The word learning abilities of Persons with Aphasia (PWAs), including their neural underpinnings, connections to other cognitive-linguistic skills, and importance for treatment outcomes, have gained increased attention in recent years (e.g., Kelly & Armstrong, 2009). This increased interest is fueled by developments in neurocognitive studies of word learning in neurologically healthy people, which have illuminated the neural mechanisms that underlie vocabulary acquisition (Davis & Gaskell, 2009). Additionally, studies have shown how verbal short-term memory and other cognitive linguistic skills play a part in word learning (Martin & Saffran, 1999). Authors in this line of research believed that the ability to learn is an essential part of aphasia recovery (Hopper & Holland, 2005).
Studies on learning new words in aphasia have previously looked at expressive versus receptive and implicit versus explicit word learning. According to research, word learning in PWAs varies greatly; some learn solely receptively, while others also acquire words more expressively (e.g., Gupta, Martin, Abbs, Schwartz, & Lipinski, 2006; Tuomiranta et al., 2013). According to Tuomiranta et al. (2013), learning modalities such as auditory-phonological versus visual-orthographic input can also affect learning and increase individual variability.
Given this wide variability in learning profiles among PWAs, individualized and adaptive therapeutic interventions become essential. The authors claim that greater than 40% of PWAs manifest language impairments (Brown & Thiessen, 2018). These impairments lead to limited functional communication skills and restricted life participation (Brown & Thiessen, 2018). Speech and Language Therapy (SLT) will be provided to PWAs to address these deficits. One of the most commonly used therapeutic tools during SLT is the use of images, among the various techniques and assistance required for assessment, restoration, and compensation employed by speech language pathologists (SLPs).
In both formal and informal evaluations, images are crucial stimuli for evaluating language competence (Heuer, 2016; Heuer & Hallowell, 2007). When it comes to deficit reduction or restoration therapy, images serve as a key instrument. Images assist individuals with cognitive linguistic deficiencies by promoting augmentative and alternative communication (AAC), either temporarily or as a substitute for spoken speech, enhancing conversation skills, and improving confrontation naming abilities (Brown & Thiessen, 2018; Johnson, Hough, King, Vos, & Jeffs, 2008; Waller, Dennis, Brodie, & Cairns, 1998; Weissling & Beukelman, 2006).
Given the challenges PWAs face in processing symbolic information related to written script, auditory, and visuographic modalities, the use of images in speech and language therapy is essential. Given the general notion that PWAs could lack symbolic processing skills, methods that rely on nonlinguistic information presentation are likely to be beneficial to these individuals (McNeil & Pratt, 2001). Difficulties in comprehending visuographic signals and graphemes, coupled with the presence of Alexia, can restrict the effectiveness of utilizing written material in therapy (Beeson & Insalaco, 1998; Brown & Thiessen, 2018; Mayer & Murray, 2002). Consequently, incorporating visual stimuli (images) in interventions for PWAs enhances linguistic output and understanding, mitigating the necessity for damaged language systems in these individuals.
Images enable SLPs to leverage the comparatively intact visual processing abilities of PWAs. It is often accepted that PWAs exhibit greater dysfunction in the left hemisphere compared to the right hemisphere. The right hemisphere may facilitate enhanced visuospatial processing (Mesulam, 1981). Furthermore, it is hypothesized that PWAs possess intact cognitive abilities, enabling them to effortlessly discern the “gist” of images in contrast to other verbal tasks.
The efficacy of employing images in aphasia rehabilitation is additionally corroborated by the Resource Allocation Theory (RAT; Brown & Thiessen, 2018; McNeil et al., 2004; McNeil, Odell, & Tseng, 1991), which clarifies how cognitive resources (e.g., attention, memory, and visuo-spatial abilities) are allocated according to the task complexity and requirements. The RAT asserts that effective communication employs the previously described cognitive resources. For PWAs, whose cognitive capacities typically remain intact despite other language difficulties, using images as one of the AAC tools enables them to utilize their preserved capabilities to enhance their linguistic skills. Nonetheless, although images provide a potent method for PWAs to use their retained cognitive capacities, the efficacy of this strategy is significantly dependent upon the types of images employed.
Influence of Images in Aphasia Rehabilitation
Despite the crucial importance of images in aphasia rehabilitation, SLPs encounter difficulties in choosing images that are both clear and meaningful. It is essential to select images that accurately convey the intended words, messages, and concepts. This may pose challenges for PWAs due to varying linguistic, cognitive, and communicative requirements. As a result, one image may not be suitable for all PWAs (Brown & Thiessen, 2018). SLPs during therapy recommend meticulous selection of images according to the individual’s specific requirements and challenges, such as language, cognition, and sensory skills (Brown & Thiessen, 2018; Lasker & Bedrosian, 2001; Lasker & Garrett, 2006).
Image types come in a variety of forms, including black-andwhite line drawings and high-context images, such as photographs. Each category of visual support communicates distinct levels and types of information. ‘Content’ explicitly denotes the level of detail shown in an image, which may vary from minimal to highly intricate details (Brown & Thiessen, 2018; Knollman-Porter, Brown, Hux, Wallace, & Uchtman, 2016). ‘Context’ refers to the details depicted exclusively specific to the environment or backdrop of an image. Images with little or no context convey little contextual information, have little content, and do not provide identifiable information about people or places. In contrast, highcontext images generally feature elaborate backgrounds that convey information about environmental settings, thereby fabricating a scene or location, and frequently include several objects or individuals (Wallace, Hux, & Beukelman, 2010).
Each image style has linguistic benefits and challenges, but these impacts seem to be primarily reliant on task needs. According to researchers, line drawings or low-context visuals are best suited for communicating single words or ideas because they are simple and nuanced (Brown & Thiessen, 2018; Ma, Boyd-Graber, Nikolova, & Cook, 2009). Complex images, particularly high-context images, possess the capacity to convey greater substance and more intricate language compared to single-object, line-drawn images that lack contextual information (Ma et al., 2009). The use of iconic or low-context pictures as stimuli may necessitate significant language processing, which may limit the benefits they provide for PWAs.
Recent technological breakthroughs have transformed the methods by which individuals capture, disseminate, and utilize photographs (Engebretsen, Hartman, Beukelman, & Hux, 2014; van Dijck, 2008). The advent of digital photography resulted in the utilization of many types of photos overall (van Dijck, 2008). Images extend beyond mere memory preservation; they have significantly influenced communication. Furthermore, the proliferation of smartphones and social media applications has facilitated the shooting and sharing of photos. This discourse on imagery examines how images can serve as useful substitutes or supplements to verbal communication for both those with the presence or absence of neurological disabilities (Beukelman, Taylor, & Ullman, 2013).
Personalized images are considered more efficient than universal icons or photographs, even when the content and context are similar. Furthermore, researchers observed that the use of personalized images contributes to improved performance when matching words to images (McKelvey, Hux, Dietz, & Beukelman, 2010). Despite the image’s benefits, there are numerous constraints, such as the feasibility of capturing the images and incorporating them into daily interactions with PWAs.
The visual attention response of PWAs to images with a human figure either task-engaged or camera-engaged has been studied. The findings of these studies show that PWAs react to visual cues by enhancing visual attention on objects when a human image is task-engaged compared to when the human image is camera-engaged, parallel to persons without neurological disorders (Brown & Thiessen, 2018; Thiessen, Beukelman, Hux, & Longenecker, 2016; Thiessen, Beukelman, Ullman, & Longenecker, 2014). Although robust findings are present, it is essential to conduct additional research to ascertain the communicative value of visual cues as represented in images.
A recent study by Pillay, Van Der Linde, Graham, and Dada (2024) aimed to evaluate the practices of speech-language pathologists regarding image selection and its use in the treatment of PWAs. The authors found that more than 80% of SLPs found the use of colored images more beneficial than black and white line drawings in treating naming abilities in PWAs. Overall, images were primarily employed to support learning and language and were less commonly used in discourse tasks with PWAs, according to the study findings.
A study was conducted by Reymond et al. (2023) to evaluate the impact of photographic and graphic images on the naming accuracy and latency of naming nouns and verbs in PWAs. In addition, researchers conducted a comparison of the naming efficacy among PWAs and control groups. There was no discernible effect of either image category on the accuracy of naming. Based on the findings, researchers demonstrated that the disparities in naming accuracy and latencies across both image types are eliminated when the images are manipulated to eliminate features that enhance naming performance. In this study, researchers used image features that facilitate naming for both nouns and verbs respectively. As a result, it is not clear which features specifically aid the naming of different word classes.
Prior to this finding, Fonseca, de Miranda, Leal, Pinho e Melo, and Pavão Martins (2021) compared word retrieval abilities through real objects versus photographs. They found superior naming performance for real objects compared to color photographs of the same entities in PWAs. This implied that the effects of multisensory richness, or strong ecological validity, on word retrieval abilities were unique to real objects.
Deepak and Goswami (2020) explored the effect of semantic cues coupled with colored images in treating word retrieval abilities among PWAs. The study reported that word retrieval skills improved among all the PWAs when semantic cues and images were used, compared to their pre-treatment phases. Authors contend that use of images augmented the responses positively when analyzed through presentation of colored images of untrained probes also. Thus, the use of colored images is observed to have been beneficial in word retrieval treatment in PWAs.
Although prior research has established robust evidence of the beneficial effects of images in the rehabilitation of aphasia, additional investigation is required in terms of different kinds of images and which type will be more beneficial to the PWAs during clinical intervention. This encompasses the evaluation of diverse image categories, the investigation of various types and formats of personalized images, and the examination of images with varying levels of content and context.
While it is acknowledged that the utilization of images has evolved over time, there is less research examining the extent of use of images by individuals with neurological problems in their daily communication. Comprehending this would assist researchers and SLPs in utilizing imagery to facilitate cognition and communication post-brain injury. Research in this area has indicated that the design of images can affect the capacity of PWAs to comprehend the related words and concepts (Brown & Thiessen, 2018; McKelvey et al., 2010).
It is imperative to continue documenting and refining research in this field due to the heterogeneity of linguistic abilities of PWAs. SLPs and researchers will benefit from such research, which will provide valuable insights on practicing more effective assessment and intervention strategies to strengthen the outcomes for PWAs.
Present Study
In PWAs, picture-naming tasks are commonly used to assess language abilities, often relying on image-based stimuli. However, naming performance can be significantly influenced by perceptual factors such as color, contrast, imageability, and luminance. Traditionally, these tasks have used black-and-white line drawings, which are based on outdated norms and may not fully support naturalistic naming. Recent research suggests that real-life or photographic images, due to their richness in visual detail and personal relevance, can enhance naming performance by aiding memory retrieval and contextual understanding (e.g., McKelvey et al., 2010). Despite the promising results of the study, the findings appear to be preliminary due to the limited sample size and specific methodological constraints.
SLPs encounter difficulties in determining the most effective image forms, as there is a wide range of available formats, such as line drawings, graphical representations, and actual photographs. The inconsistent practices and outcomes are exacerbated by the ambiguity concerning which image types are the most advantageous for assessment and intervention. The effectiveness of these various image types in PWAs has been the subject of only a limited number of studies (e.g., Brown & Thiessen, 2018), and there is ongoing debate regarding their relative benefits for diagnosis and therapy.
A major gap in the literature is the lack of studies focusing on non-English-speaking populations, particularly Kannada speakers in southern India. Given that language structure (e.g., SOV in Kannada vs. SVO in English) may influence word learning and naming (Hills, Maouene, Riordan, & Smith, 2010), it is essential to use culturally and linguistically appropriate stimuli. Developing image databases that reflect regional language norms and cultural contexts can lead to more accurate assessments and effective interventions. Culturally relevant visual stimuli may also be more familiar and meaningful to Kannada-speaking PWAs, enhancing recognition, comprehension, and engagement during therapy. Images being language-free, the scope for their use widens, as even in bilingual or cross-linguistic contexts SLPs may certainly use images to elicit responses from PWAs. Research in this line would not only support the development of augmentative tools specific to regional and cultural contexts but also contribute to the global evidence based on aphasia rehabilitation. Therefore, the present study aims to investigate the effects of different image types – real images, black-and-white line drawings, and graphical images – on word training in Kannada-speaking PWAs. To achieve this, the subsequent objectives were assessed:
1. To investigate the efficacy of image training on naming accuracy across pre- and post-treatment phases for each type of image in PWAs.
2. To investigate the superiority of images on naming accuracy in the post- and follow-up treatment phases in PWAs.
METHODS
The current study implemented a non-randomized time series single-group treatment design. The study involved purposive sampling following strict inclusion and exclusion criteria, thus making it a non-randomized study. The study adhered to a single group of PWAs receiving treatment using three different types of images, and their effects were examined across pre-, post-, and follow-up phases. Hence, the design incorporated in the study is a non-randomized time series single-group treatment design. The use of diverse image types and examining their relative effectiveness in word retrieval in PWAs through a treatment paradigm is exploratory in nature. Thereby, it paves the way to explore the possible outcomes of the effect of images on word retrieval abilities in PWAs. Ethical approval for the study was granted by the board of Father Muller Charitable Institutions (FMIEC/CCM/087/2024, Protocol no: 054/2024).
Participants
In the current study, a total of 21 PWAs post left-hemispheric stroke were recruited, out of which 16 individuals (Males=12 and Females=4) were considered for the training protocol. These participants were aged between 35 to 59 years (M=45.81, SD=9.62). Five PWAs were excluded from the study owing to cognitive impairments and severe comprehension impairments. All PWAs recruited in the study were native Kannada speakers, a regional language spoken in southern India, and were residents of Karnataka State, India. Among these participants, ‘six’ were diagnosed with Broca’s aphasia, ‘seven’ were diagnosed with anomic aphasia, ‘two’ were diagnosed with Transcortical motor aphasia, and ‘one’ was diagnosed with conduction aphasia based on the Western Aphasia Battery-Kannada version (WAB-K) (Shyamala, Vijayashree, & Kumar, 2008). All participants recruited for the study were given informed consent before participation (Appendix 1).
The WAB-K was employed to evaluate the presence or absence of aphasia in the study. The study included only those PWAs with an auditory verbal comprehension score of ‘5’ or higher and an Aphasia Quotient (AQ) less than 93.4 (AQ above 93.4 is non-aphasic). This inclusion criterion served a dual purpose: first, to verify the presence of aphasia, and second, to document the severity of comprehension impairments. This approach allowed researchers to differentiate between aphasic and non-aphasic participants and to control comprehension deficits as a variable. Additionally, WAB-K was implemented to evaluate the type of aphasia and identify the presence of apraxia of speech (AOS). The study excluded participants diagnosed with Wernicke’s aphasia, global aphasia, or AOS due to their significant comprehension impairments, loss of verbal output, or the presence of speech programming deficits associated with AOS.
The participants recruited for the study were required to be premorbidly right-handed, as the study specifically aimed to examine outcomes related to left hemisphere stroke. Additionally, participants were expected to produce at least one-word responses in the spontaneous speech section of the WAB-K. This criterion was created in accordance with the training protocol for the study, which emphasized the training of words using different variants of images. Individuals with at least some spontaneous speech will be able to benefit from the intervention, as opposed to those with no verbal output at all (e.g., global aphasia). The researcher deployed PWAs who were a minimum of six months post-onset of stroke (M=9.37, SD=2.77) to mitigate the potential impact of spontaneous recovery. The modified Kuppuswamy Socioeconomic Scale (Bairwa, Rajput, & Sachdeva, 2013) ascertained the middle socioeconomic status of the participants recruited in the study. PWAs recruited in the study had a minimum of 12 years of formal education (M=15.87, SD=1.5). In the current study, participants with a history of psychiatric or psychological disorders were excluded. This was determined through detailed case history and discussions with both the participants and their care givers.
Cognitive impairment screening was conducted using the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005). The study included only participants who scored ‘26’ or higher to ensure normal cognitive functioning among the PWAs cohort. Participants were assessed with the Boston Naming Test-Kannada (BNT-K; Girish & Shyamala, 2015). The test assesses naming abilities in PWAs, if the participant’s score was on par with the standardized normative score (M=107.31, SD=4.37), then those participants were excluded. The average BNT score of the participants was M=80.18, SD=16.96. The PWAs selected for the study included persons with a medical history of diabetes and hypertension. The demographics of the participants are provided in Table 1.
Sample Size Estimation
The sample size of the study was established through the use of G Power software (Faul, Erdfelder, Buchner, & Lang, 2009). The power (1–β error probability) was set at 0.80 and the α error probability at 0.05 during the analysis. The suitable sample size for comparable research was also determined by reviewing pertinent literature (Harnish et al., 2014). Finally, 16 PWAs were considered as the ideal sample size for the study based on the G Power analysis.
Materials and Stimulus Selection
A total of 90 images along with their orthographic forms were selected from the BNT-K and from the Linguistic Profile Test-Kannada (LPT-K; Suchitra & Karanth, 1989). Further, these 90 images were equally categorized into three groups, namely real images, graphical images, and black and white line drawings. The entire set of images’ orthographic forms consisted of words with two to three syllables. The real images were designed and incorporated using stock image platforms (www.istockphoto.com); black and white line drawing images were taken from BNT-K and the image stock platform (www.istockphoto.com); and graphical images were constructed using Getty Images (www.gettyimages.ch) (Figure 1). Computer-generated images that frequently resemble symbols or simplified drawings and are not based on actual photographs are known as graphic images. Black-and-white line drawings are illustrations created by hand or by computers that simply use lines—no color or shading. Real images are photo-graphs of real objects that were captured using a camera.
Four native Kannada speakers and five SLPs evaluated these stimuli on a two-point rating scale for simplicity, appropriateness, familiarity, imageability, frequency, trainability, and stimulability. Stimuli with a score of ‘0’ are not entirely relevant, those with a value of ‘1’ are moderately relevant, and those with a score of ‘2’ are highly relevant. Point-to-point agreement was used to pick the stimulus, and the stimulus with a score greater than 90% was deemed the final stimulus.
A total of sixty images, including real images, graphical images, and black and white line drawings, make up the final test stimuli (Appendix 2). These are separated into three sets of twenty images each, consisting of real images, graphical images, and black-andwhite images.
Procedure
The current investigation was conducted in four distinct phases. Initially, standardized tests such as the MoCA and the WAB-K were used to validate the presence of aphasia and screen for potential cognitive deficits. The training program was meticulously arranged into distinct learning sessions. All these phases are explained in detail in the following section.
Phase 1-Pre-treatment:
PWAs were first assessed using 60 images (20 real, 20 graphical, and 20 black-and-white line drawings). Each image was shown on a laptop screen for 60 seconds in a quiet room, and participants were asked to name them. Correctly named items were excluded to avoid ceiling effects, while incorrectly named or unnamed items were selected for training. From the 60, a set of 30 images (10 from each type) was finalized for treatment. This phase took about 60 minutes in a single session.
Phase 2-Treatment:
After a two-day break to minimise exposure effects, participants underwent six therapy sessions across two weeks (1–3 sessions per week, ~60 minutes each). The treatment followed the protocol of Reymond et al. (2023) with modifications, such as different image types and extra training elements.
Training followed an ABA design:
Day 1: Train real images → Day 2: Probe real images
Day 2: Train graphical images → Day 3: Probe graphical images
Day 4: Train line drawings → Day 5: Probe line drawings
Day 6: Train and probe all image sets together
Image sets were counterbalanced during probing. Sessions were led by a certified speech-language pathologist experienced in aphasia rehabilitation. These responses were carefully recorded with a high-fidelity audio recorder (Sony ICD-UX570) for later transcription and to examine the naming accuracy.
During every training session, PWAs were seated comfortably in a noise-free, distraction-free setting. They were given explicit instructions prior to the training, which instructed them to carefully listen to the names of each image and to recall them orally. Additionally, as the study was not interested in correcting their articulatory or phonological errors, no particular training was provided. However, if participants needed help or felt that they were having trouble producing, they were provided verbal feedback and a brief drill to modify their responses.
Phase 3-Post-treatment:
Immediately after therapy, participants were tested again on the trained images of three types to measure improvement.
Phase 4-Follow-up:
One month later, the same trained images were tested to examine naming abilities of all three types of images, and progress over time was recorded.
Throughout the training, PWAs were consistently prompted to name each image displayed on the laptop screen. In cases where the PWA experienced difficulty, the author provided appropriate verbal cues to aid in response formulation before proceeding to the next stimulus. Importantly, progression to subsequent stimuli occurred regardless of the accuracy of the previous response, ensuring consistent exposure to all items the same training protocol was implemented across all six individual sessions, ensuring balanced presentation and practice for each image type. Upon completion of the training sessions, a two-day gap was introduced to minimize potential adaptations or practice effects. Following this interval, the post-treatment assessment was conducted to evaluate the naming performance of PWAs based on the trained stimuli (Figure 2).
A post-treatment evaluation was carried out as phase 3 of the study. The post-treatment phase, initiated after two days of cessation of training, gauged the participants’ naming performance on the trained stimuli across all three types of images. In this interval of two days, participants were not advised to practice or train on the images that were trained in the sessions. The task for this phase was to have PWAs look closely at each image on the laptop screen and name them; each image was displayed for 10 seconds with a maximum of 60 seconds to name each image. When the PWA was unable to respond and requested to view the image, they were re-presented by the researcher. Importantly, during this phase, no verbal cues or corrective feedback were provided, ensuring that the responses accurately reflected the participants’ stored knowledge without external interference. The verbal responses of all PWAs were later transcribed and separately analyzed by two trained SLPs for naming accuracy. Inter-rater agreement was assessed using Cohen’s kappa coefficient (k)=.84 (p<.001, 95% CI: .78-.90), indicating excellent agreement between raters. The assessment in this phase was completed in one session, which took approximately 30 to 45 minutes for each PWA.
The follow-up phase is the 4th phase of the study. In this phase, the performance of the trained images was evaluated one month after the cessation of treatment. During the follow-up phase, PWAs were instructed to name each of the previously trained images as they were randomly presented on a laptop screen. Each image was displayed for a duration of 10 seconds with 60 seconds of response time. No prompts, cues, or verbal feedback were provided in response to PWAs’ naming attempts to ensure that performance reflected independent recall without any external support. In instances where PWAs expressed a need to view an image again, the researcher accommodated their request by re-display-ing the stimulus. This approach maintained consistency with the post-treatment assessment procedure while allowing participant flexibility within the assessment protocol. However, PWAs were not recommended or encouraged to practice these images at home during the one-month period. This strategy was used to minimize the potential influence of home-based training on follow-up outcomes. The duration of the assessment typically ranged between 30-45 minutes for each PWA.
Scoring
In the current study, a binary scoring system was used to evaluate naming accuracy during pre-treatment, post-treatment, and follow-up phases. Each correct response was scored ‘1’ for accurate naming or distortions or omission errors. For incorrect or ‘no response’, PWAs received a score of ‘0’.
Data Analysis
Analyzed data were tabulated and subjected to appropriate statistical analysis using SPSS software version 22.0. Initially data was subjected to a normality check using the Shapiro-Wilk test. Owing to non-skewed data (p>.05), parametric tests were applied. One-way repeated measure ANOVA with adjusted Bonferroni’s corrections was performed on the data to analyze the differences between naming accuracy across pre-, post-, and follow-up phases for each type of image. Subsequently, one-way repeated measure ANOVA (multiple comparisons) with adjusted Bonferroni’s pairwise comparisons was performed to gauge the superiority of naming accuracy across each type of image in the post-treatment phase and follow-up phase in isolation. For results of one-way repeated measure ANOVA, the effect sizes at ŋp2 ≤.009, .06 and .1 were considered as low, medium and large effects, respectively (Norouzian & Plonsky, 2018).
RESULTS
A total of 1440 responses (90 responses by each participant) were analyzed from the naming task of three types of images across pre, post-, and follow-up phases of the study. Parametric tests were applied after confirming the normal distribution of data (p>.05). The results are described in two parts: a) the improvement in naming accuracy between pre- and post-phases and post- to follow-up phases for each type of image, and b) the superiority of image type on naming accuracy in the post- and follow-up phases in isolation.
Efficacy of Image Training on Naming Accuracy between Pre- and Post-Treatment Phases and Post-versus Follow-up Treatment Phases for Each Type of Image
The naming accuracy was observed to improve from the pretreatment phase to the post- and follow-up phases. Specifically, significant differences were noted in naming accuracy from pre- to post- and follow-up when analyzing real images naming abilities [F(2,30)=116.65, p<.001] with a large effect of ŋp2=.886. Post hoc Bonferroni’s adjusted pair wise comparisons revealed significant difference in naming accuracy between pre- and post-(p=.001). On observing the central tendency measures (Means and SD), post-treatment naming accuracy (M=68.75, SD=16.27) manifested improved scores compared to the pre-treatment (M=17.50, SD=5.16) phase. The naming accuracy was observed to be almost equivalent in the post (M=68.75, SD=16.27) and follow-up phases (M=62.50, SD=19.74) and not statistically significant (p=.06) for real image naming, suggesting a good maintenance effect. Results are tabulated in Table 2.
Similarly, significant departures in naming accuracy were observed from pre to post when analyzing graphical images [F(2,30)=172.39, p=.001] and line drawings [F(2,30)=124.09, p=.001], with large effect sizes of ŋp2=.406 and ŋp2=.896, respectively. Specifically, Bonferroni’s adjusted pair wise comparisons revealed significant variations between pre- and post-graphical image naming accuracy (p=.001), wherein post-treatment graphical image naming abilities showed robust scores (M=53.15, SD=3.50) compared to pre-treatment scores (M=20.26, SD=1.89). On comparing the post- and follow-up naming accuracy, follow-up scores (M=45.0, SD=3.50) showed a significant dip compared to the post-treatment phase (M=53.15, SD=3.50) (p=.001) (Table 2).
On analyzing the performance in naming accuracy of black and white line drawing images, significant improvement was noted in post-treatment naming accuracy scores (M=50.62, SD=11.81) compared to pre-treatment scores (M=14.06, SD=4.17). On comparing the post- and follow-up naming accuracy, follow-up scores (M=36.56, SD=10.75) showed significantly declined scores compared to the post-treatment phase (M=50.62, SD=11.81) at p=.001, akin to graphical image naming accuracy (Table 2).
The means and standard deviations were analyzed for each type of image, and a clear improvement was evinced from the pretreatment phase to the post-treatment phase, indicating an upward trend in word learning augmented through images among the PWA cohort (Figure 3). Further, the naming accuracy was examined across post- and follow-up phases for all the image types, wherein a clear maintenance effect was noted merely for real images (Figure 4). Specifically, the maintenance effect was more pronounced in word learning through real images compared to graphical and black and white line drawings. Figure 3 and Figure 4 illustrate the trend in improvement, and maintenance effect on all the types of images across the treatment phases, respectively.
Superiority of Images on Naming Accuracy in the Postand Follow-up Treatment Phases
The naming accuracy for each type of image was analyzed for the effect of superiority in the word-learning process among the PWA cohort. The results of a one-way repeated measure ANOVA (multiple measure comparison) on naming accuracy scores of the post-treatment phase revealed significant departures in naming accuracy across the types of images [F(2,30)=13.91, p=.001, ŋp2=.482] with a large effect (ŋp2>.1). Wherein, the naming accuracy of real images was significantly superior to graphical and black and white line drawings at the post-treatment phase (Table 3). Whereas the naming accuracy was equivalent between graphical and black and white line drawings, indicating minimal differences between these two image types compared to real images on naming accuracy observed at the post-treatment phase among PWAs in the cohort.
Further, on exploring the naming accuracy across the types of images in the follow-up phase, the results demonstrated a clear superiority effect of real images and a statistically significant difference over graphical and black and white line drawings [F(2,30)=20.92, p=.001, ŋp2=.582] with a large effect, akin to the pattern noted in the post-treatment phase. Further, post-hoc adjusted Bonferroni’s pairwise comparison revealed a significant difference between the naming accuracy of real images and graphical and black-and-white line drawing images (p=.000), with real images showing superior accuracy scores (Table 3). But no significant difference (p=1.00) was noted between graphical images and black-and-white line drawings in naming the accuracy of the PWAs. Figure 5 depicts the effect of the superiority of image type on word learning abilities in PWAs’ cohort based on the descriptive measures.
Overall, the results of the study demonstrated a varied effect of images on word training among the PWA cohort gauged in postand follow-up treatment phases. Specifically, the real images stood out significantly among the three image types in both phases. Interestingly, the trained words through black-and-white line drawings and graphical images were not maintained through post- to follow-up phases.
DISCUSSION
Three major findings emerge from the investigation of understanding the possible effect of image training on learning words in 16 PWAs. Firstly, the results showed there was a significant improvement from pre- to post-treatment phases across real images, black and white line drawings, and graphical images in general. Secondly, when documenting the superiority of images, real images manifested significant improvement in naming accuracy of PWAs compared to black and white line drawings and graphical images. Lastly, the study also gauged the maintenance effect across the image types, and the result revealed real image training yielded a better maintenance effect than black and white line drawings and graphical images.
The current study findings indicate that image training facilitates improved word learning performance in the naming accuracy of PWAs, which is in line with the previous literature (Ho, Weiss, Garrett, & Lloyd, 2005; Ma et al., 2009). PWAs often retain strengths in visual perception, recognition memory, and intellectual functioning, making visuographic materials a practical tool for communication support (Blake, 2005; Brookshire, 2003; McNeil, 1983; Murray, 1999). These intact abilities facilitate PWAs, even in severe cases, to recognize more easily, understand, and remember familiar words (Fox & Fried-Oken, 1996).
The utility of images in the rehabilitation of PWAs is paramount due to the difficulties they often experience with symbolic processing of auditory, written, and visuographic information. Aphasia is essentially a symbolic processing problem; PWAs benefit well from remedies that use nonlinguistic modes of information presentation. Specifically, impairments in interpreting visuographic symbols and graphemes, with co-occurring conditions like alexia, can limit the effectiveness of written text in therapy (Brown & Thiessen, 2018; Mayer & Murray, 2002). Using images to stimulate verbal expression and support language comprehension, SLPs can thereby reduce reliance on the impaired language systems of PWAs. Thus, the use of images in speech and language therapy can lead to improved performance. Furthermore, improved word learning from the pre- to post-treatment phase can be elucidated based on the use of images that the PWAs are trained with. Usually, the use of images inhibits the activation of similar neighborhood lexicons. Thus, participants were able to retrieve the intended word by inhibiting the undesired words (Puttanna, Al Harbi, Fernandes, & Dsilva, 2025).
Images serve as visual representations of specific linguistic content, and as such, they must be clear and meaningful to ensure that users can quickly interpret the intended messages, maximising communication efficiency. While many factors can affect how abstract an image appears, word class is a particularly well-established influence. Specifically, concrete words and concepts, such as nouns, tend to be easier to represent visually than more abstract ones, like verbs or adjectives (Koul & Harding, 1998; Schlosser & Sigafoos, 2002). Thus, the use of nouns in the present study may have contributed to better post-treatment performance across all image types. Additionally, our findings align partly with previous evidence that reliance on visual speech perception for the treatment could potentially improve speech production, possibly through activation of speech motor automatism through different channels (Fridriksson et al., 2009).
Real images generally yielded better performance than the other two images in both post-treatment and follow-up phases. This robustness may be due to the lesser cognitive load associated with real images. As we know, real images are encountered in daily life; thus, processing this type of image may pose less cognitive load than others. In contrast, line drawings and graphical images, even though they are true representations of the image, require more cognitive load, owing to their inherent features. Similarly, studies comparing photgraphs to black and white line drawings have shown advantages for using photographic stimuli (Griffith, Dietz, & Weissling, 2014; Heuer, 2016; Reymond et al., 2023). Furthermore, a study by Usinskiene, Mouthon, Martins Gaytanidis, Toscanelli, and Annoni (2019) found positive results using visual orthographic images to improve the language of PWAs. These results shed light on the importance of the use of real images and their role in improving linguistic competency in PWAs in general.
Real images demonstrated stronger maintenance effects compared to the other two image types. This advantage may stem from the robust connection between realistic images and the memory system (Kennedy, Most, Grootswagers, & Bowden, 2024). Because real images provide clear and accurate representations of familiar objects or items, they allow individuals to easily associate them with everyday experiences and functional use. This facilitates more effective encoding and supports long-term retention. In contrast, more abstract representations such as graphical images and black-and-white line drawings tend to be less concrete, which may place greater cognitive demands on PWAs. Consequently, graphical and black-and-white line-drawing image types may lead to weaker memory consolidation and reduced maintenance effects.
CONCLUSION
The findings of this study suggest that image-based training can significantly enhance naming abilities in PWAs, supporting the role of images as an effective tool in aphasia rehabilitation. Specifically, the results highlight the potential of images to elicit robust naming responses, thereby underscoring their value in improving language outcomes. This study contributes to the growing body of evidence emphasizing the importance of image selection in both the assessment and treatment of aphasia. Given the central role that visual stimuli play in clinical practice, it is essential for SLPs to understand how different image types—such as real images, black-and-white line drawings, and graphical images—differentially impact naming performance. The present findings offer preliminary insights into this area, particularly supporting the superiority of real images in facilitating naming compared to more abstract representations. Despite these promising results, the findings should be interpreted cautiously due to certain methodological limitations. The lack of randomization and the use of a heterogeneous participant group limit the generalizability of the results.
LIMITATIONS AND FUTURE DIRECTION
The present study has few limitations that should be acknowledged. First, it employed a single-group design without a control group and lacked randomization. The absence of a comparison group limits the ability to attribute observed effects solely to the intervention. Incorporating a randomized controlled trial (RCT) design in future research would enhance the internal validity and provide more reliable insights into the performance trends associated with different image types, particularly in comparison to typically developing or neurologically healthy individuals.
Second, the study assessed the maintenance effects of the intervention only up to a one-month follow-up. While this provides some understanding of short-term retention, it does not capture the longer-term durability of the treatment effects. Extending the follow-up period to three or six months post-intervention would offer a more comprehensive view of sustainability of gains and whether the benefits of real image-based interventions persist over time.
Moreover, the current research focused broadly on word training through images without differentiating between classes of words. Future studies could examine how image type influences the naming abilities of various word classes, such as nouns, verbs, and adjectives, since these categories differ in terms of abstractness and cognitive processing demands. Determining whether certain image types (e.g., real images vs. line drawings) are more effective for particular grammatical categories could inform more tailored and effective intervention strategies. Future research with more rigorous designs, including randomized controlled trials and stratified sampling based on aphasia subtypes, is needed to validate and expand upon these initial observations.
The participant group was not stratified based on aphasia subtype. Since fluent and non-fluent aphasia differ significantly in terms of linguistic profiles and cognitive strengths, it would be valuable for future research to focus on a specific subtype. For example, studying the effect of image type exclusively within individuals with non-fluent aphasia may yield clearer patterns and improve clinical applicability. This line of research would help refine image-based therapy approaches and better match them to the unique needs of different aphasia populations.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation 
Supplement
Print 
