JSLHR

Research Article

Combined Aphasia and Apraxia of

Speech Treatment (CAAST):
Effects of a Novel Therapy
Julie L. Wambaugh,a,b Sandra Wright,a Christina Nessler,a and Shannon C. Mauszyckia,b

Purpose: This investigation was designed to examine the context of multiple baseline designs. CAAST entailed
the effects of a newly developed treatment for aphasia elaboration of participant-initiated utterances, with sound
and acquired apraxia of speech (AOS). Combined production training applied as needed to the elaborated
Aphasia and Apraxia of Speech Treatment (CAAST) targets productions. The dependent variables were (a) production
language and speech production simultaneously, with of correct information units (CIUs; Nicholas & Brookshire,
treatment techniques derived from Response Elaboration 1993) in response to experimental picture stimuli,
Training (Kearns, 1985) and Sound Production Treatment (b) percentage of consonants correct in sentence repetition,
(Wambaugh, Kalinyak-Fliszar, West, & Doyle, 1998). The and (c) speech intelligibility.
purpose of this study was to determine whether CAAST Results and Conclusions: CAAST was associated with
was associated with positive changes in verbal language increased CIU production in trained and untrained picture
and speech production with speakers with aphasia and sets for all participants. Gains in sound production accuracy
AOS. and speech intelligibility varied across participants; a
Method: Four participants with chronic aphasia and AOS modification of CAAST to provide additional speech
received CAAST applied sequentially to sets of pictures in production treatment may be desirable.

A
phasia and acquired apraxia of speech (AOS) Robin, & Schmidt, 2009). It ranges in severity from a
are neurogenic disorders of language and motor complete inability to speak to minor sound distortions.
speech, respectively. AOS seldom occurs as an iso- Stroke is the most common etiology for nonprogressive AOS
lated disorder and is typically accompanied by nonfluent (Duffy, 2013), and damage to cortical and/or subcortical
aphasia (Duffy, 2013). The relative contributions of aphasia areas of the language dominant hemisphere has been asso-
and AOS to overall communication disruption in persons ciated with AOS (Wambaugh & Shuster, 2008).
with both disorders are not well understood; singly, and Improvements in speech production have been dem-
in combination, these disorders can significantly disrupt onstrated with a variety of AOS treatments (Wambaugh,
communication. Although many persons with aphasia and Duffy, McNeil, Robin, & Rogers, 2006). Across the 59 AOS
AOS likely require treatment for both disorders, there has treatment investigations included in the AOS treatment
been limited research addressing treatments that have been guidelines systematic review, all 148 research participants
designed to target both. The purpose of the current investiga- had AOS accompanied by aphasia. However, none of the
tion was to examine the effects of a newly developed behav- AOS treatments reviewed in the AOS guidelines included
ioral treatment that targets aphasia and AOS simultaneously. direct treatment for language (Wambaugh et al., 2006).
AOS is characterized by slow rate of speech, difficul- Although it may be advantageous for researchers to
ties in sound production, and disrupted prosody (McNeil, develop and investigate treatments focused on a single dis-
order, a unilateral treatment focus may have limitations
in terms of clinical application for persons with multiple
a
disorders. Allocation of therapy time to different disorders
Veterans Affairs Salt Lake City Health Care System, Utah
b and treatments may be challenging, particularly with a lim-
University of Utah, Salt Lake City
ited number of therapy sessions. Determining whether
Correspondence to Julie L. Wambaugh: wambaugh@health.utah.edu
AOS or aphasia is the primary contributor to a particular
Editor: Rhea Paul
communication disruption may be difficult, and perhaps
Associate Editor: Kristine Lundgren
Received January 7, 2014
Revision received May 1, 2014
Accepted June 20, 2014 Disclosure: The authors have declared that no competing interests existed at the
DOI: 10.1044/2014_JSLHR-L-14-0004 time of publication.

Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014 • © American Speech-Language-Hearing Association 2191
impossible, for a clinician to ascertain in a timely manner. application require further study (e.g., optimal presenta-
For example, inability to describe a picture stimulus may tion of stimuli, most effective dosage, and social validity),
result from a language disruption or a motor speech dis- its acquisition, response generalization, and maintenance
ruption, or a combination of both, in an individual with effects have been demonstrated to be robust and rela-
AOS and aphasia. Given the high co-occurrence of aphasia tively predictable; improvement in production of treated
with AOS, the likelihood of the disorders interacting or sounds has been consistently observed in trained and un-
co-contributing to communication disruption, and the need trained items during nontreatment probes. Increases in
for efficient and clinically applicable treatments, a treatment articulatory accuracy have been maintained above baseline
that targets both disorders may be desirable. levels for the majority of speakers in post-SPT follow-up
Response Elaboration Training (RET; Kearns, 1985) probes.
is an aphasia treatment that has been modified for use with A key component of SPT in previous research has
persons with aphasia and AOS (Modified RET [M-RET]; been the targeting of specific sounds that were determined
Wambaugh & Martinez, 2000; Wambaugh, Wright, & to be problematic for the AOS speaker in pretreatment as-
Nessler, 2012). RET was developed to facilitate creative sessment. That is, treatment items have been selected on the
and generalizable language production by capitalizing on basis of a priori observation of patterns of sound errors. In
unique, patient-initiated utterances and has been shown contrast, with M-RET, there are no predetermined target
to consistently result in increased production of content in responses; treatment proceeds on the basis of speaker-initiated
discourse (Gaddie, Kearns, & Yedor, 1991; Kearns, 1985; utterances. Therefore, with the new treatment, SPT could
Kearns & Scher, 1989; Kearns & Yedor, 1991; Yedor, not be applied to preestablished target sounds. Instead, in
Conlon, & Kearns, 1993). With RET, the clinician encour- the newly developed treatment, Combined Aphasia and
ages the speaker to produce an utterance of his or her choice Apraxia of Speech Treatment (CAAST), the SPT hierarchy
in response to picture stimuli. Reinforcement, modeling, is employed with the speaker-generated utterances that are
and forward chaining are then used to expand the utterance. unique with each treatment trial.
RET and M-RET have been shown repeatedly to result in As with RET and M-RET, a primary goal of CAAST
improved production of language content in response to is to increase verbal language productivity by facilitating
trained and untrained stimuli (see Wambaugh et al., 2012, elaboration of patient-initiated productions; flexible and
for a review). generalized language use is expected (Kearns, 1985). In
Wambaugh and colleagues (Wambaugh & Martinez, keeping with the emphasis of SPT, improved speech pro-
2000; Wambaugh et al., 2012) adapted the basic RET pro- duction is also a goal of CAAST. Given the dual focus of
cedure to accommodate speech production difficulties asso- CAAST, language and speech production outcome mea-
ciated with AOS. Although the modification included some sures were required to measure the effects of treatment.
techniques that may promote improved speech production A measure of production of content in discourse was
(e.g., integral stimulation, repeated productions), M-RET chosen to elucidate potential changes in language produc-
does not include specific treatment for apraxic speech pro- tivity (Nicholas & Brookshire, 1993). Because increases
duction errors, and speech production has not been a mea- in content production have been regularly reported with
sured outcome. That is, increased language production RET and M-RET (Gaddie et al., 1991; Kearns, 1985;
remains the focus of M-RET. Kearns & Scher, 1989; Kearns & Yedor, 1991; Wambaugh
Because of the strong, experimentally sound evidence & Martinez, 2000; Wambaugh et al., 2012; Yedor et al.,
base supporting generalized language changes with RET 1993), it was deemed important that such changes be evi-
and M-RET and the limited data concerning speech changes dent with CAAST.
for other aphasia treatments, we chose to use M-RET as the Typical SPT outcome measures, such as accuracy
foundation for a new treatment that targets both language of targeted sounds/words in trained and untrained items,
and speech. We combined M-RET with an established AOS could not be utilized to evaluate the effects of CAAST be-
treatment, Sound Production Treatment (SPT; Wambaugh, cause treatment does not involve prespecified targets. Con-
2004; Wambaugh, Kalinyak-Fliszar, West, & Doyle, 1998; sequently, a measure was devised in which percentage of
Wambaugh & Mauszycki, 2010; Wambaugh & Nessler, consonants correct (PCC) was measured in the production
2004; Wambaugh, Nessler, Cameron, & Mauszycki, 2013; of sentences. The sentences were developed to be similar to
Wambaugh, Nessler, Wright, & Mauszycki, 2014). SPT is expected, participant-generated utterances in CAAST. In
an articulatory-kinematic treatment for AOS that was de- addition, single word intelligibility was selected as an out-
signed to improve articulation of targeted sounds produced come measure that could reflect positive changes in speech
in the context of words, phrases, and/or sentences. SPT production.
combines modeling, repetition, minimal pair contrast, inte- In summary, we have developed a new treatment that
gral stimulation, articulatory placement cueing, and feed- targets aphasia and AOS simultaneously. CAAST com-
back in a response-contingent hierarchy. SPT has been bines features of two existing treatments (i.e., M-RET and
studied with AOS speakers of various severities with level SPT) to increase generalizable verbal language skills and to
of speech production during treatment being adjusted to improve speech production. Not only is this approach novel
meet the speaker’s needs (e.g., monosyllabic words, multi- in terms of being a combined approach, it is also unique
syllabic words, phrases). Although aspects of SPT and its with respect to applying sound production remediation in

2192 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
the context of patient-initiated utterances. The specific ex- 3. Assessment of Intelligibility of Dysarthric Speech
perimental questions addressed in this investigation were as (AIDS; Yorkston & Beukelman, 1981);
follows: 4. consonant production probe (Wambaugh, Kalinyak-
1. Is CAAST associated with increases in the produc- Fliszar, et al., 1998);
tion of content in discourse elicited in response to 5. sentence repetition (Wambaugh, West, & Doyle,
trained and untrained stimuli? 1998); and
2. Is CAAST associated with improvements in speech 6. multisyllabic word repetition (Mauszycki & Wambaugh,
production as measured by PCC in sentences and 2008).
single word speech intelligibility?
The following behaviors deemed necessary for the
diagnosis of AOS were demonstrated by all of the partici-
pants: slow rate of speech production (including syllable
Method segregation), sound errors that were relatively consistent in
Participants type and location across repeated trials, sound errors that
Four men with chronic, stroke-induced aphasia and were predominately sound distortions, and prosodic abnor-
AOS served as participants. Participant characteristics are malities. The presence of the requisite diagnostic charac-
summarized in Table 1. Medical records indicated that each teristics and corresponding diagnosis of AOS was initially
participant’s AOS and aphasia resulted from a single epi- determined by each participant’s primary clinician (one
sode, left hemisphere stroke. of the authors) and was independently confirmed by the
The participants were native English speakers between first author. All authors have extensive experience in the
36 and 72 years of age. Each passed a pure tone hearing diagnosis, treatment, and study of AOS.
screening at 35 dB at 500, 1000, and 2000 Hz for at least The participants’ word-level intelligibility scores
one ear, unaided. All demonstrated performance within ranged from 8% to 76% as scored through orthographic
normal limits on the Test of Nonverbal Intelligence—4 transcription (Yorkston & Beukelman, 1981; see Table 2).
(Brown, Sherbenou, & Johnsen, 2010). All had self-reported Severity of AOS, estimated on the basis of intelligibility
negative histories for alcohol or substance abuse and neuro- and prevalence of sound production errors, was considered
logical conditions other than stroke; the reports were veri- to range from mild–moderate to moderate–severe across
fied through review of existing medical records. None of the participants.
the participants received any other speech/language therapy All of the participants presented with agrammatic
during the course of this study. Each participant lived in his aphasia. Participants 2, 3, and 4 received a diagnosis
own home with a significant other. of Broca’s aphasia on the basis of the Western Aphasia
The presence of AOS was determined using the diag- Battery (Kertesz, 2007). Participant 1’s performance on the
nostic criteria described by McNeil, Robin, and Schmidt Western Aphasia Battery resulted in a classification of ano-
(1997, 2009). Speech samples were obtained from each par- mic aphasia. The participants’ productive verbal language
ticipant employing the following elicitation tasks: ranged from single words to short sentences and was gen-
erally characterized as follows: Participant 1—short sen-
1. Increasing Word Length and Repeated Trials subtests tences and phrases; Participant 2—mainly single words;
of the Apraxia Battery for Adults—Second Edition Participant 3—single words, phrases, and short sentences;
(Dabul, 2000); and Participant 4—single words and phrases (for samples
2. narrative and procedural discourse tasks (Nicholas & of discourse, see Appendix A in the online supplemental
Brookshire, 1993); materials).

Table 1. Participant characteristics.

CVA Age Months post Years of Premorbid Race/

Participant Gender location/type (years) onset of stroke education handiness ethnicity Hemiparesis

Participant 1 Male L MCA ischemic 72 12 11+ R White–nH / L None

Participant 2 Male L MCA ischemic 71 65 20 R White–nH / L R UE, R LE
Participant 3 Male L MCA hemorrhagic 36 23 11 R White–nH / L R UE, R LE
Participant 4 Male L basal ganglia 54 255 14 R White–nH / L R UE, R LE
hemorrhagic with
intraventricular
hemorrhage and
L frontal lobe
hematoma

Note. CVA = cerebrovascular accident; L = left; MCA = middle cerebral artery; R = right; nH / L = non-Hispanic/ Latino; UE = upper extremity;
LE = lower extremity.

Wambaugh et al.: CAAST 2193

Table 2. Pretreatment assessment results.

Measure Participant 1 Participant 2 Participant 3 Participant 4

Western Aphasia Battery

AQ 77.1 24.3 56 55.6
Type Anomic Broca’s Broca’s Broca’s
Porch Index of Communicative Ability
Overall percentile 58 47 49 49.
Verbal percentile 59 17 51 48.
Auditory percentile 54 43 40 64.
Reading percentile 40 84 28 44.
Test of Adolescent/Adult Word Finding
Total raw score 55/107 0/107 44/107 8/107.
% comprehension 95 93 93 95.
Verb and Sentence Test
Sentence construction 6/20 0/20 0/20 4/20.
Sentence anagram with pictures 9/20 10/20 8/20 10/20.
Sentence anagram without pictures 12/20 16/20 8/20 6/20.
Test of Nonverbal Intelligence—4
Percentile 34th 42nd 19th 39th
Reading Comprehension Battery for Aphasia
Word-visual 10/10 10/10 9/10 7/10.
Word-auditory 10/10 10/10 6/10 10/10.
Word-semantic 9/10 10/10 9/10 9/10.
Functional reading 8/10 8/10 DNT 3/10.
Sentence-picture 9/19 10/10 DNT 8/10.
Assessment of Intelligibility of Dysarthric Speech (% single word) 76 8 76 64.
Estimated AOS severity Mild–moderate Moderate–severe Moderate Moderate

Note. AQ = aphasia quotient; DNT = did not test; AOS = acquired apraxia of speech.

Overall percentile scores on the Porch Index of Com- further treatment effects to be detected. In addition, A MBD
municative Ability (Porch, 2001) ranged from the 47th to across the participants’ component was included in the design
58th percentile. The participants did not exhibit symptoms in which the number of baseline measurements was extended
of dysarthria as described by Duffy (2013). across participants; the number of baseline probes was in-
creased by one session across each participant so that baseline
probes increased from five to eight for Participant 1 through
Experimental Design Participant 4. To demonstrate experimental control, applica-
General Description tion of treatment must be systematically associated with be-
Each participant received treatment applied in the havioral change with treatment phases instituted at different
context of a multiple baseline design (MBD) across behaviors. time periods (relative to baseline) across participants.
Production of correct information units (CIUs; Nicholas & The design also included maintenance and follow-up
Brookshire, 1993) was measured repeatedly with three sets measurements. Specifics of the design by each phase are
of experimental pictures in the baseline phase. Then, treat- provided in the following sections.
ment was applied sequentially to two of the picture sets.
Experimental control is demonstrated with the MBD across Baseline Phase
behaviors when systematic changes in the dependent vari- During baseline, discourse samples were elicited in re-
able coincide with application of treatment and when those sponse to the three sets of picture stimuli for each partici-
effects are replicated with subsequent applications of treat- pant. Prior to the start of the study, the minimum number
ment with that participant (Kratochwill et al., 2010). In the of baseline probes for the first participant was designated as
event that treatment results in generalized responding to five to allow application of the conservative dual-criterion
untreated behaviors (i.e., changes in untreated behaviors (CDC) method (Fisher, Kelley, & Lomas, 2003) in data
occur prior to treatment being applied directly to those be- analysis. In addition, a nonascending or descending trend
haviors), demonstration of the replication of treatment ef- in CIU production was required to be evident in the base-
fects within the participant may be difficult. We anticipated line phase prior to initiation of treatment.
that increases in CIU production might occur with untreated
experimental sets and included additional design controls. Treatment Phase
We completed additional probing prior to applying Following baseline, CAAST was applied sequentially
treatment to the second experimental set to ensure stability of to two experimental sets. During the treatment phases, the
responding with that set. Conceptually, this may be consid- experimental set under treatment was probed after every
ered an additional baseline (“A”) phase designed to allow two treatment sessions but immediately prior to the next

2194 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
treatment session. The sets that were not under treatment similar to those produced in the narrative discourse sam-
were probed less frequently throughout the treatment ples. These stimuli were developed on the basis of previous
phases. The set slated for the second treatment phase was experience with M-RET and the expectation that use of the
probed at the midpoint of the first treatment phase and sentence frame in CAAST would stimulate production of
then with increasing frequency as the first phase neared at least canonical sentences. The same sentences were used
completion. Prior to initiation of treatment with the second for Participants 2, 3, and 4, and each sentence was seven
set, additional extended baseline probing was conducted to to nine syllables in length (e.g., “The boy is riding a bike.”).
ensure behavioral stability. The third set of pictures did not Because of his less severe AOS and aphasia, more difficult
receive treatment and was probed only at the end of each sentences were devised for Participant 1 (Set 1 = 7–11 sylla-
phase of treatment. The reduced probing schedule was used bles; Set 2 = 10–19 syllables).
to avoid overexposure and potential practice effects with Within each set, the10 sentences were presented ver-
untreated sets. All probes during the treatment phases were bally, one at a time in random order, and the participant
conducted prior to the day’s treatment session. was asked to repeat the sentence as accurately as possible.
Printed stimuli were presented along with the verbal model
Maintenance and Follow-Up Phases for one of the sentence sets to counter possible word-retrieval
When treatment was applied to the second set of pic- or memory difficulties. Sentences in both sets (auditory only
tures, probes were completed with the previously treated and auditory plus orthographic) were comparable for Par-
set at the midpoint and end of the second treatment phase. ticipants 2, 3, and 4. For Participant 1, lengthier sentences
Follow-up probes were completed with all experimental sets were used for the auditory plus orthographic condition than
at 2 and 6 weeks following the last treatment session of the in the auditory only condition. The longer sentences were
second treatment phase. developed because Participant 1’s speech production accu-
racy increased substantially with the addition of printed
sentences.
Procedures
Experimental Stimuli Probes
Narrative discourse elicitation stimuli. Thirty line Production of discourse was elicited in probes through
drawings depicting common actions were used as experi- presentation of the three picture sets. The order of presen-
mental stimuli. These pictures were part of a larger set of tation of sets and items within each set was randomized. The
100 drawings that were redrawn pictures from An Object 10 pictures in each set were presented one at a time. The in-
and Action Naming Battery (Druks & Masterson, 2000). vestigator provided the following instructions: “Tell me as
The 100 pictures had been validated for a previous investi- much as you can about this picture. You can talk about this
gation with respect to adequacy of depiction of the actions. picture or anything it reminds you of.” No time limits were
Ten adults without brain damage ranging from 41 to 70 years imposed. The preceding procedures were utilized with each
of age (M age = 57.6 years; five men and five women) were picture presentation, and all probes were audio recorded.
asked to provide a one-word response that best described The 20 sentences used for eliciting speech samples
the action in each drawing; all drawings elicited the desired were administered on three separate occasions prior to the
action name or an acceptable alternative. The drawings start of treatment; for Participants 1, 2, and 4, these probes
were not used in the current investigation to elicit specific occurred prior to the final discourse probe. For Participant 3,
action names but rather were used to stimulate production these probes were administered after the final discourse probe.
of narrative discourse. The sentences were then readministered at the midpoint
The drawings for Participants 1 and 2 were selected and end of each treatment phase and at 2 and 6 weeks
on the basis of personal relevance or potential interest as posttreatment.
determined by each participant’s primary speech-language
pathologist (SLP); approximately half of the items over-
lapped both participants’ sets. The drawings for Partici- Dependent Variable
pants 3 and 4 were the same as those for Participant 2 to CIUs
provide replications and counterbalancing with the stimuli. Total number of CIUs produced in response to the
For each participant, the 30 pictures were randomly divided picture stimuli served as the primary dependent variable.
into three sets of 10 pictures each. Two lists were designated CIUs reflect the appropriateness, relevancy, and informa-
for use in treatment and measurement of acquisition effects tiveness of words produced by a speaker in relation to a
in probes. The third list remained untreated to allow for as- particular topic (Nicholas & Brookshire, 1993). Procedures
sessment of generalization effects (for lists, see Appendix B described by Nicholas and Brookshire (1993) for calculat-
in the online supplemental materials). ing CIUs were utilized.
Speech elicitation stimuli. Two sets of 10 short sen- The discourse samples from the picture probes were
tences were developed to elicit speech samples for mea- orthographically transcribed using online transcriptions
suring accuracy of articulation as reflected by PCC (see supplemented by the audio recordings. An investigator,
Appendix C in the online supplemental materials). The sen- who was blinded to the assignment of pictures to condition
tences were composed of words that were predicted to be and was not involved in treatment, independently verified

Wambaugh et al.: CAAST 2195

all transcriptions. Any discrepancies in transcription were probe (total judgments). The percentage of agreement of to-
corrected for calculation of CIUs. Total number of CIUs tal judgments (total agreements divided by total judgments)
was tabulated separately for each set of 10 pictures for each was calculated for each probe for each participant: Partici-
probe. pant 1—87.5%, Participant 2—85%, Participant 3—91.5%,
To measure the possible generalization effects of and Participant 4—93%.
treatment beyond the experimental pictures used in this Point-to-point reliability for scoring of PCC was also
study, narrative and procedural discourse samples were col- calculated for 20% of randomly selected sentence repetition
lected prior to the start of treatment and following comple- probes. A reliability examiner used the audio recordings
tion of all treatment using elicitation procedures established to rescore the selected sentence repetition probes. Average
by Nicholas and Brookshire (1993). Nicholas and Brookshire agreement for scoring of each consonant in each probe was
included additional CIU metrics beyond total CIUs, such 96.25% across the participants. Agreement across probes
as percentage of CIUs and CIUs per minute. Wambaugh, ranged from 88% to 100%.
Nessler, and Wright (2013) provided a rationale for utilizing Point-to-point reliability for scoring CIUs in the
number of CIUs rather than the other CIU metrics in the nontreatment narrative and procedural discourse samples
study of RET, which also applies to the examination of the (Nicholas & Brookshire, 1993) was calculated for 50% of the
effects of CAAST. samples. One of the pre- or posttreatment samples was ran-
domly selected for each participant for rescoring by an ex-
PCC in Sentence Repetition aminer who had not conducted the original scoring. Average
For each target sentence, the number of consonants agreement was 86.5%, with a range from 82% to 91%.
articulated correctly in content words was calculated. A
scoring template was created for each sentence in which the
target consonants were identified. Using the audio record- Treatment
ings, the examiner who conducted the PCC probe (one of CAAST combined steps from M-RET (Wambaugh
the participant’s SLPs) scored each consonant as correct/ & Martinez, 2000; Wambaugh et al., 2012) with SPT
incorrect; examiners were not limited in the number of (Wambaugh et al., 1998). In addition, use of an empty sen-
times that the audio recordings could be replayed. To be tence frame was included in the M-RET portion of treat-
scored as correct, the consonant was required to be produced ment to assist in production elicitation and expansion. The
accurately in the correct location within the word. Words therapist elicited a verbal production in response to an ac-
were not required to be in the correct order in the sentence. tion picture and then used modeling, additional prompting,
Omissions, substitutions, additions, and distortions were and forward-chaining to expand the initial utterance. The
scored as incorrect. For an addition, total possible conso- participant’s verbal responses were written in the empty
nants correct was maintained, and a point was subtracted. frame by the therapist. After the expanded utterance was
If an entire word was omitted, all consonants for that word developed, the participant was asked to repeat the utter-
were scored as incorrect. PCC was calculated on the basis ance. If there were articulatory errors in the repeated utter-
of total number of target consonants for the entire set of ance, the SPT hierarchy was then implemented. Following
10 sentences (for scoring examples, see Appendix C in the completion of the SPT steps, the next picture was pre-
online supplemental materials). sented. The detailed CAAST protocol is provided in
Appendix D in the online supplemental materials. A brief
Word Intelligibility example of a treatment trial follows:
The single word intelligibility portion of the AIDS Introduction:
(Yorkston & Beukelman, 1981) was administered prior to The therapist describes the sentence frame (completed
and following treatment. An investigator who had not ad- one time at the start of the session).
ministered treatment to any of the participants and who
was blinded to condition (pre- or posttreatment) used audio Step 1:
recordings to orthographically transcribe all AIDS sam- Therapist—Presents action picture and says, “Tell me
ples. Percentage of intelligibility was calculated from the anything about this picture; what does it remind you
transcriptions. of ?; what’s happening?”
Participant—No response.
Therapist—“You could say something like man spills
Reliability . . . or . . . drops a cup.”
Point-to-point reliability was calculated for scoring of Participant—“Spill.”
CIUs in probes. For each participant, 20% of probes were
Step 2:
quasi-randomly selected so that all phases of the study were
Therapist—“Spill, great.” Referring to the sentence
represented. The reliability examiner independently scored
frame, asks participant to indicate where to write “spill.”
the selected probes using the verified transcripts. Any
items that were counted as CIUs by one examiner and not Step 3:
the other examiner were counted as disagreements. Total Therapist—“What does the man spill?”
agreements and disagreements were tabulated for each Participant—“Milk.”

2196 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
 Step 4: to the start of the study, with 20 treatment sessions desig-
Therapist—“Milk, good, spill milk.” Referring to the nated per treatment phase (i.e., per experimental set); this
sentence frame, asks the participant where to write number of sessions was selected on the basis of previous
“milk.” experience with M-RET (Wambaugh & Martinez, 2000;
Wambaugh et al., 2012) as a best estimate for elucidating
Step 5A:
treatment effects. However, as treatment and probing pro-
Therapist—“Repeat after me . . . spill milk.”
gressed with Participant 1, treatment effects appeared to
Participant—“Pill milk.”
have stabilized after 14 treatment sessions with the first ex-
Step 5B: perimental set; consequently, treatment was terminated
Therapist—“Good try, but not quite correct. Let’s with the first set and was extended to the second set of pic-
concentrate on this sound (underlines the “s” on tures for 14 sessions. For Participants 2, 3, and 4, 20 ses-
the sentence frame) and try again . . . spill milk.” sions were completed for each treatment phase.
Participant—“Spill milk.” All SLPs (and the clinical fellow) practiced adminis-
Therapist—“That’s right. Now, let’s say it three more tration of CAAST prior to use with participants. A treat-
times.” ment administration log was created to ensure that all steps
Step 6: of the CAAST protocol were employed accurately (see
Therapist—Removes the picture and imposes a 5-s Appendix F in the online supplemental materials). The log
delay (e.g., “Wait and then I’m going to ask you to contained therapist instructions for each of the six CAAST
say it again”). treatment steps and substeps along with a grid to mark
Participant—“Milk.” completion of each step. A log was used for every treatment
Therapist—“Good try, but not quite. Watch me and session, and all sessions were audio recorded.
try it with me . . . spill milk.” Ten percent of all treatment sessions were quasi-
randomly selected (balanced across SLPs) for calculation
Each of the 10 treatment pictures (i.e., pictures in an of accuracy of administration of treatment. An examiner
experimental set) was submitted to the CAAST treatment who had not administered the selected treatment session
hierarchy, one at a time in random order in each treatment used the audio recordings to determine accuracy of admin-
session. All six steps of the treatment hierarchy were com- istration of each step of the treatment protocol. The number
pleted with each picture. After presentation of the 10 pic- of times a treatment component was correctly adminis-
tures, the expanded productions were submitted to the SPT tered was determined and divided by the number of oppor-
step/substeps (Step 5) one more time if time permitted. As tunities for occurrences of that component. Accuracy for
shown in the full treatment protocol, CAAST provides var- administration of the following CAAST components was
iations in the substeps depending on the participant’s re- as follows:
sponses. Examples of elaborated responses produced during
CAAST are shown in Appendix E in the online supplemen- 1. presentation of treatment pictures (10 per session)—
tal materials. 100%;
Participants were scheduled for treatment three times 2. accurate elicitation and reinforcement for Steps 1–6
per week and were seen in their homes, a clinic setting, or (60 per session)—98%; and
a social rehabilitation day program according to their pref-
3. accurate elicitation of elaborated production repeti-
erence: Participant 1—home, Participant 2—home and tions (30 per session)—98%.
clinic, Participant 3—home, and Participant 4—home and
social rehabilitation day program. The conditions were as
quiet and free of distractions as possible. For all partici- Results
pants, there was a comfortable seating arrangement with
table space for picture presentation. If a family member Production of CIUs With Experimental Picture Sets
chose to observe a session (a rare occurrence), he or she sat Data representing number of CIUs produced during
out of the participant’s line of sight and was asked to not probes are shown in Figures 1–4 for Participants 1–4, re-
make comments to the participant. Treatment sessions (ex- spectively. Within each figure, the separate graphs repre-
cluding probes) ranged from approximately 60 to 75 min. sent production of CIUs in response to presentation of a
Treatment was administered by SLPs who were certi- particular picture set (two treated sets and one untreated
fied by the American Speech-Language-Hearing Association set). The order of the graphs from top to bottom in each
or by a doctoral student completing a clinical fellowship year figure indicates the order of application of treatment to
(supervised by a certified SLP at least 25% of the time). Each the sets. Graphed data for the picture sets reflect total
participant was assigned two SLPs (including the clinical number of CIUs produced for all 10 pictures in the set.
fellow) who shared treatment and testing responsibilities. The scaling of the y-axes varies by participant to accom-
As noted previously, each participant received treat- modate the range of number of CIU productions across
ment applied sequentially to two sets of experimental pictures. participants.
Order of administration of the sets was counterbalanced Determining that a systematic change in behavior
across participants. Treatment length was established prior was associated with treatment requires evaluation of the

Wambaugh et al.: CAAST 2197

Figure 1. Number of correct information units (CIUs) produced in response to experimental picture sets by Participant 1.

level, trend, and variability of probe data within and across entailed creating trend and mean (level) lines based on each
treatment phases. In addition, the immediacy of change, set of baseline data. These lines were then adjusted upward
the consistency of performance, and the degree of overlap by 0.25 SDs (standard deviation of each set of baseline data)
across phases must also be considered (Kratochwill et al., in the direction of the expected treatment effect. The ad-
2010). The CDC method (Fisher et al., 2003) was used to justed trend and level lines were then extended into each cor-
aid in interpretation of treatment effects. The CDC method responding treatment phase (for details of implementation

2198 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
Figure 2. Number of CIUs produced in response to experimental picture sets for Participant 2.

of the CDC method, see Fisher et al., 2003; Swoboda, prespecified number of data points falling above both lines;
Kratochwill, & Levin, 2010). In Figures 1–4, the level line the prespecified number of points is dependent on the num-
is represented by long dashed lines, and the trend line is ber of total probe points in the treatment phase (i.e., 8 data
represented by short dashed lines. With the CDC method, points in a 9- to 10-point treatment phase; 6 data points in
interpretation of positive treatment effects relies on a a 6- to 7-point treatment phase; Fisher et al., 2003). Type I

Wambaugh et al.: CAAST 2199

Figure 3. Number of CIUs produced in response to experimental picture sets for Participant 3.

error rates (i.e., erroneous attribution of a positive treat- Appendix G in the online supplemental materials. For the
ment effect) have been shown to be well controlled with the calculation of effect sizes, the initial baseline probe values for
CDC method even with high degrees of autocorrelation a given set (i.e., probes conducted in baseline prior to the
(Fisher et al., 2003). In addition, the CDC method has been application of any treatment) and the two follow-up probe
found to have higher power than the general linear model values for that set were used. The calculated effect sizes re-
and interrupted time series analysis. flect the cumulative effects of both phases of treatment
Effect sizes (d-index; Bloom, Fischer, & Orme, 2009; on the behavior (i.e., any generalization effects would be
Cohen, 1988) were calculated as indicators of the magni- included).
tude of change associated with treatment; d-index values To interpret the effect sizes, the following bench-
are shown on the individual graphs in Figures 1–4 and in marks suggested by Beeson and Robey (2006) were used:

2200 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
Figure 4. Number of CIUs produced in response to experimental picture sets for Participant 4.

d = 2.6, small; d = 3.9, medium; and d = 5.8, large. These have utility for interpreting the magnitude of the effect sizes
values were derived from effect sizes calculated by Robey, obtained for CAAST.
Schultz, Crawford, and Sinner (1999) as part of a review Participant 1. As shown in Figure 1, Participant 1
of single-subject design treatment investigations in aphasia received CAAST applied first to Set 1 and then to Set 3,
that included a predominance of studies focused on verbal with Set 2 remaining untreated. Baseline probe values
production. Consequently, the proposed benchmarks may ranged from 28 to 58 CIUs across the three sets. Following

Wambaugh et al.: CAAST 2201

application of CAAST to Set 1, CIU production for that 8 of 10 of the probe data points in the second treatment
set increased to a maximum of 173 CIUs (highest baseline phase fell above both lines. Thus, the CDC criteria were
value for Set 1 was 47 CIUs). Upon the completion of treat- met for both treatment phases, indicating that treatment
ment with Set 1, increases in CIU production were observed was associated with systematic behavioral change. A me-
for the untrained sets: Set 2—90 CIUs, and Set 3—103 CIUs. dium effect size of d = 5.41 was obtained for the first treated
Because of the increases in CIU production with Set 3, set, and a large effect size of d = 6.11 was obtained for the
which was slated for the second treatment application, ex- second treated set. A small effect size of 3.8 was found for
tended probing was completed to establish behavioral sta- the untreated set.
bility prior to treatment application. With six additional Participant 3. As shown in Figure 3, Participant 3 re-
probes (Probe Sessions 13–18), performance with Set 3 re- ceived CAAST applied first to Set 1 followed by treatment
mained somewhat variable, but it appeared to have reached a of Set 3. Across the seven initial baseline sessions, CIU pro-
ceiling of approximately 140 CIUs. When treatment was duction across sets ranged from 21 to 53. With application
applied to Set 3, additional gains were seen, with an in- of treatment, a maximum of 106 CIUs was produced with
crease to a maximum of 174 CIUs; however, performance Set 1. Increases in CIU production were also seen for the
remained variable. Participant 1 completed 14 treatment untreated sets at the end of the first phase of treatment. Par-
sessions with Set 3 but was unable to complete a final probe ticipant 3 experienced a hospitalization following the first
because of family issues (i.e., deaths in the family). Follow- treatment phase (nonneurological in nature), and a 1-month
up probes at 2 and 6 weeks posttreatment revealed that gap in probing occurred. Additional extended probing was
gains in CIU production were maintained well above initial conducted after he returned home to ensure stability of
baseline levels for both treated sets and the untreated set: At responding with Set 3 (the set designated for second treat-
6 weeks posttreatment, Set 1 = 114 CIUs, Set 3 = 135 CIUs, ment phase). With application of treatment to Set 3, CIU
and Set 2 = 75 CIUs. production initially decreased but then increased above
As shown in the top graph of Figure 1, all of the Set 1 the extended probing levels with increases being unstable.
probes during the treatment phase fell above both CDC lines, Follow-up probing revealed high, stable levels of CIU pro-
indicating a systematic behavior change associated with duction for both treated sets: Set 1 = 165 and 144 CIUs at
treatment. Because of the observed generalization with Set 3, 2 and 6 weeks, respectively, and Set 3 = 190 CIUs at both
the CDC lines were drawn using the extended probed data 2 and 6 weeks posttreatment. Gains in production of CIUs
following completion of treatment with Set 1 (i.e., second for the untreated set also remained at high levels post-
“A” phase). As can be seen in the second graph of Figure 1, treatment (Set 2 = 250 and 152 CIUs at 2 and 6 weeks post-
only three of the six Set 3 data points in the second treat- treatment, respectively).
ment phase fell above both CDC lines, suggesting a lack of The CDC criteria for demonstration of a treatment
treatment effect for this phase. Large effect sizes were ob- effect were met for the first treatment phase (10 of 10 points
tained for the two treated sets (d = 8.89, Set 1; d = 6.9, above both lines) but not the second treatment phase (3 of
Set 3), and a medium effect size was obtained for the un- 10 data points above both lines; see Figure 3). Large effect
treated set (d = 4.12, Set 2). As noted previously, the effect sizes of d = 15.52 and d = 21.63 were found for Sets 1 and
sizes represented the magnitude of change from the initial 3, respectively (both treated sets). A medium effect size
baseline probes. of d = 5.72 was obtained for the untreated set.
Participant 2. Participant 2 received treatment ap- Participant 4. Participant 4 received CAAST applied
plied first to Set 3 and then to Set 1 (see Figure 2). Over the sequentially to Sets 2 and 3 (see Figure 4). CIU production
six initial baseline sessions, number of CIUs ranged from 3 ranged from 6 to 20 CIUs across sets over the eight initial
to 23 across sets. Following application of CAAST to Set 3, baseline sessions. When treatment was applied to Set 2,
increases in CIU production were observed for that set, CIU production increased to a high of 48 CIUs for that set
with a maximum of 53 CIUs. For untrained sets, CIU pro- (baseline high of 15 CIUs). Treatment of Set 2 was also as-
duction increased above baseline levels for Set 1 but not for sociated with increases in CIU production for the untrained
Set 2. Extended probing was completed with Set 1 prior to sets, with increases to 26 CIUs and 28 CIUs for Sets 3 and
treatment of that set. The additional probing indicated sta- 1, respectively. Extended probing with Set 3 prior to appli-
ble responding ranging between 37 and 42 CIUs in the five cation of CAAST indicated stable performance ranging
probe sessions preceding application of CAAST with Set 1. from 19 to 26 CIUs. With application of CAAST to Set 3,
With application of treatment, CIU production for Set 1 in- increases to a high of 53 CIUs were seen. Upon completion
creased to a maximum of 58 CIUs. Substantial increases in of treatment with Set 3, additional increases were seen with
CIU production were also seen for the remaining untreated untreated Set 1. Follow-up probing indicated that gains
set (Set 2), with 45 CIUs produced at the end of treatment were maintained with all sets at levels well above baseline:
of Set 1. For all sets, number of CIUs remained well above At 6 weeks posttreatment, Set 2 = 50 CIUs, Set 3 = 56 CIUs,
initial baseline levels at 2 and 6 weeks posttreatment: At and Set 1 = 55 CIUs.
6 weeks posttreatment, Set 3 = 37 CIUs, Set 1 = 43 CIUs, For both treatment phases, 9 of 10 treatment phase
and Set 2 = 40 CIUs. probe data points fell above both CDC lines, indicating a
As can be seen in Figure 2, all of the probe data points systematic change in CIU production associated with treat-
in the first treatment phase fell above both CDC lines, and ment. Large effect sizes were obtained for both treated sets:

2202 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
Set 2, d = 13.64; and Set 3, d = 16.01. A medium effect size Single Word Intelligibility
was found for the untreated set (Set 1), with d = 5.79.
Pre- and posttreatment single word intelligibility
scores from the AIDS (Yorkston & Beukelman, 1981) are
Production of CIUs in Pre- and Posttreatment shown in Appendix H in the online supplemental materials.
Discourse Samples A 12% increase in intelligibility was seen for Participant 3,
but scores remained relatively unchanged for the other
Increased CIU production in narrative and proce-
participants.
dural discourse (Nicholas & Brookshire, 1993) was seen
for Participants 2 and 3 at posttreatment: Participant 2,
12 CIUs (pre) and 22 CIUs (post); Participant 3, 129 CIUs
(pre) and 260 CIUs (post). Participants 1 and 4 demon-
Discussion
strated decreases in CIU production in posttreatment nar- This study was conducted to determine whether
rative and procedural discourse: Participant 1, 279 CIUs CAAST was associated with improvements in verbal lan-
(pre) and 104 (post); Participant 4, 54 CIUs (pre) and guage and speech production in speakers with chronic
36 CIUs (post) (see Appendix H in the online supplemental aphasia and AOS. Increases in production of content were
materials). found for all participants, whereas gains in speech produc-
tion measures varied across participants.
The outcome measure chosen to measure language
PCC in Sentence Repetition production changes was number of CIUs produced in dis-
PCC values for pretreatment, interim, and posttreat- course; this measure was selected to allow comparison
ment sentence repetition samples are shown in Table 3. of the effects of CAAST with RET (Kearns, 1985) and
Values are presented separately for the sets of sentences, M-RET (Wambaugh & Martinez, 2000; Wambaugh et al.,
and the sets are designated as being presented “with” or 2012). Substantial increases in production of CIUs in re-
“without” printed stimuli. sponse to trained picture sets were found for all participants
Participant 1 demonstrated relatively high levels of in the current study. These increases were consistent with
accuracy in the three pretreatment baseline samples, with previous RET findings (Kearns, 1985; Kearns & Scher,
PCC ranging from 82% to 90% across the samples. After 1989; Kearns & Yedor, 1991) and M-RET findings
treatment was initiated, there were only two instances when (Wambaugh & Martinez, 2000; Wambaugh et al., 2012).
PCC was slightly higher than baseline levels. Benchmarks for effect sizes are not currently available
Participant 2 displayed low levels of accuracy in base- specifically for RET and M-RET. However, effect sizes
line samples, with PCC ranging between 7% and 17% at follow-up for treated sets for the current participants
for the auditory only sentences (“without”) and between (d-index = 5.41–21.63) were in keeping with those found for
12% and 24% for the auditory plus orthographic sentences M-RET (Wambaugh et al., 2012; d-index = 0.58–19.45).
(“with”). After the first phase of treatment, PCC increased Generalization to untreated picture sets was consis-
to 24% for the auditory only sentences, but this increase tently positive in the present study. Although generalization
was not sustained; PCC at follow-up periods approximated to untreated sets has been found to occur for the majority
baseline levels for the auditory only condition. For the of participants who have received RET or M-RET, gen-
auditory plus orthographic sentences, increased PCC values eralization has been found to be absent or limited in a few
were obtained at all sampling times following the initia- instances (see Wambaugh et al., 2012, for a summary).
tion of treatment. However, accuracy levels for follow-up Small effect sizes ranging from 3.8 to 5.79 were obtained
probes were only slightly higher than the highest baseline for the untrained set for the four participants who received
level (e.g., 26%). CAAST. In comparison, less than small effect sizes were
Participant 3’s PCC values ranged from 35% to 44% found for untrained sets for four of six participants who re-
for the auditory only sentences and from 39% to 51% for ceived M-RET (Wambaugh et al., 2012). The finding that
the auditory plus orthographic condition in baseline. Al- changes in CIU production for untreated picture sets was
though PCC values exceeded baseline for most sampling positive but consistently of lesser magnitude than treated
times for the auditory only sentences after treatment began, sets is consistent with Yedor et al.’s (1993) review of gener-
the 6-week follow-up PCC value approached baseline levels alization findings with RET; across RET participants, in-
(e.g., 46%). Increases were also observed for the auditory creases in untrained items were about 33% of increases with
plus orthographic sentences, and these gains were maintained trained items.
in follow-up samples (68% at 6 weeks posttreatment). Generalization of treatment effects to the narrative
In baseline samples, Participant 4’s auditory only and procedural discourse samples elicited using Nicholas
sentence PCC values ranged from 34% to 44%. Baseline and Brookshire’s (1993) procedure was positive for two of
PCC values for auditory plus orthographic sentences the four participants. This finding is in line with M-RET
ranged from 33% to 55%. Increases that were maintained findings in that Wambaugh et al. (2012) found posttreat-
at follow-up were seen for both sets of sentences: At 6 weeks ment increases in this condition for four of six participants.
posttreatment, auditory only PCC = 60%, and auditory However for the remaining two participants, decreases
plus orthographic PCC = 78%. in CIU production were found posttreatment. Repeated

Wambaugh et al.: CAAST 2203

 2204
Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014

Table 3. Percentage of consonants correct in sentence repetition.

Participant Condition BL 1 BL 2 BL 3 Midtreatment 1 Posttreatment 1 Midtreatment 2 Posttreatment 2 2 weeks 6 weeks

Participant 1 Without printed stimuli 87 86 82 82 87 88 CNT 82 83

With printed stimuli 89 90 87 91 86 90 CNT 93 89
Participant 2 Without printed stimuli 9 17 7 15 24 13 18 14 15
With printed stimuli 20 24 12 33 27 30 26 26 26
Participant 3 Without printed stimuli 35 39 44 58 40 47 61 51 46
With printed stimuli 39 51 48 55 54 69 68 60 68
Participant 4 Without printed stimuli 44 34 43 56 50 74 76 59 60
With printed stimuli 33 55 47 79 69 76 81 80 78

Note. Bold values are greater than the highest baseline (BL) value. CNT = could not test because participant was unavailable.
measurement with these stimuli prior to treatment would error is an important component of SPT; CAAST offered
be desirable in future investigations to elucidate within- fewer opportunities for practice than is typical with SPT.
participant variability with this measure (Cameron, This may be the reason why improvements in speech pro-
Wambaugh, & Mauszycki, 2010). Without such infor- duction were not found with all participants. However, it is
mation, gains or losses associated with treatment cannot unknown as to whether SPT would result in improvements
be reliably determined. in the speech production outcome measures chosen for this
Differences in participant characteristics, treatment study. That is, on the basis of our measures, we cannot state
variations, and study designs may account for slight differ- that the effects of CAAST on speech production differed
ences in CIU production findings across the various RET from SPT.
and M-RET reports and this investigation of CAAST. The speech outcome measures for CAAST may or
However, it appears that CAAST outcomes relative to CIU may not have reflected the practice that occurred during
production were quite similar to those of RET and M-RET. CAAST; an analysis of all of the productions and sound er-
An anonymous reviewer suggested that the increases rors that occurred during treatment is beyond the scope
in CIU production with untrained sets may have occurred of this initial report. PCC in repetition of sentences (which
because the sentences used in the PCC probes served as were predicted to be similar to those generated during
models for responding. This did not appear to be the case, CAAST) was chosen as a general measure of speech produc-
as PCC probes were completed prior to the final baseline tion accuracy. For the participants who did not demonstrate
picture description probe for Participants 1, 2, and 4; conse- change on this measure (Participant 1 and Participant 2),
quently, any benefit from these sentences should have been there may have been issues with stimuli difficulty that con-
apparent during baseline. For Participant 3, all PCC probes tributed to the lack of findings. Specifically, Participant 1
were completed after the final picture description probe. made relatively few errors on the sentences prior to treat-
However, there were no gains in CIU production in the first ment, leaving little room for improvement. Conversely,
two picture description probes following treatment applica- Participant 2 displayed a great deal of difficulty in repeat-
tion (after the PCC probes), which speaks against any effect ing sentences of seven to nine syllables. Although the PCC
from the PCC sentences. It was also suggested that use of measure was focused on speech production, the participants’
the CAAST sentence frame may have caused the partici- aphasia certainly played a role in ability to repeat accurately.
pants to adjust their responding in the picture probes. We Shorter sentences or phrases would likely have been more
argue that flexibility in responding is a goal of CAAST appropriate for Participant 2 and may have allowed improve-
(and of RET), and if a different approach to narrative dis- ments in PCC to be evident. Interpretation of the clinical
course occurred with treatment, this is compatible with a significance of changes in PCC requires further study.
treatment effect (note that no sentence frames are used with On the other hand, stimuli length may have not played
RET). a role in the outcomes. The second speech production out-
Speech production outcomes were not as robust as come measure, single word speech intelligibility, was clearly
the CIU outcomes for CAAST. With respect to accuracy of not impacted by the factor of length. However, this measure
consonant production in repeated sentences, clear improve- could also have been affected by the participants’ aphasia.
ments were seen with auditory only (without written stimuli) It would have been preferable if we had conducted repeated
and auditory + orthographic (with written stimuli) sentences administrations of the AIDS (Yorkston & Beukelman, 1981)
for two participants (Participant 3 and Participant 4). A to establish individual patterns of variability for the partici-
third participant demonstrated gains in repetition that were pants’ intelligibility scores. However, the most parsimonious
not maintained (Participant 2), and the fourth participant explanation for the nonrobust speech production outcomes
(Participant 1) did not exhibit any notable increases. An im- is that insufficient AOS treatment occurred.
provement in single word speech intelligibility was found Future development of CAAST will include additional
for only one of the participants (Participant 3). SPT trials with the participants’ self-generated utterances.
CAAST includes components of the apraxia treat- In addition to modifying the current outcome measures in
ment, SPT. As noted previously, SPT has typically been terms of difficulty (sentence repetition) and repeated admin-
applied to prespecified target items (e.g., specific sounds in istrations (intelligibility), other speech outcome measures
words or specific words with multiple sound targets). Con- will be explored. For example, intelligibility of self-generated
sequently, SPT effects have been measured in terms of im- utterances may be an appropriate outcome measure for
provement of targeted items (Wambaugh, 2004; Wambaugh CAAST.
& Mauszycki, 2010; Wambaugh & Nessler, 2004; Wambaugh, Additional language outcome measures will also be
Nessler, Cameron, & Mauszycki, 2013). With CAAST, the evaluated with CAAST. Analyses of language changes asso-
participant generates novel utterances, and, as such, there are ciated with RET and M-RET (beyond amount of content)
no prespecified targets for treatment. Therefore, direct com- have revealed qualitative improvements (e.g., increases in
parison of findings from the current study with previous novel content, increases in mean length of utterance, and in-
SPT findings is not possible. We had been concerned that creases in phrase and sentence production) for some partici-
there might not be sufficient opportunities for instruction pants. Given the increases in CIUs found with CAAST and
and practice with erroneous productions with CAAST. That the overlap of treatment techniques with RET and M-RET,
is, repeated practice with productions that are frequently in similar qualitative changes would be expected with CAAST.

Wambaugh et al.: CAAST 2205

 Design issues relative to the generalization effects Druks, J., & Masterson, J. (2000). An Object and Action Naming
found with CAAST also warrant attention in future investi- Battery. Philadelphia, PA: Taylor & Francis.
gations. That is, generalization to untrained behaviors may Duffy, J. R. (2013). Motor speech disorders. Substrates, differential
diagnosis, and management (3rd ed.). St. Louis, MO: Elsevier
result in difficulty in demonstrating experimental control
Mosby.
through replication of treatment effects. In the current in- Fisher, W. W., Kelley, M. E., & Lomas, J. E. (2003). Visual aids
vestigation, we were able to reestablish stable (albeit higher) and structured criteria for improving visual inspection and in-
extended baselines with untreated behaviors prior to apply- terpretation of single-case designs. Journal of Applied Behavior
ing the second phase of treatment. For two participants Analysis, 36, 387–406.
(Participant 2 and Participant 4), treatment effects with the Gaddie, A., Kearns, K. P., & Yedor, K. (1991). A qualitative
second treated behavior were clearly demonstrated (accord- analysis of Response Elaboration Training effects. Clinical
ing to CDC criteria). For Participant 1 and Participant 3, Aphasiology, 19, 171–183.
although behavioral change appeared to occur with the sec- Kearns, K. P. (1985). Response Elaboration Training for patient
ond treatment application, changes did not meet CDC cri- initiated utterances. In R. H. Brookshire (Ed.), Clinical apha-
siology (pp. 196–204). Minneapolis, MN: BRK.
teria, which reduced the experimental control afforded by
Kearns, K. P., & Scher, G. P. (1989). The generalization of Re-
the MBD-behaviors design. Because generalization was sponse Elaboration Training effects. Clinical Aphasiology,
anticipated, we included the MBD-participants design to 18, 223–245.
provide additional experimental control. With baselines ex- Kearns, K. P., & Yedor, K. (1991). An alternating treatments com-
tended across participants, behavioral change did not occur parison of loose training and a convergent treatment strategy.
until treatment was implemented in all cases. However, the Clinical Aphasiology, 20, 223–238.
increase of one additional probe per participant is acknowl- Kertesz, A. (2007). The Western Aphasia Battery—Revised.
edged as being minimal. In future investigations, it would San Antonio, TX: Pearson.
be preferable to implement the MBD-participants design Kratochwill, T. R., Hitchcock, J., Horner, R. H., Levin, J. R.,
using more rigorous extensions of baselines across partici- Odom, S. L., Rindskopf, D. M., & Shadish, W. R. (2010).
Single-case designs technical documentation. Retrieved from
pants (e.g., increases of 2–3 probe sessions per participant).
What Works Clearinghouse website: http://ies.ed.gov/ncee/
As in the development of any new treatment, there wwc/pdf/wwc_scd.pdf
are many aspects of CAAST that require investigation. On Mauszycki, S. C., & Wambaugh, J. L. (2008). The effects of rate
the basis of the current findings, it appears that combining control treatment on consonant production accuracy in mild
treatment techniques from RET (and M-RET) with those apraxia of speech. Aphasiology, 22, 906–920.
of SPT did not adversely affect outcomes in terms of CIU McNeil, M. R., Robin, D. A., & Schmidt, R. A. (1997). Apraxia
production. Although not all participants demonstrated im- of speech: Definition, differentiation, and treatment. In
provements in speech production, it appears that CAAST M. R. McNeil (Ed.), Clinical management of sensorimotor
may also have speech production benefits for some par- speech disorders (pp. 311–344). New York, NY: Thieme.
ticipants. However, additional speech production practice/ McNeil, M. R., Robin, D. A., & Schmidt, R. A. (2009). Apraxia
of speech: Definition, differentiation, and treatment. In
training may be necessary to effect consistent positive
M. R. McNeil (Ed.), Clinical management of sensorimotor
changes in speech outcome measures. speech disorders (2nd ed., pp. 249–268). New York, NY:
Thieme.
Nicholas, L. W., & Brookshire, R. H. (1993). A system for quan-
Acknowledgments tifying the informativeness and efficiency of the connected
speech of adults with aphasia. Journal of Speech and Hearing
This research was supported by the Department of Veterans
Affairs, Rehabilitation Research and Development. Thanks are Research, 36, 338–350.
extended to Catharine DeLong for her assistance with this project. Porch, B. (2001). Porch Index of Communicative Ability: Administra-
tion, scoring and interpretation (Vol. 2, 4th ed.). Albuquerque,
NM: PICA Programs.
Robey, R. R., Schultz, M. C., Crawford, A. B., & Sinner, C. A.
References (1999). Single-subject clinical outcome research: Designs, data,
Beeson, P. M., & Robey, R. R. (2006). Evaluating single-subject effect sizes, and analyses. Aphasiology, 13, 445–473.
treatment research: Lessons learned from the aphasia literature. Swoboda, C. M., Kratochwill, T. R., & Levin, J. R. (2010). Con-
Neuropsychology Review, 16, 161–169. servative dual-criterion method for single-case research: A guide
Bloom, M., Fischer, J., & Orme, J. G. (2009). Evaluating practice for visual analysis of AB, ABAB, and multiple-baseline designs
—Guidelines for the accountable professional (6th ed.). Boston, (WCER Working Paper No. 2010-13). Retrieved from Univer-
MA: Pearson Higher Education. sity of Wisconsin—Madison, Wisconsin Center for Education
Brown, L., Sherbenou, R. J., & Johnsen, S. K. (2010). Test of Non- Research website: http://www.wcer.wisc.edu/publications/
verbal Intelligence—4. Austin, TX: Pro-Ed. workingPapers/papers.php
Cameron, R. C., Wambaugh, J. L., & Mauszycki, S. C. (2010). In- Wambaugh, J. L. (2004). Stimulus generalization effects of Sound
dividual variability on discourse measures over repeated sam- Production Treatment for apraxia of speech. Journal of Medi-
pling times in persons with aphasia. Aphasiology, 24, 671–684. cal Speech Language Pathology, 12, 77–97.
Cohen, J. (1988). Statistical power analysis for the behavioral sci- Wambaugh, J. L., Duffy, J. R., McNeil, M. R., Robin, D. A., &
ences (2nd ed.). Hillsdale, NJ: Erlbaum. Rogers, M. (2006). Treatment guidelines for acquired apraxia
Dabul, B. L. (2000). Apraxia Battery for Adults—Second Edition. of speech: A synthesis and evaluation of the evidence. Journal
Austin, TX: Pro-Ed. of Medical Speech Language Pathology, 14, xv–xxxiii.

2206 Journal of Speech, Language, and Hearing Research • Vol. 57 • 2191–2207 • December 2014
Wambaugh, J. L., Kalinyak-Fliszar, M. M., West, J. E., & Doyle, Wambaugh, J. L., Nessler, C., Wright, S., & Mauszycki, S. C.
P. J. (1998). Effects of treatment for sound errors in apraxia (2013). Sound Production Treatment: Effects of blocked and
of speech. Journal of Speech, Language, and Hearing Research, random practice. American Journal of Speech-Language
41, 725–743. Pathology, 23, S225–S245. doi:10.1044/2014_AJSLP-13-0072
Wambaugh, J. L., & Martinez, A. L. (2000). Effects of modified Wambaugh, J. L., & Shuster, L. I. (2008). Nature and manage-
Response Elaboration Training with apraxic and aphasic ment of neuromotor speech disorders accompanying aphasia.
speakers. Aphasiology, 14, 603–617. In R. Chapey (Ed.), Language intervention strategies in aphasia
Wambaugh, J. L., & Mauszycki, S. C. (2010). Sound Produc- and related neurogenic communication disorders (5th ed.,
tion Treatment: Application with severe apraxia of speech. pp. 1009–1042). Philadelphia, PA: Lippincott, Williams, &
Aphasiology, 24, 814–825. Wilkins.
Wambaugh, J. L., & Nessler, C. (2004). Modification of Sound Wambaugh, J. L., West, J. E., & Doyle, P. J. (1998). Treatment
Production Treatment for apraxia of speech: Acquisition and for apraxia of speech: Effects of targeting sound groups.
generalisation effects. Aphasiology, 18, 407–427. Aphasiology, 12, 731–743.
Wambaugh, J. L., Nessler, C., Cameron, R., & Mauszycki, S. Wambaugh, J. L., Wright, S., & Nessler, C. (2012). Modified
(2013). Treatment for acquired apraxia of speech: Examina- Response Elaboration Training: A systematic extension with
tion of treatment intensity and practice schedule. American replications. Aphasiology, 26, 1407–1439.
Journal of Speech-Language Pathology, 22, 84–102. Yedor, K. E., Conlon, C. P., & Kearns, K. P. (1993). Measure-
Wambaugh, J. L., Nessler, C., & Wright, S. (2013). Application ments predictive of generalization of Response Elaboration
of Response Elaboration Training to procedural discourse. Training. Clinical Aphasiology, 21, 213–223.
American Journal of Speech-Language Pathology, 22, Yorkston, K. M., & Beukelman, D. R. (1981). Assessment of Intel-
S409–S425. ligibility of Dysarthric Speech. Austin, TX: Pro-Ed.

Wambaugh et al.: CAAST 2207

Copyright of Journal of Speech, Language & Hearing Research is the property of American
Speech-Language-Hearing Association 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.