% 14.17%
0.23% 2.11% 29.31% 6.33% 13.95%
0.31% 2.55% 27.07% 5.62% 15.73%
854
853
979
To examine the distributions of specific errors further, a series of onefactor analyses of variance (ANOVA) with repeated measures for experimental conditions (Winer, 1971) on arcsin-transformed specific error proportion scores was completed. Of the ANOVAs completed on the four prominent error categories, only three (semantic paraphasias, unrelated words, and phonemic paraphasias) indicated that significant differences existed among experimental conditions. For semantic paraphasias (F(2, 38) = 8.69, p < .OOl), a NewmanKeuls test revealed that subjects produced significantly greater proportions of semantic paraphasias in the semantic condition (36.2%) than in the neutral (29.4%) and phonemic conditions (27.2%). No significant difference was observed in the proportions of semantic paraphasias between the phonemic and the neutral conditions. For unrelated words (F(2,38) = 7.39, p < .Ol), results of the NewmanKeuls test indicated that subjects produced significantly smaller proportions of unrelated words in the semantic condition (19.3%) than in the phonemic (26.0%) and neutral (24.6%) conditions. No significant difference was observed in the proportions of unrelated words between the phonemic and the neutral conditions. For phonemic paraphasias (F(2, 38) = 4.80, p < .OS), the NewmanKeuls test revealed that subjects produced significantly greater proportions
PRESTIMULATION
CUES AND NAMING
503
of phonemic paraphasias in the phonemic condition (22.5%) than in the neutral (19.6%) and semantic (18.6%) conditions. No significant difference was observed in the proportions of phonemic paraphasias between the semantic and the neutral conditions. Of all the other specific error types, the only error type that appeared to be affected by experimental condition was the part-whole error. However, whereas part-whole errors were produced by only eight subjects and comprised less that one-half of a percent of all of the specific errors made in this study, they appear to contribute little to the understanding of the effects of phonemic and semantic cues and therefore will not be discussed further in this article. DISCUSSION Results indicate that aphasic naming performance varies as a function of the type of information provided by phonemic and semantic prestimulation cues. Naming performance under the phonemic condition was characterized by increased naming accuracy and an error distribution characterized by increased proportions of phonemic paraphasias. By contrast, naming performance under the semantic condition was characterized by increased naming accuracy and an error distribution characterized by increased proportions of semantic paraphasias and decreased proportions of unrelated word errors. The Effects of Phonemic Prestimulation
Cues
Results of this study indicate that subjects produced more correct responses under the phonemic than the neutral condition and thus confirm the general findings of previous studies on phonemic cues (Li & Canter, 1983; Li & Williams, 1989; Love & Webb, 1977; Pease & Goodglass, 1978; Podraza & Darley, 1977; Rochford & Williams, 1962). Nonsignificant differences in the proportions of combined nonspecific and no responses between the phonemic and the neutral conditions indicate that increases in the number of correct responses were not the result of an increased number of naming attempts under the phonemic condition. Subjects produced 126 more correct responses under the phonemic condition despite producing an identical number of specific responses under the phonemic and neutral conditions. Increased proportions of phonemic paraphasias in response to phonemic cues do not appear to be simply quantitatively related to increased successful word-finding attempts, but instead may reflect qualitative differences in word-finding and word-production processes. By comparison, increases in the number of correct responses under the semantic prestimulation condition occurred without concomitant increases in the proportion of phonemic paraphasias. Results of the current study support the hypothesis that phonemic cues
504
STIMLEY AND NOLL
influence that stage of the naming process involving phonological aspects of selecting and producing word level responses. Interpreted within the framework of cascade models of visual confrontation naming, activation at the phoneme level by the phonemic cue facilitates activation by the semantic system of phonological word forms that share the phoneme presented in the cue. Activation at the phoneme level by the phonemic cue also inhibits activation of phonological word forms that do not share the phoneme presented in the cue. “Noise” within the visual confrontation naming system is reduced not only by further visual processing of the target item (Riddoch & Humphreys, 1987), but also by inhibition of inappropriate phonological forms. As a result, the disordered naming system is more likely to make a response based on the appropriate phonological form. The observed increases in phonemic paraphasias also support a specific hypothesis made by Stemburger (1985) that activation from the semantic system is channeled through a speech output lexicon to phonemic level representations and that verbal responses are generated from these phonemic level representations rather than directly from the phonological forms located in the speech output lexicon. While reflecting increased accuracy in selecting and/or activating correct phonological word forms, increases in phonemic paraphasias also reflected the more generalized consequences of increased activation levels of particular phonemic level representations. Phonemic paraphasias may result from subjects producing verbalizations that have not been completely and accurately specified. The increases in phonemic paraphasias seen in this study suggest a failure to inhibit inappropriate but sufficiently activated phonemic level representations that may exist either before or after the phonological word form has been correctly selected. Howard and Orchard-Lisle (1984) note that phonemic cues could result in increased naming accuracy whether naming deficits result from failure to initiate verbal responses, raised thresholds in the output lexicon, and/or deficiencies in the verbal semantic system. Results of this study suggest that phonemic cues facilitate accessof phonological representations without specifying how or why access would have been insufficient without the cue. The Effects of Semantic Prestimulation Cues Results of this study indicate that subjects produced more correct responses under the semantic than under the neutral condition and thus confirm the general findings of previous studies on semantic cues (Pease & Goodglass, 1978; Li & Williams, 1989; Rochford & Williams, 1962). As under the phonemic condition, increases in naming accuracy are not explained by differences in the number of naming attempts made under the semantic condition. Subjects produced 106 more correct responses
PRESTIMULATION
CUES AND
NAMING
505
under the semantic condition despite producing 19 fewer specific responses. Findings of the current study also partially confirm those of the Podraza and Darley (1977) investigation of semantic prestimulation cues. Semantic cues in both studies resulted in increased semantic paraphasias even though the forms of the semantic cues were considerably different. In contrast to the findings of Podraza and Darley’s study, the error distribution under the semantic condition in the current study was characterized further by decreased proportions of unrelated paraphasias. An examination of data sets revealed that only one of Podraza and Darley’s subjects produced a sufficient number of unrelated word errors under the neutral condition to avoid a floor effect under the semantic condition. This single subject demonstrated the effect observed in the current study by producing 14 fewer unrelated paraphasias under the semantic condition than under the neutral prestimulation condition. Perhaps by coincidence, this subject was the only subject in the previous study to demonstrate a significant increase in naming accuracy in response to semantic cues. Results of this study support the hypothesis that semantic cues influence the access of information in the semantic system. Interpreted within the framework of cascade models of visual confrontation naming, activation of semantically related representations by the semantic cue facilitates the activation of semantically related representations by the visual recognition/categorization system and increases the amount of relevant information available on which to base the selection of the phonological word forms. “Noise” within the system is reduced not only by further visual processing of the target item (Riddoch & Humphreys, 1987), but also by the inhibition of semantically unrelated representations within the semantic system. As a result, subjects are more likely to make correct responses and semantic paraphasias and less likely to make unrelated word errors. Once a picture has been recognized, word-finding difficulties can arise either at the level of the semantic system or at the level of the speech output lexicon (Ellis & Young, 1988). One consequence of a breakdown at either of these levels is that the correct phonological representations of the target words are not sufficiently activated to allow production of correct verbal responses. Although semantic cues ultimately affect access of phonological word forms, semantic cues would not have resulted in the pattern of responses observed unless they had their initial influence within a defective semantic system. If the semantic system were already working properly, increased activation of semantic representations by semantic cues would not have significantly facilitated the access of phonological forms of target words and their close semantic associates (as indicated by increases in frequency of occurrence of correct responses and semantic paraphasias) and would not have significantly inhibited the access
506
STIMLEY AND NOLL
of phonological forms of unrelated words (as indicated by decreases in frequency of occurrence of unrelated word errors). Results of this study suggest that semantic cues facilitate accessof phonological representations by increasing the accuracy and completeness of semantic representations used in the access process. Implications
Regardless of the type of cue used in this study, cues ultimately increased the efficiency and the accuracy of accessing correct phonological word forms. Interpreted within the framework of interactive visual confrontation naming models, the results of this study suggest that phonological word form access is facilitated from the top-down by phonemic cues and from the bottom-up by semantic cues. In the current study, not all of the subjects demonstrated the effects that were reported for the group. The use of error analysis in studies based on matched groups of subjects from different aphasia diagnostic categories may facilitate identification of subject characteristics associated with different error distributions. Such studies may lead to a better understanding of naming disorders as they exist in the different types of aphasia. Although previous studies (e.g., Pease & Goodglass, 1978; Li & Williams, 1989) revealed considerable overlap between aphasia groups in their quantitative response to phonemic and semantic cues, similar studies using error analysis may be able to detect qualitative response differences between groups. APPENDIX: RESPONSE TYPES AND DESCRIPTIONS Correct Responses
Responses in which the subject correctly named the picture (minor dysarthric distortions were permitted). Phonemic Paraphasias
Responses that were approximations of the target word, with one or more of the phonemes in error (“taw” for saw). To be considered a phonemic paraphasia of the target word, a response needed to contain more than 50% of the target word’s sounds. Phonemically
and Semantically Related Errors
Responses that resulted from the deletion of specific sounds from the target word and that were also real words semantically related to the target (“toast” for toaster). Semantic Paraphasias
Responses that were superordinates (“plant” for tree), in-class coordinates (“spoon” for fork), contextual associates (“kitchen” for pan) of
PRESTIMULATION
CUES AND NAMING
507
the pictured object, or names of the material from which the object was made (“wax” for candle). Phonemically
Flawed Semantic Paraphasias
Responses that were approximations of semantic paraphasias, with one or more phonemes in error (“miger” for lion). Morphologically
and Semantically Related Errors
Responses that could be considered to be a semantic paraphasia of the target consisting of the root morpheme (the target word) plus other morphemes that substantially changed the meaning of the response (“screwdriver” for screw). Pluralized target words were not considered to be morphologically and semantically related errors. Part- Whole Errors
Responses that named a part of the picture rather than the whole picture (“door” for truck). Circumlocutions
Responses that were related to the target but did not involve an attempt to name the object in the picture consisting of an adjective (“heavy” for hammer), an adverb (“loudly” for drum), an -ing form of the verb in isolation (“drinking” for glass), or a long phrase of other multiword responses related to the target (“you wear it on your head” for crown). Unrelated Words
Recognizable word responses that showed no obvious phonological or semantic resemblance to the target words or object. Neologisms
Responses that were phonologically flawed to the point of not being clearly recognized as an attempt to produce an English word (“gloket” for hat). Perseverations
Whole word responses that had been used to name one of the previous five pictures. To be scored as a perseveration, responses that could have been scored as phonemic paraphasias, phonemically and semantically related errors, semantic paraphasias, phonemically flawed semantic paraphasias, morphologically and semantically related errors, part-whole errors, and circumlocutions had to have occurred previously as part of an established pattern of perseveration. To be considered part of an established pattern
508
STIMLEY AND NOLL
of perseveration, the response had to have been scored as a perseveration at least once within the last five responses. Correct responses were always scored as correct whether they appeared as part of an established pattern of perseveration or not. For unrelated words and neologisms, logic does not suggest that the repetition of a response in either of these error types was anything other than a perseveration. Repetition of unrelated words or neologisms within the five stimulus item limit were coded as perseverations. “No responses” and nonspecific responses were never coded as perseverations. Nonspecific Responses Responses in which the subject responded with a general comment or question (“I have one of those”), repeated part of the prestimulation cue, substituted a vague (general) word for the target word (“thing” for bed), produced a broken-off noun phrase with a modifier (It’s a big . . .), negated a specific response before it was made (It’s not a dog), attempted to spell the word, or produced open-syllable responses which did not contain more than 50% of the sounds in the target word. If a subject produced one or more nonspecific responses followed by a specific response within the first 15 set following the presentation of the stimulus, the nonspecific responses were discounted and the first specific response was scored. No Responses Responses in which the subject either failed to make a response (said nothing) or stated an inability to respond. REFERENCES Ellis, A. W. 1985. The production of spoken words: A cognitive neuropsychological approach. In A. W. Ellis (Ed.), Progress in the psychology of language. Hillsdale, NJ: Erlbaum. Vol. 2. Ellis, A. W., & Young, A. W. 1988. Human cognitive neuropsychology. Hillsdale, NJ: Erlbaum. Forster, K. I. 1976. Accessing the mental lexicon. In R. J. Walker (Ed.), Explorations in the biology of language. Montgomery, VT: Bradford Books. Francis, W. N., & Kucera, H. 1982. Frequency analysis of English usage: Lexicon and grammar. Boston: Houghton Mifflin. Howard, D., & Orchard-Lisle, V. 1984. On the origin of semantic errors in naming: Evidence from the case of a global aphasic. Cognitive Neuropsychology, 1, 163-190. Kertesz, A. 1982. Western Aphasia Battery. New York: Grune & Stratton. Kohn, S. E., & Goodglass, H. 1985. Picture-naming in aphasia. Brain and Language, 24, 266-283. Li, E. C., & Canter, G. J. 1983. Phonemic cueing: An investigation of subject variables. In R. Brookshire (Ed.), Clinical aphasiology conference proceedings. Minneapolis: BRK Pub.
PRESTIMULATION
CUES AND NAMING
509
Li, E. C., & Williams, S. E. 1989. The efficacy of two types of cues in aphasic patients. Aphasiology, 3, 619-626. Love, R., and Webb, W. 1977. The efficacy of cueing techniques in Broca’s aphasia. Journal of Speech and Hearing Disorders, 42, 170-178. Morton, J. 1985. Naming. In S. Newman and R. Epstein (Eds.), Current perspectives in dysphasia. New York: Churchill Livingston. Pease, D. M., & Goodglass, H. 1978. The effects of cuing on picture naming in aphasia. Cortex, 14, 178-189. Podraza, B. L., & Darley, F. L. 1977. Effect of auditory prestimulation on naming in aphasia. Journal of Speech and Hearing Research, 20, 669-683. Riddoch, M. J., & Humphreys, G. W. 1987. Picture Naming. In M. J. Riddoch & G. W. Humphreys (Eds.), Visual object processing: A cognitive neuropsychological approach. Hillsdale, NJ: Erlbaum. Rochford, G., & Williams, M. 1962. Studies in the development and breakdown on the use of names. I. The relationship between nominal dysphasia and the acquisition of vocabulary in childhood. Journal of Neurology, Neurosurgery, and Psychiatry, 25, 222227. Schuell, H. 1965. Minnesota Test for Differentia! Diagnosis of Aphasia. Minneapolis: Univ. of Minnesota Press. Snodgrass, J. G., & Vanderwart, M. 1980. A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174-215. Stemburger, J. P. 1985. An interactive activation model of language production. In A. W. Ellis (Ed.), Progress in the psychology of language. Hillsdale, NJ: Erlbaum. Vol. 2. Williams, S. E., & Canter, G. J. 1982. The influence of situational context on naming performance in aphasic syndromes. Brain and Language, 17, 92-106. Winer, B. J. 1971. Statisfical principles in experimental design. New York: McGraw-Hill.