Perceptions of People with Aphasia about Supporting Reading with Text-to-speech Technology: A Convergent Mixed Methods Study

Abstract

Introduction:

Text-to-speech (TTS) technology is a possible reading support for people with aphasia; however, adoption for functional reading remains rare, and people with aphasia may have limited knowledge about TTS. Given this situation coupled with inherent communication challenges associated with aphasia, the purpose of this convergent mixed methods study was to explore the perceptions of participants about desired features, benefits, and drawbacks of TTS technology after having engaged in one-on-one education and guided practice activities.

Methods:

Nineteen adults with chronic aphasia completed a single education and guided practice session followed by close-ended questions and participation in a semi-structured interview to explain preferences, concerns, beliefs, and opinions about potential TTS technology benefits and drawbacks. Three participants had previously used TTS technology for functional reading; all others had some prior exposure but did not use a system for functional purposes.

Results:

Seventeen of 19 participants expressed TTS technology interest after education and guided practice activities. Participants endorsed selection of a preferred voice, control of speech output rate, and highlighting as priority features. Frequently endorsed benefits were improved comprehension and increased reading independence; some participants believed they would succeed in reading a greater variety of materials, communicate more with others, participate in more reading activities, and/or read faster. The greatest concern was mastering device operation; other concerns related to understanding the voice output, needing another person’s help for system use, and matching the speech output rate to a preferred reading rate.

Conclusions:

Overall, most participants had positive perceptions about possible benefits afforded by TTS technology. Practitioners need to provide opportunities for people with aphasia to learn about and explore TTS systems to determine whether adoption is desired.

1.0. Introduction

Loss of proficient word decoding and reading comprehension is common with acquired aphasia (Webster et al., 2020). Despite remediation, many people with aphasia experience persistent challenges and struggle to resume independent reading (Keelor et al., 2013; Kjellén et al., 2017). Chronic reading challenges result in decreased quality of life, participation in desired and essential activities, and engagement with others (Keelor et al., 2013; Lynch et al., 2013; Worrall et al., 2011). This has prompted exploration of compensatory strategies via assistive technology (e.g., text-to-speech (TTS) technology) as one technique for supporting reading (Brown et al., 2019; Caute et al., 2016; Harvey et al., 2013; Hux et al., 2020; Knollman-Porter et al., 2019; Wallace et al., 2019). However, successfully implementing technology-based compensations is dependent on several personal and environmental factors (Lasker & Bedrosian, 2001; Scherer, 1993). People with aphasia must be willing to learn and capable of mastering independent device operation, and the benefit derived must be enough to encourage continued use. Also, devices must have sufficient adaptability of user-adjustable features to meet individual needs. Concerns about the adaptability, challenges, and benefits people with aphasia perceive about TTS systems and features prompted this convergent mixed methods study.

TTS technology has the potential to support reading by people with acquired literacy challenges in much the same way it can help people with developmental reading problems process text faster and more accurately (Schmitt et al., 2019; Young et al., 2019). Researchers have noted that, given the option, most people with aphasia prefer combined auditory and written content over single-modality access when processing sentence- and paragraph-length material (Brown et al., 2019; Knollman-Porter et al., 2019; Wallace et al., 2019). As an example, Brown and colleagues (2019) had 27 people with aphasia select images to correspond with sentences presented in auditory-only, written-only, or combined auditory and written modalities; two-thirds preferred the combined presentation. Similarly, Wallace and colleagues (2019) presented 20 people with aphasia with auditory-only, written-only, and combined auditory and written forms of short and long narratives; 70% preferred the combined modality. Knowing this preference provides incentive to continue TTS technology exploration as a support for people with aphasia.

Another incentive for continued exploration is that TTS devices can promote better and more efficient post-aphasia comprehension by providing a means of bypassing the decoding process necessary with traditional reading (Knollman-Porter et al., 2019). Knollman-Porter and colleagues documented this when they found that, given combined modality exposure to newspaper articles, 28 people with aphasia answered multiple choice questions with a mean accuracy over 70%; accuracy given only auditory input was significantly worse, and accuracy given only written input was roughly comparable but required significantly longer reviewing time. Improved comprehension is not a universal phenomenon, however. Neither Harvey and colleagues (2013) nor Caute and colleagues (2016) found improved comprehension when people with aphasia processed TTS content rather than printed texts; however, participants expressed greater reading confidence, read more challenging material, or processed content faster with TTS than with written-only texts.

TTS technology implementation by people with aphasia remains relatively rare despite the appeal of eliminating reliance on independent word decoding. When interviewing people with aphasia, Knollman-Porter and colleagues (2015) found only one person who routinely used TTS technology as a reading support. Similarly, among 10 people interviewed by Webster and colleagues (2018) about post-aphasia reading, none reported using a TTS device. In a more recent survey of 82 people with aphasia, only about 17% used any technology to support reading (Webster et al., 2020).

Negative opinions and perceptions about technology are possible reasons why people with aphasia only infrequently adopt TTS systems as reading supports. Attempts to implement technology-based strategies have revealed the importance of understanding the perceptions, desires, and beliefs of people with disabilities (Lasker & Bedrosian, 2001; McCall, 2012). Practitioners need to consider how strategies or supportive devices align with everyday needs, residual abilities, technological competence, and environmental supports and demands. Simply having a technological solution available does not ensure a person will want to pursue that option, have sufficient confidence about independent operation, or have adequate support when problems arise. Regarding TTS technology, researchers have examined effectiveness for people with aphasia and have solicited opinions about specific features (e.g., highlighting (Brown et al., 2020); rate of speech (Hux et al., 2020); computer-generated voice output (Hux et al., 2017)), but they have not inquired about anticipated problems or benefits when implementing devices during functional activities. Obtaining this type of information can be challenging, but understanding the perceptions of people with aphasia is critical from a clinical standpoint.

The communication problems of people with aphasia limit the type and amount of information researchers can elicit from them. However, despite inherent challenges, performing qualitative research can be a means of obtaining answers to clinical questions (Simmons-Mackie & Lynch, 2013). Interviewing people with aphasia has served as the basis for data collection in many existing studies (e.g., Berg et al., 2017; Harmon et al., 2018; Rohde et al., 2012), even though questions persist about data quality and accuracy (Simmons-Mackie & Lynch, 2013). Providing receptive and expressive communication supports—such as presenting questions in simultaneous written and spoken forms and with supportive images, using confirmatory statements to ensure understanding, altering question style, and offering ideas for confirmation or refutation—are helpful for improving reliability (Luck & Rose, 2007; Wilson & Kim, 2019).

Limited familiarity with TTS technology is another potential barrier to eliciting opinions and perceptions from people with aphasia. Extant research suggests that people with aphasia have limited exposure to assistive technology tools supporting reading (Knollman-Porter et al., 2015; Webster et al., 2018, 2020). To address this, providing education and an opportunity to manipulate TTS systems before asking people for their perceptions is appropriate. Because TTS systems vary substantially regarding user-adjustable features (e.g., voice selection, speech rate, highlighting) and basic attributes (e.g., cost, platform, display size, material uploading requirements), exposure to multiple devices is necessary to ensure knowledgeable and informative feedback.

The combination of communication problems and possible limited familiarity with TTS technology prompted our provision of structured, aphasia-friendly education, an opportunity to manipulate devices, and adoption of a convergent mixed methods design to structure the research. Our goal was to learn about the perceptions of people with aphasia regarding desired features, benefits, and drawbacks of the technology. Complementary quantitative and qualitative strands maximized our elicitation of information and provided us with mutually confirmatory evidence about response validity. Using convergent mixed methods acknowledged the importance of soliciting and understanding stakeholders’ opinions in decisions relevant to them (Charlton, 2000; Paul, 2016) while simultaneously recognizing the difficulty of conversing with people with aphasia. Specific quantitative, qualitative, and mixed methods research questions appear in Figure 1.

Figure 1.

Quantitative, qualitative, and mixed methods research questions.

2.0. Methods

2.1. Study Design

Convergent mixed methods research requires simultaneous quantitative and qualitative data collection (Creswell & Piano Clark, 2018). We used quantitative data from responses to close-ended questions and qualitative data from conversational interactions and interviews conducted with recommended alternative questioning style and supportive communication techniques (Luck & Rose, 2007; Wilson & Kim, 2019). Both strands had equal weight. Quantitative data were etic in that they reflected our biases as researchers in formulating questions and selecting items for inclusion on pre-determined lists of response options (Creswell & Piano Clark, 2018; Creswell & Poth, 2018). In contrast, qualitative data were emic in nature—that is, they provided insiders’ views about the topic with minimal external guidance (Creswell & Poth, 2018). We converged results from the separate strands to corroborate and validate findings and to maximize our understanding of participants’ views. Furthermore, we adopted a research perspective of pragmatism (Creswell & Piano Clark, 2018; Tashakkori & Teddlie, 2003) that allowed us to focus on data collection methods that worked best for people with communication challenges. The approach was practical and consistent with applied research.

We used an integrated approach to data collection and analysis (Creswell & Piano Clark, 2018). We first asked participants to respond to close-ended questions on the Reading and Technology Experiences questionnaire. Specific answers to several of these questions extend beyond the purpose of this study and appear elsewhere rather than in the results reported herein; however, we used participant responses to guide follow-up questions asked during semi-structured interviews. By integrating quantitative and qualitative findings, we identified ways the datasets converged and diverged. This yielded a more complete understanding of participants’ perspectives than would have been possible given only quantitative or only qualitative results.

2.2. Participants

Study participants included 12 males and 7 females from a convenience sample of community-dwelling adults with chronic aphasia. Participants had received in the past or were currently receiving speech and language services at a university clinic and/or were members of local aphasia support groups. They ranged from 35 to 78 years old (M = 62.37; SD = 11.69) and were between 2.17 and 22.83 years post-stroke (M = 10.64; SD = 7.09). Educational achievement ranged from 12 to 19 years (M = 15.26; SD = 2.23), and all participants self-reported functional literacy prior to acquiring aphasia. Eighteen had acquired aphasia from a left hemisphere stroke; one participant acquired aphasia from non-progressive encephalopathy. All spoke American English and had adequate hearing and vision to participate in the research activities. Adequate hearing was confirmed through perception in at least one ear of 1000, 2000, and 4000 Hz tones presented at 40dB; for 5 participants who wore hearing aids, adequate hearing was based on self or family report of an audiological examination or hearing aid adjustment within the previous 8 months and confirmation that they could hear conversational speech in a quiet environment. Adequate vision was ensured through accurate performance on a name identification task presented with 20-point Times New Roman font. Pre-stroke illiteracy, not speaking English, and inadequate hearing or vision to pass the screening measures were the only exclusion criteria.

Standardized assessment results provided objective measures of language and cognitive abilities. We administered standardized tests during a separate session occurring 1 to 21 weeks prior to performance of experimental tasks. Because participants were at least 2 years post-stroke, we did not repeat any tests administered within the previous 6 months. Furthermore, because we were seeking participants with variable severities and types of aphasia and acquired reading challenges, we did not impose restrictions regarding performance on these measures. As such, Aphasia Quotient scores on the Western Aphasia Battery – Revised (Kertesz, 2006) encompassed a wide range (i.e., 14.5 to 91.5) and corresponded with very severe to mild fluent or nonfluent aphasia. Participants also performed the Spoken Sentence Comprehension, Spoken Paragraph Comprehension, and Written Sentence Comprehension subtests of the Comprehensive Aphasia Test (Swinburn et al., 2005) and the Paragraph Factual subtest of the Reading Comprehension Battery for Aphasia – 2^nd edition (Lapointe & Horner, 1998) to provide information about their spoken and written sentence and paragraph comprehension. Again, scores reflected a wide range of ability. Finally, participants completed the Cognitive Linguistic Quick Test - Plus (Helm-Estabrooks, 2017) as a measure of basic cognitive functions supporting language processing. Participants’ demographic information as well as scores on standardized aphasia, reading, and cognitive assessments appear in Appendix A.

2.3. Data Collection

We received institutional review board approval prior to recruiting participants. All participants reviewed and signed an informed consent form. The study procedures through which we progressed as researchers and the products developed during execution of the research protocol appear in Figure 2.

Figure 2.

Study procedures and products diagram for researcher progression through the convergent mixed methods design.

Experimental activities occurred in three phases within a single session for each person. Session activities included: (a) completing the Reading and Technology Experiences questionnaire; (b) participating in one-on-one education and guided practice activities about TTS technology; and (c) responding to close-ended questions and engaging in a semi-structured interview. Research sessions lasted between 60 and 90 minutes. We video and audio recorded the entirety of each session using a Canon HF R700 or R800 video camera. We aimed the camera to capture gestures and pointing responses referencing the educational materials or TTS devices.

Participants first responded to close-ended questions on the Reading and Technology Experiences questionnaire, the text of which appears in Appendix B. This questionnaire included yes/no, frequency-of-use (e.g., everyday, 1–3 days per week), and multiple choice questions about technologies used, post-aphasia reading habits (e.g., materials read, strategies favored), and familiarity with computer-generated speech. The final item queried participants about their willingness to consider using TTS technology as a reading support. Questions incorporated aphasia-friendly principles by using large and clear font, ample white space, sentences with simple structures, and images to supplement written text (Rose et al., 2011). We presented the written questions in sequential order, read them aloud, provided supplemental explanations to facilitate understanding, recorded in writing participants’ responses, and verified our comprehension of responses through confirmatory statements. As desired and able, participants verbalized elaborations to supplement close-ended responses.

Participants next watched and listened to an educational presentation about TTS technology. We created a 19-slide PowerPoint© presentation and accompanying script to structure the information provided. Slides addressed topics relating to TTS technology including potential uses; types of reading materials accommodated; computer platforms supporting device use; user-adjustable features controlling voice output, speaking rate, and highlighting options; and cost. All written content used aphasia-friendly principles along with images to support comprehension. The accompanying script mirrored slide content and included multiple queries about participant comprehension and the desire for information repetition. Reading from the script ensured consistency regarding the manner and extent of information presented. Review of the educational material took between 4 and 14 minutes with the difference attributable to the extent of unsolicited verbalizations participants made in response to presented information.

Participants engaged in guided practice and manipulation of TTS systems following the educational presentation. For user-adjustable features, we first demonstrated control operations and then supported participants as they performed the same or comparable steps. Each participant practiced the following feature manipulations: voice and rate of speech options with Natural Reader© software on an iPad©, highlighting and material upload options with Read&Write™ software on a Windows© Surface™, and pausing and replaying options with Kurzweil® software on a Windows© laptop computer. Including multiple devices and TTS programs for demonstration and guided practice ensured participants experienced different computer platforms and a range of user-adjustable and standard features. The time devoted to guided practice varied from 14 to 35 minutes (M = 23.18, SD = 4.64) dependent on two factors: (a) the amount of explanation and demonstration support needed for successful feature selection or operation and (b) the extent of participant comments or questions.

We asked yes/no, Likert-type scale, and multiple choice questions as well as semi-structured interview questions after the education and guided practice activities to solicit reactions to TTS systems and features. The written texts of questions appear in Appendix C. All close-ended questions appeared in printed form on separate pieces of paper using black, 18-point, Times New Roman font. Five-point Likert-type scales provided for indicating and recording responses had descriptive words (e.g., “not at all helpful,” “very helpful”) and images (e.g., thumbs-up and thumbs-down icons) at endpoints; the remaining points had descriptive words (e.g., “a little helpful,” “moderately helpful,” “helpful”) but no accompanying images. We presented the written questions, read them aloud, marked participant responses, and checked our comprehension through confirmatory statements. We encouraged participants to elaborate on responses as desired and possible.

We embedded the close-ended questions within a semi-structured interview during which we probed for preferences, concerns, and opinions about potential benefits and drawbacks of TTS technology as a personal reading support or as a support for others with aphasia. Questions solicited participants’ reactions to and feelings about TTS technology having had an opportunity to learn about and manipulate devices and features. As evident in Appendix C, open-ended semi-structured interview questions always preceded related questions provided in close-ended format. Initial semi-structured interview questions were general in nature, with follow-up open- and close-ended questions serving to solicit response clarifications and expansions. We included questions about the perceived importance of user-adjustable and basic system features, the amount and type of help participants anticipated needing to achieve operational competence, and any perceived benefits or challenges associated with implementing TTS technology as a reading support. We concluded the interview by again asking participants about their willingness to consider using TTS technology as a reading support. This paralleled the question we asked at the end of the Reading and Technology Experiences questionnaire.

Some participants (i.e., P3, P8, P13, and P19) could verbalize minimal, if any, relevant content because of apraxia, verbal perseveration, and/or paraphasic productions. Utterances these participants produced spontaneously during the research activities were largely uninformative regarding novel content. Communication conveyed via gestures, facial expressions, vocal intonation, and yes/no responses to our interpretive attempts was informative, however. Expressive challenges of other participants varied in severity but still likely reduced the generation of spontaneous comments during open-ended questioning.

2.4. Data Analysis

We tabulated responses to yes/no, Likert-type scale, and multiple choice questions. Descriptive statistics served to summarize these data. For yes/no questions, we computed positive versus negative response percentages; for each Likert-type scale rating, we computed the mean, standard deviation, and range across participants and counted the number of participants selecting responses of 1 or 2 (e.g., little or no benefit) versus 3 (e.g., medium benefit) versus 4 or 5 (e.g., substantial benefit); and for multiple choice questions, we tallied how many participants selected each response option.

We prepared the qualitative data for analysis by transcribing participants’ verbal and nonverbal behaviors. We transcribed verbatim in separate cells of a spreadsheet each communication turn along with descriptions of gestures (e.g., head nods/shakes, pointing behaviors, shoulder shrugs) when appropriate. Then, we designated two members of our research team as the primary and secondary qualitative data coders (i.e., KH and SEW, respectively). The primary coder was unacquainted with the participants; hence, she was unaware of their thoughts or feelings about TTS technology. This was important because a critical feature of qualitative methodology is ensuring that researchers attend to participants’ beliefs and perceptions rather than inserting preconceived expectations or biases into the analysis process (Creswell & Poth, 2018; Merriam & Tisdell, 2016; Moustakas, 1994; Padilla-Diaz, 2015). For this study, we needed to suspend judgment and suppress preconceptions based on our experiences as academicians and speech-language pathologists familiar with working with people with aphasia.

The coders systematically and repeatedly reviewed each transcript for familiarization and to identify points of interest (Creswell & Poth, 2018; Moustakas, 1994; Padilla-Diaz, 2015). In alignment with phenomenological research that explores the perspectives of people who share a common experience (i.e., having reading challenges associated with aphasia and participating in the education and guided practice activities), we first identified narrow units of meaning and then progressed to broader themes indicative of consistencies within and across participants. This required us to assign equal importance to all participant statements and behaviors reflective of salient experiences and perceptions regarding TTS technology.

The primary and secondary coders independently analyzed by hand the datasets of 5 participants to generate a list of potential codes. We then discussed each code, used consensus to resolve any disagreements in interpretation, and established a definition reflective of our shared understanding. The primary coder used these defined codes to analyze the remaining transcripts. Whenever necessary, she consulted other research team members about the meaning of ambiguous behaviors, statements, or exchanges. In these instances, we used consensus to determine communicative intent by discussing the confusion and reviewing the appropriate session recording to extract salient contextual information. Ambiguous or confusing exchanges occurred on 31 of 3608 communicative turns (i.e., <1%) taken by participants. After coding all transcripts, we grouped data elements according to assigned codes and formed them into themes representing consistent meaning across participants.

We combined the quantitative and qualitative findings for examination and identification of commonalities and discrepancies. We developed joint display tables to illuminate findings as well as consistencies and inconsistencies between datasets. From these, we developed our understanding of the perceptions of people with aphasia regarding TTS systems.

3.0. Results

Results appear in accordance with themes emerging from the data. Identified themes included TTS technology experience and interest, important features of TTS systems, benefits associated with TTS technology, and concerns associated with TTS technology. Numbers corresponding to the research questions listed in Figure 1 appear in parentheses following each theme heading. Within each of the following sections, we first present quantitative and then qualitative results. We then integrate the findings for comparison, confirmation, and provision of a robust understanding of participants’ perceptions.

3.1. TTS Technology Experience and Interest (QUAN 1, QUAL 1a, and MIXED 1)

We queried participants both at the start and conclusion of the experimental session about experience with and interest in TTS technology as a reading support. Table 1 provides a visual display of pertinent quantitative and qualitative findings. Determination of prior experience with and initial interest in TTS technology came from responses to the Reading and Technology Experiences questionnaire. Determination of interest after the education and guided practice activities came from close-ended responses at the end of each participant’s session.

Table 1.

Experience with and Interest in TTS Technology Before and After Engagement in the Education and Exploration Activities

Participant	Prior experience	Initial interest	Final interest	Subtheme quotes about initial interest
P1	Yes	Yes	Yes	• Interest based on prior functional experience  ∘ Researcher: Some people read and listen at the same time. P1: That’s what I do until I get too tired.  ∘ Researcher: You like it? P9: Okay, little.  ∘ Researcher: Does it help you read social media? P12: Yeah, me little bit, butyeah, oh yeah, yes.
P9	Yes	Yes	Yes
P12	Yes	Yes	Yes
P4	Research	Yes	Yes
P6	Research	Yes	Yes	• Interest based on prior research exposure  ∘ P7: I would like to try it because iťs helpful, very helpful.  ∘ P11: I mean, I liked when we 'ere doing the research like that. You know, if you can do that [listen and read] at the same time, you know, nyou verson with the voice and everything, while you were trying to read together at the same time. I thought it was, I thought it was pretty good.
P7	Research	Yes	Yes
P8	Research	Yes	Yes
P11	Research	Yes	Yes
P13	Research	Yes	Yes	• Interest tempered by concerns  ∘ Researcher: think you might use text-to-speech? P14: Little bit, but [it might be] tough [to operate].
P14	Research	Yes	Yes
P15	Research	Yes	Yes	• Uncertain about interest  ∘ R esearcher: Some people read and listen at the same time, so they can see the words and hear it. P10: Yeah, at times that wouldprobably help. Researcher: Do you think you might use text-to-speech? P10: I honestly don ’t know.  ∘ Researcher: Do you think you might use text-to-speech? P18: Maybe.... Iguess [using it to read the]paper and all that stuff [would be] okay.
P17	Research	Yes	Yes
P19	Research	Yes	Yes
P5	Research	Mayb e	Yes
P18	Research	Maybe	Yes	• Unsupported reading is sufficient  ∘ P15: Um I’m pretty good about [reading by] myself
P3	Research	IDK	Yes
P10	Research	IDK	Yes	• No interest  ∘ Researcher: So you can see the text and hear it at the same time. P2: Bad for all the time. Researcher: You don ’t like it? P2: No.
P16	Other	IDK	No
P2	Research	No	No

Note. Prior experience “Yes” indicates functional use of TTS technology; “Research” indicates TTS technology exposure was only from participation in one or more previous studies coordinated by our research team; “Other” indicates TTS exposure occurred but was only from unspecified structured activities. IDK = I don’t know.

All participants responded positively about having tried or seen TTS technology at least once. However, 16 of the 19 positive responses (84.21%) came from participants who did not use TTS systems in functional activities and, instead, had TTS experience from participating in previous research studies with us or trying a system during unspecified structured activities. Hence, only 3 study participants (i.e., P1, P9, and P12; 15.79%) had experience using TTS technology for functional reading.

Ten participants (52.63%)—in addition to the 3 (15.79%) already using a system for functional purposes—expressed interest prior to the education and guided practice activities in trying TTS technology. Two other participants (i.e., P5 and P18; 10.53%) indicated they might be interested, and 3 (i.e., P3, P10, and P16; 21.05%) expressed uncertainty. Only 1 participant (i.e., P2) had no interest. Qualitative comments made before the education and guided practice activities confirmed variable interest across participants. The 3 people already using a system (i.e., P1, P9, and P12) indicated moderate to strong enthusiasm about continued use. In contrast, the person with no initial interest was consistent in conveying this opinion. Among remaining participants, some expressed considerable enthusiasm for the possible benefits afforded, but others were more reticent because of limited familiarity, doubts about the types of materials they could access, or concerns about mastering device operation. Of note, 1 person (i.e., P15) provided contradictory information via the quantitative and qualitative procedures. He responded positively to a yes/no question about interest but verbalized little initial desire to explore the technology given his satisfaction with reading without external supports.

Seventeen of the 19 participants (89.47%) expressed interest in further pursuit of TTS technology after completing the education and guided practice experiences. P2, who expressed no initial interest, persisted in this view. He did not believe TTS technology would help him and said, I don’t see that [i.e., TTS technology] as needed to [i.e., by] me, to explain his lack of interest. Additionally, P16, who was uncertain about her initial interest, decided the technology was not something she wished to pursue after trying it during guided practice. Her rationale was that she did not believe she could operate the technology given paresis of her right hand.

3.2. System Features (QUAN 2, QUAL 1b, and MIXED 1)

Table 2 is a joint display of the quantitative and qualitative findings regarding feature importance. Frequency of endorsement data came from tallying participants who selected a feature as important during closed-ended questioning; subtheme quotes came from comments made during guided practice or semi-structured interviews.

Table 2.

Endorsement Frequency and Quotes about System Feature Importance

Feature/subtheme	Frequency of endorsement (N = 19)	Subtheme quotes
Voice output	19	P5: [The man ’s voice is] very grating. Like, I canť listen to that. P6: I like the female voices a lot. P9: Man, [Iprefer] a man one. P10: I couldnť take that talking....I can’t figure what he’s saying. P12: Me too [I’m a woman]. Me and good, but [it’s] not me. Woman fine. P17: [The British accent is] too much.
Rate of speech	17	P1: The speed is important because if it is too fast, you canť understand it. And it’s better to have it at, um, a rate that you can listen to and that you can understand it well. P5: I think too slow is really hard for me. …So slow that I can’t like remember what it said. P10: [If it is too fast], by the time I think of what they are saying, I’m forgetting it. P12: Slow down. I help gone [i.e., it does not help me when the rate is too fast]. Have to forget it all done [i.e., I forget what was said by the time it’s done]. Researcher: So you forget it if it’s too fast? P12: Yeah, yeah.
Highlighting	16	P5: I think I listen better if [there is highlighting]. P11: I think on the highlight stuffbefore [i.e., single word highlighting], I was having a hard time. Not here. The other one before. Every word. It seemed like too much. But sometimes for like a whole, like, a sentence, that’s not so bad. But every one of those '.e., word highlighting] was hard like on the highlights. P15: This one I want [i.e., points to no highlighting option]. P17: I can see it [i.e., the word being read], you know, [when there is highlighting].
Replaying content	10	P2: [Replaying is important] because [I] want to replay it to hear.
Pausing output	9	P11: If you wanted to do something, uh, maybe when you put some notes or something there, you can stop and do something over here, and then do it again [i.e., go back to reading/listening].
Cost	8	P1: If it were too expensive, I probably wouldn ’t get it. P2: Money. How much is the cost? P14: But this money, a lot or a little bit. I don’t know. And money scares me too.

Participants universally endorsed preferred voice selection as an important feature. A majority also identified control of speech output rate (n = 17; 89.47%) and highlighting (n = 16; 84.21%) as priorities. About half selected pausing content (n = 9; 47.37%), replaying content (n = 10; 52.63%), and device cost (n = 8; 42.11%) as salient features. Qualitatively, comments shed light on participants’ perceptions about the relative importance of various features. Participant comments were primarily about voice, rate of speech, and highlighting features; few comments addressed system cost or pausing and replaying options.

Quantitative and qualitative data integration revealed consensus about the importance of having options for certain features but not about which settings were preferable. For example, all participants wanted to select a preferred voice, but opinions varied regarding which was best. Similarly, almost universal endorsement occurred for controlling speech output rate, but comments about the optimal speed revealed differing opinions. For highlighting, some participants thought that highlighting words or sentences would help with tracking place, but others found this distracting. In contrast to the spontaneous and highly opinionated comments about voice output, speech rate, and highlighting, statements about pausing and replaying capabilities were more matter-of-fact and simply acknowledged a possible use for the feature.

3.3. Benefits of TTS Technology (QUAN 3, QUAL 1c, and MIXED 1)

We asked participants about potential benefits associated with TTS technology. The joint display of quantitative and qualitative findings appearing as Table 3 specifies the number of participants endorsing a benefit through close-ended questioning and provides representative quotes explaining opinions. Figure 3 provides individual data about participants’ expressed benefits via quantitative and/or qualitative methods, identification of the benefit each person selected as most important, and evidence about quantitative and qualitative data consistency.

Table 3.

Endorsement Frequency and Quotes about Benefits of Using Text-to-speech Technology

Benefit/subtheme	Frequency of quantitative endorsement (N = 19)	Subtheme quotes
Comprehension	14	P5: Sometimes [with unsupported reading], Igo back gain and again and again because I canť get the information. So, but, um, [with] text-to-speech I can listen to that more clearly, I guess. P7: I can do that [i.e., understand the content]. That’s what I’m saying. This is good because I can listen. P11: I think some of the things for me, it would be more understanding when I could do both [i.e., listen and read] together on the text and speech.
Reading independence	14	P1: It’s easy [to operate by my’ lf]. P10: I would like to get it by myself
Types of materials	11	P1: It would allow me to read magazines. P7: I like that cause [I want to read] the sports. P11: I would do [i.e., have] more things that I could read, whether different articles or…even the small books or the, um, novels or something like that. P17: You co ’ld, um, [read] anythingyou want with, um, [a text-to- eech system]. It’s good. It’s good.
Communication with others	9	P4: [Name of a friend with aphasia] and me could do it [i.e., communicate via text-to-speech] together. P15: You can see it in here [i.e., you can know more about activities with friends]…. You know, like this volunteer something [i.e., pulls out a hospital volunteer luncheon invitation].
Participation in reading activities	8	P10: If you can get involved with other people and do things.
Faster reading	7	P5: Probably maybe readingfaster….There's a difference between just reading and just reading text-to-speech….Just reading, I know it’s hard….And so, text-to-speech, maybe reading faster.
Other	3	Lessen fatigue - P5: But this [i.e., text-to-speech] might helpyou read longer. Practice speech - P11: You could do your speech on that too if you wanted to practice something. Helps people with vision problems - P1: If you couldn ť see, you could [still] read.

Figure 3.

Participants’ quantitative and qualitative endorsements of important and most important features.

Most participants endorsed more than one benefit to TTS technology when asked close-ended questions. Fourteen of the 19 participants (73.68%) believed they would understand more written content given TTS technology access. An equal number (n = 14; 73.68%) believed TTS technology would promote greater reading independence. Eleven participants (57.89%) thought the technology would allow them to read different materials than they currently could. In contrast, only 9 participants endorsed communicating more with others via the written modality as a benefit (47.37%); 8 believed the technology would promote participation in more activities involving reading (42.11%); and 7 believed it would help them read faster (36.84%).

We next asked participants about the amount of personal benefit they predicted experiencing regarding reading comprehension and speed when using TTS technology and the extent to which they believed the technology would help other people with aphasia. For analysis, we first computed the mean Likert-scale response across participants and then determined frequency counts when collapsing the data into three groups: those selecting scale ratings of 1 or 2 to indicate little or no anticipated benefit, those selecting a rating of 3 to indicate some benefit, and those selecting ratings of 4 or 5 to represent substantial benefit. Across participants, the mean scale value for presumed personal comprehension benefit was 3.68 (SD = 1.20; range: 1 (none) – 5 (a lot)); collapsing into three groups revealed that only 2 participants anticipated little or no comprehension benefit, 5 anticipated some benefit, and the remaining 12 anticipated substantial benefit. Regarding personal reading speed benefit, the mean assigned Likert-type scale value across participants was 3.03 (SD = 1.23; range: 1 (none) – 5 (a lot)); collapsed into three groups, the data revealed that 4 participants anticipated little or no speed benefit, 11 anticipated some speed benefit, and the remaining 4 anticipated substantial benefit. Finally, given a Likert-type scale with endpoints labeled not at all helpful (i.e., 1) and very helpful (i.e., 5), 3 participants selected values of 3, and the remaining 16 selected values of 4 or 5 when asked about TTS technology supporting the reading of people with aphasia other than themselves; the mean scale value across participants was 4.29 (SD = 0.73).

Participants selected from those benefits that they identified as important the one they believed was most important. Two presumed benefits—increased communication with others and reading independence—were most frequently endorsed as most important; 5 participants (26.32%) each endorsed these as most important during quantitative data collection. No other benefit received more than three endorsements as being most important. A few comments made during semi-structured interviews supported the importance of increased communication and reading independence, although more frequent comments referenced reading comprehension, speed, and material diversity. Overall, comments reflected the belief that content provision via simultaneous written and spoken modalities would improve engagement in preferred reading activities and increase the variety of materials read. Additionally, 4 participants (P1, P5, P9, and P14) endorsed possible benefits not introduced during quantitative data collection. These benefits included (a) helping people who could not read because of vision deficits, (b) reducing fatigue associated with single modality reading, and (c) providing a means of practicing spoken language by repeating words or sentences presented via a TTS device.

Performing a person-by-person comparison of quantitative and qualitative datasets confirmed relative consistency in benefit endorsement by those participants with adequate communication to expressive themselves verbally. This consistency was evident in that participants tended to talk about the same benefits they identified as important via close-ended questioning; only 2 participants (i.e., P1 and P9) commented positively about a potential benefit they had not previously endorsed as important. However, participants did not typically comment about all benefits they designated as important during close-ended questioning. The tendency was to verbalize about the most important benefit rather than benefits simply deemed important.

3.4. Concerns about TTS Technology (QUAN 4, QUAL 1d, and MIXED 1)

We used comparable procedures to elicit perceptions about TTS technology concerns as we used for perceived benefits. The joint quantitative and qualitative display appearing as Table 4 specifies how many participants endorsed each concern included on the predetermined list supplied during close-ended questioning and provides quotes representing expressed opinions. Figure 4 provides data about concerns expressed via quantitative versus qualitative methods, identification of the concern each person deemed as most important, and evidence about data consistency across the quantitative and qualitative strands.

Table 4.

Endorsement Frequency and Quotes about Concerns with Text-to-speech Technology

Concern/subtheme	Frequency of quantitative endorsement (N = 19)	Subtheme quotes
Technology/ operation problems	9	P2: Even the third [i.e., three]year old [can do it, but I have trouble]. P11: It’s not like I know all that [technology] stuff now, and iťll be easy for me. I'll have to remember or figure out some things that I might not, you know, [know how to do] for that kind of stuff. P14: Some of it [technology] good, but some of it bad. P16: I do [i.e., can only use] one hand. P17: I forget, Iforget [how to work technology].
Understanding voice output	8	P5: Some voices are like ugh. P10: I can ’t figure out what he ’s saying….I couldn ’t take that talking.
Needing help using system	8	P1: Doing it on the computer, I mign need help. P10: [Technology is] always ha. J….I i ould have a hard time catching on to it, and remembering it. My husband will sit and tell me how to do something. Then he goes away, and I think I can ’t do it.
Matching reading speed	7	P2: I [read] fast and slow, fast and slow [on different parts]. P6: [I may not] get it [i.e., the speaking rate] close enough to [what I want].
Cost	5	P9: Money. How much money? P10: I donť think [spouse’s name] could afford it, though.
Other	4	Vision issues – P1: Doing it on the Computer, I can ’t see some of [the w yrds]….There are some things on the Computer that don ’t blow up ii.e., enlargefontsize]. Will not improve unsupported reading – P1: I could read faster or understand it better [before my stroke]. If I could read it on my own, [that’s what I would like]. P5: Iknow it’s hard butI really want to get back to that [i.e., reading without support]. Limitations of current text-to-speech system – P9: [P9 shows researcher content on his phone] Text and speaking words. Mad at texting, um, graduationpictures. Researcher: …So maybe notbeing able to use it when you want to use it? MU9: Yes. Equipment needs – P11: Do you have to get a computer? Or, how do I do it?
Looking different from others	3	P14: Friendsgood….But, uh, [when I’m on] vacation or something, it’s a lot of people, and [that makes] me, uh, shaky too.

Figure 4.

Participants’ quantitative and qualitative endorsements of important and most important concerns.

Between one-third and one-half of participants endorsed concerns about technology/operation problems (n = 9, 47.37%), understanding computer-generated voice output (n = 8, 42.11%), needing another person’s support for system use (n = 8, 42.11%), and matching speech output rate to preferred reading rate (n = 7, 36.84%). Five participants (26.32%) endorsed concern about system cost, and 3 (15.79%) endorsed concern about appearing different from others. Across participants, the issue most frequently identified as most important was technology/operation problems (n = 5; 26.32%).

The greatest number of comments elicited during semi-structured interviews provided elaboration about technology/operation concerns and the anticipated need for assistance from others. This was consistent with multiple participants identifying these issues during quantitative data collection. Comments about technology/operation problems primarily targeted a lack of confidence using a computer-based system; only 1 person (i.e., P29) expressed a different operation concern—that is, difficulty stemming from limited motor control of her dominant hand. Verbalizations about needing assistance were similar to those about technology/operation problems in that they referenced a lack of sufficient computer knowledge for independent use. The decision to code an utterance as a technology/operation problem versus needing help from others reflected whether participants relied on others to resolve issues or attempted to manage them independently. Acknowledging the potential overlap of these subthemes, we identified the dominant concern as doubts about successful, independent TTS system operation.

Four participants raised TTS technology concerns during semi-structured interviews that were not among those preselected for inquiry. One concern mirrored the benefit P1 identified about TTS technology providing a possible advantage for people with vision problems. P1 said she could not see sufficiently to read text appearing on computer monitors unless it was enlarged; hence, her concern was that some TTS systems might not provide text enlargement. Two participants (i.e., P1 and P5) wanted to regain lost literacy skills and did not believe using TTS technology would help in this effort. Participant P11 was concerned about having the equipment needed to support TTS technology. Finally, Participant P9, who already used a TTS system for functional purposes, was concerned that it worked only with certain document types.

Inconsistencies appeared between the quantitative and qualitative datasets regarding two issues. First, a lack of qualitative data emerged about understanding voice output despite relatively frequent endorsement of this via close-ended responses. Second, the opposite scenario appeared regarding system cost—that is, only 5 people responded positively to a yes/no question about system cost being a concern, but 3 of these people (i.e., P2, P9, and P14) as well as 2 others (i.e., P10 and P12) raised the concern during the interview. Of note, when we had asked earlier whether system cost was important for consideration rather than a concern, 8 participants (42.11%) responded positively (See Table 2).

Person-by-person comparison of quantitative and qualitative datasets regarding concerns revealed a similar pattern as observed for benefits; specifically, participants made more spontaneous verbalizations during interviews about concerns endorsed during quantitative questioning than about issues not endorsed. Furthermore, comments were particularly likely regarding the issue a participant identified as most important. This consistency provided validation and confirmation that the quantitative and qualitative procedures were effective in eliciting accurate perceptions about study participants’ concerns.

4.0. Discussion

Recognizing the needs and perspectives of people who may use assistive technology devices is vital prior to acquisition. In accordance with the Technology Acceptance Model, motivation to adopt a device stems from a person’s attitude about utilizing it, perceptions about helpfulness, and beliefs about ease of mastery and implementation (Taherdoost, 2018). Although potential advantages exist when people with aphasia implement TTS technology (Brown et al., 2019; Hux et al., 2020; Knollman-Porter et al., 2019; Wallace et al., 2019), decoding and comprehension benefits are not universal, and not all prospective users are likely to view the technology favorably. A desire to understand the attitudes and perceived potential uses, benefits, and challenges reported by adults with aphasia about TTS technology prompted this research.

Analysis, integration, and synthesis of quantitative and qualitative datasets revealed consistency among participants regarding certain aspects of TTS technology but also unique perspectives about other aspects. To summarize, although only 3 of the 19 participants used a TTS device for functional reading purposes, 10 additional participants expressed interest prior to engaging in education and guided practice, and 4 more expressed interest after the activities (i.e., QUAN 1, QUAL 1a, and MIXED 1 findings). Important features included selecting a preferred speech output voice; controlling speech output rate, pausing, and content replay; and device cost. Participants expressed unique opinions about optimal feature settings (i.e., QUAN 2, QUAL 1b, and MIXED 1 findings). The most frequently expressed TTS system benefits were understanding more written content, having greater reading independence, reading more diverse material types, increasing communication with others, participating in more reading activities, and reading faster than currently possible. The extent of anticipated benefit to comprehension and speed were modest (i.e., 3.68/5.00 and 3.30/5.00, respectively), although the presumed benefit to people with aphasia in general was substantial (i.e., 4.29/5.00) (i.e., QUAN 3, QUAL 1c, and MIXED 1 findings). Finally, TTS technology concerns centered on technology/operation problems, understanding voice output, needing support from others for system operation, and matching speech output rate to preferred reading rate (i.e., QUAN 4, QUAL 1d, and MIXED 1 findings).

The following sections address the importance of (a) providing opportunities for people to learn about and manipulate TTS systems, (b) tailoring user-adjustable features to individual preferences, and (c) addressing system operation concerns. We conclude with discussions about incongruent findings from converging the quantitative and qualitative datasets and methodological limitations restricting generalization.

4.1. Education and Guided Practice Opportunities

Opportunities to learn about and manipulate TTS systems are critical for people who may benefit from the technology (Lasker & Bedrosian, 2001; McCall, 2012). Without such experience, some people may be open to and curious about the possibilities afforded; however, others may have concerns that inappropriately dissuade them from considering the option. Including people with aphasia in decision-making processes concerning them is critical for ensuring client-centered services (Charlton, 2000; Paul, 2016). Our finding of greater interest following education and guided practice reinforces this phenomenon and suggests that some people with limited experience may lack sufficient knowledge to realize potential benefits. However, in one instance, education about and guided practice with TTS technology had the opposite effect—that is, the person realized she did not have adequate motor control to operate a system independently, and, therefore, any reading comprehension benefit was likely to be outweighed by increased reliance on others for system use. The lesson to clinicians is recognizing the importance of providing opportunities to engage with a technology prior to asking for a decision about whether its use will be advantageous (Lasker & Bedrosian, 2001; McCall, 2012). Perceptions may change in positive or negative directions based on perceived benefits and concerns illuminated through education and guided practice. An exception to needing such exposure regarding TTS technology is when a person believes current reading ability—with or without supports—is adequate to meet wants and needs.

4.2. Individual Feature Preferences

Consideration of individual preferences about TTS features emerged as an important issue. This is not surprising given the acknowledged need to match features to individual preferences when selecting assistive technology supports for people with disabilities (McBride, 2011). In the current study, participants expressed strong and unique opinions about preferred voice outputs, rates of speech presentation, and highlighting options.

The communication challenges of adults with aphasia may make the setting of user-adjustable features particularly important. For example, some study participants indicated not understanding certain voice outputs; thus, selecting a preferred voice may directly affect auditory comprehension, although this contradicts evidence of comparable comprehension accuracy when people with aphasia match sentences presented via two commonly available synthetic speech voices to pictures (Hux et al., 2017). People without communication challenges may also have voice output preferences; however, they are likely to understand nonpreferred as well as preferred voices because of their robust language comprehension skills. Another example of the importance of adjusting features to a preferred setting relates to the concern participants expressed about aligning speech output rate with reading speed. Decreased processing speed when people with aphasia perform listening or reading tasks (Neto & Emília Santos, 2012) may have prompted this concern. Seemingly small deviations from a desired rate may have more substantial comprehension effects on people with aphasia than they do on people without aphasia. Practitioners need to ensure that user-adjustable settings match those preferred by a person with aphasia to maximize acceptance and functional use of assistive technology.

4.3. Difficulties Using Technology Independently

A prominent concern of study participants was successful and independent TTS system use. This emerged both in the quantitative and qualitative data as concerns about technology problems interfering with device management and concerns about having to rely on other people for operational assistance. Although some research supports the notion that assistive technology use by adults with disabilities decreases caregiver burden (e.g., Mortenson et al., 2013), reliance on family member caregivers exerts a large toll (Feinberg et al., 2011). Caregivers of older adults or adults with disabilities report experiencing high time demands for care and frustration with having to provide cues and support for effective technology use (Mortenson et al., 2012). In turn, many adults with aphasia may feel uncertain about staying current with technology changes and may rely on others to manage, monitor, and update electronic devices. Hence, the discomfort conveyed regarding TTS system operation appears appropriate. This finding supports the importance of selecting TTS devices that are relatively simple to navigate and use. To date, researchers have not examined the ease with which people with aphasia learn to upload and open materials, adjust user-modifiable settings, navigate through texts presented on TTS devices, and manage technology problems. Determining operational success when people with aphasia use commercially available TTS systems is a necessary next step regarding device recommendations.

Extant research about technology use by older adults suggests that adequate exposure and training can alleviate many system operation concerns. For example, a study implementing assistive device training for adults with stroke revealed significantly improved functioning and satisfaction post-intervention (Chui & Man, 2004). Given that older adults report fear of dependence and lack of training as barriers to assistive technology use (Yusif et al., 2016), practitioners need to design and adapt training to match communication, motoric, and learning challenges of adults with aphasia wishing to master TTS device operation. Also, clinicians should promote generalization by encouraging functional use and recognize the potential need for multiple sessions devoted to adjusting features to match needs in environments in which use is likely.

4.4. Incongruent Findings

Quantitative and qualitative dataset inconsistencies were relatively rare in the present study; however, a few noteworthy ones emerged. In particular, participants made few comments during interviews about concerns understanding the computer-generated voice output of TTS systems despite endorsing this issue during close-ended questioning. This may have occurred because participants listened to several voice output options while learning about and practicing device use, and many expressed strong likes and dislikes for certain voices. Having heard some voices they judged unacceptable may have prompted selection of understanding voice output as a concern during quantitative questioning. However, despite different preferences across participants, all acknowledged the acceptability of at least one demonstrated voice. This experience may have dampened concern about understanding voice output to the extent that it did not arise during the interview portion of experimental sessions.

A second instance of inconsistency between quantitative and qualitative datasets was in reference to system cost. During quantitative data collection, 8 participants identified system cost as an important factor, but only 5 endorsed it as a concern. Then, during qualitative data collection, 5 participants—including 2 who had not identified cost quantitatively as an issue—verbalized concerns about it. This inconsistency may have resulted because we did not provide specific information about the range of system costs. Further exploration of the extent to which cost is an important factor may provide additional information about concerns people with aphasia have about TTS technology. Much TTS software is available for download with no or minimal cost; other systems can exceed $1,000. Whether free or inexpensive software provides adequate feature flexibility to meet the needs of people with aphasia remains unknown.

4.5. Limitations

Potential study limitations involve participant sample characteristics and our selection of response modes for quantitative data collection. A concern with the study sample is their heterogeneity regarding aphasia type and reading impairment severity. We purposefully did not restrict inclusion to people with specific language or reading characteristics, but this choice may have affected perceptions about TTS technology. Support for this comes from extant research about the effects of single and dual modality stimulus presentation on comprehension by people with differing aphasia severities (Brown et al., 2019). Specifically, at the sentence level, people with mild or moderate aphasia participating in Brown and colleagues’ study had minimal comprehension differences given auditory only, written only, or combined written and auditory stimuli; however, for people with more severe aphasia, combined modalities resulted in better performance than either modality in isolation. Experiencing or not experiencing a comprehension benefit is likely to influence a person’s perception about the value of TTS technology. Future researchers may wish to explore the perceptions of people with differing impairment profiles to determine whether distinctions exist among aphasia population subgroups.

Another sample concern is that many participants had TTS technology exposure through engagement in previous studies with us. This experience may have biased them toward greater interest in the technology than people without such exposure. In turn, the perceptions expressed may have been more positive than would have occurred with people without prior TTS technology exposure.

Another possible, yet necessary, study limitation reflects our selection of yes/no options and Likert-type scale ratings as response modes to some questions included in the quantitative research strand. Being forced to provide a yes/no response eliminates the conveyance of nuanced answers; therefore, the obtained data may over- or under-reflect endorsement of an idea. In a similar fashion, answers obtained using Likert-type scales require careful analysis. This is because a value representing a descriptive statistic—such as the group mean—for a scale rating can mask the distribution of scores yielding that value. Participant responses may have been unimodal and clustered near the mean; they may have been bimodal with responses representing only extreme scale values; or they may have been distributed evenly across the continuum of scale values. We attempted to minimize this problem by presenting mean, standard deviation, and range data for each question and by stating the frequency with which participants selected responses within specified ranges. Still, accurate interpretation of these data can be difficult. A final limitation with these quantitative data is that they came from responses to the Reading and Technology Experiences questionnaire for which validity testing does not exist.

4.6. Conclusion

Few people with aphasia use TTS technology when reading functionally, but most express positive perceptions about possible benefits associated with it as a form of assistive technology support. Having control over user-adjustable features—such as voice and highlighting selection and setting speech output rate to a desired speed—was important to people with aphasia. As such, practitioners need to consider carefully the quality and variability of these features during device selection. Providing people with aphasia with a way to learn about TTS systems and practice adjusting features to desired settings is likely an important aspect of the selection process. Also, because the most frequently expressed concern was managing the technology and gaining competence with system operation, clinicians need to ensure they provide adequate learning and practice opportunities to promote independence.

We used a convergent mixed methods design for this research because the language problems of people with aphasia can hinder conveyance of preferences and opinions. Having two data collection methods maximized our likelihood of discerning participants’ true perceptions about TTS technology. However, despite following recommendations about optimal interviewing methods (Luck & Rose, 2007; Wilson & Kim, 2019), communication problems may still have limited the information successfully conveyed. This was especially likely when participants tried to discuss benefits or concerns we had not previously introduced. Future researchers should pay particular attention to investigating the benefits and concerns our study participants expressed spontaneously to ensure capturing the range of opinions and beliefs people with aphasia hold about TTS technology.

Supplementary Material

Appendix A

Appendix A: Demographic and Standardized Test Information about Individual Participants

Highlights.

• Text-to-speech devices are a possible aphasia reading support, but adoption is sparse

• Voice selection, control of speech output rate, & highlighting are priority features

• Better comprehension, independence & communication with others are perceived benefits

• Successful mastery of device operation is the greatest concern