Comprehension, Processing Time, and Modality Preferences When People With Aphasia and Neurotypical Healthy Adults Read Books: A Pilot Study

Abstract

Background:

Many people with aphasia (PWA) want to read books. Text-to-speech (TTS) technology sometimes provides comprehension and processing time benefits when PWA read short, multisentence passages. Currently, no research examines the effect of TTS support when PWA read books.

Aims:

This study's primary purpose was to examine comprehension accuracy and total processing time of PWA and neurotypical healthy adults (NHAs) when reading book sections in read-only versus TTS-supported conditions. A secondary aim was to examine condition preference and perceived degree of understanding by people in both participant groups.

Method and Procedure:

Ten PWA and 10 NHAs alternated between read-only and TTS-supported conditions to read a book. Participants answered comprehension questions and provided feedback about their reading experience, condition preference, and desire to use TTS technology for future book reading.

Outcomes and Result:

Overall, PWA exhibited less accurate comprehension and slower processing times compared to NHAs in both conditions. No significant comprehension accuracy difference occurred between conditions for either group. However, four PWA exhibited a 10% or greater increase in comprehension accuracy when receiving TTS support. A significant processing time difference occurred with PWA processing text faster with TTS support, whereas NHAs did not demonstrate processing time differences. Most PWA preferred the TTS condition and expressed a desire to use TTS technology in the future. Most NHAs expressed the opposite preference.

Conclusions:

TTS support during book reading promotes faster processing without compromising comprehension for PWA. Clinicians should discuss with PWA the relative importance of comprehension accuracy, processing time, and comfort with technology when determining whether using TTS support during book reading is desirable.

Reading books after acquiring aphasia is often challenging (Knollman-Porter et al., 2015) because of high text-processing demands (Webster et al., 2018, 2020). The difficulty and increased processing time associated with decoding and comprehending lengthy written texts can lead to frustration and ultimate abandonment of attempts to read books (Knollman-Porter & Julian, 2019; Parr, 2007). In turn, the loss of a previously enjoyed activity can contribute to a decline in the quality of life experienced by people with aphasia (PWA; Bahia & Chun, 2014; Knollman-Porter et al., 2015; Parr, 2007).

Reading books affords many benefits both for PWA and neurotypical healthy adults (NHAs). It is a common means of acquiring knowledge and information, improving language skills, fostering creative thinking, recognizing and developing social perspectives, learning about current events, maintaining personal identity, developing empathy, relaxing, and experiencing enjoyment (Dogan, 2014; Kjellen et al., 2017). Reading books for pleasure also provides a sense of self-control and protects from loneliness and isolation (Rane-Szostak & Herth, 1995; Worrall et al., 2011). Even though PWA report reading less than they did prior to the onset of aphasia, a continued desire to read books often persists (Knollman-Porter & Julian, 2019). Therefore, developing supports that facilitate comprehension by PWA while decreasing processing time during independent book reading is clinically relevant.

Reading and Aphasia

Book reading is a complex task that involves multiple linguistic and cognitive processes. PWA may struggle to decode and assign meaning to single words, form semantic associations, and process syntactic structures found in sentences and multisentence texts (Cistola et al., 2021; DeDe, 2012, 2013; Webster et al., 2018, 2020). Further complication comes from changes in critical cognitive skills such as processing speed, verbal memory, and attention. Clinical practitioners have designed treatment procedures to address these various aspects of linguistic and cognitive function (McNeil et al., 1991; Murray et al., 1997). However, a recent systematic review by Purdy et al. (2019) revealed that outcomes from such interventions are inconsistent, and no treatment to date has facilitated a return to preaphasia levels of reading performance. Research is, however, beginning to emerge on treatments to facilitate book reading for PWA. Cocks et al. (2013) employed an individual strategy-based reading memory treatment resulting in improved comprehension, reading rate, and confidence for a person with mild aphasia when reading personally relevant books. In addition, Caute et al. (2019) implemented a 14-hr, technology-assisted training approach that resulted in improved comprehension and reading confidence when PWA engaged in book reading. Although preliminary outcomes on book reading treatments are positive, continued exploration about using text-to-speech (TTS) technology as a compensatory support when PWA read books is needed.

Using TTS Technology to Support Reading

Technology advances have led to new methods of accessing written material both by PWA and NHAs. Specifically, researchers have explored the use of TTS technology as an alternative accessing method when traditional reading is not desired or possible (Brown et al., 2019; Knollman-Porter et al., 2019, 2022; Wallace et al., 2019). TTS technology augments written text with synthesized speech output to present content simultaneously through auditory and visual modalities and reduce or eliminate the need for PWA to rely on others to read to them or to revise materials to match preserved language and reading skills (Brown et al., 2019; Hux et al., 2020).

Reading Comprehension and Processing Time

TTS systems are widely available, but the extent to which PWA benefit from the support in terms of comprehension accuracy and processing time is mixed and not fully understood. For example, groups of PWA participating in research activities have exhibited significantly better comprehension of single sentences (Brown et al., 2019) and 10-sentence narratives (Wallace et al., 2019) given TTS support versus no reading support. In contrast, however, no significant group comprehension difference occurred when PWA read five- to six-sentence narratives (Wallace et al., 2019) or nine- to 14-sentence newspaper articles (Knollman-Porter et al., 2019, 2022) under the same conditions. Similar conflicting findings exist regarding processing time benefits when PWA use TTS technology. For example, although most researchers have documented that PWA need significantly less total processing time and experience no negative effect on comprehension accuracy when reading with TTS support versus no support (Harvey et al., 2013; Knollman-Porter et al., 2019, 2022), Wallace et al. (2019) did not find a significant processing time difference when PWA read five- to six-sentence narratives using TTS support.

An explanation for conflicting research findings may reflect the variability in reading comprehension and processing time benefit evident across the PWA who have participated in previous studies. Specifically, across multiple studies, some PWA have exhibited substantial benefit given TTS support, but others appear to have benefitted minimally or not at all (Brown et al., 2019; Knollman-Porter et al., 2019, 2022; Wallace et al., 2019). This variability across participants may have been of sufficient magnitude to negate some group differences. Other possible explanations for conflicting findings relate to the length of material participants have read for various experimental tasks or the types of questions asked to measure their comprehension. Comprehension and processing time benefits may not be apparent when PWA read only short written texts or when they have to prove their comprehension sufficiency by answering questions that place substantial demand on memory and recognition recall. Further confounding the task of explaining conflicting findings is the fact that researchers have not yet investigated the effect of TTS technology on the reading comprehension and processing time of NHAs. Also unknown is the effect both on PWA and NHAs when reading lengthy texts such as those found in books.

TTS Use Preferences

Opinions and preferences about using TTS technology as a reading support is another important clinical consideration (Dalemans et al., 2007; Knollman-Porter et al., 2019; Worrall et al., 2011). The majority of PWA who have gained experience with TTS technology through research participation indicate a strong preference for the support regardless of the length of reading material (Brown et al., 2019; Harvey et al., 2013; Knollman-Porter et al., 2019, 2022; Wallace et al., 2019). Despite not having explored book reading with TTS support, PWA participating in previous studies involving TTS technology have expressed anticipated benefits when reading longer, functional materials. They have mentioned decreased fatigue, faster reading speed, greater independence, greater diversity of materials read, and improved communication with others as possible benefits of having TTS support (Caute et al., 2018; Hux et al., 2021; Knollman-Porter et al., 2022; Wallace et al., 2021). Because of these predicted benefits, further examination of TTS support with book-length material is an important next step to understanding the extent to which TTS technology can compensate for functional reading deficits associated with aphasia.

Study Purpose

Many PWA want to read books (Knollman-Porter et al., 2015, 2019) and at least some appear to benefit from having TTS technology as a reading support for relatively short materials (Brown et al., 2019; Harvey et al., 2013; Knollman-Porter et al., 2019, 2022; Wallace et al., 2019). Furthermore, benefits such as improved comprehension and decreased processing time appear to occur more frequently when PWA read content containing more than just a few sentences; such advantages, however, are far from universal and do not necessarily extend to the reading of book-length material. We also do not know how NHAs feel about reading with the simultaneous spoken output provided by TTS systems, nor do we know how it affects their reading comprehension and speed. Because researchers often compare the behavior of people with acquired disorders to NHAs to better understand the extent of experienced deficits or the advantage associated with a compensatory strategy, we sought to investigate the effects of TTS support on people from both of these populations. Hence, we examined the comprehension accuracy and total processing time of NHAs and PWA when reading book sections that alternated regarding the provision of TTS support. We also examined the preferences of people in both participant groups regarding TTS support. Finally, we examined correlations between aphasia severity and comprehension and processing time when participants with aphasia read with and without TTS support. Specific research questions included the following:

1. How does comprehension accuracy vary when PWA and NHAs process written content from book sections in read-only and TTS-supported conditions?

2. How does processing time vary when PWA and NHAs process written content from book sections in read-only and TTS-supported conditions?

3. What are the relations among reading comprehension, processing time, and aphasia severity?

4. What are the preferences of PWA and NHAs regarding the read-only and TTS-supported conditions when processing written content from book sections, and how do they foresee using TTS technology in the future during book reading activities?

Method

Institutional review boards from the participating institutions approved all methods and procedures prior to participant recruitment and data collection.

Participants

Inclusionary criteria both for PWA and NHAs included being between 19 and 90 years of age, speaking American English as a primary language, passing hearing and vision acuity screenings, and having access to a personal computer or tablet with reliable Internet and Webex connectivity. The vision screening required accurate identification of a participant's first name each time it appeared in 12 rows of five names each in 8-point Calibri font. This font type and size mirrored the default setting on NaturalReader—the TTS application used for the current study—and signified adequate vision to complete the experimental tasks. Regarding hearing, participants C, D, R, and S regularly wore hearing aids, had received audiological services within the past year, and demonstrated adequate hearing of conversational speech with their aids. All other study participants passed a hearing screening confirming perception of 1000-, 2000-, and 4000-Hz tones presented at 40 dB in at least one ear.

Participants With Aphasia

Participants with aphasia included five men and five women. They ranged in age from 37 to 76 years (M = 63.2, SD = 13.48), were between 26 and 294 months poststroke or stable encephalopathy (M = 151.4, SD = 99.50), and had completed between 12 and 19 years of education (M = 15.7, SD = 2.58). All were right-hand dominant prior to acquiring aphasia. At the time of study participation, only participant E was receiving speech-language pathology services; his goals focused exclusively on spoken expression.

We used results from standardized aphasia batteries and subtests performed within the past year to describe the language and cognitive abilities of participants with aphasia. Assessment results served descriptive rather than exclusionary purposes. Assessment measures included the Aphasia Quotient portion of the Western Aphasia Battery–Revised (WAB-R; Kertesz, 2006), the Comprehension of Spoken Paragraphs of the Comprehensive Aphasia Test (CAT; Swinburn et al., 2004), the Paragraph Factual subtest of the Reading Comprehension Battery for Aphasia–Second Edition (RCBA-2; LaPointe & Horner, 1998), and the Cognitive Linguistic Quick Test–Plus (Helm-Estabrooks, 2017).

WAB-R Aphasia Quotient scores ranged from 37.3 to 98.4 (M = 71.81, SD = 17.06). In accordance with cutoff scores provided in the WAB-R manual, two participants exhibited mild aphasia (i.e., scores between 76 and 93.8; Kertesz, 2006), six exhibited moderate aphasia (i.e., scores between 51 and 75), and one exhibited severe aphasia (i.e., scores between 26 and 50). The remaining participant scored above the threshold for an aphasia diagnosis using WAB-R criteria but had a clinical diagnosis of aphasia and demonstrated language difficulties during conversation and lengthy reading tasks. Three participants displayed characteristics of fluent aphasia, and seven participants displayed characteristics of nonfluent aphasia based on WAB-R aphasia quotient scores (Kertesz, 2006). RCBA-2 Paragraph Factual subtest scores ranged from 6 to 10 (M = 9.1, SD = 1.28), and CAT Comprehension of Spoken Paragraphs subtest scores ranged from 2 to 4 (M = 3.3, SD = 0.82). Demographic and assessment information for participants with aphasia appears in Table 1; participants are listed from lowest to highest WAB-R Aphasia Quotient score.

Table 1.

Demographic and testing information for people with aphasia.

								WAB-R	CAT	RCBA-2	CLQT+
Participant	Race	Gender	Age (years)	Education level (years)	Time postonset (months)	Aphasia type	Fluency type	Aphasia Quotient (100)	Spoken Paragraph (4)	Paragraph Factual (10)	Composite Severity (4)
A	White	Female	76	16	267	Broca's	Nonfluent	37.3	2	9	3
B	White	Male	54	12	186	Broca's	Nonfluent	60.74	3	8	2.4
C	White	Male	52	18	26	Transcortical motor	Nonfluent	61.1	2	6	2.8
D	White	Female	67	14	269	Broca's	Nonfluent	70.9	4	10	3
E	White	Female	76	12	117	Anomic	Fluent	71.4	3	9	3.6
F	White	Male	52	16	163	Broca's	Nonfluent	71.5	4	10	3.4
G	White	Male	72	18	63	Transcortical motor	Nonfluent	73.5	4	10	3.2
H	White	Male	74	19	294	Transcortical motor	Nonfluent	82.2	4	9	3.4
I	White	Female	37	18	79	Anomic	Fluent	91.1	3	10	3.6
J	White	Female	72	14	50	Anomic	Fluent	98.4	4	10	*

Note. The asterisk indicates data not available. WAB-R = Western Aphasia Battery–Revised (Kertesz, 2006); CAT = Comprehensive Aphasia Test (Swinburn et al., 2004); RCBA-2 = Reading Comprehension Battery of Aphasia–Second Edition (LaPointe & Horner, 1998); CLQT+ = Cognitive Linguistic Quick Test–Plus (Helm Estabrooks, 2017).

Neurotypical Healthy Participants

Participants who were NHAs had no history of neurological or developmental impairment. They included four men and six women. We selected NHAs who were within 1 year of education and 6 years of age of matched participants with aphasia. The NHAs ranged in age from 36 to 81 years (M = 64.2, SD = 13.11) and had completed between 12 and 19 years of education (M = 15.8, SD = 2.61).

We assessed the language, cognitive, and reading comprehension abilities of the NHA participants via the Telephone Interview for Cognitive Status (TICS; Brandt & Folstein, 2003) and the Paragraph Factual subtest of the RCBA-2 (LaPointe & Horner, 1998). The TICS is a brief, standardized test of attention, memory, language, and executive function (Brandt & Folstein, 2003). NHAs needed to score above 34 points—the score corresponding to unimpaired functioning—to qualify for participation. Obtained TICS scores ranged from 35 to 39 (M = 37, SD = 1.41). Demographic and assessment information for NHA participants appears in Table 2.

Table 2.

Demographic and testing information for neurotypical healthy adults.

Participant	Race	Gender	Age (years)	Education level (years)	TICS	RCBA-2
					Raw score	Paragraph Factual (10)
K	White	Male	74	14	35	10
L	White	Female	81	12	36	10
M	White	Male	72	16	36	10
N	White	Female	68	14	36	10
O	White	Female	60	12	37	9
P	White	Male	71	19	37	10
Q	White	Female	55	18	38	10
R	White	Female	36	18	38	10
S	White	Male	54	17	39	10
T	White	Female	71	18	39	10

Note. TICS = Telephone Interview of Cognitive Status (Brandt & Folstein, 2003); RCBA-2 = Reading Comprehension Battery of Aphasia–Second Edition (LaPointe & Horner, 1998).

Materials

Study materials included technology equipment and software; an informational binder containing study materials and printed instructions for computer use, accessing reading materials, and video conferencing with the researchers; and the reading materials, comprehension questions, calendars, and reading experience questionnaire.

Technology Equipment and Software

We loaned each participant a Samsung Galaxy Tab A7 tablet and a charging cord for use throughout research participation. We installed Gmail, Google Drive, and NaturalReader applications on each tablet, removed all other applications from the home screen, set the screen brightness to the maximum level, and locked each tablet in portrait mode. We selected NaturalReader as the TTS application because the included features (e.g., voice rate and font size) were easy to modify, and the software was compatible with Android devices and free of charge. We activated sentence highlighting within NaturalReader for all participants because prior research had shown that PWA most preferred this type of highlighting when reading short narratives (Brown et al., 2021). In addition, we programmed all systems to use the NaturalReader “Will” voice. We selected this voice based on overall intelligibility as perceived by the authors and previous results suggesting that the majority of PWA prefer TTS presented with a male voice (Knollman-Porter et al., 2022). We also loaned each participant an adjustable stand for tablet placement. This provided hands-free access and maintained tablet orientation in portrait mode. Participants also used their own personal computer or tablet with access to Webex as a means of interacting remotely with the researchers.

Informational Binder

We provided participants with a binder containing instructions for tablet use, accessing reading sections via Google Drive and NaturalReader, manipulating tablet features (i.e., font size, volume, and TTS output rate), and video conferencing via Webex. The binder also included a set of eight calendars for documenting reading progress (i.e., time read, pages read, and number of reading sessions) throughout the study. A zippered pouch contained four laminated, 3.75 × 6.75 in., colored cards (i.e., one each in orange, yellow, green, and blue), along with one “Yes” and one “No” card for responding to reading comprehension questions. All printed material incorporated aphasia-friendly formatting (Brennan et al., 2005; Rose et al., 2003) through the use of 18-point standard font, abundant white space, and visual-graphic supports to accompany printed instructions. Figure 1 presents a summary of binder contents.

Figure 1.

Binder contents.

Reading Materials

Fannie Flagg's A Redbird Christmas (2004), a 207-page novel with an average Flesch–Kincaid grade level of 5.1, served as the book for this study. We divided the novel into eight sections in accordance with the existing chapters and converted each section to a PDF document. This yielded a total of 100 pages of text, with each section ranging in length from 8 to 15 pages. We determined section content in such a way as to minimize length differences while still maintaining the chapter divisions established by the novel author. Table 3 provides details about each reading section.

Table 3.

Description of reading sections.

Reading section	Corresponding novel chapters	Number of PDF pages	Total sentences	Total words	Flesch–Kincaid grade level
1	1 & 2	12	409	5,565	5.2
2	3 & 4	8	253	4,223	5.6
3	5	13	438	5,788	4.3
4	6–8	15	434	7,452	6.2
5	9–12	11	420	5,905	5.1
6	13–16	11	426	5,833	4.8
7	17–19	15	670	8,969	4.5
8	20–22	15	542	8,008	5.2

We purchased chapter summaries of the target novel from the website for the Aphasia Center of California (i.e., http://www.aphasiacenter.net/the-book-connection/; Aphasia Center of California, n.d.). We combined chapter summaries to correspond with our established reading sections and edited them for length and content to facilitate participant understanding. Across sections, the edited reading summaries averaged 2.25 pages (SD = 0.43, range: 1.5–2.5 pages) using a 16-point standard font. On average, they contained 477 words, 15.25 bullet points, and had a Flesch–Kincaid grade level of 4.05. The summaries provided a means for participants to review with the experimenter the content they had read independently for each assigned section of the novel after having performed the associated comprehension testing.

Comprehension Questions

We wrote 15 multiple-choice questions relating to the content presented in each of the eight reading sections. In general, PWA understand explicit content better than inferential information during paragraph reading tasks (Webster et al., 2018). Therefore, because of the preliminary nature of this study and the comprehension demands associated with reading a novel, questions addressed only explicit content directly expressed in the written text. A researcher-generated PowerPoint presentation served as the means of presenting the comprehension questions for each reading section. We created the presentation slides in accordance with those developed by Knollman-Porter et al. (2022) for remote administration to study participants. Each presentation included introductory slides to establish that a participant had the correct materials for responding and was ready to answer the comprehension questions. Next, each comprehension question appeared on a separate slide in 40-point font. The sentence stem was at the top of a page, and four possible response options in a 2 × 2 colored grid comprised the remainder of the page; an example appears in Appendix A. The colors in the grid matched the response cards provided in the informational binder. The color coding allowed participants to indicate the selection of their desired response by holding up the corresponding color card for the researcher to see rather than having to verbalize an answer. A verification slide followed each question slide to give participants a means of confirming or disconfirming their selected response.

Reading Experience Questionnaires

We created questionnaires for each reading condition to query participants regarding their experiences using the tablet and reading application (see Appendix B). The questionnaires included yes/no, multiple-choice, open-ended, and Likert-type rating scale questions. Within a PowerPoint presentation, one question appeared per slide using 44-point font and visual support images.

We also asked participants two final questions about the two experimental conditions at the end of the final comprehension testing and check-in session. The final questions were as follows:

1. When considering your entire experience during this study, do you prefer the read-only or TTS condition? Why?

2. If a TTS system were available to you, would you use it to read books? Why or why not?

Procedure

Participants engaged in pre-experimental training, independent reading, and comprehension testing and check-in sessions.

Pre-experimental Training

We met in person with each participant to review the study materials and provide instructions about tablet use, steps for accessing reading materials in the read-only and TTS-supported conditions, and steps for manipulating reading and tablet features (i.e., font size, volume, and TTS output rate). We provided modeling and standby support until a participant demonstrated independence in performing the procedures. We then had the participant demonstrate proficiency using practice stories for the read-only and TTS-supported conditions. Participants also received in-person training to learn how to document their time spent reading. We provided visual models and gave participants the opportunity to practice using the visually displayed time on the tablet to document reading starting and ending times on the provided reading calendar. If a participant could not demonstrate accurate copying of digitally displayed starting and ending times, we requested that a care partner perform this task. We used the documented reading times to calculate the total time a participant spent reading each book section.

We met with each participant either in person or online via Webex two additional times over the next 24–72 hr to ensure competence operating the experimental technology and accessing the study materials; standby support from a care partner was permitted for technology setup and accessing reading materials. Appendix C lists the skills required to confirm operational and material access competency.

Independent Reading Sessions

We e-mailed a PDF of each reading section to a participant's tablet on the first day accessing that particular section was allowed. Participants alternated between the read-only and TTS-supported conditions to review the reading sections. Half of participants began with the read-only condition, whereas the other half began with the TTS-supported condition using NaturalReader. To simulate typical reading experiences, participants performed independent reading sessions at home and at desired times. They could read or read and listen at their own pace, reread portions or all of a reading section, and modify technology features (e.g., font size, volume, and TTS output rate) at their discretion; however, other than for accessing a reading section or operating the technology equipment, they had to complete reading activities independently and without the assistance of another person.

Participants had 7 days following receipt of a reading section to review as much of the material as possible. If a participant completed the reading section in fewer than 7 days, they notified us via phone or e-mail to schedule a comprehension testing and check-in session. If a participant did not contact us or finish reading the material within 7 days, we initiated contact to schedule the reading comprehension and check-in session. Only two instances occurred in which a participant reported not completing the assigned reading within 7 days. In these instances, we extended the allotted time by 2 days to adjust for conflicts associated with national holidays. In all cases, participants reported completing the assigned reading prior to the comprehension testing and check-in session. Subsequent to the comprehension testing and check-in session, we e-mailed a PDF of the next reading section to the participant. This continued until the participant had read or read and listened to all eight reading sections comprising the target novel.

Comprehension Testing and Check-In Sessions

We performed a reading comprehension and check-in session either within 48 hr of a participant indicating completion of an assigned reading section or after a participant had access to the reading section material for 7 days, whichever came first. Because memory abilities could influence performance on comprehension testing when administration followed a substantial delay after having completed the reading task, we strove to complete comprehension testing sessions on the day the participant reported reading completion. We completed 78% of the testing sessions on the day the participant reported finishing the reading, 20% within 24 hr, and 2% within 48 hr of reading completion.

Participants used their personal computer or tablet device to connect with a researcher via Webex for the comprehension testing and check-in session. To document participants' verbal and gestural responses, we video- and audio-recorded the sessions using the Webex built-in record feature. The session included five activities: (a) soliciting and recording responses to comprehension questions, (b) reviewing a summary of the reading section, (c) reviewing correct responses to all comprehension questions, (d) completing the reading experiences questionnaire, and (e) reviewing the reading calendar. At the end of the final session, we asked participants for their opinions about reading with and without TTS support.

Comprehension testing. We used the PowerPoint presentation to solicit participant responses to the 15 comprehension questions associated with a reading section. To respond, participants either verbally expressed their answer or held up the color card corresponding with their desired answer. After each question, we verified the participant's response via the PowerPoint presentation. Participants could not refer to the reading materials or any previously taken notes when answering comprehension questions. After completion of the testing, we informed participants about their response accuracy.

Reviewing section summaries. We read aloud the visually presented written summary associated with a reading section to solidify understanding of the material by participants with aphasia prior to advancing to the next reading section. NHA participants read the written summaries independently. Participants could ask questions about the summary content, as desired.

Comprehension question review. The next activity was a review of the correct responses to the 15 comprehension questions presented earlier. This served to further solidify participants' understanding of the reading material. We reviewed the comprehension questions regardless of the accuracy of participant responses.

Reading experience questionnaire. We asked participants questions regarding tablet use and reading experience with the experimental condition used to access the most recent reading section materials. We read questions aloud while participants viewed them simultaneously on the PowerPoint presentation. Participants responded either verbally or with gestures to signify desired responses. For open-ended questions, participants provided verbal answers to the best of their ability. We prompted them to discuss any challenges they experienced reading the materials either independently or with TTS support, based on the assigned condition.

Reviewing of the reading calendar. Participants held up their reading calendar, so we could view it and capture an image of it on the computer monitor. If available and as needed, a participant's friend or family member interpreted or clarified calendar information. We defined time for reading section completion as the total number of minutes participants reported spending with the section.

Final session questions. We asked participants two final questions about the two experimental conditions at the end of the final comprehension testing and check-in session. Following response to these questions, we also provided participants with information about the total time they spent completing the assigned reading and their average comprehension accuracy for each condition. Participants could change responses to final session questions following presentation of cumulative accuracy and processing time data for each condition.

Data Analysis

Study variables of interest included (a) participants' overall percentage of correct responses to comprehension questions in each experimental condition, (b) total processing time in minutes in each experimental condition, and (c) participants' condition preference at the end of the final check-in session. We selected the criterion of scores within 10 percentage points of one another as indicative of comparable performance for clinical significance based on methods from prior TTS technology research (Hux et al., 2017). A 10% difference in comprehension accuracy between stimulus conditions is a relatively conservative criterion; hence, we felt this criterion was both justifiable and clinically relevant.

Comprehension Accuracy and Processing Time

We used R software Version 4.1.1 to perform the statistical analyses. We fit separate two-way mixed analysis of variance (ANOVA) models to the study variables of comprehension accuracy and processing time; group (i.e., PWA and NHAs) was a two-level between-subjects factor, and condition (i.e., read only and TTS) was a two-level within-subject factor. We logit transformed (i.e., log odds) the comprehension accuracy variable to account for the proportional response and associated variance concerns secondary to floor and ceiling effects at 0 and 100, respectively. We log transformed the processing time variable to account for skewness. We first tested for a significant interaction effect between the group and condition factors. Significant results prompted group comparisons within each condition. We checked and verified ANOVA model assumptions. All tests used a probability level of .05 to determine significance. We further analyzed the PWA group data by calculating and determining the significance of Pearson product–moment correlations among aphasia quotient scores on the WAB-R, average comprehension accuracy scores in both experimental conditions, and average processing times in both experimental conditions.

Participant Condition Preferences and Anticipated Future Use

We totaled the number of participants with aphasia and the number of NHA participants who preferred each experimental reading conditions when considering their experience throughout the study. Finally, we tallied the number of participants who expressed a desire to use or not use a TTS system to read books in the future, assuming access to TTS technology.

Results

Comprehension Accuracy

ANOVA computation revealed no interaction effect between group and condition, F(1, 18) = 2.236, p = .152; however, a significant main effect for group was evident regardless of condition (estimated NHAs/PWA odds ratio = 5.37, SE = 0.059), F(1, 18) = 27.62, p < .001. The percentage of correct responses to comprehension questions by PWA was greater in the TTS-supported condition (M = 78.5, SD = 14.96) than the read-only condition (M = 73.9, SD = 17.44). In contrast, comprehension accuracy was greater for NHA participants in the read-only condition (M = 97.6, SD = 2.07) than in the TTS-supported condition (M = 96.4, SD = 2.91). Figure 2 displays individual PWA scores, group average, and group median comprehension accuracy scores for each condition. The figure does not show individual scores for NHA participants because they all performed similarly and with near 100% accuracy.

Figure 2.

Participants' percent comprehension accuracy across condition.

Four participants with aphasia (i.e., B, F, G, and H) demonstrated a clinically significant increase (i.e., 10% or greater) in comprehension accuracy given TTS support compared to no reading support despite the lack of a statistically significant difference for experimental condition or the interaction between participant group and experimental condition. Participants G and F achieved the greatest improvement (i.e., 20% and 23% increases, respectively). The magnitude of change reaching the level deemed clinically significant never occurred in the opposite direction—that is, no participant increased comprehension accuracy by 10% or greater in the read-only condition compared to the TTS-supported condition.

Total Processing Time

Total time processing was measured in minutes as documented by study participants. PWA spent more time processing the reading materials in the read-only condition (M = 595.60, SD = 414.12) than the TTS-supported condition (M = 373.10, SD = 277.39). In contrast, NHA participants spent less time processing materials in the read-only condition (M = 116.40, SD = 18.72) than TTS-supported condition (M = 137.40, SD = 27.06). ANOVA computation revealed a significant interaction effect between group and condition, F(1, 18) = 7.464, p = .0006. Within the PWA group, the estimated mean ratio of the read-only to TTS-supported condition was 1.633 (SE = 0.1798), meaning that, for this group, the estimated mean time for the read-only condition was 1.633 times longer than for the TTS condition. This difference was statistically significant, t(18) = 4.454, p = .0003. For NHAs, the difference was not significant, estimated read-only/TTS ratio = 0.852, SE = 0.0938, t(18) = −1.456, p = .1626. Figure 3 presents individual data for PWA, group average, and group median total processing time for each condition. The figure does not present individual data for NHA participants because of the performance comparability within and across conditions.

Figure 3.

Participants' total processing time across conditions.

Correlations

Calculation and significance testing of the Pearson correlations appearing in Table 4 revealed a significant positive correlation between WAB-R Aphasia Quotient scores and read-only and TTS-supported comprehension accuracy scores; thus, participants who achieved higher Aphasia Quotients tended to respond with greater accuracy on comprehension questions than participants who achieved lower scores. A significant positive correlation also occurred between the read-only and TTS-supported comprehension accuracy scores, suggesting higher accuracy in one condition correlated with higher accuracy in the other condition. In contrast, a significant correlation between processing time and WAB-R Aphasia Quotients did not occur in either experimental condition, thus suggesting accuracy on the standardized testing did not correspond to the processing time taken by participants with aphasia when reading book sections. However, a significant positive correlation did occur between the read-only and TTS conditions regarding processing times, indicating that longer processing time in one condition correlated with longer processing time in the other condition.

Table 4.

Pearson correlation coefficients among Western Aphasia Battery–Revised (WAB-R) and experimental task performance.

Variable	1	2	3	4
1 WAB-R Aphasia Quotient
2 Read-only comprehension accuracy	.669*
3 TTS-supported comprehension accuracy	.686*	.881**
4 Read-only processing time	−.369	−.189	.381
5 TTS-supported processing time	−.426	−.343	−.601	.762*

Note. TTS = text-to-speech.

^* p < .05.

^** p < .01.

Participant Condition Preferences

We asked participants two final questions at the end of the experiment to reveal overall preferred condition and the likelihood of using TTS technology for future book reading activities. For condition preference, six of 10 PWA (i.e., B, C, D, F, G, and J) expressed an overall preference for the TTS-supported condition, two (i.e., E and H) preferred the read-only condition, and two (i.e., A and I) preferred both conditions equally. In contrast, only one NHA participant (i.e., O) expressed a preference for the TTS-supported condition; eight (i.e., K, M, N, P, Q, R, S, and T) preferred the read-only condition, and one (i.e., L) preferred both conditions equally.

Nine of 10 PWA participants (i.e., A, B, C, D, E, F, G, I, and J) indicated a desire to use TTS support for future book reading endeavors, whereas one (i.e., H) desired to read without the support of TTS technology. In contrast, one of 10 NHA participants (i.e., O) indicated a desire to use TTS support for future book reading, four (i.e., Q, R, S, and T) desired to read without TTS support, and five (i.e., K, L, M, N, and P) indicated that they might use TTS support for future reading.

Discussion

The primary purpose of this study was to examine the comprehension accuracy and total processing time of PWA and NHAs when reading sections of a book in alternating read-only and TTS-supported conditions. We also examined preferences when PWA and NHAs processed written content in each condition. To gather attitudes about the potential use of TTS systems for book reading, we inquired about the desirability of using the technological support for future reading activities; specifically, we asked participants which condition they preferred and if they would consider using TTS technology in the future if it were available to support their reading of books.

Comprehension Accuracy

A significant difference in comprehension accuracy was evident between the participant groups; however, a significant difference was not present between the read-only and TTS-supported conditions for either group. The lack of a statistically significant difference between the two presentation conditions aligns with findings from past research exploring comprehension of five- to six-sentence narratives and nine- to 14-sentence newspaper articles by PWA (Knollman-Porter et al., 2019, 2022; Wallace et al., 2019). A potential reason for the confirmatory result in this study was the small sample size and the presence of substantial variation among the participants in aphasia type and reading impairment severity. This may have resulted in insufficient power to reveal a significant difference across conditions.

Individual differences in comprehension were evident among the participants with aphasia despite the lack of a significant finding for the group as a whole. As has occurred in previous studies, some participants with aphasia exhibited a clinically significant improvement in comprehension accuracy given TTS support compared to reading without access to assistive technology. Specifically, four participants (i.e., B, F, G, and H) exhibited a 10% or greater improvement in comprehension accuracy in the TTS-supported condition compared to the read-only condition. This suggests that the provision of TTS support facilitates improved comprehension for some PWA. Given this finding, exploration of TTS-supported reading of books warrants further investigation with a larger sample size than used herein. Exploration of the influence of aphasia type and the degree of reading impairment on the comprehension benefit afforded by TTS support is also worthy of investigation.

Total Processing Time

A significant difference in total processing time emerged between the two participant groups and a significant difference between the read-only and TTS-supported conditions within the group comprised of PWA. The latter finding aligns with results from previous studies in which PWA have processed paragraph-length content significantly faster with TTS support than without (Harvey et al., 2013; Knollman-Porter et al., 2019, 2022).

An additional finding of importance regarding processing time was the magnitude of difference in the read-only condition observed between the participant groups. On average, participants with aphasia spent approximately 10 hr over the course of the study to process materials in the read-only condition; individual total processing times ranged from 2.9 to 25.0 hr. This contrasted with an average of 2.0 hr of processing time spent by NHA participants in the read-only condition; their processing times ranged from 1.5 to 2.6 hr. Hence, the slowest processor in the NHA group was faster than the fastest processor in the group of PWA, and the slowest participant with aphasia took almost 10 times longer than the slowest NHA reader. This discrepancy reinforces the notion that reading book-length material requires considerable effort from PWA and places a substantial burden on them because of the time required to decode and comprehend content. The fact that some PWA abandon time-consuming and difficult reading tasks (Knollman-Porter et al., 2015) is not surprising given the time commitment required. In addition, the fact that processing time did not correlate with WAB-R Aphasia Quotient scores suggests that severity of aphasia does not influence the speed with which PWA can process written content when reading books. For this reason, clinicians must examine beyond the administration and interpretation of standardized test battery results the reading behaviors of PWA to determine the impact of processing time on lengthy reading tasks.

The time discrepancy between participant groups was less substantial but still large in the TTS-supported condition. On average, the participants with aphasia devoted 6.2 hr to process content in the TTS-supported condition; their times ranged from 2.9 to 14.5 hr. This contrasted with the NHA participants who spent an average of 2.3 hr processing content in the TTS-supported condition; processing times ranged from 1.6 to 3.1 hr. Hence, some overlap, albeit small, occurred between the groups, but the slowest reader among the participants with aphasia took approximately 4.5 times longer than the slowest reader among the NHA participants. These results serve to strengthen previous findings suggesting that access to TTS technology benefits PWA because it lessens the time needed to process written content (Hux et al., 2021; Wallace et al., 2021). The multimodal presentation of book content at a desired speed and with sentence-by-sentence highlighting appears to decrease the burden on PWA caused by decoding and comprehension deficits. Thus, the technology increases overall reading efficiency and increases the likelihood that a PWA can engage in previously enjoyed reading activities. Still, decreased processing time in the TTS condition was not universally observed among the participants with aphasia. This suggests that TTS support promotes clinically significant decreases in processing times for some, but not all, PWA. Specifically, two participants (i.e., E and H) took longer to process written content in the TTS condition than the read-only condition. Of note, one of these participants—that is, H—had the shortest total processing time (i.e., 2.9 hr) of all participants with aphasia in the read-only condition, and his total processing time was relatively close to the average time exhibited by NHA participants (i.e., 2 hr). H's processing time lengthened by less than 1 hr given TTS support. Given this fact, H's response to TTS support appears to have more closely approximated that of the NHA participants than the participants with aphasia. Potential reasons for this may relate to the type of aphasia he exhibited (i.e., transcortical motor), the nature of his relatively preserved versus impaired reading skills, his prior reading habits, his reading environment, and/or his comfort with reading books independently.

Most NHA participants took more time to process the reading materials in the TTS-supported condition than in the read-only condition. Although the processing time difference was not statistically significant, the narration and highlighting features of TTS support may have been distracting to NHA participants, thus reducing their attention and/or increasing their rereading of text passages. The lack of underlying decoding and comprehension deficits among the NHA participants appears to have provided sufficient cognitive resources to minimize the visible impact of any such challenges. Given that the everyday reading practices of NHAs are variable and show inconsistent patterns on the basis of demographic groupings (Parr, 1992), further systematic exploration of the effects of TTS output on their reading behaviors may be informative.

Condition Preferences, Perceived Comprehension, and Projected TTS Use

Participant Condition Preferences

Taking individual preferences into account is important when selecting reading support strategies because assistive technology features can affect enjoyment, interaction, and perceived competence with high-interest materials (Cordova & Lepper, 1996). To implement effective strategies, clinicians must ensure that instructional supports align with a client's personal preferences and goals. Similar to previous studies about using assistive technology to support post-aphasia reading (Brown et al., 2019; Knollman-Porter et al., 2019; Wallace et al., 2019), a clear, universally endorsed preference for or against TTS-supported reading did not emerge in this study. Instead, 60% of the participants with aphasia expressed a preference for having TTS support available when reading books, 20% expressed a preference for not having TTS support, and 20% were ambivalent.

Participants' ease with using and accessing material via technology, comprehension accuracy, and total processing time were likely contributors to preference decisions. Anecdotally, one participant (i.e., E) reported that she sat at her dining room table to keep the tablet plugged in when reading the study materials but that this contrasted with her prestroke habit of sitting in a comfortable chair in her living room to read. This change in positioning may have influenced her selection of reading without TTS support as her preferred condition. In addition, E and the other participant with aphasia who preferred the read-only condition (i.e., H) were the only two PWA who took longer to process content in the TTS-supported condition than the read-only condition.

One person (i.e., I) who preferred both conditions equally presented with comparatively mild deficits compared with the other participants with aphasia. This may have led her to feel relatively comfortable with unsupported reading. Another participant (i.e., C) changed his preference from the read-only to the TTS-supported condition following presentation of objective data about his performance in each condition. C reported believing his performance across conditions was comparable prior to learning that improved comprehension and less processing time occurred given TTS support. When informed about his performance, he changed his answer to the preference question and also indicated that he would consider using TTS technology for future book reading tasks. His change in attitude about TTS support shows the importance of providing objective performance feedback to PWA when asking them about support preferences.

The NHA participants contrasted markedly with the participants with aphasia regarding their reading condition preference. Eighty percent of the NHA participants versus 20% of the participants with aphasia indicated preference for the read-only condition. Anecdotal comments from NHA participants suggested that their preference stemmed from a desire to read from a hard-copy book rather than on a tablet, a dislike of simultaneously reading and listening to written content, and a desire to read at their preferred rate.

Use of TTS Technology to Support Future Book Reading

Participants with aphasia were nearly unanimous in their desire to use TTS technology to read books in the future; nine of the 10 participants indicated this desire. Potential reasons likely relate to the perceived and/or actual boost in comprehension accuracy, the decrease in total time required to process written content, the benefit provided by simultaneous auditory output and written content enhanced by highlighting to facilitate tracking and place marking, and the enjoyment associated with using technology.

Participants in the NHA group were fairly evenly split between not wanting to use a TTS system for future book reading and feeling ambiguous about the technology; four participants gave negative responses, five expressed ambivalence, and only one indicated a desire to use TTS technology in the future. Anecdotally, some NHA participants made spontaneous comments about not liking the TTS output because they found it distracting, they wanted to read faster than the selected output rate, they tended to “zone out” with the simultaneous auditory and written content, they felt uncomfortable with the reading environment of the TTS-supported condition, or they preferred holding and reading from a hard-copy book rather than a tablet.

Expecting all people to prefer a single way of reading books is not realistic given variability in likes and dislikes, reading ability, and prior experiences. Similarly, not all PWA are likely to prefer the same compensatory strategy to support book reading. Differences in decoding ability, comprehension accuracy, and processing speed as well as individual preferences and experiences are factors clinicians need to consider when selecting compensatory strategies (Brown et al., 2019, 2021; Knollman-Porter et al., 2019, 2022). Some PWA may prefer to maximize comprehension accuracy, whereas others may view processing time as a more important factor when reading. Consequently, exploring both factors and sharing objective data about a person's reading performance is essential to facilitating good clinical decisions when introducing and explaining the advantages and disadvantages of various reading support strategies (Worrall et al., 2011).

Limitations and Future Directions

This study was preliminary in nature. The large quantity of material we asked participants to read, our uncertainty about the time PWA would need to read the study materials, and the large time demand associated with managing this project prompted us to limit the sample size to 10 people per participant group and to not restrict the type of aphasia or reading challenges PWA exhibited. Because of this, generalizing the results is difficult. Future researchers may wish to include larger sample sizes and select participants with aphasia who have similar language and reading profiles. Doing so may reveal patterns of behavior regarding those people most or least likely to benefit from TTS support when reading books.

Participants performed all reading tasks associated with this study independently and in their preferred natural environment. Because of this, either the participant or a care partner documented starting and ending times for reading sessions. Although the documented processing time results are consistent with findings from past research, we cannot verify that the time data collected were accurate.

We also encouraged participants to read at their own pace, and we provided them unlimited access to each reading section during each 7-day period. We did this to facilitate a natural reading experience and respect the demands of everyday life. Unlimited access to study materials may have affected the observed differences in comprehension accuracy and total processing time across experimental conditions, however. Although we asked participants during each check-in session to confirm the experimental condition used to access the most recent reading material, we could not directly assess the manner in which participants engaged with the book content. Some may have relied primarily on the spoken TTS output when it was available, whereas others may have largely ignored the spoken output and relied primarily on the written text; still others may have relied equally on the written and auditory content presentations and used the TTS output in a multimodal fashion. Future researchers may wish to implement eye-tracking technology or other research methodologies to reveal ways in which PWA and NHAs interact with content presented in read-only and TTS-supported conditions. Findings from these types of studies may reveal differences between PWA and NHAs that facilitate a greater understanding of the reading behaviors of PWA and the influence of reading support strategies such as using TTS technology.

We limited the assessment of reading comprehension to 15 questions per reading section. We did this because a study by Webster et al. (2018) found that PWA comprehend main ideas easier than details, and we wanted to focus on reading for meaning versus the recall of details that may not directly influence overall understanding of a book. However, the result was a limiting of the total number of data points per person to 60 per condition. Having a greater number of data points would have provided more robust findings.

A final limitation involves the technology used to present study materials. As technology use grows and evolves, software updates are common. During this study, a NaturalReader update occurred that caused a popup box to appear unexpectedly on the tablet screen, along with a change to previously selected personalized settings. When notified about the update, we assisted participants to ensure they retained access to study materials and could reset preferred TTS settings. However, routine updates such as this may affect a person's comfort and success using TTS technology, especially if difficulty resolving or mitigating these issues results.

Conclusions

PWA desire to read books, but they often abandon the activity because of the cognitive effort and time burden associated with decoding and comprehending written text. At this time, reliably effective treatments to restore word decoding, reading comprehension, and reading speed to pre-aphasia levels do not exist. However, TTS technology shows promise as a compensatory support to help at least some PWA read and understand book-length material within reasonable time periods, thus paving the way for a return to previously enjoyed reading activities. Furthermore, most PWA who have had experience with TTS technology express interest in using it for future book reading activities. However, TTS technology is not universally beneficial. The simultaneous auditory and written output can be disruptive and distracting to some PWA and most NHAs—especially if a mismatch exists between the rate of spoken output and a person's desired reading speed. Because both potential advantages and disadvantages exist, clinicians need to engage in thoughtful discussions with PWA before promoting the technology. This includes ensuring sufficient opportunities to practice and master system operation, experiment with adjustments to modifiable features, and understand how TTS support affects comprehension accuracy and processing speed for a particular PWA.

Data Availability Statement

Data are available upon request from the authors.

Appendix A

Sample Comprehension Question From PowerPoint Presentation

Appendix B

Weekly Questionnaire

Appendix C

Skills Required to Confirm Operational and Material Access Competency

Note. Participants also had to demonstrate mastery accessing study materials for the read only condition.

Funding Statement

Research reported in the article was supported by the National Institute on Deafness and Other Communication Disorders under Award 1R15DC015579. The content is solely the responsibility of the authors and does not necessarily represent the view of the National Institutes of Health.