Disability and Rehabilitation: Assistive Technology

ISSN: 1748-3107 (Print) 1748-3115 (Online) Journal homepage: http://www.tandfonline.com/loi/iidt20

Consistency and idiosyncrasy of semantic categorization by individuals with traumatic brain injuries Jessica Anne Brown, Karen Hux, Carrie Kenny & Trisha Funk To cite this article: Jessica Anne Brown, Karen Hux, Carrie Kenny & Trisha Funk (2015) Consistency and idiosyncrasy of semantic categorization by individuals with traumatic brain injuries, Disability and Rehabilitation: Assistive Technology, 10:5, 378-384 To link to this article: http://dx.doi.org/10.3109/17483107.2014.921250

Published online: 20 May 2014.

Submit your article to this journal

Article views: 30

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iidt20 Download by: [University of Birmingham]

Date: 05 November 2015, At: 18:37

http://informahealthcare.com/idt ISSN 1748-3107 print/ISSN 1748-3115 online Disabil Rehabil Assist Technol, 2015; 10(5): 378–384 ! 2014 Informa UK Ltd. DOI: 10.3109/17483107.2014.921250

ORIGINAL RESEARCH

Consistency and idiosyncrasy of semantic categorization by individuals with traumatic brain injuries Jessica Anne Brown1, Karen Hux1, Carrie Kenny1, and Trisha Funk2 1

Department of Special Education and Communication Disorders, Barkley Memorial Center, University of Nebraska-Lincoln, Lincoln, USA and Quality Living, Omaha, NE, USA

Downloaded by [University of Birmingham] at 18:37 05 November 2015

2

Abstract

Keywords

Purpose: Information about semantic categorization consistency may help practitioners to implement augmentative and alternative communication (AAC) options for people with traumatic brain injuries (TBIs). The researchers sought to determine the consistency and extent of general consensus agreement with which adults with TBI organize semantic information. Methods: The researchers compared the semantic categorization consistency of 10 participants with severe TBI to 10 neurotypical adults matched on age and gender. Participants performed a semantic categorization task three times over a 1-month period. The experimental task consisted of two stages: (a) sorting ordinate exemplars into superordinate categories and (b) sorting subordinate exemplars into the previously established ordinate categories. Results: Results showed that participants with TBI were less consistent across trials and more idiosyncratic than neurotypical peers in placing exemplars within categories. Although some participants with TBI achieved higher general consensus agreement scores with experimental task repetition, their performance did not reach levels comparable to those of neurotypical participants. Conclusions: Individually, semantic categorization patterns of some people with severe TBI conform to those of neurotypical adults; patterns of others do not. Some, but not all, survivors demonstrate increased consistency given task repetition. These findings have implications for AAC design and instruction for people with TBI.

AAC, semantics, traumatic brain injury History Received 5 December 2013 Accepted 1 May 2014 Published online 20 May 2014

ä Implications for Rehabilitation 





Clinicians should evaluate the manner in which an individual with TBI categorizes semantic information rather than assuming that he/she intuitively uses the hierarchical superordinate– ordinate–subordinate categorization pattern common to neurotypical adults. Clinicians should evaluate the consistency with which an individual with TBI categorizes semantic information before determining the manner of organizing content within an AAC system or device. When individuals with TBI display idiosyncratic and/or inconsistent patterns of semantic organization, clinicians should explore the possibility that repeated exposure to specific lexical items or direct instruction about categorization strategies will normalize and/or stabilize performance.

Introduction Many people who have experienced traumatic brain injuries (TBIs) have cognitive deficits persisting well beyond the initial stages of recovery and interfering substantially with communication abilities [1]. As a result, some individuals with TBI need to rely on augmentative and alternative communication (AAC) systems and strategies to supplement or replace natural speech. However, effectively communicating via AAC systems requires cognitive competencies that people with TBI may lack. For

Address for correspondence: Jessica Anne Brown, MS, Department of Special Education and Communication Disorders, Barkley Memorial Center, University of Nebraska-Lincoln, Lincoln, NE 68583-0738, United States. E-mail: [email protected]

example, researchers have found word retrieval, semantic organization, memory, theory of mind, and flexibility of thought challenges that affect the way survivors perform semantic categorization and word association tasks [2,3]. These idiosyncrasies in semantic organization may present barriers to the successful implementation and use of AAC, especially if adults with TBI vary their semantic organization across times, settings, or communicative situations or have patterns of semantic organization distinct from those typical of adults without disabilities. As such, the purpose of this study was twofold: (a) to examine the consistency over time with which adults with histories of severe TBI organize semantic information typical of that included in many AAC systems and (b) to compare the pattern of hierarchical structure adults with severe TBI use to organize semantic information with that of neurotypical adults.

Semantic categorization consistency

Downloaded by [University of Birmingham] at 18:37 05 November 2015

DOI: 10.3109/17483107.2014.921250

Common organizational strategies for AAC include basing systems on (a) common context, episodes, or topics [4–7]; (b) grammatical functions [4,8]; (c) semantic categories [4,9]; or (d) alphabetic arrangement [10]. Of these, semantic categorization is the strategy toward which professionals often gravitate when designing AAC for people with TBI, most likely because of the consistency with which neurotypical adults use hierarchical arrangements to organize semantic material [11,12]. Professionals may assume that the uniformity of hierarchical organization among neurotypical adults will make application of this framework appropriate and beneficial to people with TBI attempting to master AAC navigation and use. However, given the idiosyncrasies with which people with TBI perform semantic categorization tasks, this may not be the most efficient strategy for them. Indeed, structuring AAC support in a manner that contradicts a user’s internal semantic organization goes against the long-held tenet of striving for maximally transparent and intuitively obvious system design [13]. Memory impairments, concrete reasoning, and challenges with new learning may present additional obstacles to the effective use of AAC systems by people with TBI. The crux of the concern given these types of deficits is that a person will struggle with an organizational framework requiring substantial effort to master and retain during the learning process. To circumvent the effects of such challenges, AAC systems for survivors of TBI should minimize demands on new learning and require little training [13]. One method of minimizing learning demands and compensating for the persistent cognitive challenges and semantic organization idiosyncrasies of survivors of TBI is to personalize AAC devices, making the organization salient, the content personally relevant, and the navigation easy to learn and remember. Specifically, a professional can explore a person’s unique semantic organization and then use this to design an AAC system. However, this attempt at personalization does not guarantee success, because researchers do not know whether the unique organizational frameworks of people with TBI are consistent over time. Even the most specialized system will be ineffective if an individual’s categorization varies across time, settings, communication partners, or communication situations. Cognitive challenges and semantic idiosyncrasies of people with TBI make investigation of the consistency of their organizational strategies important, especially when considering the development of AAC applications and instruction about their use. In addition, professionals need to know about the likelihood that categorizing semantic information in a hierarchical arrangement will be congruent with the internal semantic networks of survivors. As such, the purpose of the present study was to examine these issues with a group of adult survivors of severe TBI.

Methods Participants Adults with TBI Participants included 10 male adults with histories of severe TBI. Severe TBI was defined in accordance with guidelines stipulated by Fortuny, Briggs, Newcombe, Ratcliff, and Thomas [14] regarding the duration of impaired consciousness; for the injury to be severe, the person must have experienced a loss of consciousness for more than 1 d or a state of post-traumatic amnesia lasting more than 1 week. All participants resided in a transitional living facility, were between the ages of 23 and 59 years (M ¼ 40.2; SD ¼ 12), and

379

were between 2 and 118 months post-injury (M ¼ 25; SD ¼ 41.6). Participants had between 12 and 18 years of formal education (M ¼ 13.5 years; SD ¼ 2.21). According to self-report and medical chart review, participants did not have developmental or learning disabilities, neurological impairments, or psychiatric issues prior to their injuries. All were native speakers of American English. Participants completed screening measures prior to enrollment to ensure adequate hearing, language, reading, and visual–spatial skills to perform the experimental tasks. All participants passed a hearing screening at 30 dB SPL for 500 Hz, 1000 Hz, and 4000 Hz in at least one ear. To rule out the possibility of aphasia, all participants performed the Aphasia Quotient portion of the Western Aphasia Battery-Revised (WAB-R) [15] and scored above 93.8. The researchers administered the Reading Subtest of the Wide Range Achievement Test 3 (WRAT3) [16] to confirm participants had at least a seventh grade reading level. Performance within normal limits on a line bisection task served to ensure the absence of any substantial visual neglect. Adults without TBI Neurotypical participants included 10 male adults. All were native speakers of America English and reported having no history of learning disabilities, neurological injuries, or psychiatric issues. Participants passed a hearing screening at 30 dB SPL for 500 Hz, 1000 Hz, and 4000 Hz in at least one ear prior to enrollment in the study. The neurotypical participants ranged in age from 21 to 55 years (M ¼ 36.8 years; SD ¼ 12.61) and had achieved between 12 and 18 years of education (M ¼ 14.5 years; SD ¼ 2.46). Computation of t-tests confirmed that the neurotypical adults did not differ significantly from the adults with TBI on age (t ¼ 0.618; p ¼ 0.5444) or education (t ¼ 0.935; p ¼ 0.3620). Materials Stimulus words The stimulus words comprised items within a hierarchical framework of semantic categories and were identical to those used by Wallace et al. [17]. To create the stimuli, Wallace and colleagues first selected five categories (i.e. people and animals, events, house, hobbies, and town), each of which was superordinate to five ordinate categories; this yielded a total of 25 ordinate-level categories. Then, Wallace and colleagues used a focus group of 12 adults without communication impairments to sort 175 potential subordinate words into the 25 ordinate categories. Final stimulus items included the 125 words that the focus group participants most consistently assigned to each ordinate category such that each of the 25 ordinate categories included five subordinate words. Because individual variations in the placement of words occurred among focus group participants, the final organization – as shown in Table 1 – represents a general consensus for semantic categorization rather than an absolute categorization. Additional details about the procedures for identifying and categorizing the stimulus items is available in Wallace and colleagues’ publication [17]. For this study, the researchers printed each of the five superordinate category labels, 25 ordinate stimulus words, and 125 subordinate stimulus words onto individual, unlined, 3  5-in. index cards using black, 20-point, Arial font. The superordinate category labels appeared on pink index cards, the ordinate stimuli appeared on blue index cards, and the subordinate stimuli appeared on yellow index cards as a means of reducing confusion between category levels during the performance of the experimental tasks.

Wedding: gown, reception, bride, groom, church Birthday: party hats, card, party favors, wrapping paper, presents Vacation: hotel, rental car, cruise, train, airplane Holidays: New Year’s Eve, Halloween, Valentine’s Day, Fourth of July, St. Patrick’s Day Graduation: diploma, tassel, degree, commencement ceremony, cap and gown Sports: volleyball, baseball, boxing, running, hockey Arts and crafts: sketching, sculpting, knitting, scissors, glue Reading: mystery books, newspaper, biography, novel, magazine Watching TV: game show, soap opera, action movie, comedic movie, TV show Games: backgammon, checkers, chess, poker, dominos Restaurants: Italian food, Mexican food, diner, Chinese food, fast food Grocery store: dairy, meat, vegetable, fruit, bakery Mall: music store, toy store, shoe store, clothing store, jewelry store School: English class, recess, gym class, math, science Post office: postage stamp, envelope, PO Box, mailbox, mail truck Kitchen: refrigerator, microwave, sink, cupboards, dishwasher Bathroom: shampoo, toothpaste, toothbrush, bath towel, mirror Bedroom: dresser, clothes closet, night stand, blanket, alarm clock Living room: rug, carpet, fireplace, coffee table, recliner Yard: garden, driveway, swing set, lawnmower, sprinklers

Events Hobbies Town House

Table 1. Superordinate, ordinate, and subordinate stimulus words.

Downloaded by [University of Birmingham] at 18:37 05 November 2015

Family: aunt, uncle, cousin, grandma, grandpa Friends: Beth, Alan, neighbor, Ryan, Katie Professionals: police officer, businessman, fireman, nurse, lawyer Farm animals: goat, sheep, pig, pony, horse Pets: cat, gerbil, lizard, turtle, gold fish

J. A. Brown et al.

People and animals

380

Disabil Rehabil Assist Technol, 2015; 10(5): 378–384

Procedures Experimental sessions The study included three experimental sessions occurring at least 1 week apart and lasting no more than 1 h per session. The examiner conducted each of the three sessions in two stages. Both stages included the same procedures but with different levels of stimuli. Hence, for Stage 1, participants organized ordinate stimulus words within superordinate categories, and, for Stage 2, participants organized subordinate stimulus words within ordinate categories. Using a laptop computer, the examiner recorded online all participant responses in a data collection table. The table organization mirrored the physical presentation of the superordinate and ordinate stimulus cards appearing in front of the participant. For Stage 1, the examiner placed the superordinate category stimulus cards on a wooden table in view of the participant and read the category labels aloud. Then, she handed the participant an ordinate-level stimulus card and read that word aloud. The participant’s task was to place the stimulus card below the superordinate category label to which he felt it had the closest association. The examiner recorded the response by modifying the data collection table to reflect each of a participant’s placement choices as it occurred. This procedure was repeated until the participant had placed the 25 ordinate-level stimulus words in the superordinate categories. No restrictions limited the number of ordinate words that could appear within a given superordinate category. The examiner did not impose a time limit, and a participant could change answers at any point prior to completion of Stage 1. All superordinate and ordinate cards remained visible to the participant throughout the subsequent Stage 2 procedures. For Stage 2, the examiner handed the participant 1 of the 125 subordinate stimulus cards and read the word aloud. The participant selected the ordinate level category – either verbally or by pointing – to which he believed the subordinate item was most closely associated. To prevent confusion from a large number of stimulus cards being present simultaneously, the subordinate words were not left on the table following selection of the desired ordinate category; instead, the examiner recorded each subordinate word placement in the data collection table and then removed the stimulus card from the participant’s sight. This procedure was repeated until the participant had indicated the desired location of the 125 subordinate stimulus words in the 25 ordinate categories. Once all subordinate stimuli were categorized, the examiner showed the participant her computer screen displaying the data collection table and allowed the participant to make any desired changes to his categorization choices. No restrictions limited the number of subordinate words that could appear within a given ordinate category, and the examiner did not impose a time limit. Each participant performed the experimental task three times to allow for the evaluation of the consistency of semantic organization over time. The examiner used the same stimuli during all experimental sessions but randomized the order of card presentation within the ordinate and subordinate stimulus sets. Between 7 and 14 d separated each experimental session to prevent participants from performing the task solely on the basis of recalling prior selections. Data analysis Overall ordinate and subordinate consistency proportions A participant’s responses during the first session served as a basis for comparison with responses during the second and third sessions. As such, Session 1 served as a means of establishing a participant’s baseline categorization pattern, and

Semantic categorization consistency

Downloaded by [University of Birmingham] at 18:37 05 November 2015

DOI: 10.3109/17483107.2014.921250

the researchers did not assign any score to that dataset. Scores computed following Sessions 2 and 3 served to quantify the extent to which a participant’s placement of ordinate and subordinate stimulus cards was consistent with his Session 1 organization. The researchers computed an overall ordinate consistency proportion (OOCP) for each participant. To do this, a participant earned one point for each ordinate stimulus card in Session 2 that he placed in the same superordinate category as selected in Session 1. The raw score, with a possible range from zero to 25 points, reflected the extent of ordinate consistency between Sessions 1 and 2. Second, a participant received one point for each ordinate stimulus card in Session 3 that he placed in the same superordinate category as selected in Session 1 and one point for each card placed in the same superordinate category as selected in Session 2. The resulting raw score, with a possible range from zero to 50 points, reflected the extent of ordinate consistency between Sessions 1 and 3 and between Sessions 2 and 3. Third, the researchers computed the OOCP by summing the ordinate consistency raw scores from Sessions 2 and 3 and dividing by 75; the resultant proportion, ranging from zero to one, reflected the participant’s overall consistency in categorizing ordinate-level stimuli across the three experimental sessions. The researchers followed comparable procedures to compute the overall subordinate consistency proportion (OSCP) for each participant. First, a participant earned one point for each subordinate stimulus card in Session 2 that he placed in the same ordinate category as selected in Session 1. The raw score, with a possible range from zero to 125 points, reflected the extent of subordinate consistency between Sessions 1 and 2. Second, a participant received one point for each subordinate stimulus card in Session 3 that he placed in the same ordinate category as selected in Session 1 and one point for each card placed in the same ordinate category as selected in Session 2. The resulting raw score, with a possible range from 0 to 250 points, reflected the extent of subordinate consistency between Sessions 1 and 3 and between Sessions 2 and 3. Third, the researchers computed the OSCP by summing the subordinate consistency raw scores from Sessions 2 and 3 and dividing by 375; the resultant proportion, ranging from zero to one, reflected the participant’s overall consistency in categorizing subordinate-level stimuli across the three experimental sessions.

381

Interjudge reliability Twenty-five percent of experimental sessions were randomly selected and recorded to allow for computation of interjudge reliability regarding data collection accuracy. A second member of the research team used the recordings to duplicate the data collection tables generated by the examiner during the actual experimental sessions. Referencing these tables, the researchers used a point-by-point agreement formula to compute interjudge reliability and found it to be 100% for ordinate stimulus categorization and 99.9% for subordinate stimulus categorization.

Results Overall ordinate and subordinate consistency proportions The range of OOCP scores was from 0.61 to 1.00 (M ¼ 0.808; SD ¼ 0.116) for participants with TBI and from 0.81 to 1.00 (M ¼ 0.920; SD ¼ 0.066) for neurotypical participants. 5 of the 10 participants with TBI did not achieve OOCP scores within the range of scores achieved by the neurotypical participants. Hence, as a group, the OOCP scores achieved by participants with TBI were lower than and approximately twice as variable as the scores achieved by neurotypical participants. A similar pattern with somewhat lower mean values for both participant groups emerged for the OSCP scores. Specifically, the participants with TBI earned OSCP scores ranging from 0.29 to 0.94 (M ¼ 0.716; SD ¼ 0.217), and the neurotypical participants earned scores ranging from 0.79 to 0.98 (M ¼ 0.909; SD ¼ 0.061). 5 of the 10 participants with TBI did not achieve OSCP scores within the range of scores achieved by neurotypical participants. The variability of OSCP scores for participants with TBI was approximately 3.5 times greater than the variability of these scores for neurotypical participants. Computation of two t-tests allowed determination of the presence of significant differences between the participant groups for the OOCP and OSCP scores. Both computations revealed significant group differences (OOCP: t ¼ 2.645, p ¼ 0.0165; OSCP: t ¼ 2.714, p ¼ 0.0142), with the neurotypical participants achieving higher consistency proportion scores on an average than the participants with TBI. Computation of Cohen’s d revealed large effect sizes associated with both these findings (OOCP: d ¼ 1.247; OSCP: d ¼ 1.025). Agreement with general consensus proportions

Ordinate and subordinate agreement with general consensus categorization The examiner compared each participant’s ordinate and subordinate responses separately from Sessions 1, 2, and 3 to the general consensus categorization obtained by Wallace and colleagues [17]. To do this, the examiner assigned one point to each ordinate stimulus card placed in the same superordinate category as the general consensus. This resulted in a raw score ranging from 0 to 25 points for each session. Dividing the scores by 25 yielded an ordinate agreement with general consensus proportion (OA) ranging from 0 to 1 for each experimental session. Likewise, the examiner assigned one point to each subordinate stimulus card placed in the same ordinate category as the general consensus subordinate stimuli. This resulted in a raw score ranging from 0 to 125 points for each session. Dividing the scores by 125 yielded a subordinate agreement with general consensus proportion (SA) ranging from 0 to 1 for each experimental session. The examiner utilized the OA and SA scores from each session to evaluate the extent to which a participant’s responses conformed to the general consensus responses over time regardless of response consistency across the three experimental sessions.

The mean, range, and standard deviation values for Sessions 1, 2, and 3 OA and SA scores achieved by participants in both groups are shown in Table 2. Across all sessions, participants with TBI achieved lower mean consensus proportions, wider ranges of

Table 2. Mean, range, and standard deviation values for sessions 1, 2, and 3 OA and SA scores for participants with and without TBI. OA scores

SA scores

Session number

Participants without TBI

Participants with TBI

Participants without TBI

Participants with TBI

1 Mean Range SD

0.904 0.80–1.00 0.076

0.816 0.64–1.00 0.125

0.896 0.74–0.95 0.060

0.737 0.36–0.91 0.177

2 Mean Range SD

0.956 0.84–1.00 0.063

0.844 0.64–1.00 0.124

0.921 0.82–0.97 0.051

0.827 0.47–0.97 0.142

3 Mean Range SD

0.972 0.92–1.00 0.037

0.880 0.76–1.00 0.098

0.945 0.90–0.98 0.028

0.850 0.60–0.96 0.105

Downloaded by [University of Birmingham] at 18:37 05 November 2015

382

J. A. Brown et al.

proportion scores, and larger standard deviation values than the neurotypical participants. Within their respective groups, participants tended to receive higher average OA and SA proportion scores as they progressed through the three experimental sessions. Despite this tendency for increased scores over the time, the mean OA and SA scores achieved by the participants with TBI in Session 3 remained lower than the respective scores achieved by the neurotypical participants in Session 1. Within sessions, mean scores for the ordinate categorization task were closer to the consensus than were mean scores for the subordinate categorization task for both participant groups. Computation of a 2  3 repeated measures ANOVA served to reveal whether significant differences existed in OA scores. Results confirmed a significant main effect both for group (F ¼ 7.400, p ¼ 0.0140, d ¼ 1.282) and for experimental session (F ¼ 6.488, p ¼ 0.0039); the interaction effect did not reach significance (F ¼ 0.243, p ¼ 0.7859). To determine which experimental sessions differed significantly, the researchers performed post-hoc analyses involving the computation of paired t-tests. Only the difference between Sessions 1 and 3 OA scores reached significance (t ¼ 4.616, p ¼ 0.0002), with the scores for Session 3 reflecting greater agreement with the general consensus than the scores for Session 1. Computation of Cohen’s d revealed a large effect size associated with this finding (d ¼ 2.118). The researchers performed a second 2  3 repeated measures ANOVA to determine the existence of significant differences regarding SA scores. Results confirmed a significant main effect both for group (F ¼ 6.694, p ¼ 0.0186, d ¼ 1.219) and for experimental session (F ¼ 14.198, p50.0001); the interaction effect approached but did not reach significance (F ¼ 2.835, p ¼ 0.0719). To determine which experimental sessions differed significantly, the researchers again performed post-hoc analyses involving the computation of paired t-tests. Significant differences emerged between Sessions 1 and 2 (t ¼ 3.236, p ¼ 0.0043) and between Sessions 1 and 3 (t ¼ 4.366, p ¼ 0.0003); the difference between Sessions 2 and 3 approached but did not reach significance (t ¼ 1.952, p ¼ 0.0658). For both groups, participants tended to perform the subordinate categorization task in a manner more similar to the consensus categorization given repetition of the experimental sessions. The computation of Cohen’s d revealed large effect sizes for both significant findings (Sessions 1 and 2: d ¼ 1.525; Sessions 1 and 3: d ¼ 2.058). The researchers examined OA and SA patterns of individual participants with TBI to explore further the tendency of these participants to approximate more closely the general consensus categorization over time. Specifically, the researchers used the OA and SA mean and standard deviation values of the neurotypical participants to serve as points of comparison and to allow for determination of the number of participants with TBI who progressed toward consensus performance over time. These OA and SA data are depicted graphically in Figure 1(a) and (b), respectively. Visual inspection of Figure 1(a) reveals that participants with TBI fell into one of the two subgroups. Specifically, half of the participants (#1, #6, #8, #9, and #10) achieved OA scores within one standard deviation of the neurotypical participants’ mean score for at least two of the three experimental sessions; the remaining five participants with TBI never achieved a score above one standard deviation below the mean and typically scored two or more standard deviations below the mean of the neurotypical participants. A similar pattern of two subgroups emerged regarding the SA scores (Figure 1b). 3 of the 10 participants with TBI (#6, #8, and #10) scored within one standard deviation of the neurotypical participants’ mean SA score for at least two of the three experimental sessions, and an additional three participants with

Disabil Rehabil Assist Technol, 2015; 10(5): 378–384

TBI (#2, #3, and #9) reached this level for at least one of the three experimental sessions. The remaining four participants with TBI persisted in achieving SA scores substantially lower than the neurotypical participants’ mean score across all sessions.

Discussion Research regarding semantic categorization consistency may help practitioners to understand some of the obstacles negatively affecting AAC use by adults with TBI. In particular, inconsistency in categorizing semantic exemplars or using a categorization schema other than the superordinate-subordinate hierarchy prevalent among neurotypical adults may hinder an individual’s success in locating vocabulary items within an AAC device. Therefore, the purpose of this study was to determine the consistency and extent of general consensus agreement with which adults with severe TBI organized semantic information typical of that included in many AAC systems. Overall, results showed that individuals with TBI were less consistent and more idiosyncratic than their neurotypical peers in placing exemplars within ordinate and superordinate categories. Furthermore, although some individuals with TBI achieved higher general consensus agreement scores with repetition of the experimental task, their performance did not reach a level comparable with that of their peers. Two observed patterns within the study results – participants’ consistency across sessions and participants’ agreement with the general consensus – warrant further discussion. Consistency across sessions As a group, study participants with TBI did not reach semantic categorization consistency levels comparable with their neurotypical counterparts. This confirms the findings of other researchers (e.g. [2,3]) who have examined semantic organization among survivors of TBI and found numerous idiosyncratic and egocentric factors affecting behavior. Further exploration of the consistency of semantic organization in the present study revealed two subgroups among the participants with TBI. One subgroup demonstrated comparable consistency as the neurotypical participants (i.e. they achieved consistency scores within the same range as the neurotypical participants), whereas the other did not. This suggests that semantic categorization may not be an obstacle for some survivors of TBI when considering AAC applications, but it may be for others. Hence, professionals need to determine whether an individual demonstrates difficulty with consistency when exploring potential implementation of AAC. Those with difficulty consistently organizing semantic information – 5 of the 10 participants with TBI in the current study – may have trouble efficiently locating stored messages within AAC systems. Questions remain, however, about whether those who demonstrate inconsistent semantic categorization behaviors can learn a neurotypical categorization schema. Future researchers should explore whether training methods for semantic categorization consistency are effective with this population. A limitation of the current study regarding the consistency findings was that the researchers used stimuli developed for a previous project (i.e. [17]). Using these stimuli did not account for an individual’s personal vocabulary knowledge and frequency of word use in daily conversation. At times, participants with TBI in the current study asked for clarification about the meaning of words presented, thus suggesting they were unfamiliar with some of the stimulus words. This raises concerns about the appropriateness of the stimulus set and limits generalization of the findings. Individuals with TBI may demonstrate different categorization patterns than those observed herein when given only personally relevant and familiar vocabulary.

Downloaded by [University of Birmingham] at 18:37 05 November 2015

DOI: 10.3109/17483107.2014.921250

Semantic categorization consistency

383

Figure 1. (a) Across-session progression of OA scores for individual participants with TBI in reference to mean and standard deviation values of participants without TBI. (b) Across-session progression of SA scores for individual participants with TBI in reference to mean and standard deviation values of participants without TBI.

Agreement with general consensus of categorization A second major finding from this research was that the participants with TBI demonstrated a different pattern of semantic organization than the traditional superordinate–subordinate hierarchy typical of normal adults. The robustness of hierarchical organization among neurotypical adults [11,12] makes this finding important, because it is likely to cause AAC facilitators to neglect consideration that a survivor might have an alternative semantic schema. Using a hierarchical organization pattern as the basis for data analysis in the current study revealed that both groups of study participants shifted toward increased agreement with the general consensus over time; however, the extent of this shift differed between those with and without TBI. Further exploration of the shift toward agreement with the general consensus pattern again revealed two subgroups among the participants with TBI – those tending to make this shift and those not doing so. Apparently, familiarity with the stimulus

words prompted some of the participants with TBI to modify their semantic organization strategy over time. To provide effective and efficient treatment, practitioners need information about the amount of exposure necessary for this shift to occur among adults with TBI. Hence, future researchers need to explore for whom such exposure by itself is adequate for prompting movement toward typical hierarchical schemas, those for whom additional instruction might prompt this change, and, perhaps, those for whom this change never occurs regardless of intervention. Underlying cognitive deficits affecting the theory of mind or flexibility of thought may have contributed to the tendency for some participants with TBI to fail to progress towards the general consensus pattern; however, cognitive processes were not measured in the current study, so only speculation about this contribution is possible. Future research is needed to determine the effects of underlying cognitive deficits on categorization choices and the feasibility of survivors mastering new semantic schemas.

384

J. A. Brown et al.

In addition to agreement with general consensus categorization, the plausibility of semantic categorization choices made by people with TBI should be a consideration when electing to use an individual’s personal categorization pattern for AAC purposes. Because AAC systems are commonly used as tools for enhancing interactions among a wide variety communication partners, message location that is transparent and understandable to more than just a principal user is beneficial. However, this does not mean that agreement with a general consensus organizational schema is the only viable means of categorizing messages within an AAC system. In the current study, participants who did not achieve agreement proportions within the normal range may still have made plausible categorization choices given that words often have multiple meanings and unique associations based on a person’s past experiences. Recognition of the acceptability of alternate categorization schemas is important when designing AAC systems.

Downloaded by [University of Birmingham] at 18:37 05 November 2015

General conclusions The consequence of inconsistent semantic categorization on efficient and effective AAC use by individuals with TBI is an area of research requiring further investigation. The results presented herein highlight the possibility that a person’s semantic categorization pattern and consistency may be idiosyncratic following TBI and, hence, may affect AAC use. As such, clinicians should be aware of the possibility that people with TBI may display unique and inconsistent categorization schemas. Providing treatment to increase the consistency of semantic categorization through direct instruction and repeated exposure to target words may be appropriate.

Disabil Rehabil Assist Technol, 2015; 10(5): 378–384

2. 3. 4. 5. 6. 7.

8.

9. 10. 11. 12. 13.

Acknowledgements The authors thank Dr. Snell and the staff and residents of QLI, for their support and assistance with this research.

14.

Declaration of interest

15.

The authors report no declarations of interest.

16.

References 1. Constantinidou F, Wertheimer JC, Tsanadis J, et al. Assessment of executive functioning in brain injury: collaboration between

17.

speech-language pathology and neuropsychology for an integrative neuropsychological perspective. Brain Injury 2012;26:1549–63. Hux K, Beukelman DR, Dombrovskis M, Snyder R. Semantic organization following traumatic brain injury. J Med Speech-Lang Pathol 1993;1:121–32. Snyder C, Hux K. Traumatic brain injury survivors’ ability to reduce idiosyncrasy in semantic organization. J Med Speech-Lang Pathol 2000;8:187–97. Beukelman DR, Mirenda P. Augmentative and alternative communication. 3rd ed. Baltimore: Paul H, Brookes Publishing Company; 2005. Goossens’ C. Aided communication intervention before assessment: a case study of a child with cerebral palsy. Augment Altern Commun 1989;5:14–26. Goossens’ CA, Crain S. Augmentative communication: intervention resource. Wauconda (IL): Don Johnson Inc.; 1986. Musselwhite CR, St. Louis KW. Aided communication systems and strategies. In: Musselwhite CR, St. Louis KW, eds. Communication programming for persons with severe handicaps vocal and augmentative strategies. 2nd ed. Austin (TX): Pro-Ed; 1988:177–265. Brandenburg S, Vanderheiden GC. Communication board design and vocabulary selection. In: Bernstein LE, ed. The vocally impaired: clinical practice and research. Orlando (FL): Grune & Stratton; 1998:84–135. Garrett K, Beukelman D, Low-Morrow D. A comprehensive augmentative communication system for an adult with Broca’s aphasia. Augment Altern Commun 1989;5:55–61. Damico JS, Muller N, Ball MJ, eds. The handbook of language and speech disorders. West Sussex: Wiley Blackwell; 2010. Raposo A, Mendes M, Marques JF. The hierarchical organization of semantic memory: executive function in the processing of superordinate concepts. Neuroimage 2012;59:1870–8. Rosch E. Cognitive representations of semantic categories. J Exp Psychol: General 1975;104:192–233. FriedOken M, Doyle M. Language representation for the augmentative and alternative communication of adults with traumatic brain injury. J Head Trauma Rehabil 1992;7:59–69. Fortuny LA, Briggs M, Newcombe F, et al. Measuring the duration of post traumatic amnesia. J Neurol Neurosurg Psychiatry 1980;43: 377–9. Kertesz A. Western Aphasia Battery – Revised (WAB-R). San Antonio (TX): Pearson Education; 2007. Wilkinson GS, Robertson GJ. WRAT4: Wide Range Achievement Test professional manual. Lutz (FL): Psychological Assessment Resources; 2006. Wallace SE, Hux K, Beukelman DR. Navigation of a dynamic screen AAC interface by survivors of severe traumatic brain injury. Augment Altern Commun 2010;26:242–54.

Consistency and idiosyncrasy of semantic categorization by individuals with traumatic brain injuries.

Information about semantic categorization consistency may help practitioners to implement augmentative and alternative communication (AAC) options for...
878KB Sizes 0 Downloads 3 Views