Attentive Reading With Constrained Summarization-Written, A Multi-Modality Discourse-Level Treatment For Mild Aphasia

APHASIOLOGY
2021, VOL. 35, NO. 1, 100–125

https://doi.org/10.1080/02687038.2019.1686743
Attentive Reading with Constrained Summarization-Written,

a multi-modality discourse-level treatment for mild aphasia
Jessica A. Obermeyera, Yvonne Rogalskib and Lisa A. Edmondsc
a
Communication Sciences and Disorders in the School of Health and Human Sciences, University of North
Carolina at Greensboro, Greensboro, NC, USA; bSpeech-Language Pathology and Audiology in the School of
Health Sciences and Human Performance, Ithaca College, Ithaca, NY, USA; cCommunication Sciences and
Disorders in the Department of Biobehavioral Sciences, Teachers College, Columbia University, New York
City, NY, USA
ABSTRACT ARTICLE HISTORY

Background: Attentive Reading with Constrained Summarization- Received 29 August 2018
Written (ARCS-W) is a treatment in development that was adapted Accepted 24 September 2019
to improve written and spoken discourse in people with mild aphasia KEYWORDS
by targeting the cognitive requirements of discourse production. The Discourse; mild aphasia;
current research represents the second report of ARCS-W and aimed aphasia treatment; discourse
to refine the protocol and replicate previous results. writing
Aims: Determine if and to what extent ARCS-W affects the following
abilities at post-treatment and one-month post-treatment:
1. Micro and macrolinguistic written discourse abilities.
2. Micro and macrolinguistic spoken discourse abilities.
3. Other measures of language, including confrontation naming
(spoken and written), sentence production and functional
communication.
Methods & Procedures: ARCS-W was administered to two partici-
pants with mild aphasia in a pre- to post-treatment design.
Treatment required attentive reading followed by constrained sum-
marization of novel news articles in the written and spoken modal-
ities. Constraints were: use specific words, use complete sentences,
stay on topic. Treatment outcomes evaluated micro (correct informa-
tion units, complete utterances, grammatically complex utterances)
and macrolinguistic (main events and main ideas, global coherence)
discourse structure, confrontation naming, sentence production, and
functional communication.
Outcomes & Results: Both participants demonstrated improvement in
untrained discourse at micro and macro levels. Participant 1 improved
lexical specificity in confrontation naming and the proportion of main
events and main ideas conveyed in discourse, and Participant 2
improved topic maintenance and lexical retrieval in discourse.
Conclusions: There are few treatments for people with mild aphasia
that work at a discourse level and these findings provide further
evidence that ARCS-W may be a treatment option to improve written
and spoken discourse in this population.
CONTACT Jessica A. Obermeyer jaoberme@uncg.edu Communication Sciences and Disorders in the School of
Health and Human Sciences, University of North Carolina at Greensboro, 300 Ferguson Building, 524 Highland Ave.,
Greensboro, NC 27402, USA
Supplemental data for this article can be accessed here.
© 2019 Informa UK Limited, trading as Taylor & Francis Group
APHASIOLOGY 101
Introduction
Aphasia is an acquired language disorder that impacts listening, reading, speaking, and/or
writing (National Aphasia Association, n.d). As a result of these impairments, people with
aphasia (PWA) have reduced functional communication that impacts their ability to
participate in conversation, maintain social relationships and regain their independence
(Hilari & Northcott, 2006; Vickers, 2010). People with mild aphasia (PWmA) typically report
substantial communication impairments even though they score well on standardized
tests (Armstrong, Fox, & Wilkinson, 2013; Elman & Bernstein-Ellis, 1995). This may be
because impairments are often most evident in discourse (Frankel, Penn, & Ormond-
Brown, 2007; Fromm et al., 2017), though there are few treatments that address spoken
discourse-level language abilities (e.g., Armstrong et al., 2013) and even fewer that
address written discourse (e.g., Behrns, Hartelius, & Wengelin, 2009). The current study
represents the second report of Attentive Reading with Constrained Summarization-
Written (ARCS-W; Obermeyer & Edmonds, 2018) with two additional PWmA based on
Western Aphasia Battery-Revised criteria (Kertesz, 2006). Treatment efficacy can be estab-
lished through five phases identified by Robey (2004). The purpose of Phase I is to
establish therapeutic protocols, dosage, and population, and evaluate the therapeutic
effect of the treatment. Phase II studies should further refine the treatment protocol,
therapeutic outcomes, dosage, and magnitude of efficacy. The current study represents
elements of both of these phases by aiming to evaluate the preliminary efficacy of ARCS-
W (Phase I), while also refining the treatment protocol, outcome measures and the
appropriate population (Phase II).
Broadly, discourse can include words or connected utterances that carry out commu-
nication actions such as conversation, writing an email, giving directions or telling a story.
The majority of daily interactions take place at a discourse level and written discourse is
becoming a substantial component of social (e.g., text messaging, social networking, and
email) and functional (e.g., banking and shopping) interactions (Dietz, Ball, & Griffith,
2011). This creates a greater need for treatments that address written discourse for PWmA.
While written discourse can be expressed via different modalities (e.g., handwriting,
typing) that have varied motoric and cognitive demands (Ardila, 2004), participants in
the current study were allowed to select the modality with which they were most
comfortable. Throughout this paper, we will use the term “written discourse” to refer to
both handwritten and typed discourse.
Written discourse is typically not addressed in research or in the clinical environment for
two main reasons: 1) insurance coverage is often exhausted by the time someone with aphasia
is ready to work on written discourse, 2) many clinicians and PWA perceive written discourse as
too difficult or time consuming to address in therapy (Behrns, Ahlsén, & Wengelin, 2008).
However, evidence suggests that PWA can demonstrate more linguistic complexity in written
discourse, making it an avenue for treatment that could facilitate greater functional commu-
nication and ability to interact socially (Behrns, Wengelin, Broberg, & Hartelius, 2009; Dietz
et al., 2011; Ulatowska, Hildebrand, & Haynes, 1978). The majority of these reports have focused
on people with mild or moderate aphasia due to the complexity of producing written
discourse. Behrns et al. (2009) found that PWA produced fewer words in written narratives
than spoken, but their written output had more syntactic complexity (e.g., clauses per T-unit)
and was easier to understand when rated by unfamiliar listeners. Ulatowska et al. (1978) found
102 J. A. OBERMEYER ET AL.
that written discourse was more complex than spoken discourse for both PWA and a control
group, and written output had higher communicative adequacy (based on ratings of specifi-
city and intelligibility) for PWA when compared to their spoken discourse. The combination of
these findings reveals that for some PWA, especially those with mild aphasia, written discourse
can allow for more complexity and communicative adequacy, likely due to the reduced time
pressure in written communication and the ability to monitor tangible written output.
However, there continues to be a lack of evidence-based treatments that address written
discourse output for PWA. This is problematic since improving written discourse could increase
the potential for social interaction, functional independence, and returning to work.
Attentive Reading and Constrained Summarization (ARCS; Rogalski & Edmonds, 2008;
Rogalski, Edmonds, Daly, & Gardner, 2013) is a discourse level treatment that aims to improve
both microlinguistic (words and sentences) and macrolinguistic (organizational framework)
discourse structure by targeting lexical specificity and topic maintenance during constrained
summarization of novel material. In this context, constraint requires participants to produce
a summary while following specific guidelines (e.g., no pronouns, no opinion) and monitor-
ing their performance. In ARCS, attention and intention are recruited through reading with
the intent to summarize using lexically specific items instead of imprecise words or pronouns
(e.g., bicycle instead of thing). The “no opinion” constraint works at the macrolinguistic level
by encouraging attention-to and maintenance-of the summarization topic. During this
process, participants monitor their ability to produce summaries while following the guide-
lines. The results of ARCS treatment have been reported in a single person with primary
progressive aphasia (Rogalski & Edmonds, 2008) and two people with Wernicke’s aphasia
(Rogalski et al., 2013). Attentive Reading with Constrained Summarization-Written, ARCS-W,
was adapted from this treatment for the purpose of improving written and spoken discourse
production in PWmA (Obermeyer & Edmonds, 2018). Consistent with ARCS, ARCS-W treat-
ment addresses the cognitive and linguistic requirements for discourse production by
targeting intentional word selection, cognitive flexibility, goal-directed behavior, and self-
monitoring. The addition of written discourse allows for an emphasis on executive functions
by providing the opportunity to set specific goals and monitor progress toward those goals.
By using a tangible written product, participants can monitor their output and determine if
they followed their constraint guidelines with the potential for more accuracy and awareness
than spoken discourse allows.
ARCS-W requires constrained summarization of novel material in writing and speaking
during each treatment session. Constraints include three summarization guidelines; the
first two target micro and macrostructure and are consistent with ARCS (e.g., no non-
specific words, stay on topic), and the third is prescribed based on participants’ individual
needs (e.g., no incomplete sentences, listen/look for repeated information). Additionally,
ARCS-W differs from ARCS by requiring participants to identify keywords, or the most
important words, in the text. This step ensures that participants have accurately compre-
hended the content. The keywords are then used to shape the participants’ summaries.
ARCS and ARCS-W differ from many traditional language therapies because treatment
stimuli (i.e., summarized material) are not repeated and no specific linguistic items are
targeted. Instead, the process of discourse production is the treatment target. This is
accomplished by emphasizing the goal of summarizing with specific lexical items for the
purpose of successful communication of content while self-monitoring the implementa-
tion of constraints.
APHASIOLOGY 103
Obermeyer and Edmonds (2018) reported on three PWmA who improved their written
and spoken discourse at a micro and macrolinguistic level after engaging in ARCS-W.
Microlinguistic level changes were evaluated at the word and utterance level with correct
information units (CIUs; Nicholas & Brookshire, 1993) and complete utterances (CUs;
Edmonds, Nadeau, & Kiran, 2009). CIUs are words that are accurate and relevant to the
stimuli without being repeated, and CUs are utterances that are both accurate/relevant
and contain basic sentence structure. Additionally, the macrolinguistic level changes were
evaluated using measures that captured the extent to which participants were able to
produce the most essential information related to the discourse stimuli.
The three participants reported in Obermeyer and Edmonds (2018) demonstrated
different patterns of discourse improvement, and gains in untrained discourse tasks
were evident in both modalities. While these results were promising, more research on
ARCS-W is needed to determine if these results are replicable and to further refine the
treatment protocol. Based on our previous work, the ARCS-W protocol was modified for
this study by adding a weekly homework activity to promote generalization and by
requiring participants to summarize all material independently from memory without
the assistance of keywords. In the previous version of ARCS-W, participants identified
keywords in segments of text, wrote them down and then used them as a guide while
producing their spoken and written summary. For the current version of ARCS-W, key-
words were still used to plan the summaries but were removed during summarization.
The purpose of this was to emphasize the planning component inherent in summariza-
tion and to provide more opportunities for independent retrieval of lexical items. See
Appendix A for treatment protocol with highlighted changes.
The current study evaluated the effect of ARCS-W in two PWmA within an experimen-
tally controlled pre- to post-treatment design. Additionally, this study sought to refine the
ARCS-W protocol and to replicate previous results (Obermeyer & Edmonds, 2018) by
answering the following questions: To what extent does ARCS-W affect
(1) micro and macrolinguistic written discourse abilities at post-treatment and one-
month post-treatment?
(2) micro and macrolinguistic spoken discourse abilities at post-treatment and one-
(3) other measures of language, including confrontation naming (spoken and written),
sentence production and functional communication at post-treatment and one-
Research questions 1 and 2 were evaluated using the microlinguistic measures of percent
correct information units (%CIUs), percent complete utterances (%CUs) and grammatical
complexity. Macrolinguistic measures included global coherence and participants’ ability to
convey the most important information related to the stimuli, which was measured using %
main events (Capilouto, Wright, & Wagovich, 2005, 2006) and %main ideas. It was hypothe-
sized that participants would demonstrate improvement in written and spoken discourse at
microlinguistic and macrolinguistic levels due to the multi-modality approach of this treat-
ment and our previous findings (Obermeyer & Edmonds, 2018). A wide range of discourse
measures were selected to capture the impairment and potential improvement patterns of
both participants, but we did not anticipate that each participant would improve on all of
these measures. It was also hypothesized that written and spoken confrontation naming
abilities could improve as was observed in P2 (one of the three participants) from
Obermeyer and Edmonds (2018). Sentence production improved within the discourse context
(%CUs) for participants in the first ARCS-W study, so we hypothesized that isolated sentence
production could also improve after ARCS-W treatment. Functional communication by self-
report may improve since ARCS-W addresses discourse production and self-monitoring, which
are essential in everyday communication.
Methods
Research design
An experimentally controlled within-subjects pre- to post-treatment design was used.
This design included four phases: pre-treatment testing, treatment, post-treatment test-
ing, and one-month post-treatment testing. All study procedures were approved by the
Teachers College, Columbia University Institutional Review Board.
Participants
Participants were right-handed, monolingual speakers of English who had mild aphasia,
determined using the WAB-R criteria of an Aphasia Quotient at or above 76 (Kertesz,
2006). Aphasia was due to left hemisphere stroke (>9 months post), and participants
demonstrated the ability to write at the phrase level (i.e., a sequence of two or more words
arranged in a grammatical structure) in at least 50% of written discourse based on the
written picture description in the WAB-R. We did not set a minimum level of total output
(i.e., total number of words) as part of our inclusion criteria, because we were more
concerned with participants’ ability to write at a phrase level. Exclusionary criteria
included history of learning disability, neurogenic disorder/disease other than stroke,
alcohol or drug abuse and depression or other mental health issues. Participants were
given the option to complete assessment and treatment via handwriting or typing on
a qwerty keyboard. Once they selected a modality, all written tasks for assessment and
treatment were completed in that modality. See Table 1 for pre-treatment test scores.
Participant 1 (P1) was a 66-year-old African American male 45-month post left middle
cerebral artery (MCA) infarct. He was diagnosed with mild anomic aphasia. He presented with
agraphia characterized by errors in the lexical (e.g., semantic, lexical) and nonlexical routes (e.g.,
reduced accuracy writing nonwords to dictation) and scored within normal limits on the
Cognitive Linguistic Quick Test (CLQT; Helm-Estabrooks, 2001), which was used as a cognitive
screen. P1 demonstrated a 91% composite score on the Reading Comprehension Battery for
Aphasia-2 (RCBA-2; LaPointe & Horner, 1998), which was important to ensure that P1 could
comprehend the written treatment material. P1 was a retired human resources manager. Prior
to his stroke, he had been active in his church but reported that he no longer attended. He
lived with his wife and ambulated independently. P1 chose to complete assessment and
treatment via handwriting with his nondominant (left) hand. See Appendix B for an example of
pre- and post-treatment discourse.
Participant 2 (P2) was a 49-year-old Asian Pacific American male 48-month post left
posterior parieto-temporal infarct. He was diagnosed with mild anomic aphasia. He
APHASIOLOGY 105
Table 1. Pre-Treatment Language Assessment Results for Both Participants.

Measure Participant
Western Aphasia Battery-Revised P1 P2
Aphasia Quotient 79.3 86.7
Language Quotient 78.3 81.5
Cortical Quotient 81.9 83.46
Raven’s Coloured Progressive Matrices 33/37 28/37
Reading Comprehension Battery for Aphasia
Word-Visual 100% 100%
Word-Auditory 100% 100%
Word-Semantic 100% 100%
Functional Reading 70% 90%
Synonyms 90% 90%
Sentence-picture 100% 100%
Paragraph picture 80% 50%
Paragraph factual 100% 90%
Paragraph inferential 90% 100%
Morpho-syntax 80% 90%
Overall Score 91% 91%
Cognitive Linguistic Quick Test
Attention WNL Moderate
Memory Mild Moderate
Language WNL Mild
Executive Function Mild WNL
Visuospatial WNL Mild
Composite WNL Mild
Note. WNL = Within normal limits.
presented with agraphia characterized by impairment in the lexical (e.g., semantic errors
and neologisms) and the nonlexical route (e.g., reduced accuracy writing nonwords to
dictation) and had a mild composite score on the CLQT (Helm-Estabrooks, 2001). Like P1,
P2 scored 91% on the composite score of the RCBA-2 (LaPointe & Horner, 1998), indicating
that he would be able to comprehend the treatment material. P2 worked as a stockbroker
prior to his stroke and had not been able to return to work. He lived with family. He selected
the typing modality for assessment and treatment, for which he was able to use both hands.
See Appendix C for an example of pre- and post-treatment discourse.
Treatment dosage
A priori intended dosage included 6 pre-treatment testing visits, 24 treatment visits (2–3
times a week), 4 post-treatment testing visits and 4 follow-up testing visits one-month
after treatment. Testing and treatment visits lasted approximately 1 h and 30 min each.
P1 received the intended dosage. P2 participated in fewer treatment and assessment
sessions due to holiday traveling. In total, P2 attended 6 pre-treatment assessment visits,
21 treatment visits, 5 post-treatment assessment visits and no one-month post-treatment
testing due to extended travel.
Discourse assessments
For all discourse tasks described below, half of the stimuli were completed in the spoken
modality and half in the written modality at pre-/post-treatment and one-month post-
treatment testing.
Story retell task

Stimuli from the Discourse Comprehension Test (DCT; Brookshire & Nicholas, 1997) were
used as a measure of story retelling (Doyle et al., 1998). The DCT includes 12 pre-recorded
stories, but only 8 were administered to ease the burden of testing. Participants listened
to the stories, answered yes/no questions, heard the stories again and then retold half the
stories via speaking and half via writing.
Nicholas and Brookshire task

Additional discourse included the Nicholas and Brookshire (Nicholas & Brookshire, 1993)
stimuli and protocol with half of the discourse prompts completed verbally (Set B) and
half in writing (Set A). The single picture descriptions, six-panel picture descriptions and
two requests for personal information were included, but procedural discourse was not
due to the substantial differences between procedural discourse and other discourse
types regarding information content, communicative purpose, and structure (Ulatowska,
Allard, & Chapman, 1990).
Article summarization task

This task required summarization of abridged news articles (from websites such as
http://newsela.com) that were between 200 and 250 words and were written at the
sixth-grade level based on the Flesch Kincaid scale (Kincaid, Fishburne, Rogers, &
Chissom, 1975) (See Obermeyer & Edmonds, 2018 for more detailed description).
During the article summarization task, the clinician instructed the participant to read
the article and then write a summary of what was read without time constraints.
Participants completed four article summarizations at each of the following time points:
pre-treatment assessment, post-treatment and one-month post-treatment for a total of
three administrations of four article summarizations.
Control task
Nonword writing to dictation of 34 items from the Johns Hopkins Dyslexia and Dysgraphia
Battery (Goodman & Caramazza, 1986) was administered at the same time points as the
article summarization tasks to establish experimental control. At pre-treatment, the control
task was administered four times on separate visits over two weeks, and both participants
demonstrated stable baselines. P1’s nonword writing accuracy was 26%, 20%, 26%, and
26% across the four administrations, and P2’s nonword writing accuracy was 5%, 2%, 8%,
and 8%. P2 demonstrated a change of less than 10 percentage points across each admin-
istration, and his performance did not change from the 3rd to 4th administration.
Non-discourse assessments
Object and Action Naming Battery (OANB; Druks & Masterson, 2000)
The OANB assessed participants’ confrontation naming (half written, half spoken) of
nouns and verbs following Furnas and Edmonds (2014).
APHASIOLOGY 107
The Sentence Production Test (SPT; Wilshire, Lukkien, & Burmester, 2014)
The SPT evaluated spoken and written sentence descriptions of pictured events that
ranged in grammatical and lexical complexity. The SPT included a total of 20 pictures
and each participant described the picture first via speaking and then writing.
Functional communication
The Communicative Effectiveness Index (CETI; Lomas et al., 1989) was used to assess
functional communication via communication partner report and the Aphasia
Communication Outcome Measure (ACOM; Hula et al., 2015) was used to assess func-
tional communication via self-report. The CETI requires the communication partner to
rate the person with aphasia’s functional communication across 16 communication
scenarios of varying difficulty. Contrary to Lomas et al. (1989), we did not show
participants their pre-treatment ratings at post-treatment (see Obermeyer & Edmonds,
2018 for more detail). The ACOM (Hula et al., 2015) requires participants to rate their
own communication ability on a visual scale of 59 functional scenarios (e.g., reading
nutrition information on food labels) involving verbal expression, comprehension and
writing (Doyle & Hula, 2012). Based on their responses, a t-score was generated that
allowed their pre-treatment and post-treatment responses to be compared to the mean
of 329 people with aphasia.
Data recording
Each assessment and treatment session was audio and video recorded for the purposes of
data collection and treatment fidelity.
Transcription
Trained research assistants (RAs) or the first author completed orthographic transcriptions
using audio and video recordings of discourse sampling. RAs were M.S. students majoring
in speech language pathology. For transcription purposes, utterances were divided into
C-units, which include a main clause with any subordinate clauses (Loban, 1976).
Transcription instruction with RAs included C-unit segmentation training in which three
discourse transcripts had to be divided into C-units with greater than 90% accuracy. RAs
then had to achieve point-by-point reliability greater than 90% on at least two training
transcripts. Point-to-point transcription reliability was conducted on words, pauses and
utterance breaks. To calculate the inter-rater reliability, the total number of agreements
was divided by the total possible. P1 reliability was completed for 40% of transcripts with
93.33% reliability. P2 reliability was completed for 20% of transcripts with 92.82% relia-
bility. In addition, C-unit division was evaluated separately on 30% of transcripts: for P1,
reliability was 91.0% and for P2, reliability was 88.7%. Transcription disagreements were
resolved by referring to the original media recordings to reach a consensus on the most
accurate transcription.
Discourse analysis
Microlinguistic measures
Percent correct information units (%CIUs) are words that are intelligible in context, accurate
in relation to the stimuli or topic and relevant to, and informative about the content of the
picture/story/topic (Nicholas & Brookshire, 1993).
Complete Utterances (%CU) contain subject + verb + (object) structure and are also
relevant to the topic (Edmonds et al., 2009). Grammatical, morphological, and phonemic
errors are not penalized. For example, if a participant said Small 1-bedroom apartment to
describe where he lives, it would be relevant but does not include basic sentence
structure, and so it is not a CU, whereas I live in a small 1-bedroom apartment contains
an SVO structure and is relevant.
Grammatical Complexity was rated using Altman, Goral, and Levy (2012) scale defined
as: 1 = incomplete sentence, 2 = simple and complete (contains all required elements), 3 =
complex structure with incomplete subordinate or coordinate clause, 4 = complete
coordinate structure, 5 = complete containing a subordinate structure. Since utterances
were defined as C-units, an utterance with complete coordinate structures was divided
into two utterances, in which case, the first clause was not coded and the second would
be coded as a 4 (e.g., complete coordinate structure). For example, (1) The boy is running
with his kite (2) but the dog is about to catch him, would be divided into two C-units and
receive a score of 4. Utterances combined with the coordinating conjunction “and” were
coded individually. For example, the following utterances would both be scored a 2, (1)
The boy was flying a kite (2) and he is having fun, due to the over reliance of “and” as a filler
between utterances.
Macrolinguistic measures
The percent of main events (%ME), which are defined as the events that contain all
essential information related to a stimuli, in addition to causal relationships and can
contain one or more verbs (Capilouto et al., 2005, 2006), were used to evaluate the
proportion of the most important information that participants were able to convey.
Main Events were evaluated for selected N&B stimuli using the identified events published
by Capilouto et al. (2005; 2006) including: cat in tree, birthday, argument, and directions.
For the article summarization task, main events were selected using a protocol similar to
Capilouto et al. (2005; 2006) in which three speech language pathologists read the
summarization article and then identified the most important events in the text, which
could contain one or more verbs. The agreed upon items were used as main events. There
was a total of six main events in each summarization article. For example, in an article
discussing the migration of monarch butterflies, one of the identified main events was
“New construction/building is destroying the butterflies’ habitat”. An example of an event
with more than one verb and a causal relationship is “If the butterflies’ habitat is
destroyed, they/the butterflies could be gone forever”.
Main ideas (MIs) were evaluated for the Discourse Comprehension Test (DCT) story
retelling stimuli, and were extrapolated from the comprehension questions evaluating
main ideas in the DCT. Because the predetermined information from the story retell stimuli
were not identified in a manner similar to the main events from the N&B stimuli and the
article summarization task, we refer to them as MIs in keeping with how they are referred to
APHASIOLOGY 109
in the DCT question protocol. However, the purpose of this measure is similar to that of the
main events – to capture if participants were able to convey the most important informa-
tion related to the story. The DCT includes four main idea questions, two overt and two
implied. Those four questions were used to create main ideas for the DCT stimuli. For
example, in the story about a man going to a baseball game, one of the overt main idea
questions from the DCT is “Did George go to a baseball game?”. Thus, the main idea
developed for the recall task was “George went to a baseball game”. In the case of implied
main ideas, the implied information, or the overt information it was extrapolated from, was
considered correct. For example, in a story about two cousins with a painting business, the
implied main idea that “Fred and Ben were painting the wrong house” was accepted as
complete. However, if the participant said the overt information instead, for example, “Fred
and Ben were painting the neighbor’s house/The Nelson’s house instead of the Foster’s
house,” this too was accepted as complete.
To capture the proportion of information participants were able to convey, MEs and
MIs were scored as complete (score of 1) if they contained all predetermined information
or incomplete (score of .5) if only part of the information was present or correct. For
example, one of the events identified by Capilouto et al. (2005; 2006) for the birthday
picture is “The dog is hiding under the sofa”. If a participant said “the dog is hiding”, this
utterance would be scored a .5, because it does not contain all the required information.
Similarly, if an utterance contained a single word-finding error it would be scored a .5, as
long as the remaining information was correct. For example, both of the following
utterances would be scored as a .5, “The dog is hiding under the chair” or the “The dog is
hiding under there”. The following utterance, which has two errors (e.g., lexical retrieval
error and underspecified information), would not be given any credit, “The boy is hiding
under there”. This represents a divergence from the scoring methods used by Capilouto
et al. (2005; 2006), who evaluated the percentage of MEs without incomplete scores.
Additionally, if participants could not convey the complete ME or MI in a single utterance,
but were able to convey it over two consecutive utterances, each utterance would be
scored as .5 for a total score of 1 (i.e., complete). For example, if a participant produced the
following, “The dog is under the sofa. He is hiding from the woman”, each utterance
would be scored a .5 for a total score of 1 (complete) for this main event (i.e., the dog is
hiding under the sofa). This scoring criteria was created so that it could be used across the
three types of discourse stimuli and to capture potential increases in the proportion of
important information the participants were able to convey.
Main Events represent different degrees and amounts of information that can result in
them being characterized as micro level, macro level or a combination of the two
structural levels (see Capilouto et al., 2006; Richardson & Dalton, 2016). The primary
purpose of evaluating the main events and ideas utilized in the study was to determine
if participants were able to produce a higher proportion of the information deemed
essential to the stimuli. As such, changes in %ME and %MI were considered a change in
macrolinguistic discourse structure. However, it is important to recognize the intercon-
nection between word level elements and the ability to convey this information
(Richardson & Dalton, 2016). Additionally, the MEs and the MIs used for the different
stimuli in this study do not represent a one-to-one relationship and are reported sepa-
rately for each set of stimuli.
Global coherence was evaluated using a four-point scale (Wright & Capilouto, 2012;
Wright, Capilouto, & Koutsoftas, 2013; Wright, Koutsoftas, Capilouto, & Fergadiotis, 2014).
Each utterance was scored based on how related it was to the discourse topic, with 1
indicating an off-topic utterance and 4 an on-topic utterance. For example, if a participant
was describing where they live and said “I live in a house in New York”, it would be rated
a 4 (on topic) and the following utterance, “The place I live is busy”, would be rated a 3 for
being related to the topic but missing information that must be inferred (i.e., not clear if
house or neighborhood is busy). “I used to give lessons in the house” would be rated a 2
for being egocentric but somewhat related to the overall topic, and the utterance “I gave
lessons at various levels” would be rated a 1 (off topic). An overall global coherence score
was calculated by adding the scores for each utterance and dividing it by the total
number of utterances in the discourse sample.
Discourse coding
Trained research assistants, blinded to time of testing, completed discourse coding for
each discourse measure. Research assistants were M.S. students in speech language
pathology. Prior to coding independently, each research assistant had to achieve greater
than 90% reliability on each coding measure on at least two practice transcripts. Point-by-
point inter-rater reliability was completed by the first author for each discourse measure
on 40% of the transcripts. Reliability for P1 was 99.49% for words, 90.21% for CIUs, 88.96%
for CUs, 88.26% for grammatical complexity, 87.56% for global coherence and 91.24% for
MEs and MIs. Reliability for P2 was 98.77% for words, 92.06% for CIUs, 88.60% for CUs,
90.01% for grammatical complexity, 90.01% for global coherence and 90.32% for MEs and
MIs. Coding discrepancies were discussed between the original coder and the reliability
coder. If an agreement could not be reached, a third trained coder, who was working on
the project and met the previously described training criteria, was asked to resolve the
issue.
Therapy protocol
News articles used for treatment were obtained from current event websites such as
www.newsela.com and were adapted to a sixth-grade reading level. To promote engage-
ment, each participant’s interests were taken into consideration for selection of article
topics. At the beginning of each session, participants were provided with two to three
article choices. During the first treatment step, the clinician read the entire news story
aloud while the participant followed the text. Participants then read a segment (one-three
sentences) twice with the intent to summarize and identified keywords in the segment.
Once keywords were identified, the participant wrote/typed the keywords and compared
them with the keywords that the clinician had identified. If there was a disagreement
between the clinician and participant, it was resolved through discussion. The participant
was then given an opportunity to look at their keywords and plan a verbal summary of the
material. Planning was completed verbally or silently without feedback from the clinician.
Once ready, the keywords were covered/removed, and the participant produced a verbal
summary from memory of the segment. The same step was completed (look at keywords,
plan, cover keywords, summarize from memory) in writing. During summarization, the
APHASIOLOGY 111
participant was prompted to follow prescribed constraints 1) avoid nonspecific words

(e.g., pronouns), 2) stay on topic and 3) use complete sentences. These steps were
repeated until the participant reached the end of the article. Then the participant reread
and summarized the article from memory in its entirety verbally and then in writing and
rated the completeness of their summary on a scale of 1–5 (e.g., 1 being “not at all
complete” and 5 representing “entirely complete”). The purpose of this step was for the
participant to focus on the content of their discourse in relation to the article. See
Appendix A for complete protocol.
Homework activity
The weekly homework assignment required participants to read an assigned article,
identify the keywords and then write a summary implementing the same constraints
used in the therapy sessions (e.g., use specific words, stay on topic, use complete
sentences). During homework, participants could refer to the written text (which they
could not do during therapy). This accommodation was made to increase the likelihood
that participants would write correct words and demonstrate better understanding of the
text while completing their summaries independently. During the first therapy session of
the week, the clinician and participant reviewed the homework summary together, and
the clinician provided feedback about whether the participant followed their constraints
and the completeness of the summary.
Treatment fidelity
A trained research assistant completed treatment fidelity by watching video recordings of
50% of treatment sessions and identifying completed treatment steps on a checklist.
Treatment fidelity was 97% for P1 and 98% for P2. See supplemental material for a copy of
the treatment fidelity checklist.
Data analysis of discourse assessments

Story retell and N&B tasks
There is not an established threshold for determining change in discourse measures
reported in the aphasia literature, largely due to differences in populations, stimuli, and
methodologies. For the current study, we selected an a priori benchmark of change for
discourse measures from N&B and story retell, except global coherence, as an increase of
10 percentage points or greater. This threshold was selected because it is consistent with
previous treatment literature (e.g., Altman et al., 2012; Edmonds et al., 2009; Peach &
Reuter, 2010), exceeds the 5.5% change in %CIUs that Brookshire and Nicholas (1994)
reported as possible due to repeated exposure to their discourse elicitation stimuli, and
because the participants had mild aphasia and relatively high pre-treatment abilities.
Global coherence was interpreted using data from Wright et al. (2014), who reported
a standard deviation (SD) of 0.22 based on the single picture N&B stimuli and 0.18 for
a storytelling task. Therefore, an increase greater than or equal to twice the SD was
considered a change. Using this benchmark an increase of 0.44 for N&B and 0.36 for the
story retelling task was considered improvement.
Article summarization task

For the article summaries and nonword writing to dictation control tasks, percentage
accuracy and effect sizes are reported. Effect sizes were calculated using a variation of
Cohen’s d reported by Beeson and Robey (2006) and Busk and Serlin (1992), which involves
averaging performance for each variable at pre-treatment (four different article summariza-
tions), post-treatment (the same four article summarizations completed at pre-treatment)
and one-month post-treatment (the same four article summarizations completed at pre- and
post-treatment). Effect sizes were calculated from pre-to post-treatment using the following
formula: Meanposttreatment-Meanpretreatment/SDpretreatment, and from pre- to the one-month
follow-up using the following formula: Meanone-month follow-up-Meanpre-treatment/SDpre-treatment.
Effect sizes were interpreted using the benchmarks (2.6 = small, 3.9 = medium, 5.8 = large)
reported by Robey, Schultz, Crawford, and Sinner (1999), which have been applied to written
outcomes at the single word level (Beeson & Robey, 2006). To our knowledge, there are no
effect size benchmarks for discourse-level spoken or written language. The published
benchmarks are designed to interpret results from single subject design research, which
the current study is not. Additionally, they are not meant to evaluate generalization mea-
sures or untrained stimuli such as the tasks used in this study (e.g., Beeson & Robey, 2006;
Kendall et al., 2008); therefore, effect sizes and the a priori benchmarks (i.e., change of
10 percentage points, or 2 SDs) are used to interpret change in the article summarization
tasks.
Data analysis for non-discourse assessments

We used the following criteria as thresholds of change for our remaining measures: Greater
than a two standard deviation increase on the O&A (standard deviation = 2.72 for nouns,
4.10 for verbs based on errors produced by healthy adults aged 71–80; Druks & Masterson,
2000), and one standard deviation for the ACOM (change of 10 points on the T-score).
Communication partner responses on the CETI were compared from pre- to post-treatment
and pre- to one-month post-treatment using a Wilcoxon Rank Test. Lastly, performance on
the SPT (percentage of sentences produced correctly) was compared from pre- to post-
treatment and pre- to one-month post-treatment using a nonparametric McNemar’s test to
evaluate the difference in proportions correct at each testing point.
Results
P1: research question 1: effect of ARCS-W on written discourse
Written N&B task (Table 2)
Both macrolinguistic measures showed improvement based on our threshold. Global
coherence went from 2.77 to 3.34, which was maintained at 3.31 one-month after
treatment. The %MEs was 41.67% at pre-treatment, 66.67% at post-treatment and
79.17% at one-month post-treatment. Microlinguistic measures did not improve.
Written story retell task (Table 2)

Percent of grammatically complex utterances did not meet the threshold for improve-
ment from pre- to post-treatment but did at one-month post-treatment (28.57%-42.86%).
APHASIOLOGY 113
Table 2. P1 Written Discourse Assessments from Pre- to Post-Treatment and at 1-month Post-Treatment.
Task Written N&B (1993) Pre-tx Post-tx Pre-to Post Change 1-month Post Pre-to 1-month Post Change
Total Words 166 253 +87 318 +152
%CIUs 84.94 88.54 +3.6 76.42 −8.52
CIUs/Min 7.18 8.14 +0.96 6.63 −0.55
Total Utterances 25 40 +15 40 +15
%Complete Utterances 80.0 77.5 −2.5 75.0 −5.0
%GramComplex 20.0 8.0 −12.0 14.0 −6.0
%Main Events 41.67 66.67 +25.0 79.17 +37.5
Global Coherence 2.77 3.34 +0.57 3.31 +0.54
Story Retelling Pre-tx Post-tx Pre-to Post Change 1-month Post Pre-to 1-month Post Change
Total Words 314 346 +32 418 +104
%CIUs 83.12 80.64 −2.48 80.19 −2.93
CIUs/Min 7.18 7.53 +0.35 7.92 +0.74
Total Utterances 46 49 +3 48 +2
%Complete Utterances 67.39 73.47 +6.08 75.0 +7.61
%GramComplex 28.57 30.95 +2.38 42.86 +14.29
%Main Ideas 50.0 56.25 +6.25 68.75 +18.5
Global Coherence 3.45 3.49 +0.04 3.4 −0.05
Note. Pre-tx = pre-treatment, post-tx = post-treatment, 1-month post = one month post-treatment; N&B = Nicholas and
Brookshire (1993); %CIUs = percent of correct information units (Nicholas & Brookshire, 1993), %GramComplex = percent
of grammatically complex utterances (Altman et al., 2012).
Change of 10 percentage points from pre- to post-treatment are in boldface, except for global coherence in which
a change of two standard deviations (.44 for N&B, .36 for story retelling) or more are in boldface (Wright et al., 2014).
The %MIs present did not meet the improvement benchmark from pre-treatment to post-
treatment but did show an improvement from pre-treatment (50.0%) to one-month post-
treatment (68.75%).
Written article summarization task (Table 3)

Here the averages of each discourse measure for the four individual article summaries
completed during pre-treatment, re-administered at post-treatment and then again one-
month post-treatment are reported. Pre- to post-treatment change in %CIUs were not
significant using the percentage threshold or effect size calculation (d = .99). At one-month
post-treatment, %CIUs were 62.93% (pre-treatment = 50.97%), which met our percentage
change criteria (10 percentage points) but did not produce an effect (d = 1.28). The change in
%CUs did not meet improvement criteria from pre-to post-treatment (d = −0.07), but did
improve from pre-treatment to one-month post-treatment (31.55%-53.48%) with a change of
21.93 percentage points that approached a small effect (d = 2.58) (Robey et al., 1999). No
change was observed in the percent of grammatically complex sentences, %ME, or global
coherence.
Control task (Table 3)

P1’s performance on the nonword writing to dictation control task did not improve from
pre- to post-treatment (d = .50), or pre- to one-month post-treatment (d = .25) based on
his effect sizes and the percentage change criteria.
Table 3. P1 Written Article Summarization Tasks and Control Task at Pre-, Post-Treatment and
1-month Post-Treatment Assessment Periods.
Microlinguistic Macrolinguistic Control
% accurate nonwords
Probe administration %CIU %CU %GramCom TW TU %ME Average GC (N = 34)
Pre-tx 1 41.38 33.33 0 29 6 0 2.33 26.47
Pre-tx 2 47.37 25 0 38 8 14.29 2.25 20.59
Pre-tx 3 63.41 42.86 0 41 7 16.67 3 26.47
Pre-tx 4 51.72 25 0 29 4 7.14 2.25 26.47
Average 50.97 31.55 0 34.25 6.25 9.53 2.46 25
St Dev. 9.31 8.5 0 6.18 1.71 7.53 0.36 2.94
Post-tx 1 63.63 33.33 16.67 33 6 8.33 3.17 29.41
Post-tx 2 56.14 16.67 16.67 57 6 14.29 2.67 20.59
Post-tx 3 62.86 50 0 35 4 8.33 2.5 23.53
Post-tx 4 58.33 28.57 0 60 7 7.14 2.43 20.59
Average 60.24 32.14 8.34 27.5 5.75 9.52 2.69 23.53
St Dev. 3.6 13.81 9.62 7.05 1.26 3.23 0.33 4.16
1-month post 1 38.78 12.5 0 49 8 0 1.5 26.47
1-month post 2 80 71.43 0 60 7 35.17 3 20.59
1-month post 3 74.29 80 0 35 5 25 3.2 23.53
1-month post 4 59.02 50 0 61 8 7.14 2.5 32.35
Average 62.93 53.48 0 51.25 7 16.83 2.55 25.74
St Dev. 18.4 30.09 0 12.12 1.41 16.13 0.76 5.02
Note. %CIUs = percent correct information units; %CUs = percent complete utterances; %GramCom = percent grammatically
complex utterances; TW = total words; TU = total utterances; %ME = percent main events; Average GC = average global
coherence; pre-tx = pre-treatment; post-tx = post-treatment; 1-month post = one month post-treatment; st dev. = standard
deviation.
Table 4. P1 Spoken Discourse Assessments from Pre- to Post-Treatment and 1-month Post-Treatment
Assessment.
Task Spoken N&B (1993) Pre-tx Post-tx Pre-to Post Change 1-month Post Pre-to 1-month Post Change
Total Words 744 1100 +356 698 −46
%CIUs 46.24 50.45 +4.21 48.71 +2.47
CIUs/Min 18.11 20.19 +2.08 22.62 +4.51
Total Utterances 66 111 +45 51 −15
%GramComplex 5.88 8.91 +3.03 22.22 +16.34
%Main Events 45.83 41.67 −4.16 66.67 +20.84
Global Coherence 2.92 2.16 −0.76 2.98 +0.06
Task Spoken Story Retell Pre-tx Post-tx Pre-to Post Change 1-month Post Pre-to 1-month Post Change
Total Words 602 665 +63 644 +42
%CIUs 46.01 60.75 +14.74 52.95 +6.94
CIUs/Min 18.32 26.93 +8.61 20.0 +1.68
Total Utterances 68 70 +2 62 −6
%GramComplex 22.95 29.31 +6.36 42.31 +19.36
%Main Ideas 53.13 65.63 +12.5 65.63 +12.5
Global Coherence 2.88 3.25 +0.37 3.15 +0.27
Note. Pre-tx = pre-treatment, post-tx = post-treatment, 1-month post = one-month post-treatment; N&B = Nicholas and
Brookshire (1993); %CIUs = percent of correct information units (Nicholas & Brookshire, 1993), %GramComplex = percent of
grammatically complex utterances (Altman et al., 2012).
APHASIOLOGY 115
P1: research question 2: effect of ARCS-W on spoken discourse (Table 4)

Spoken N&B task
The change in percent of grammatically complex utterances did not meet the improvement
threshold from pre- to post-treatment but did at one-month post-treatment (5.88%-22.22%).
The %MEs did not meet the 10-percentage point increase threshold from pre- to
post-treatment but did when comparing pre-treatment to one-month post-treatment
(45.83%-66.67%).
Spoken story retell task

Relevant words met the 10-percentage point threshold from pre-treatment (%CIUs = 46.01%)
to post-treatment (%CIUs = 60.75%), but this increase was not maintained one-month post.
A similar pattern was observed with %CUs (pre-treatment = 52.94%; post-treatment = 64.29%;
one-month post-treatment = 59.68%). The percent of grammatically complex utterances did
not meet the improvement criteria when comparing from pre-to post-treatment but did when
comparing pre-treatment performance to one-month after treatment (22.95%-42.31%).
Global coherence met the threshold of improvement from pre-treatment to post-
treatment (2.88–3.25) but was not maintained. P1 conveyed a higher %MIs at post-
treatment (65.63%) than at pre-treatment (53.13%), which was maintained (65.63%) and
met the 10 percentage point criteria.
P1: research question 3: effect of ARCS-W on non-discourse assessments (Table 5)

P1 demonstrated higher accuracy on confrontation naming of written (from pre-
treatment to one-month post-treatment) and spoken nouns that met the predetermined
threshold for improvement (greater than 2 SD change, 1 SD = 2.72). One-month post-
treatment, P1’s spoken confrontation naming accuracy was within normal limits (Druks &
Masterson, 2000). P1’s self-rated functional communication (ACOM; Hula et al., 2015)
t-score was 56.57 at pre-treatment and 69.76 at post-treatment. This change met the
criteria for improvement (1 SD, 10 points) but was not maintained one-month post-
treatment (t-score = 60.6).
Table 5. Non-discourse Assessments for Participant 1 and Participant 2.

Measure P1 P2
O&A Pre-tx Post-tx 1-month post Pre-tx Post-tx
Written nouns 62/81 (76.5%) 67/81 (82.7%) 74/81 (91.4%) 59/81 (72.8%) 62/81 (76.5%)
Spoken nouns 72/81 (88.9%) 78/81 (96.3%) 80/81 (98.8%) 76/81 (93.8%) 74/81 (91.4%)
Written verbs 39/50 (78.0%) 41/50 (82.0%) 44/50 (88.0%) 38/50 (76.0%) 42/50 (84.0%)
Spoken verbs 48/50 (96.0%) 45/50 (90.0%) 48/50 (96.0%) 43/50 (86.0%) 45/50 (90.0%)
Sentence Production Test Pre-tx Post-tx 1-month post Pre-tx Post-tx
Written 12/20 (60.0%) 10/20 (50.0%) 14/20 (70.0%) 14/20 (70.0%) 15/0 (75.0%)
Spoken 13/20 (65.0%) 12/20 (60.0%) 15/20 (75.0%) 11/20 (55.0%) 12/20 (60.0%)
Functional Communication Pre-tx Post-tx 1-month post Pre-tx Post-tx
CETI 81.25 80.28 NT 85.6 87.5
ACOM 56.57 69.76 60.6 57.48 62.15
Note. O&A = Object and Action Naming Battery (Druks & Masterson, 2000); CETI = Communicative Effectiveness Index
(Lomas et al., 1989); ACOM = Aphasia Communication Outcome Measure (Hula et al., 2015); Pre-tx = pre-treatment,
post-tx = post-treatment, 1-month post = one-month post-treatment.
Table 6. P2 Written Discourse Assessments from Pre-to Post-Treatment.

Task Written N&B (1993) Pre-tx Post-tx Pre-to Post Change
Total Words 107 122 +15
%Correct Information Units 84.11 86.07 +1.96
CIUs/Min 3.29 2.73 −0.56
Total Utterances 15 17 +2
%Complete Utterances 66.67 76.47 +9.80
%GramComplex 0.00 13.33 13.33
%Main Events 37.50 54.17 +16.67
Global Coherence 3.27 3.75 +0.48
Task Written Story Retell Pre-tx Post-tx Pre-to Post Change
Total Words 123 171 +48
CIUs/Min 2.54 2.26 −0.28
Total Utterances 14 17 +3
%GramComplex 0.08 0.06 −0.02
%Main Ideas 34.38 40.63 +6.25
Note. Pre-tx = pre-treatment, post-tx = post-treatment; N&B =Nicholas and Brookshire (1993); %CIUs = percent of correct
information units (Nicholas & Brookshire, 1993), %GramComplex = percent of grammatically complex utterances (Altman
et al., 2012).
P2: research question 1: effect of ARCS-W on written discourse

Written N&B task (Table 6)
At pre-treatment P2 did not produce any grammatically complex utterances, and he
produced 13.33% at post-treatment. This change met the 10-percentage point threshold
for improvement.
Both macrolinguistic measures met the criteria for improvement with average global
coherence at 3.27 pre-treatment and 3.75 at post-treatment, and the %MEs were at
37.50% during pre-treatment and 54.17% at post-treatment.
Written story retell task (Table 6)

P2’s production of relevant words (%CIUs) was 74.80% at pre-treatment and 84.80% at
post-treatment, which met the 10-percentage point benchmark for improvement.
The change in average global coherence from pre- to post-treatment (3.08–3.65), met
the threshold for improvement.
Written article summarization task (Table 7)

The change in relevant words (%CIUs) met our percentage change threshold for improve-
ment with pre-treatment %CIUs at 58.18% and post-treatment %CIUs at 69.23%. However,
there was no effect based on the calculated effect size (d= 0.80) due to high variability
during pre-treatment assessment (SD = 13.84). A similar pattern was observed for %CUs
which were at 16.67% pre-treatment, and 54.17% at post-treatment with an effect size of
d = 1.12 due to high variability at pre-treatment (SD = 33.34).
Global coherence was 2.45 at pre-treatment and 3.27 at post-treatment, which was
a change greater than 2 SDs; however, the corresponding effect size of d = 1.95 did not
support a treatment effect. No change was observed on the remaining discourse
APHASIOLOGY 117
Table 7. P2 Written Article Summarization Tasks and Control Tasks at Pre- and Post-Treatment
Assessment Periods.
Microlinguistic Macrolinguistic Control
% accurate nonwords
Probe administration %CIUs %CUs %GramCom TW TU %ME Average GC (N = 34)
Pre-tx 1 77.78 66.7 33.3 28 3 7.14 3.0 5.88
Pre-tx 2 57.89 0.0 0.0 19 3 8.33 2.0 2.94
Pre-tx 3 50.0 0.0 0.0 18 2 0.0 2.5 8.82
Pre-tx 4 47.06 0.0 0.0 17 3 0.0 2.33 8.82
Average 58.18 16.67 8.33 20.5 2.75 3.87 2.45 6.62
St Dev. 13.84 33.34 16.66 5.07 .50 4.49 .42 2.81
Post-tx 1 87.04 66.7 20.0 54 6 35.17 3.2 2.94
Post-tx 2 61.54 100.0 0.0 26 2 16.67 4.0 5.88
Post-tx 3 35.0 0.0 0.0 20 2 0.0 2.5 14.71
Post-tx 4 93.33 50.0 0.0 15 2 0.0 3.5 8.82
Average 69.23 54.17 5.0 28.75 3.0 12.96 3.27 8.09
St Dev. 26.64 41.67 10.0 17.42 2.0 16.76 .63 5.03
Note. %CIUs = percent correct information units; %CUs = percent complete utterances; %GramCom = percent gramma-
tically complex utterances; TW = total words; TU = total utterances; %ME = percent main events; Average GC = average
global coherence; pre-tx = pre-treatment; post-tx = post-treatment; st dev. = standard deviation
measures (i.e., percent of grammatically complex utterances and main events) based on
the percentage change and effect size criteria.
Control task (Table 7)

P2’s performance on the nonword writing to dictation control task did not improve from
pre- to post-treatment (d = .52), based on his effect size and the percentage change criteria.
Table 8. P2 Spoken Discourse Assessments from Pre- to Post-Treatment.

Task Spoken N&B (1993) Pre-tx Post-tx Pre-to Post Change
Total Words 1039 552 −487
CIUs/Min 28.53 30.42 +1.89
Total Utterances 89 50 −39
%GramComplex 23.94 29.27 +5.33
%Main Events 54.17 50.0 −4.17
Task Spoken Story Retell Pre-tx Post-tx Pre-to Post Change
Total Words 654 471 −183
CIUs/Min 25.55 34.82 +9.27
Total Utterances 48 35 −13
%GramComplex 35.90 42.86 +6.96
%Main Ideas 62.50 81.25 +18.75
Note. Pre-tx = pre-treatment, post-tx = post-treatment; N&B =Nicholas and Brookshire (1993); %CIUs = percent of correct
information units (Nicholas & Brookshire, 1993), %GramComplex = percent of grammatically complex utterances (Altman
et al., 2012).
P2: research question 2: effect of ARCS-W on spoken discourse (Table 8)

Spoken N&B task
P2 produced a higher percentage of relevant words at post-treatment (61.78%) compared
to pre-treatment (42.44%), which met our 10-percentage point threshold for improvement.
Of the macrolinguistic measures, global coherence met the criteria for change with an
increase from 2.83 to 3.30.
Spoken story retell task

Percent CIUs were 42.97% at pre-treatment and 62.85% at post-treatment, which met the
threshold for improvement. Percent CUs also increased (52.08–65.71%) and met the
criteria for improvement.
P2 demonstrated change on both macrolinguistic measures that met the threshold for
improvement with global coherence increasing from 2.76 to 3.27, and %MIs increasing
from 62.50% at pre-treatment to 81.25% at post-treatment.
P2: research question 3: effect of ARCS-W non-discourse assessments (Table 5)

No change meeting the predetermined criteria for improvement was noted on confronta-
tion naming, sentence production or partner/self-rated functional communication.
Discussion
This study sought to refine the ARCS-W protocol and replicate the results observed in study
1 (Obermeyer & Edmonds, 2018) with two additional PWmA. Each participant’s response to
treatment is discussed below including patterns that were observed across participants. This
is followed by a discussion of the clinical implications and limitations of this work.
Before treatment, P1 demonstrated reduced lexical retrieval in naming tasks and in
spoken discourse output, which was effortful. Lexical retrieval in written discourse was
more accurate prior to treatment, as indicated by higher percent correct information units,
than spoken discourse, but output was sparse, especially in the N&B tasks. After treatment,
P1 demonstrated an increase that met the threshold for improvement in his lexical retrieval
of objects in speaking and writing. P1 produced a higher percentage of main events on 2/3
tasks and main ideas on 2/2 discourse tasks, indicating that P1 was producing a higher
percentage of important information related to the stimuli. Additionally, P1 produced more
grammatically complex utterances in 3/5 discourse tasks across modalities based on the
improvement criteria. See Appendix B for an example of pre- and post-treatment discourse.
At pre-treatment, P2 presented with reduced lexical retrieval and difficulty maintaining
topic (global coherence) in spoken and written discourse. Additionally, written discourse
was extremely effortful and time intensive due to his difficulty spelling. As a result, P2
demonstrated a large discrepancy between his written and spoken discourse production
(See Tables 6 and 8). However, his pattern of improvement was similar in both modalities,
with the most substantial and consistent increases observed in %CIUs (4/5 tasks) and
global coherence (5/5 tasks) across discourse types and modalities.
Measures of word level content and global coherence have been shown to be highly
inter-related (Marini & Urgesi, 2012; Wright et al., 2014) in discourse. P2’s pre-treatment
APHASIOLOGY 119
spoken discourse was characterized by a large number of words and difficulty maintaining
topic (i.e., global coherence) which influenced his %CIUs, since many words were consid-
ered irrelevant (i.e., not a CIU; Nicholas & Brookshire, 1993). At post-treatment, P2 had better
topic maintenance based on our benchmark of improvement, and this resulted in fewer
irrelevant words. Additionally, P2 demonstrated more lexical specificity, which resulted in
higher %CIUs (change of 10 percentage points or greater) in 2/2 spoken discourse tasks.
P2’s written discourse performance was characterized by fewer total words across all writing
tasks and considerable difficulty spelling. Combined, these factors made writing time-
consuming and laborious. Similar to spoken discourse, P2 demonstrated the most consistent
improvements (based on the threshold for improvement) in %CIUs and global coherence. In
written discourse, these changes appeared to be a result of more successful lexical retrieval,
which was evidenced by fewer neologisms and higher proportions of relevant content (%CIUs
on 2/3 written tasks). However, written discourse production continued to be extremely
effortful for P2 and output remained sparse. See Appendix C for an example of pre- to post-
treatment discourse.
The results of this study provide further evidence that ARCS-W is a potential treatment
option for PWmA interested in improving written and spoken discourse, however, more
research is required to establish efficacy. While each participant responded differently to
treatment, there were some discernible patterns that applied to both. First, written dis-
course produced higher scores across outcome measures related to micro-level relevant
content production (e.g., %CIUs, %CUs) at pre-treatment than spoken discourse, a finding
consistent with previous literature (Behrns et al., 2009; Ulatowska, Baker, & Stern, 1979). This
observation is clinically relevant because the outcome measures used in this study were
developed to evaluate spoken, not written, discourse, and in this case, did not appear to
capture the extent of each participants’ impairments. Potential reasons contributing to the
higher written discourse scores include that participants were less likely to add words they
were unsure of, rarely demonstrated word finding behaviors in writing (e.g., circumlocution)
and produced less written output than spoken output. It is possible that participants were
more intentional with their written discourse output due to the time and effort required,
which is consistent with reports that people with aphasia produce fewer words in written
discourse (e.g., Behrns et al., 2009), require more time to produce written discourse (e.g.,
Ulatowska et al., 1979) and that writing requires more explicit word selection than speaking
(e.g., Behrns et al., 2009; Van Gelderen & Oostdam, 2002).
Another pattern observed from pre- to post-treatment was that both participants
made gains in the macrolinguistic measures of the percent of main events and main
ideas (P1) and global coherence (P2). This finding is meaningful since macrolinguistic
elements of discourse production are not typically targeted in discourse level treatment
even though they are required for all types of discourse production (see Boyle, 2011).
Additionally, macrolinguistic measures are often omitted from discourse analysis for
people with aphasia even though a variety of studies have reported that people with
aphasia can, and often do, demonstrate impaired macrostructure (see Capilouto et al.,
2006; Christiansen, 1995; Wright et al., 2014). This finding provides evidence that macro-
linguistic structure should be incorporated into both aphasia treatment and discourse
analysis.
Functional communication was evaluated via self-report and communication partner
report. Self-rated functional communication improved for P1, but not P2, based on the
pre-determined criteria. Neither communication partner reported overall changes on the

CETI, but both communication partners provided higher ratings for discourse and read-
ing/writing-related tasks (e.g., P1: “having a one-on-one conversation” (7.2/10-8.6/10), and
“describing or discussing something in depth” (6.2/10-8.8/10), P2: “ability to understand
writing” (5.1–8.1 out of 10)). Thus, future questionnaires should have more questions
related to reading and writing and discourse production given the focus of ARCS-W.
Further, it would be informative to add more functional performance-based outcome
measures, such as written email correspondence, to capture functional change in the
modalities being targeted in treatment in addition to partner- and self-reported scales.
Clinical implications
The findings presented in this study replicate those reported in study one with two
additional participants with mild aphasia, indicating that ARCS-W may be a treatment
option to improve the written and spoken discourse of PWmA who can write at the phrase
level. This study is unique in that all of the outcome measures reported are a result of
generalization to discourse or other language measures, and no trained stimuli were
evaluated. Additionally, the participants in this study presented with chronic mild aphasia,
for which few treatment options exist, and who present several challenges in selecting
outcome measures and evaluating treatment effect. These factors should be considered
when interpreting the results presented here.
Another important observation from this study was the treatment response of P2. Based
on our predetermined criteria, he improved his written and spoken discourse production;
however, his spelling abilities did not improve and continued to be time consuming and
effortful, especially when he attempted to use the non-lexical route (e.g., when he was
unsuccessful retrieving a word from the orthographic output lexicon). This observation is
consistent with the theoretical premise of ARCS-W, which requires that participants have
relatively strong skills in isolation since specific linguistic processes are not targeted and not
likely to improve. This observation also adds to potential writing inclusion criteria, which
currently only requires participants to be able to write at the phrase level with no specific
spelling or total output criteria. Although P2 made gains in treatment, his dysgraphia was an
ongoing frustration that made the assessment and treatment tasks difficult. Had P2 received
spelling focused treatment prior to ARCS-W, he might have made greater gains.
In this study, both participants made more substantial improvements in spoken dis-
course compared to written discourse. There are many potential reasons for this observa-
tion. One is that all outcome measures used were originally created to evaluate spoken
discourse. As such, they may be better suited for measuring spoken discourse while written
discourse could require alternative methods. Pre-treatment measures were typically higher
in written discourse than spoken discourse (except %MEs, %MIs, and grammatical complex-
ity). This finding is not a result of extremely high proficiency in writing versus speaking.
Instead, it seems to be related to the modality differences exhibited between speaking and
writing (e.g., fewer words, less revisions in written discourse, explicit word selection in
written discourse, etc.). Additionally, participants did not engage in functional written
discourse tasks (e.g., email, journaling, etc.) as often as spoken discourse (e.g., conversation,
storytelling, etc.), which could have impacted generalization patterns.
APHASIOLOGY 121
The current study included two changes to the treatment protocol reported in study
one (Obermeyer & Edmonds, 2018) (i.e., removing keywords during summarization and
adding homework). It is difficult to determine what effect the changes had on participant
outcomes since these two participants presented with very different pre-treatment
abilities than the cohort in study one. However, based on the treatment protocol, these
changes did appear advantageous and consistent with the desired outcomes. During
treatment, removing keywords did not negatively impact the participants’ ability to
complete the task, and it appeared to have the desired effect of requiring participants
to plan, store and then independently retrieve information.
The purpose of adding a homework task was to encourage generalization to the home
environment. Speaking is required in a variety of contexts and, potentially, with a variety of
partners. However, at pre-treatment, both participants in this study reported that they rarely
engaged in writing or typing at home, even though they identified writing/typing as some-
thing they wanted to improve. Both participants were compliant with the homework task, and
P1 often requested additional homework (this request was not granted), which suggests that
he found the task meaningful. While the amount of writing completed at home was not
measured, P1’s wife reported that during treatment he had composed and sent a letter to the
motor vehicle administration to dispute a fine. She reported that this was a personally
meaningful achievement and not a task he would have attempted before treatment. It is
reasonable to assume that the written modality allowed P1 to take the time he wished to
compose his argument and that the treatment provided him with confidence in his own
writing ability. In the future, adding functional homework tasks and/or treatment components
could promote written discourse skills, likely to be completed in the home, more directly.
Limitations
The written article summarization task proved problematic in this study, similar to study 1
(Obermeyer & Edmonds, 2018). This task required participants to read and summarize a current
event news article in writing. Participants completed this task for four novel articles at pre-
treatment and then those four articles were re-administered at post-treatment and one-month
post-treatment. This protocol represented a shift from the first study in which novel articles
were used at each time point. The four pre-treatment summarization articles were repeated to
minimize the variability observed in study one. However, in the current study, the participants’
response to the article summarization tasks continued to be extremely variable, and the task
itself was more difficult for this cohort. This issue elucidates one of the difficulties inherent in
discourse-level outcome measures, which is their variability. Additionally, there are no trained
stimuli used in this study, which should be considered during interpretation.
Second, both participants in this study had very little written output in discourse tasks,
especially at pre-treatment. Written discourse production was more time intensive and
effortful than spoken discourse production in this study, which could have influenced the
overall amount participants were willing to produce. The additional time required to
produce written discourse could also interact with fatigue and influence total output.
Number of words produced has implications for how written data is interpreted.
Brookshire and Nicholas (1994) reported that approximately 300–400 words are required
for good test–retest reliability in spoken discourse. That benchmark was not met for either
participant at pre-treatment in written discourse. Given the lack of normative data in
written discourse, it is difficult to determine to what extent this impacts the current
findings, but it should be considered, and written data should be interpreted carefully.
The current study was an experimentally controlled single subject pre- to post-treatment
design, which limits the generalization of this work and the strength of the findings. Further,
one of the primary tenants of the ARCS-W treatment is that stimuli are novel, as such, this
treatment is difficult to evaluate using traditional designs for small numbers of participants
in which there are sets of trained and untrained stimuli assessed during pre-treatment,
treatment and post-treatment phases. Further, testing a discourse level treatment with
untrained discourse prompts presents challenges for selecting a priori outcomes and
identifying the active ingredient of the treatment, since linguistic breakdowns can manifest
in unique ways for individuals with aphasia in the discourse context.
Summary and conclusions

In conclusion, the two participants in this study demonstrated modest improvements in
their ability to produce untrained written and spoken discourse. These findings replicate
the results reported by Obermeyer and Edmonds (2018) and provide more preliminary
evidence that ARCS-W is a potential treatment option for PWmA who are interested in
improving their written and spoken discourse. Future studies should evaluate the non-
linguistic cognitive skills that are suspected to play a role in discourse production and
treatment outcomes, investigate normative data for written discourse and consider
including functional discourse tasks (e.g., email, conversation).
Acknowledgments
We would like to thank the following members of the Aphasia Rehabilitation and Bilingual Research
lab at Teachers College, Columbia University who assisted on this project: Heather Swanson,
Jaquelyn Franquez and Patricia Reyes. Additional thanks to Carol Hammer, Mira Goral and
Dolores Perin for their contributions. Lastly and importantly, we would like to thank the participants
who generously devoted their time and energy to this research study.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
Altman, C., Goral, M., & Levy, E. S. (2012). Integrated narrative analysis in multilingual aphasia: The
relationship among narrative structure, grammaticality, and fluency. Aphasiology, 26, 1029–1052.
doi:10.1080/02687038.2012.686103
Ardila, A. (2004). There is not any specific brain area for writing: From cave-paintings to computers.
International Journal of Psychology, 39, 61–67. doi:10.1080/00207590344000295
Armstrong, E., Fox, S., & Wilkinson, R. (2013). Mild aphasia: Is this the place for an argument? American
Journal of Speech-Language Pathology, 22, S268–S278. doi:10.1044/1058-0360(2012/12-0084)
Beeson, P. M., & Robey, R. R. (2006). Evaluating single-subject treatment research: Lessons learned
from the aphasia literature. Neuropsychology Review, 16, 161–169. doi:10.1007/s11065-006-9013-7
Behrns, I., Ahlsén, E., & Wengelin, Å. (2008). Aphasia and the process of revision in writing a text.
Clinical Linguistics & Phonetics, 22, 95–110. doi:10.1080/02699200701699603
APHASIOLOGY 123
Behrns, I., Hartelius, L., & Wengelin, Å. (2009). Aphasia and computerised writing aid supported
treatment. Aphasiology, 23, 1276–1294. doi:10.1080/02687030802436892
Behrns, I., Wengelin, Å., Broberg, M., & Hartelius, L. (2009). A comparison between written and spoken
narratives in aphasia. Clinical Linguistics & Phonetics, 23, 507–528. doi:10.1080/02699200902916129
Boyle, M. (2011). Discourse treatment for word retrieval impairment in aphasia: The story so far.
Aphasiology, 25, 1308–1326. doi:10.1080/02687038.2011.596185
Brookshire, R. H., & Nicholas, L. E. (1994). Speech sample size and test-retest stability of connected
speech measures for adults with aphasia. Journal of Speech, Language, and Hearing Research, 37,
399–407. doi:10.1044/jshr.3702.399
Brookshire, R. H., & Nicholas, L. E. (1997). Discourse comprehension test: Test manual. Minneapolis,
MN: BRK Publishers.
Busk, P. L., & Serlin, R. C. (1992). Meta-analysis for single-case research. In T. R. Kratochwill & J. R. Levin
(Eds.), Single-case research designs and analysis: New directions for psychology and education (pp.
187–212). Hillsdale, NK: Lawrence Erlbaum Associates.
Capilouto, G., Wright, H. H., & Wagovich, S. A. (2005). CIU and main event analyses of the structured
discourse of older and younger adults. Journal of Communication Disorders, 38, 431–444.
doi:10.1016/j.jcomdis.2005.03.005
Capilouto, G. J., Wright, H. H., & Wagovich, S. A. (2006). Reliability of main event measurement in the
discourse of individuals with aphasia. Aphasiology, 20, 205–216. doi:10.1080/02687030500473122
Christiansen, J. A. (1995). Coherence violations and propositional usage in the narratives of fluent
aphasics. Brain and Language, 51, 291–317. doi:10.1006/brln.1995.1062
Dietz, A., Ball, A., & Griffith, J. (2011). Reading and writing with aphasia in the 21st century:
Technological applications of supported reading comprehension and written expression. Topics
in Stroke Rehabilitation, 18, 758–769. doi:10.1310/tsr1806-758
Doyle, P., & Hula, W. (2012). The Aphasia Communication Outcome Measure: Highlighting
patient-reported outcomes in the context of clinical outcome measures, limitations of measures
and issues of construct dimensionality, and results from development of an item bank and
computer adaptive test. CREd Library, 1. doi:10.1055/s-0028-1082882
Doyle, P. J., McNeil, M. R., Spencer, K. A., Goda, A. J., Cottrell, K., & Lustig, A. P. (1998). The effects of
concurrent picture presentations on retelling of orally presented stories by adults with aphasia.
Aphasiology, 12, 561–574. doi:10.1080/02687039808249558
Druks, J., & Masterson, J. (2000). An object & action naming battery. Hove: Psychology Press.
Edmonds, L. A., Nadeau, S. E., & Kiran, S. (2009). Effect of Verb Network Strengthening Treatment
(VNeST) on lexical retrieval of content words in sentences in persons with aphasia. Aphasiology,
23, 402–424. doi:10.1080/02687030802291339
Elman, R. J., & Bernstein-Ellis, E. (1995). What is functional? American Journal of Speech-Language
Pathology, 4, 115–117. doi:10.1044/1058-0360.0404.115
Frankel, T., Penn, C., & Ormond-Brown, D. (2007). Executive dysfunction as an explanatory basis for
conversation symptoms of aphasia: A pilot study. Aphasiology, 21, 814–828. doi:10.1080/
02687030701192448
Fromm, D., Forbes, M., Holland, A., Dalton, S. G., Richardson, J., & MacWhinney, B. (2017). Discourse
characteristics in aphasia beyond the western aphasia battery cutoff. American Journal of Speech-
Language Pathology, 26, 762–768. doi:10.1044/2016_AJSLP-16-0071
Furnas, D. W., & Edmonds, L. A. (2014). The effect of computerised verb network strengthening treatment
on lexical retrieval in aphasia. Aphasiology, 28, 401–420. doi:10.1080/02687038.2013.869304
Goodman, R. A., & Caramazza, A. (1986). The Johns Hopkins University dyslexia and dysgraphia
batteries. [Published in Beeson, P. M., & Hillis, A. E. (2001). Comprehension and production of
written words. In R. Chapey (Ed.), Language intervention strategies in adult Aphasia (pp. 572–595).
Baltimore: Lippencott, Williams, & Williams.
Helm-Estabrooks, N. (2001). CLQT: Cognitive Linguistic Quick Test. San Antonio, TX: The Psychological
Corporation.
Hilari, K., & Northcott, S. (2006). Social support in people with chronic Aphasia. Aphasiology, 20,
17–36. doi:10.1080/02687030500279982
Hula, W. D., Doyle, P. J., Stone, C. A., Hula, S. N. A., Kellough, S., Wambaugh, J. L., . . . Jacque, A. S.
(2015). The Aphasia Communication Outcome Measure (ACOM): Dimensionality, item bank
calibration, and initial validation. Journal of Speech, Language, and Hearing Research, 58,
906–919. doi:10.1044/2015_JSLHR-L-14-0235
Kendall, D. L., Rosenbek, J. C., Heilman, K. M., Conway, T., Klenberg, K., Rothi, L. J. G., & Nadeau, S. E.
(2008). Phoneme-based rehabilitation of anomia in aphasia. Brain and Language, 105, 1–17.
doi:10.1016/j.bandl.2007.11.007
Kertesz, A. (2006). Western Aphasia Battery–Revised (WAB-R). Pro-Ed. Austin, TX.
Kincaid, J. P., Fishburne, R. P., Jr, Rogers, R. L., & Chissom, B. S. (1975). Derivation of new readability
formulas. (automated readability index, fog count and flesch reading ease formula) for navy enlisted
personnel (No. RBR-8-75). Millington, TN: Naval Technical Training Command Millington TN
Research Branch.
LaPointe, L. L., & Horner, J. (1998). RCBA-2: Reading comprehension battery for Aphasia. Pro-Ed. Austin, TX.
Loban, W. (1976). Language development: Kindergarten through grade twelve (Research Report No.
18). Urbana, IL: National Council of Teachers of English. doi: 10.1084/jem.143.4.741
Lomas, J., Pickard, L., Bester, S., Elbard, H., Finlayson, A., & Zoghaib, C. (1989). The communicative
effectiveness index: Development and psychometric evaluation of a functional communication
measure for adult aphasia. Journal of Speech and Hearing Disorders, 54, 113–124. doi:10.1044/
jshd.5401.113
Marini, A., & Urgesi, C. (2012). Please get to the point! A cortical correlate of linguistic
informativeness. Journal of Cognitive Neuroscience, 24, 2211–2222. doi:10.1162/jocn_a_00283
National Aphasia Association. Aphasia Definitions (n.d.). Retrieved from https://www.aphasia.org/
aphasia-definitions/.
Nicholas, L. E., & Brookshire, R. H. (1993). A system for quantifying the informativeness and efficiency
of the connected speech of adults with aphasia. Journal of Speech, Language, and Hearing
Research, 36, 338–350. doi:10.1044/jshr.3602.338
Obermeyer, J. A., & Edmonds, L. A. (2018). Attentive reading with constrained summarization
adapted to address written discourse in people with mild aphasia. American Journal of Speech-
Language Pathology, 27, 392–405. doi:10.1044/2017_AJSLP-16-0200
Peach, R. K., & Reuter, K. A. (2010). A discourse-based approach to semantic feature analysis for the
treatment of aphasic word retrieval failures. Aphasiology, 24, 971–990. doi:10.1080/
02687030903058629
Richardson, J. D., & Dalton, S. G. (2016). Main concepts for three different discourse tasks in a large
non-clinical sample. Aphasiology, 30, 45–73. doi:10.1080/02687038.2015.1057891
Robey, R. R. (2004). A five-phase model for clinical-outcome research. Journal of Communication
Disorders, 37, 401–411. doi:10.1016/j.jcomdis.2004.04.003
Robey, R. R., Schultz, M. C., Crawford, A. B., & Sinner, C. A. (1999). Single-subject clinical-outcome
research: Designs, data, effect sizes, and analyses. Aphasiology, 13, 445–473. doi:10.1080/
026870399402028
Rogalski, Y., & Edmonds, L. A. (2008). Attentive Reading and Constrained Summarization (ARCS)
treatment in primary progressive aphasia: A case study. Aphasiology, 22, 763–775. doi:10.1080/
02687030701803796
Rogalski, Y., Edmonds, L. A., Daly, V. R., & Gardner, M. J. (2013). Attentive Reading and Constrained
Summarization (ARCS) discourse treatment for chronic Wernicke’s aphasia. Aphasiology, 27,
1232–1251. doi:10.1080/02687038.2013.810327
Ulatowska, H. K., Allard, L., & Chapman, S. B. (1990). Narrative and procedural discourse in Aphasia. In
Y. Joanette & H. H. Brownell (Eds.), Discourse ability and brain damage (pp. 180–198). New York,
NY: Springer.
Ulatowska, H. K., Baker, T., & Stern, R. F. (1979). Disruption of written language in aphasia. In Studies in
Neurolinguistics, 4, 241–268. doi:10.1016/B978-0-12-746304-9.50013-6
Ulatowska, H. K., Hildebrand, B. H., & Haynes, S. M. (1978). A comparison of written and spoken
language in Aphasia. In Clinical Aphasiology: Proceedings of the Conference 1978 (pp. 223–235).
Minneapolis, MN: BRK Publishers.
APHASIOLOGY 125
Van Gelderen, A., & Oostdam, R. (2002). Improving linguistic fluency for writing: Effects of explicit-
ness and focus of instruction. L1-Educational Studies in Language and Literature, 2, 239–270.
doi:10.1023/A:1021304027877
Vickers, C. P. (2010). Social networks after the onset of aphasia: The impact of aphasia group
attendance. Aphasiology, 24, 902–913. doi:10.1080/02687030903438532
Wilshire, C. E., Lukkien, C. C., & Burmester, B. R. (2014). The sentence production test for aphasia.
Aphasiology, 28, 658–691. doi:10.1080/02687038.2014.893555
Wright, H. H., & Capilouto, G. J. (2012). Considering a multi-level approach to understanding
maintenance of global coherence in adults with aphasia. Aphasiology, 26, 656–672.
doi:10.1080/02687038.2012.676855
Wright, H. H., Capilouto, G. J., & Koutsoftas, A. (2013). Evaluating measures of global coherence
ability in stories in adults. International Journal of Language & Communication Disorders, 48,
249–256. doi:10.1111/jlcd.2013.48.issue-3
Wright, H. H., Koutsoftas, A. D., Capilouto, G. J., & Fergadiotis, G. (2014). Global coherence in younger
and older adults: Influence of cognitive processes and discourse type. Aging, Neuropsychology,
and Cognition, 21, 174–196. doi:10.1080/13825585.2013.794894
Youse, K. M., & Coelho, C. A. (2005). Working memory and discourse production abilities following
closed-head injury. Brain Injury, 19, 1001–1009. doi:10.1080/02699050500109951
Appendix A. Attentive Reading with Constrained Summarization-Written (ARCS-W) Treatment Steps.

Treatment Step Participant Action Clinician Action
Review homework activity during the first session of the week
Step 1 Follow along with the text Read the entire article aloud
Read 1–3 sentences, twice for comprehension
Step 2 Identify and write down key words Write down key words in the segment
Step 3 Compare keywords with the clinician’s Compare keywords with the participant’s
Provide feedback on the key words identified
Step 4 Use key words to plan verbal summary Remove keywords when the participant is
ready to summarize
Step 5 Verbally summarize the segment from memory, Provide feedback regarding constraints, and
while following constraints content (keywords)
Step 6 Use key words to plan written summary Remove key words when participant is
ready to summarize
Step 7 Summarize what was read in writing from memory Provide feedback regarding constraints and
and then read it to the clinician and check for errors content.
Step 8 Read the entire article
Step 9 Summarize the entire article verbally and in writing Provide feedback on spoken and written
from memory summary
Note. Bold items indicate changes from the protocol reported in Obermeyer and Edmonds (2018).
Copyright of Aphasiology is the property of Taylor & Francis Ltd and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.

Attentive Reading With Constrained Summarization-Written, A Multi-Modality Discourse-Level Treatment For Mild Aphasia

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Attentive Reading With Constrained Summarization-Written, A Multi-Modality Discourse-Level Treatment For Mild Aphasia

Uploaded by

Copyright:

Available Formats

APHASIOLOGY

2021, VOL. 35, NO. 1, 100–125

Attentive Reading with Constrained Summarization-Written,

ABSTRACT ARTICLE HISTORY

Table 1. Pre-Treatment Language Assessment Results for Both Participants.

Story retell task

Nicholas and Brookshire task

Article summarization task

participant was prompted to follow prescribed constraints 1) avoid nonspeciﬁc words

Data analysis of discourse assessments

Article summarization task

Data analysis for non-discourse assessments

Written story retell task (Table 2)

Written article summarization task (Table 3)

Control task (Table 3)

P1: research question 2: eﬀect of ARCS-W on spoken discourse (Table 4)

Spoken story retell task

P1: research question 3: eﬀect of ARCS-W on non-discourse assessments (Table 5)

Table 5. Non-discourse Assessments for Participant 1 and Participant 2.

Table 6. P2 Written Discourse Assessments from Pre-to Post-Treatment.

P2: research question 1: eﬀect of ARCS-W on written discourse

Written story retell task (Table 6)

Written article summarization task (Table 7)

Control task (Table 7)

Table 8. P2 Spoken Discourse Assessments from Pre- to Post-Treatment.

P2: research question 2: eﬀect of ARCS-W on spoken discourse (Table 8)

Spoken story retell task

P2: research question 3: eﬀect of ARCS-W non-discourse assessments (Table 5)

pre-determined criteria. Neither communication partner reported overall changes on the

Summary and conclusions

Appendix A. Attentive Reading with Constrained Summarization-Written (ARCS-W) Treatment Steps.

You might also like