26
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=paph20 Download by: [University of Oregon] Date: 25 March 2016, At: 17:05 Aphasiology ISSN: 0268-7038 (Print) 1464-5041 (Online) Journal homepage: http://www.tandfonline.com/loi/paph20 Treatment of Underlying Forms in a discourse context Laura Murray PhD , Anne Timberlake & Rebecca Eberle To cite this article: Laura Murray PhD , Anne Timberlake & Rebecca Eberle (2007) Treatment of Underlying Forms in a discourse context, Aphasiology, 21:2, 139-163, DOI: 10.1080/02687030601026530 To link to this article: http://dx.doi.org/10.1080/02687030601026530 Published online: 20 Feb 2007. Submit your article to this journal Article views: 489 View related articles Citing articles: 7 View citing articles

context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=paph20

Download by: [University of Oregon] Date: 25 March 2016, At: 17:05

Aphasiology

ISSN: 0268-7038 (Print) 1464-5041 (Online) Journal homepage: http://www.tandfonline.com/loi/paph20

Treatment of Underlying Forms in a discoursecontext

Laura Murray PhD , Anne Timberlake & Rebecca Eberle

To cite this article: Laura Murray PhD , Anne Timberlake & Rebecca Eberle (2007)Treatment of Underlying Forms in a discourse context, Aphasiology, 21:2, 139-163, DOI:10.1080/02687030601026530

To link to this article: http://dx.doi.org/10.1080/02687030601026530

Published online: 20 Feb 2007.

Submit your article to this journal

Article views: 489

View related articles

Citing articles: 7 View citing articles

Page 2: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

# 2007 Psychology Press, an imprint of the Taylor & Francis Group, an informa business

http://www.psypress.com/aphasiology DOI: 10.1080/02687030601026530

Treatment of Underlying Forms in a discourse context

Laura Murray, Anne Timberlake and Rebecca Eberle

Indiana University, Bloomington, IN, USA

Background: Previous research indicates that Thompson and colleagues’ (Thompson,2001; Thompson & Shapiro, 2005) Treatment of Underlying Forms (TUF) canefficiently remediate agrammatic sentence-processing deficits. The theoretical basis ofTUF is that training production of complex, noncanonical sentence structures canconcomitantly improve production of untrained, syntactically related but simplersentence structures. Whereas this generalisation to untrained syntactic forms has beenwell established within constrained, sentence-level tasks, which exploit the sameresponse modality used during training, TUF’s generalisation potential in terms ofcross-modal effects and discourse-level improvements requires further exploration.Aims: DM, a 52-year-old male with an agrammatic Broca’s aphasia profile, wasprovided with a modified version of TUF, which targeted his writing skills and includeda Discourse Training Module that allowed direct rehearsal of targeted syntactic frameswithin a discourse context. The hypotheses tested were as follows: (a) DM wouldimprove his written production of trained sentence structures and demonstrategeneralisation to untrained exemplars of targeted sentence structures as well asuntrained, syntactically related syntactic structures; (b) written sentence productiontreatment would facilitate gains in DM’s spoken production of trained and related,untrained sentence structures; and (c) DM would exhibit improved sentence productionabilities in discourse post-treatment.Methods & Procedures: A single subject, multiple baseline across-behaviours design wasimplemented to evaluate acquisition of trained sentence types (object- and subject-extracted embedded who-question sentences), to discern generalisation to untrainedsentence types (object- and subject-extracted matrix questions, passives) and discourse,or both, and to identify maintenance of treatment effects. Each week, DM completedtwo 90-minute sessions of modified TUF as well as written sentence productionhomework.Outcomes & Results: DM displayed a pattern of sentence acquisition typical of TUFrecipients, generalising gains in complex sentence production to the production ofuntrained, less complex, theoretically related structures. Gains in written productiongeneralised to spoken production of the same structures, and improvements acrosspredominately pragmatic versus morphosyntactic discourse variables were also noted.Conclusions: The treatment outcomes of a modified, written version of TUF werecomparable to those in previous studies (e.g., Ballard & Thompson, 1999), andindicated that training written sentence production can evoke substantial cross-modalgeneralisation to speech. Despite inclusion of a Discourse Training Module, pragmaticversus morphosyntactic aspects of DM’s discourse showed most improvement.Therefore, continued investigation of TUF is recommended to determine whether itcan efficiently treat structural aspects of discourse production, or what modificationswill ensure generalisation to discourse contexts in a broader spectrum of aphasicpatients.

Address correspondence to: Laura Murray PhD, Department of Speech and Hearing Sciences, Indiana

University, 200 S. Jordan Avenue, Bloomington, IN 47405, USA. E-mail: [email protected]

APHASIOLOGY, 2007, 21 (2), 139–163

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 3: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

A growing literature has established that Treatment of Underlying Forms (TUF;

previously called Linguistic Specific Treatment) represents an efficient approach to

remediating sentence-processing deficits in individuals with agrammatic Broca’s

aphasia (Ballard & Thompson, 1999; Jacobs & Thompson, 2000; Thompson,

Ballard & Shapiro, 1998; Thompson et al., 1997; Thompson, Shapiro, Kiran, &

Sobecks, 2003; Thompson, Shapiro, Tait, Jacobs, & Schneider, 1996). Given the

lexical (e.g., difficulties using verbs with complex argument structures) and syntactic

symptoms (e.g., problems in processing sentences with movement or a noncanonicalform) frequently associated with agrammatism (Grodzinsky, 1995; Kim &

Thompson, 2004; Mauner, Fromkin, & Cornell, 1993; Thompson et al., 1997),

TUF consists of three principal constituents: (a) practising, in a sentence context, the

identification of the verb, its argument structures, and the thematic roles of the

various arguments; (b) explicitly training the movement operations required to

transform canonical sentences, or sentences in deep structure, into their non-

canonical or surface structure counterparts; and (c) facilitating integration of

thematic roles and movement operations by requiring production of the targeted,noncanonical structures. Based on early TUF outcomes (e.g., Thompson et al., 1997)

as well as data generated from treatment protocols for other linguistic (e.g., Gierut,

2001; Kiran & Thompson, 2003) and motor speech disorders (e.g., Maas, Barlow,

Robin, & Shapiro, 2002), TUF additionally prescribes that training should begin

with more rather than less complex sentence structures to enhance generalisation (for

a complete review of TUF and its theoretical underpinnings, see Thompson, 2001;

Thompson & Shapiro, 2005).

Indeed, Thompson and her colleagues have consistently documented thatagrammatic individuals who receive TUF display improved production of not only

the trained, complex sentence structures but also novel sentence structures that are

syntactically related to and less complex than the trained structures (e.g., Ballard &

Thompson, 1999; Thompson et al., 1998, 2003). Whereas this robust generalisation

to untrained sentence types is encouraging, to evaluate the breadth of TUF’s clinical

efficiency additional exploration of its generalisation potential is required. That is,

for the most part, TUF studies have examined generalisation by probing patients’

production or comprehension of untrained structures within constrained, sentence-level tasks, which exploit the same response modality used during training.

Accordingly, further research is needed to establish the extent to which TUF can

foster generalisation to not only untrained language modalities but also less

structured, communicative contexts such as discourse.

Only two prior investigations have thus far examined TUF’s cross-modal

generalisation effects. Jacobs and Thompson (2000) found that for all four of their

patients with agrammatic aphasia, spoken production treatment facilitated

improvements in writing trained sentence types; additionally, all patients maintainedwriting accuracies above baseline levels during follow-up testing. Murray, Ballard,

and Karcher (2004) also examined whether spoken TUF training would enhance

written sentence production skills. In their study, however, writing was treated

indirectly through weekly written homework assignments. Despite these homework

activities, their patients displayed only modest improvements in writing trained

sentence types. These less positive writing outcomes most likely reflected that

patients in this study were not merely agrammatic, but also displayed characteristics

consistent with other aphasia profiles (e.g., lexical-semantic access problems;graphemic output deficits), which no doubt confounded their writing progress. In

140 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 4: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

light of these limited data, additional exploration of TUF’s cross-modal effects is

needed not only to evaluate the reliability of Jacobs and Thompson’s (2000) positive

findings, but also to determine whether cross-modal gains can extend to linguistically

related untrained sentence types or discourse contexts.

In all previous TUF studies in which sentence production was treated, the trained

response modality has been speech. This focus on verbal output is consistent with the

aphasia treatment literature at large, in which spoken language protocols vastly

outnumber those available for remediating writing problems. This is unfortunate,given that many patients with aphasia must exploit writing to augment or replace

their speech because writing is their most intact output modality, their spoken

language deficits are resilient to treatment, or concomitant motor speech symptoms

limit their intelligibility (Beeson, Rising, & Volk, 2003; Murray & Karcher, 2000;

Robson, Marshall, Chiat, & Pring, 2001). Of the available, empirically evaluated,

writing protocols, few address writing beyond the isolated word level. Exceptions

include a few studies in which writing was trained by focusing on verb retrieval as

well as the construction of a grammatical frame, including verb morphologicalmarkers (Mitchum, Haendiges, & Berndt, 1993; Murray & Karcher, 2000);

concomitant changes in spoken language were also monitored. Subsequent to these

treatments, patients with agrammatic symptoms demonstrated improvements in the

syntactic and lexical content of their verbal output, although these speech gains were

not as robust as the writing improvements (e.g., longer utterances, improved open-

to closed-class word ratios, larger proportions of grammatical utterances). In sum,

this limited evidence suggesting that sentence production training in one modality

can generalise to the production of equivalent structures in another certainly meritsfurther investigation.

As previously mentioned, the extent to which TUF can positively affect discourse

skills has not yet been fully established. Determining the discourse-level effects of

TUF is important because discourse taps every linguistic function (i.e., phonology,

morphology, syntax, semantics, pragmatics; Cherney, 1998; Jacobs, 2001). Discourse

measures therefore provide insight into the integrated functioning of those linguistic

operations targeted by TUF within the linguistic system as a whole in a more

flexible, responsive, and naturalistic framework than that afforded by the sentence-level probes utilised within the TUF protocol. Furthermore, changes in discourse

performance have been found to predict listener judgements of meaningful social

change in the recovery of aphasic adults’ communication abilities (Ross & Wertz,

1999).

To date, however, only a limited number of TUF investigations have included

discourse measures (Ballard & Thompson, 1999; Jacobs & Thompson, 2000;

Thompson et al., 1996, 1997, 2003). In each of these studies, the structural analysis

method of Thompson and colleagues (Thompson et al., 1995) was applied toevaluate morphosyntactic (e.g., proportion of grammatical utterances; mean number

of embeddings per utterance) and lexical changes (i.e., open- to closed-class word

ratios; verb and verb argument use) in spoken discourse. Data from these

investigations indicate that facilitation of structural aspects of discourse can occur

subsequent to TUF, but that the nature and degree of these generalisation effects

vary substantially across studies and research participants. For example, after

providing TUF to seven agrammatic patients, Thompson et al. (1996) found

statistically significant improvements in two of five morphosyntactic measures: theproportions of simple and complex utterances. Whereas 11 of 18 measures of verb

TUF IN A DISCOURSE CONTEXT 141

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 5: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

use and verb argument structure demonstrated significant positive change, these

improvements were less consistent across individual study participants than those

observed in sentence complexity. In contrast, only three of four participants in a later

study displayed modest gains in two of the five morphosyntactic variables analysed,

mean utterance length and proportion of complex sentences (Jacobs & Thompson,

2000), and none of Ballard and Thompson’s (1999) five participants demonstrated

substantial improvement on any structural variable.

More consistent and substantial generalisation effects have been identified whenpragmatic discourse analyses have been applied. Using Nicholas and Brookshire’s

(1993) system to calculate correct information units (CIUs), intelligible words that

are accurate and informative within the given communicative context, Ballard and

Thompson (1999) found that following TUF, four of their five agrammatic

participants demonstrated increased informativeness (i.e., CIUs/min), efficiency (i.e.,

%CIUs or proportion CIUs to total words), or both in their spoken discourse, even

though, as previously mentioned, none experienced similar improvements in

morphosyntactic or lexical aspects of their discourse. Likewise, by reanalysing thespoken discourse samples of five agrammatic individuals who had participated in

two previous TUF studies (Jacobs & Thompson, 2000; Thompson et al., 1996),

Jacobs (2001) found statistically significant increases in CIUs/min, and a generally

positive, though statistically insignificant, trend towards increased %CIUs.

Additionally, both Ballard and Thompson (1999) and Jacobs (2001) included

measures of social validity in which naive listeners rated their participants’ dis-

course samples on perceptual constructs like coherence, fluency, grammaticality,

effectiveness, and informativeness. Higher ratings of the post-treatment dis-course samples indicated that the pragmatic improvements described above were

perceptible.

In summary, although TUF has the capacity to engender positive discourse

changes in agrammatic patients, these effects, particularly for structural aspects of

discourse, have thus far been neither particularly large nor consistent across

individuals. Part of this is due, no doubt, to heterogeneity in the TUF participants

who have varied in terms of the severity of their agrammatism as well as the presence

and severity of concomitant cognitive, linguistic, and motor speech symptoms:Individuals with more compromised communicative or cognitive abilities tend to

demonstrate less positive TUF and, relatedly, discourse effects. Discourse general-

isation effects also might be modulated, however, by discourse sampling procedures.

For instance, even though it is frequently recommended to sample a variety of

discourse genres (e.g., procedural, narrative, descriptive) because of the different

cognitive and linguistic demands of each genre (Cherney, 1998; Li, Ritterman, Della

Volpe, & Williams, 1996; Shadden, Burnette, Eikenberry, & DiBrezzo, 1991), most

previous TUF studies have examined only narrative samples (Ballard & Thompson,1999; Jacobs, 2001; Thompson et al., 1997, 1998, 2003). Thus more, or possibly less,

robust generalisation effects might be observed if samples from additional discourse

tasks were analysed.

Generalisation to discourse might be additionally enhanced by modifying TUF so

that treatment exercises extend beyond constrained, sentence-level practice and

target connected discourse more directly. For example, embedding syntactic

production practice within a discourse framework would allow rehearsal of targeted

linguistic structures or operations within a more communicative, rather than merelyproductive, context. Increasing the communicative value of treatment activities

142 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 6: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

should increase patient motivation and interest, particularly if the ‘‘message’’ or

communicative content is, at least to an extent, left to the patient’s discretion.

Indeed, research within the social psychology literature suggests that choice, or the

degree to which individuals perceive an act to be of their own volition, can impact

one’s commitment level (Freedman & Steinbruner, 1964; Kiesler & Sakamura, 1966).

Likewise, some extant aphasia treatments, such as Response Elaboration Training

(Kearns, 1985), have elicited positive discourse outcomes by emphasising client-

initiated responses as the primary content of therapy.Retention of syntactic movement procedures also might be enhanced if learning

occurred not only within the context of the 10 to 20 sentences used in traditional

TUF, but also in a richer, more meaningful discourse environment. Craik and

Lockhart’s (1972) Levels of Processing Hypothesis suggests that words are more

effectively encoded when learned in context versus isolation; whereas syntactic

movement operations are typically viewed as more ‘‘procedural’’ in nature, at least

prior to brain damage (Lewis & Vasishth, 2005), practising these operations in a

larger context could facilitate recall. Relatedly, structured discourse activities such asdescribing pictures or retelling stories, although still removed from many day-to-day

communicative functions, appear more naturalistic than the TUF protocol in which

a sentence production priming paradigm is used to train a limited set of sentence

stimuli. As Davis and Wilcox (1985) proposed, the more a treatment environment is

reflective of natural contexts and behaviour, the more likely it is that generalisation

of treated behaviours to extra-therapeutic environments will occur. Further,

introducing heterogeneity into treatment activities and stimuli can better mimic

the variability of real-life communicative environments. Although therapy cannotpossibly prepare patients adequately for every novel situation, practice in dealing

with novelty itself might better promote generalisation of new skill sets.

Some initial research suggests that discourse-level treatments may prove effective

for remediating agrammatism. For example, Peach and Wong (2004) constructed a

story-retelling therapeutic paradigm to target syntax directly. Their agrammatic

participant was told a short fable and then asked to retell and elaborate upon it; a

clinician transcribed the participant’s story and then invited him to revise each

utterance, afterwards providing corrective syntactic feedback. Peach and Wongproposed that story retelling not only entails a linguistic response, but also places

demands on cognitive processes; that is, their participant was forced not just to

produce a sentence, but also to extrapolate, integrate, distil, and sequence conceptual

information into a series of novel syntactic constructions. With treatment, the

participant displayed increased utterance grammaticality and complexity during the

story-retell task, and substantial gains on expressive subtests of an aphasia battery.

However, Peach and Wong did not assess whether improvements in retelling the

fables targeted in treatment were present in novel discourse samples. Thus, thegeneralisation effects of this discourse-based approach have not yet been adequately

explored.

Accordingly, to extend the generalisation effects of TUF, in the current study,

DM, an individual with chronic, agrammatic aphasia, received a modified version of

TUF that included traditional TUF procedures, an appended Discourse Training

Module, and homework activities. Because of DM’s facility with writing as well as

his diagnosis of apraxia of speech, TUF was additionally adapted to target written

sentence production. The goals of the present research were threefold. First, givenTUF’s previously documented treatment and generalisation effects (e.g., Thompson

TUF IN A DISCOURSE CONTEXT 143

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 7: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

et al., 2003), we hypothesised that DM would: (a) improve his writing of trained

sentence structures, requiring both embedding and wh-movement; (b) demonstrate

generalisation to untrained exemplars of targeted sentence structures and to related,

but less complex, syntactic structures (e.g., wh-movement only) containing trained or

untrained constituents; and (c) demonstrate no generalisation to sentences with noun

phrase versus wh-movement. Second, because some cross-modal generalisation

effects have been reported (e.g., Jacobs & Thompson, 2000), we hypothesised that

TUF training in writing would facilitate gains in DM’s verbal output. Lastly, given

our modifications to the TUF paradigm (i.e., inclusion of the Discourse Training

Module to target trained syntactic structures within a discourse context), we

hypothesised that DM would exhibit improved sentence production abilities in

discourse post-treatment.

METHOD

Participant

DM is a 52-year-old, right-handed, monolingual English-speaking male with

chronic, agrammatic aphasia and apraxia of speech due to a left hemisphere stroke

suffered in February 2003. Whereas prior to his stroke he was self-employed as an

engineering consultant, he has as yet been unable to return to work because of his

communication limitations. Prior to the onset of the research project, DM had

received 2 months of individual speech-language therapy, which primarily targeted

remediating his apraxia of speech difficulties and encouraging use of multiple

communication modes. He had also been attending a weekly aphasia support group

for approximately 6 months. These group sessions constituted an important source

of social stimulation for DM, and focused on facilitating social interaction between

group members through the use of multimodal communication strategies. Because

these sessions did not address syntax or the pragmatic use thereof, and because DM

had been in consistent attendance for some time pre-treatment and clearly valued the

opportunity to participate in the group, his withdrawal from aphasia support group

during the research project was considered unnecessary and inappropriate.

When DM was approximately 2 years post-stroke, a pre-treatment test battery

was administered to evaluate his sensory, linguistic, cognitive, and motor speech

abilities (see Table 1). DM passed pure tone hearing and vision (i.e., a picture-

matching task) screenings, indicating that neither basic auditory nor visual problems

were significantly contributing to his current communicative or cognitive difficulties.

His profile on the Western Aphasia Battery (WAB; Kertesz, 1982) was consistent

with moderately severe Broca’s aphasia. That is, his spoken and written output was

characterised by non-fluent, agrammatic utterances and almost exclusively limited to

single word utterances. His repetition was impaired, but he demonstrated only

modest difficulty on spoken and written confrontation naming subtests. Whereas

DM’s auditory comprehension appeared relatively intact, his reading comprehension

declined as stimulus length and complexity increased.

To further examine DM’s syntactic abilities, the Verb and Sentence Test (VAST;

Bastiaanse, Edwards, & Rispens, 2002), Northwestern Sentence Comprehension Test

(NST; Thompson, 1996), and Reading/Grammar subtest of the Test of Adolescent

and Adult Language (TOAL; Hammill, Brown, Larsen, & Wiederholt, 1994) were

administered. DM’s performance on the comprehension subtest of VAST, which

144 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 8: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

assesses auditory comprehension of verbs and active, passive, and clefted sentence

structures, was near perfect. In contrast, he displayed significant difficulties on

VAST production tasks, particularly items designed to elicit passive or reversible

sentences or wh-questions. Although he performed VAST’s Action Naming subtest

TABLE 1Summary of pre-treatment testing

Test Raw score

Western Aphasia Battery

Spontaneous Speech Composite 11/20

Comprehension Composite 200/200

Repetition Composite 49/100

Naming Composite 81/100

Reading 78/100

Writing 56/100

Aphasia Quotient 68.0

Verb and Sentence Test

Verb and Sentence Comprehension 117/120

Action Naming 31/40

Sentence Completion 11/20

Sentence Construction 0/20

Sentence and Wh-Anagrams 42/60

Northwestern Sentence Comprehension Test

Active 20/20

Passive 19/20

Subject Relative 18/20

Object Relative 20/20

TOAL Reading and Grammar Subtest 0/25

DCT Silent Reading Comprehension (3 stories) 24/24

Working Memory Protocol

True/false 41/42

Word recall 23/42

Test of Everyday Attention Scaled score

Map search one minute 41 12

Map search two minute 64 10

Elevator counting with distraction 10 12

Visual elevator (Accuracy) 7 7

Visual elevator (Timing) 19.0 0

Elevator counting with reversal 6 11

Telephone search 4.3 7

Telephone Search while counting 11.4 4

Apraxia Battery for Adults – 2

Diadochokinetic rate 7 (mild impairment)

Increasing word length-a 4 (mild impairment)

Increasing word length-b 3 (moderate impairment)

Utterance time for polysyllabic words 31 (mild impairment)

Repeated trials 7 (moderate impairment)

TOAL 5 Test of Adolescent and Adult Language; DCT 5 Discourse Comprehension Test; Scaled Score

on the Test of Everyday Attention has M 5 10, SD 5 3 based on a sample of 154 non-brain-damaged

adults.

TUF IN A DISCOURSE CONTEXT 145

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 9: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

fairly well, he had difficulty with subtests that involved completing a sentence with a

target verb. His most common production errors on the VAST included omitting or

misplacing auxiliary verbs (e.g., ‘‘Man catching soccer’’ for ‘‘The boy is catching the

ball’’ on the Sentence Construction subtest; ‘‘What the farmer pushing is in the

stable?’’ for ‘‘What is the farmer pushing in the stable?’’ on the Wh-Anagrams

subtest). As on the VAST, DM displayed nominal difficulty comprehending active,

passive, subject-relative, or object-relative sentences on the NST. However, a

different comprehension profile was observed on the TOAL Reading/Grammarsubtest, on which DM had to identify two sentences with identical meanings but

different syntactic frames; he immediately reached ceiling level, failing to answer

correctly any of the first four questions. In contrast, on the Discourse Comprehension

Test (DCT; Brookshire & Nicholas, 1997), which required extracting explicitly stated

or implied main ideas and incidental details from a written narrative, his

performance was excellent, suggesting that DM’s difficulties with reading

comprehension were, at least in part, syntactic in nature.

An auditory verbal working memory test (Tompkins, Bloise, Timko, &Baumgaertner, 1994) and the Test of Everyday Attention (TEA; Robertson, Ward,

Ridgeway, & Nimmo-Smith, 1994) were given to examine DM’s memory and

attention abilities, respectively. His working memory test performance indicated

significant difficulty with memory tasks involving language; that is, the number of

recall errors made by DM fell more than two standard deviations above the mean for

healthy adults (M 5 6.4, SD 5 4.6; Tompkins et al., 1994) and slightly above the

mean for individuals with left hemisphere brain damage (M 5 16.8, SD 5 10.8). On

the TEA, he performed well tasks that primarily stressed basic or auditory attentionabilities (e.g., Map Search, Elevator Counting, Elevator Counting with Reversal),

but displayed difficulty on subtests with a timed component (e.g., Visual Elevator,

Telephone Search While Counting). The Apraxia Battery for Adults – 2 (ABA;

Dabul, 2000) was administered to confirm the presence and characterise the severity

of apraxia of speech. DM’s ABA performance indicated that he had apraxia of

speech that was considered mild to moderate in severity.

Prior to treatment a total of four 5-minute discourse samples (i.e., one spoken and

one written descriptive sample; one spoken and one written procedural sample) werealso obtained. Both descriptive and procedural discourse tasks were utilised to

enhance the size and diversity of the collected samples (Li et al., 1996; Shadden et al.,

1991). For descriptive samples, DM described Norman Rockwell’s The Discovery

(1958), an illustration showing a small boy discovering Santa’s clothing in his

parents’ chest of drawers. For procedural samples, DM explained how to catch a

fish; the fishing topic was selected because of DM’s active participation in this sport.

Spoken discourse samples were audiotaped and transcribed, and all spoken and

written samples were analysed in terms of the following structural and pragmaticvariables (Ballard & Thompson, 1999; Jacobs, 2001; Murray et al., 2004): (a) total

number of words and words per minute (Nicholas & Brookshire, 1993); (b) total

number of CIUs and CIUs/minute (Nicholas & Brookshire, 1993); (c) proportion of

CIUs (%CIU; Nicholas & Brookshire, 1993) in which the total CIU count was

divided by the total word count; (d) ratio of open to closed class words (Saffran,

Berndt, & Schwartz, 1989); (e) number of utterances (Saffran et al., 1989); (f) mean

length of utterance (MLU); (g) proportions of grammatically complete and

grammatically complex utterances (Thompson et al., 1995); and (h) number ofsubstantive verb productions (Berndt, Haendiges, Mitchum, & Sandson, 1997). Note

146 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 10: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

that DM’s stereotypical word fillers (e.g., ‘‘well’’, ‘‘good’’) were omitted when

calculating MLU and the ratio of open- to closed-class words, except when they were

judged to have semantic significance. Additionally, repetitions of substantive verbs

were excluded from the substantive verb analysis when they lay adjacent to one

another (e.g., ‘‘wait, wait, wait.’’).

Analysis of DM’s pretreatment discourse samples confirmed a nonfluent,

agrammatic language profile (see Table 2) and thus that he was a suitable TUF

candidate. Both his speech and writing were significantly restricted in terms of

amount and rate of output (e.g., low word and utterance counts; extremely slow

words per minute values) and syntactic form (e.g., no grammatically complete

utterances across all four discourse samples; reduced utterance lengths).

Additionally, DM displayed high open- to closed-class word ratios, and in fact

TABLE 2Pre-treatment, post-treatment, and follow-up language sample data

Measure Probe

Written

descript.

Spoken

descript.

Written

procedural

Spoken

procedural

Written

mean

Spoken

mean

Mean across

all samples

Total

words

Pre-treatment 6 25 13 93 9.5 59 34.25

Post-treatment 9 65 18 125 13.5 95 54.25

Follow-up 11 59 23 105 17 82 49.5

CIUs Pre-treatment 4 7 13 40 8.5 23.5 16

Post-treatment 9 41 14 67 11.5 54 32.75

Follow-up 11 35 17 67 14 51 32.5

Words/min Pre-treatment 1.2 5 2.6 18.6 1.9 11.8 6.85

Post-treatment 1.8 13 3.6 25 2.7 19 10.85

Follow-up 2.2 11.8 4.6 21 3.4 16.4 9.9

CIUs/min Pre-treatment 0.8 1.4 2.6 8 1.7 4.7 3.2

Post-treatment 1.8 8.2 2.8 13.4 2.3 10.8 6.55

Follow-up 2.2 7 3.4 13.4 2.8 10.2 6.5

% CIUs Pre-treatment 75 28 100 43.0 87.5 35.5 61.5

Post-treatment 100 63.1 77.7 53.6 88.85 58.35 73.6

Follow-up 100 59.3 73.9 63.8 86.95 61.55 74.25

Open class:

Closed class

Pre-treatment 3.5 (12:0) 12 15.33 7.75 (27.33) (10.71)

Post-treatment 10 16.33 18 8.88 14 12.61 13.30

Follow-up 12 (36:0) (23:0) (75:0) (17.5) (55.5) (36.5)

Total

utterances

Pre-treatment 3 7 11 25 7 16 11.5

Post-treatment 4 19 11 31 7.5 25 16.25

Follow-up 7 14 12 27 9.5 20.5 15

MLU Pre-treatment 3.33 1.57 1.27 2.36 2.30 1.97 2.13

Post-treatment 3.25 3.26 2.09 3.29 2.67 3.28 2.97

Follow-up 2 3.71 2.42 3.48 2.21 3.60 2.90

% gram.

utterances

Pre-treatment 0 0 0 0 0 0 0

Post-treatment 0 0 0 0 0 0 0

Follow-up 0 0 0 0 0 0 0

Total #

substantive

verbs

Pre-treatment 0 0 1 5 0.25 2.5 1.5

Post-treatment 0 0 2 10 0.5 5 3

Follow-up 0 0 2 13 0.5 6.5 3.75

TUF IN A DISCOURSE CONTEXT 147

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 11: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

produced no closed-class words during his spoken, descriptive discourse sample. He

used few substantive verbs, despite showing relatively good verb confrontation

naming on the VAST. His discourse efficiency and informativeness also appeared

compromised (e.g., low CIUs/min and %CIUs).

Experimental stimuli

Stimuli consisted of 20 pairs of black-and-white line drawings previously used byThompson and colleagues (Ballard & Thompson, 1999; Thompson et al., 1997,

2003). Each pair of drawings depicted the two potential iterations of a semantically

reversible sentence (e.g., ‘‘The artist chased the thief’’ vs ‘‘The thief chased the

artist’’) and served as part of a syntax-priming paradigm designed to elicit five

sentence types: object-extracted embedded who-question sentences (OE; e.g., ‘‘I

know who the artist chased’’); subject-extracted embedded who-question sentences

(SE; e.g., ‘‘I know who chased the thief’’); object-extracted matrix who-questions

(OM; e.g., ‘‘Who has the artist chased?’’); subject-extracted matrix who-questions(SM; e.g., ‘‘Who has chased the thief?); and passive sentences (e.g., ‘‘The thief was

chased by the artist’’). Ten of the drawing pairs, called ‘‘trained’’ stimuli, were

utilised in treatment, and 10 were left ‘‘untrained’’. The treatment phase of the study

also utilised printed word and phrase cards depicting the various elements of the

trained sentence forms, including subject noun phrases (e.g., ‘‘the artist’’), object

noun phrases (e.g., ‘‘the thief’’), verb phrases (e.g., ‘‘chased’’), and other words

needed to complete OE and SE sentences (i.e., ‘‘I know’’ and ‘‘who’’). Additional

stimuli for the Discourse Training Module of the treatment protocol consisted ofcolour and black-and-white photographs gleaned from current news media.

Study design

A single subject, multiple baseline across-behaviours design (Richards, Taylor,

Ramasamy, & Richards, 1999) was implemented to evaluate acquisition of the

trained sentence types, to discern generalisation to untrained sentence types,

discourse, or both, and to identify maintenance of treatment effects. This studydesign allows the participant to serve as his or her own control: A treatment effect is

established if behaviours targeted for treatment are affected while behaviours not

targeted remain at baseline levels. The design also permits distinguishing between

treatment-related generalisation and loss of experimental control by probing

untreated behaviours that are related or unrelated, theoretically, to the treated

behaviours.

DM’s spoken and written productions of all five sentence types were probed

during baseline and treatment phases of the study. These probes alloweddetermination of the acquisition of trained sentence types (i.e., OE and SE

sentences) as well as the generalisation of treatment effects to untrained, less

complex, syntactically related sentence types (i.e., OM and SM questions). Given

that passive sentences involve noun phrase versus wh-movement and thus are

considered syntactically unrelated to the trained sentence types (Thompson &

Shapiro, 2005), production of passives served as the control behaviour and was not

expected to alter during treatment. To examine exposure effects and generalisation

to untrained exemplars of OE and SE sentences, one ‘‘untrained’’ set of 10 drawingpairs was only used to probe all five sentence types during baseline and maintenance

148 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 12: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

phases of the study. Treatment and generalisation effects were considered to be

present if DM improved greater than 30% over his baseline performance.

DM completed two baseline probes; a third baseline was collected only for spoken

production due to concerns that the second spoken baseline may have been

contaminated by a new clinician’s improper application of contrastive stress. During

the treatment phase of the study, written production of OE and SE sentences was

provided. Treatment continued until 10 complete written and spoken probes of each

of the five sentence types had been collected.

Probe tasks

Written and spoken production of five sentence types (i.e., OE, SE, OM, SM, and

passives) was probed using Thompson et al.’s (1997) syntax-priming paradigm.

Within this priming task, the examiner provided a description of one of the pictures

in a randomly selected picture pair by modelling aloud the target sentence structure

(e.g., ‘‘I know who the thief chased’’). Next, DM was asked to produce a similarsentence to describe the other picture (e.g., ‘‘I know who the artist chased’’). A total

of 10 exemplars of each sentence type were elicited; therefore a complete probe

consisted of 50 responses. During written sentence production probes, DM was

instructed to provide a handwritten response. During spoken sentence production

probes, he was requested to say his response aloud; his spoken responses were

audiotaped and transcribed. Written and spoken probe responses were scored as

correct if they contained minor orthographic or apraxic errors, respectively, or

lexical mistakes that did not confound the intelligibility of the targeted syntacticstructure (e.g., ‘‘I know who the painter chased’’ or ‘‘I know who the artist chassed’’

would be considered correct versions of ‘‘I know who the artist chased’’).

Written and spoken sentence probes were given two to three times during

baseline, at the beginning of every treatment session, once immediately after

treatment was terminated, and then once again 4 weeks after treatment had

ended. Because DM lived out of town and could attend therapy sessions only twice

per week, treatment time was at a premium; hence, during the treatment phase,

written and spoken sentence production probes were administered so that only halfa probe was collected at the beginning of each session (i.e., 5 exemplars of each of the

5 sentence types to elicit a total of 25 written and 25 spoken responses), but by the

end of every second session a complete probe had been collected (i.e., 10 exemplars of

each of the 5 sentence types to elicit a total of 50 written and 50 spoken responses). To

control for exposure effects, only the 10 ‘‘trained’’ picture pairs were probed during

the treatment phase. During other study phases, both ‘‘trained’’ and ‘‘untrained’’

picture pairs were probed (i.e., 50 in response to ‘‘trained’’ pictures and 50 in response

to ‘‘untrained’’ pictures for a total of 100 written and 100 spoken productions).

Treatment

DM received two 90-minute therapy sessions per week. Approximately 30 minutes of

each session were dedicated to each of the following components: collection of probe

data followed by administration of TUF and then the Discourse Training Module.

Only written sentence production was trained during treatment. DM also received

regular written weekly homework assignments to facilitate home-based practice oftherapeutic techniques.

TUF IN A DISCOURSE CONTEXT 149

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 13: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

TUF procedures, based on those of Murray et al. (2004), were as follows:

(a) Initial sentence probe. As in the sentence-priming probe task, the clinician first

modelled the target sentence structure aloud for one member of a randomlyselected ‘‘trained’’ picture pair. DM then had to supply, in writing, a similar

sentence for the other member of the picture pair. OE and SE sentences were

presented in pseudorandom order: All 10 ‘‘trained’’ picture pairs were presented

twice per session, once to elicit OE sentences and once to elicit SE sentences.

Regardless of the accuracy of DM’s response, training proceeded through the

following steps for each targeted OE and SE sentence.

(b) Review of active sentence. Using the word/phrase cards, the clinician arranged

the active form of the targeted sentence (e.g., ‘‘The artist chased the thief’’)beneath the appropriate picture. The remaining word/phrase cards required to

construct the target OE or SE sentence (i.e., ‘‘who’’, ‘‘I know’’) were set off to

one side but remained in view. DM was asked to copy this sentence form and

then, in an attempt to increase his awareness of verb and thematic roles, to

point to the word/phrase cards representing the verb or ‘‘action’’, the agent

(e.g., ‘‘Show me the person doing the chasing’’), and the theme (e.g., ‘‘Show me

the person being chased’’).

(c) Clinician-model of target sentence. The clinician showed DM how to form thetarget OE or SE sentence by bringing forward the word/phrase cards previously

placed off to the side of the treatment area and positioning them within the

active sentence already in place. To facilitate DM’s understanding of the

different sentence types during training sessions, OE sentences were referred to

as ‘‘who move’’ sentences and SE sentences were referred to as ‘‘who stay’’

sentences. Next, DM was asked to copy, in written form, the target OE or SE

sentence, and once more to point to the cards representing the verb, agent, and

theme.(d) Anagram of target sentence. The clinician scrambled the word/phrase cards,

moved them to the side of the picture, and then instructed DM to rearrange

them below the picture to form the correct OE or SE sentence. DM was then

asked to recopy, in writing, the sentence, and finally to identify, once more, the

verb, agent, and theme word/phrase cards.

(e) Final sentence probe. The clinician hid the word/phrase cards from DM’s view

and step (a) was repeated.

Within each step of the TUF protocol, DM was given feedback as to the accuracy

of his responses.

Once all trained OE and SE sentences had been reviewed via the TUF protocol,

the Discourse Training Module was implemented. During discourse training, DM

was asked to write a five-sentence story containing at least one sentence in the targetOE or SE syntactic frame about a current newspaper or magazine photograph

selected by the clinician. He was encouraged to write factual or imaginative

summaries of the action taking place in these pictures. Within a treatment session,

one to two pictures were reviewed, with new pictures presented each session. During

the process of constructing each picture description, the clinician provided support,

modelling, and feedback as needed regarding DM’s syntax and orthography, in a

manner somewhat akin to that described for ‘‘loose’’ training protocols (e.g.,

Kearns, 1985). For example, if DM was having trouble generating a sufficientnumber of sentences, the clinician would draw his attention to an event, character, or

150 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 14: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

object that he had yet to describe or discuss (e.g., ‘‘What’s going on towards the

bottom of the picture?’’). If DM’s sentence attempt was missing closed-class words

or verbs, the clinician would draw blanks where appropriate within his response and

ask him to try to fill in the blanks; when these blanks failed to elicit a correct

response, the clinician supplied part of speech cues (e.g., ‘‘It’s a little word’’ or ‘‘It’s

an action’’) followed by forced choices if necessary (e.g., ‘‘Is it ‘is’ or ‘are’?’’). When

DM made spelling errors, the clinician modelled the correct spelling given that (a)

the focus of intervention was on productive syntax versus grapheme retrieval skills,and (b) DM was bothered by his misspellings and wanted them corrected. Once DM

and the clinician had worked through all five sentences, DM was asked to recopy the

final version of the description.

To facilitate practice of written sentence production, DM was assigned 2 hours of

written homework weekly, during which he practised writing each of the 20 trained

sentences (i.e., 10 OE and 10 SE) at least twice. It was left up to DM whether he

wanted to complete all homework practice within one or across several sessions.

Following Murray et al.’s (2004) homework protocol, DM was given a set of the 10‘‘trained’’ picture cards. On the back of each picture card was an anagram (i.e.,

target sentence constituents plus one distractor word in a scrambled order) and,

hidden by removable covers, the target OE and SE sentences. A sheet of step-by-step

instructions for constructing the target sentences was also included (i.e., first write

the action, the person doing the action, and the person to whom the action is being

done; next write the target ‘‘who move’’ or ‘‘who stay’’ sentence). If DM encountered

difficulty in writing the target sentence, he was instructed to consult the anagram on

the back of the picture to help him formulate the sentence. The complete targetsentences were provided on the back of each picture card, allowing DM the

opportunity to check his writing accuracy and to copy the correct sentence if he

made errors. Without any clinician prompting, DM religiously completed his

homework in a notebook, which he brought to each session. The clinician confirmed

his completion of the exercises and offered encouragement and general feedback

about his home practice.

So that the passive sentence structure could be established as a true baseline

against which to measure the effects of the treatment on the wh-movement sentencetypes, it was necessary to establish that DM was in fact able to acquire the passive

structure. Accordingly, during the weeks between immediate post-treatment and

follow-up testing, written production of passives was directly trained using the above

described TUF protocol (e.g., initial production probe followed by review of the

active sentence and instruction on how to form the passive out of the active

sentence). Stimuli consisted of the 10 ‘‘trained’’ picture pairs, and training took place

across three 30-minute weekly sessions. A complete mini-probe of the ‘‘trained’’

stimuli, consisting of 10 written and 10 spoken passive sentences, was obtained priorto each passive training session to monitor acquisition of this sentence type.

Post-treatment testing

Four discourse samples (i.e., one spoken and one written descriptive sample; one

spoken and one written procedural sample) were elicited using the same stimuli and

procedures as those described for collecting pre-treatment samples. Four more

discourse samples, again similarly elicited, were obtained during follow-up testing 4weeks after treatment (except for direct training of passive sentences) ceased. All

TUF IN A DISCOURSE CONTEXT 151

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 15: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

discourse samples were transcribed (where applicable) and analysed using the

procedures described for the pre-treatment discourse sampling.

Reliability

To determine inter-rater agreement in scoring the sentence probe data, 10% of the

written and spoken probe data were randomly selected and then rescored by a

second judge. An acceptable 98% point-to-point inter-rater agreement was achieved

for both spoken and written probe data. The second judge also rescored two

randomly selected discourse samples (one written and one spoken). The average

point-to-point inter-rater agreement across all analysed language variables was 96%,

ranging from 100% agreement for the percentage of grammatical utterances and

number of substantive verbs to 90% agreement for MLU. Discussion was used to

settle inter-judge disagreements that arose when scoring probe data or analysing

discourse data.

To examine intra-rater agreement, another 10% of the probe data and two

additional discourse samples (one written and one spoken) were randomly selected

and rescored by the original judge. For the spoken and written probe data, point-to-

point intra-rater agreement was 98%. The average intra-rater agreement across all

discourse sample variables was 97%, ranging from 100% for several variables (e.g.,

number of words, words per minute, number of utterances, number of substantive

verbs) to 92% for the proportion of open- to closed-class words.

RESULTS

Written production of trained sentence types

DM’s correct responses on the sentence production priming probes are displayed in

Figure 1. Across all sentence types, DM maintained stable baselines, unable to write

correctly any exemplars of any structure. Incorrect responses occurred due to some

movement and verb morphology errors, occasional reversal of agent and theme

roles, frequent omission or distortion of the OE and SE ‘‘I know who’’ frame, and

consistent omission of auxiliary verbs and articles. With treatment, in which writing

OE and SE sentences was directly trained, DM quickly improved from 0% to 100%

accuracy for OE sentences; moreover, he maintained 100% accuracy during follow-

up testing for both ‘‘trained’’ and ‘‘untrained’’ exemplars of OE sentences. Similar,

although slower progress was observed for SE sentences: Whereas for OE sentences

he had achieved 100% accuracy across two consecutive sessions as of the sixth probe,

a comparable performance level was not achieved for writing SE sentences until

post-treatment testing.

Generalisation to writing untrained sentence types

As predicted, while OE and SE sentences were being trained, DM displayed

improved accuracy in writing the correct sequencing of thematic elements in OM and

SM sentence types (see Figure 1). However, this improvement was only evident

when an alternate scoring was applied. That is, DM did not acquire the auxiliary

verb ‘‘has’’, required of both OM and SM structures, and thus did not, strictly

construed, improve from baseline levels in either structure. In contrast, when the

152 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 16: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

Figure 1. Number of correct questions and sentences (max. 10) written and spoken by DM during

baseline, treatment, and follow-up phases. Note that ‘‘Trained’’ and ‘‘Untrained’’ refers to whether DM

had been exposed or not, respectively, to the stimulus set during the treatment phase of the study. ‘‘Alt.’’

refers to data reflecting the alternate scoring system in which inclusion of the auxiliary verb ‘‘has’’ was not

required to score written or spoken production of an OM or SM question as correct.

TUF IN A DISCOURSE CONTEXT 153

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 17: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

absence of the auxiliary was discounted, DM displayed substantial improvements in

both OM and SM sentences, improving from 0% accuracy during baseline to 100%

accuracy during both treatment and follow-up testing. Furthermore, his production

of ‘‘untrained’’ exemplars of OM and SM sentences similarly improved from 0%

accuracy during baseline to 100% correct during post-treatment testing, when the

alternate scoring system was used. As with the OE and SE sentence types, DM

acquired the syntactically more complex OM sentence type more quickly than the

subject-extracted type.DM’s performance pattern in writing passive sentences was also consistent with

our initial hypotheses: He was unable to write passives correctly during baseline, and

his 0% accuracy remained stable throughout OE and SE training (see Figure 1). DM

only showed improvements in writing these types of sentences when direct training of

passives was provided during the weeks prior to the final follow-up testing.

Following this brief training period, DM’s written accuracy rose to 90% during the

last follow-up probe.

Generalisation to spoken sentence production

DM displayed stable or falling baseline performances, ranging from 0% to 20%

correct, in his spoken production of OE, SE, SM, and passive sentences (see

Figure 1). For OM sentences, however, his accuracy quickly rose from 0% to 100%

from the first to the second baseline probe; this improvement during the second

baseline probe was particularly remarkable given that he appeared capable of

producing the auxiliary verb ‘‘has’’, a linguistic skill he had not demonstrated in hiswriting or other verbal performances. Accordingly, because there were concerns that

the novice clinician who administered the second baseline probe may have been

giving improper stress to the auxiliary, a third baseline was obtained by the original

clinician. During this third probe, DM’s accuracy returned to a level comparable

with his first baseline performance; accordingly, to avoid further possible confounds,

the original clinician collected all remaining probe data. However, if the alternate

scoring procedure was applied to OM sentences (i.e., a response was considered

correct if the thematic elements were appropriately sequenced, regardless of thepresence or absence of the auxiliary verb), DM exhibited steadily improving

performance across the three baseline probes. Thus, experimental control was lost

for spoken production of OM sentences, as even nominal exposure to the clinician’s

modelling of these sentences during baseline probes resulted in substantial

improvements in DM’s production, which he maintained throughout all phases of

the study.

For the remaining sentence types, DM displayed the expected patterns of cross-

modal generalisation: Sentence types that were directly trained during the writingtreatment (i.e., OE and SE sentences), or that were syntactically related to the trained

sentence types (i.e., SM sentences), showed improvement, whereas the sentence type

that was syntactically unrelated to the trained sentence types (i.e., passives) did not.

For example, DM’s spoken production of both ‘‘trained’’ and ‘‘untrained’’

exemplars of OE and SE sentences improved to 100% and 90% accuracy levels,

respectively, by the follow-up phase of the study. Similar to his written sentence

performance pattern, OE sentences were acquired faster than SE sentences. DM’s

spoken productions of SM sentences, when scored using the alternate procedure (i.e.,auxiliary verb disobligated), also showed steady improvement across the treatment

154 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 18: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

phase of the study and by follow-up ranged from 70% to 100% correct across both

‘‘trained’’ and ‘‘untrained’’ exemplars. As with written production, DM was unable

to produce passives in the spoken modality until their written production was

directly trained following the first post-treatment probe. Once trained in writing,

DM’s spoken production of ‘‘trained’’ and ‘‘untrained’’ passive exemplars improved

to 90% accuracy on the final follow-up probe.

Discourse

Table 2 presents discourse analyses results at pre-treatment, post-treatment, and

follow-up phases, and Table 3 shows the percentage of change observed across the

discourse variables from pre-treatment to post-treatment and follow-up. DM

TABLE 3Percent change from pre-treatment mean for discourse variables

Measure Probe

Written

mean % Change

Spoken

mean % Change

Mean across

all samples % Change

Total words Pre-treatment 9.5 – 59 – 34.25 –

Post-treatment 13.5 42% 95 61% 54.25 58%

Follow-up 17 78% 82 39% 49.5 45%

CIUs Pre-treatment 8.5 – 23.5 – 16 –

Post-treatment 11.5 35% 54 130% 32.75 105%

Follow-up 14 65% 51 117% 32.5 103%

Words/min Pre-treatment 1.9 – 11.8 – 6.85 –

Post-treatment 2.7 42% 19 61% 10.85 58%

Follow-up 3.4 78% 16.4 39% 9.9 45%

CIUs/min Pre-treatment 1.7 – 4.7 – 3.2 –

Post-treatment 2.3 35% 10.8 130% 6.55 105%

Follow-up 2.8 65% 10.2 117% 6.5 103%

% CIUs Pre-treatment 87.5 – 35.5 – 61.5 –

Post-treatment 88.85 2% 58.35 64% 73.6 20%

Follow-up 86.95 21% 61.55 66% 74.25 19%

Open class:

Closed class

Pre-treatment 7.75 – (27.33) – (10.71) –

Post-treatment 14 83% 12.61 254% 13.30 24%

Follow-up (17.5) 126% (55.5) 103% (36.5) 240%

Total

utterances

Pre-treatment 7 – 16 – 11.5 –

Post-treatment 7.5 7% 25 56% 16.25 41%

Follow-up 9.5 36% 20.5 28% 15 30%

MLU Pre-treatment 2.30 – 1.97 – 2.13 –

Post-treatment 2.67 16% 3.28 66% 2.97 40%

Follow-up 2.21 24% 3.60 83% 2.90 36%

% gram.

utterances

Pre-treatment 0 – 0 – 0 –

Post-treatment 0 0% 0 0% 0 0%

Follow-up 0 0% 0 0% 0 0%

Total #substantive

verbs

Pre-treatment 0.25 2.5 – 1.5 –

Post-treatment 0.5 100% 5 100% 3 100%

Follow-up 0.5 100% 6.5 160% 3.75 150%

TUF IN A DISCOURSE CONTEXT 155

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 19: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

displayed increases in all pragmatic measures (including total word counts, number

of CIUs, words/min, CIUs/min, and %CIUs) from pre- to post-treatment, averaged

across all sample types (i.e., spoken and written, descriptive and procedural). This

positive trend was evident, to varying degrees, in 19 of 20 comparisons between

individual pre- and post-test variables in corresponding samples (e.g., total word

count changes across spoken descriptive samples); these improvements were, for the

most part, maintained during follow-up testing. However, DM’s discourse gains

appeared markedly larger for spoken versus written samples.For morphosyntactic variables, treatment effects were less clear-cut. Across

sample types, DM displayed and maintained modest increases in utterance counts

and MLU following treatment. However, this positive trend appeared to be driven

principally by improvements in his spoken, as opposed to written, discourse. The

ratio of open- to closed-class words, which gauged the proportion to which DM was

using ‘‘content’’ as opposed to ‘‘function’’ words, actually rose post-treatment,

indicating that he was using increasing proportions of open-class words, and thus

had moved further from the norm for non-brain-damaged adults (M 5 0.91,SD 5 0.08 on the Cinderella story retell task; Thompson et al., 1995). It is probable

that this increase was linked, at least in part, to larger word and utterance counts in

his post-treatment and follow-up samples, rather than representing an independent

treatment effect. Despite these increases, DM produced no written or spoken

utterances that could be scored as grammatical (i.e., a complete sentence);

accordingly, an analysis of the proportion of complex sentences was abandoned.

Across all testing sessions, no substantive verb productions were elicited in any

written or spoken descriptive discourse sample. In contrast, DM produced a greaternumber of substantive verbs in post-treatment and follow-up versus pre-treatment

procedural samples, particularly in the spoken modality. Clinicians from DM’s

aphasia support group also reported, on an informal basis, an increase in his

substantive verb productions during group sessions; that is, he progressed from

producing no substantive verbs prior to treatment to an estimated average of three

productions per group session by the end of the study.

DISCUSSION

Previous research has shown that Treatment of Underlying Forms (TUF) can

efficiently improve the sentence production abilities of patients with agrammatic

aphasia (e.g., Ballard & Thompson, 1999; Jacobs & Thompson, 2000). TUF’s

efficiency lies in its generalisation effects: Patients not only improve production of

trained complex sentence types, but also show generalisation to untrained less

complex sentence types, syntactically related to the trained structures. Some

questions remain, however, as to whether the positive effects of TUF generalisemore broadly to (a) patients’ discourse-level communication abilities or (b)

untrained output modalities. That is, a limited literature has thus far examined the

effects of TUF on spoken discourse, and the modest improvements cited have been

inconsistent across patients and discourse variables (Ballard & Thompson, 1999;

Jacobs & Thompson, 2000; Thompson et al., 1996, 1997, 2003). Only one study has

evaluated and reported cross-modal generalisation from trained spoken sentence

production to written sentence production (Jacobs & Thompson, 2000).

Accordingly, the current study examined the effects of TUF when combined witha Discourse Training Module designed to enhance use of trained sentence structures

156 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 20: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

in connected language contexts. Specific hypotheses were as follows: (a) DM, an

individual with agrammatic Broca’s aphasia, would improve his written production

of directly trained complex sentence structures with wh-movement; (b) DM would

demonstrate generalisation to untrained exemplars of targeted sentence structures as

well as related, but less complex, syntactic structures containing trained or untrained

constituents, but not to passives (i.e., sentences with NP vs wh-movement); (c) cross-

modal generalisation from trained writing to untrained spoken production would be

observed; and (d) augmenting TUF with a Discourse Training Module wouldfacilitate improvements in discourse measures.

Overall, DM responded to TUF in a manner consistent with that previously

reported for agrammatic patients (e.g., Ballard & Thompson, 1999; Thompson et al.,

2003). He displayed clear acquisition of the OE and SE syntactic structures targeted

in the writing treatment. If his written responses were strictly scored, DM appeared

to display none of the predicted generalisation to the less complex, but syntactically

related, OM and SM structures; however, if the presence of the auxiliary verb ‘‘has’’

was not considered essential for accurate OM and SM production, a robust writingimprovement pattern was evident for both sentence types. Because no auxiliary verb

was present in the directly trained OE and SE sentences, and because TUF focuses

on the movement of thematic constituents within a sentence (Thompson & Shapiro,

2005), it is reasonable to assume that TUF would not facilitate auxiliary verb

production. Therefore, the second, less stringent method of scoring OM and SM

sentences was viewed as a more accurate measure of TUF’s generalisation effects, at

least in terms of facilitating integration of thematic roles and movement operations

in untrained, but syntactically related, sentence types.As also predicted, DM displayed no generalisation to production of passives.

Within single-subject, multiple baseline designs, this lack of a treatment effect for

passives allows improvements in the other four sentence types to be ascribed to the

experimental treatment rather than general therapy effects (Richards et al., 1999).

Subsequent to treatment, passives were directly trained to ensure that DM was

capable of acquiring NP-movement sentences; given that his production of passives

quickly progressed when directly targeted, improvements in wh-movement sentences

can be assertively attributed to TUF. Accordingly, DM’s results indicate that awritten version of TUF, like its previously investigated spoken counterpart (e.g.,

Ballard & Thompson, 1999; Thompson et al., 1997), can improve production of not

only the trained complex sentence structures but also untrained, less complex but

syntactically related structures.

An unexpected finding was that DM acquired OE more quickly than SE

sentences, even though OE sentences are associated with greater syntactic complexity

(Gibson, 1998; Grodzinsky, 2000). Similarly, and perhaps as a result, DM showed

quicker acquisition of the untrained OM versus SM sentences. Murray andcolleagues (2004), who observed a similar acquisition pattern in several of their

participants, offered a plausible, non-linguistic explication of this performance

pattern. That is, according to the TUF protocol, training OE sentences requires the

physical movement of the ‘‘agent’’ word card from one end of the nascent sentence to

the other; training SE sentences requires no such physical movement because these

sentences do not syntactically involve agent movement. For this reason, in the

Murray et al. (2004) study as well as the present investigation, OE and SE sentence

types were introduced in therapy as ‘‘who move’’ and ‘‘who stay’’, respectively.Murray et al. suggested, and we concur, that the physical movement required while

TUF IN A DISCOURSE CONTEXT 157

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 21: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

constructing an OE sentence may help increase that structure’s saliency, just as

gesture and pointing have previously been shown to promote verbal skills in

individuals with aphasia (Richards, Singletary, Gonzalez-Rothi, Koehler, &

Crosson, 2002; Rose, 2006; Rose, Douglas, & Matyas, 2002).

Missing from several previous TUF studies (e.g., Murray et al., 2004; Thompson

et al., 1997, 1998) was an investigation of whether aphasic participants could

produce acquired sentence structures in response not only to the materials used in

training, but also to novel stimuli. If participants can formulate a familiar sentence

type using unfamiliar constituents, there is further support that production of a

syntactic frame rather than a particular set of sentences has been acquired. In the

current study, an ‘‘untrained’’ stimulus set (picture pairs similar to the ‘‘trained’’

stimuli but involving, for the most part, novel agents, actions, and themes) was only

presented to DM during baseline, post-treatment, and follow-up phases. In response

to this ‘‘untrained’’ set, DM displayed substantial improvement from baseline to

post-treatment and follow-up in producing sentences with wh-movement, suggesting

that he had acquired the ability to produce specific syntactic structures irrespective

of their lexical constituents. It should be noted, however, that response time was not

measured. That is, it was the clinician’s perceptual impression that DM took longer

to produce ‘‘untrained’’ sentences than their ‘‘trained’’ counterparts. However, even

if this were the case, the delay in producing these ‘‘untrained’’ sentences would be

difficult to interpret given DM’s verb retrieval and spelling difficulties, which also

could contribute to slower sentence construction times.

Although only written sentence production was directly trained, DM’s spoken

sentence production improved in a manner similar to that observed in his writing.

That is, while treatment targeted written production of OE and SE sentences, DM

demonstrated improved spoken production of sentences with wh-movement (i.e.,

OE, SE, and SM sentences1 containing both ‘‘trained’’ and ‘‘untrained’’ sentence

constituents) with no concurrent change in his spoken production of passive

sentences; once written production of passives was treated, however, his spoken

production of this sentence type also improved. These findings suggest that, as in

writing, DM acquired verbal production of syntactic frames versus a specific

sentence set. These results also extend previous findings regarding TUF’s cross-

modal generalisation effects: Whereas spoken versions of TUF have evoked gains in

written sentence production (Jacobs & Thompson, 2000; Murray et al., 2004), the

current study demonstrates that cross-modal generalisation from writing to speech

may also occur. It is important to note, however, that although DM never received

explicit spoken language training, the sentence production priming paradigm used

during probe and treatment sessions involved verbal models from the clinician.

Additionally, DM routinely spoke aloud as or after he wrote; this unelicited verbal

output was allowed, given that clinically reproving it seemed inappropriate and

potentially detrimental to the therapeutic process. Although DM was never provided

with any feedback regarding his spoken productions, the ‘‘purity’’ of our written

intervention is certainly debatable. Further research is needed to determine whether

the robust cross-modal generalisation effects observed in the current study could be

replicated using a more strict, writing intervention (e.g., providing written vs spoken

1 Recall that due to rising baselines, experimental control was lost for spoken production of OM

sentences, and thus discussion of these data is inappropriate.

158 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 22: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

modelling), or with study participants who have less spontaneous verbal repetition of

their written productions.

Finally, it was predicted that pairing TUF with a Discourse Training Module

would encourage improved sentence production in discourse contexts. In particular,

morphosyntactic gains were expected given the previously documented effects of

TUF on discourse (Thompson & Shapiro, 2005) and that our Discourse Training

Module required production of trained sentence types within a less structured,

discourse-level task. Although DM achieved better performance on certainmorphosyntactic variables (e.g., MLU, number of substantive verbs), he displayed

no or negative change on other variables. For instance, regardless of discourse task

or modality, DM continued to produce no grammatically complete utterances

following treatment. Additionally, whereas open- to closed-class word ratios were

expected to decrease, DM displayed larger ratios as the size of his discourse samples

increased without concomitant increases in his use of closed-class words. In contrast,

more robust improvements were noted across pragmatic discourse measures: DM’s

discourse samples were longer (e.g., larger word and utterance counts), moreinformative (e.g., larger CIU counts), and more efficient (e.g., significantly larger

%CIUs) following treatment, and further, these gains were maintained during

follow-up testing.

DM’s pattern of discourse progress accords, at least in part, with previous TUF

research. In studies that examined morphosyntactic variables, some patients,

especially those with more severe agrammatic profiles like DM, were more apt to

show improvements in MLU and verb use than in open- to closed-class word ratios

or the proportion of grammatical or grammatically complex sentences (Ballard &Thompson, 1999; Jacobs & Thompson, 2000; Thompson et al., 1996, 1997, 2003). In

the only earlier TUF investigation to include both morphosyntactic and pragmatic

discourse measures (Ballard & Thompson, 1999), gains weighted towards the

pragmatic variables. Likewise, Jacobs (2001) reported significant improvements in

discourse informativeness and efficiency that, importantly, were discernible to naı̈ve

listeners. However, not all prior TUF findings correspond well with our discourse

results. Compared to DM, many previous TUF participants have displayed more

remarkable morphosyntactic improvements in their discourse, particularly in termsof grammatical complexity (Jacobs & Thompson, 2000; Thompson et al., 1996, 1997,

2003). These variable morphosyntactic outcomes suggest that despite TUF’s focus

on sentence structure, its grammatical rather than pragmatic discourse effects appear

most sensitive to participant (e.g., severity of agrammatism, presence and severity of

concomitant symptoms), stimulus (e.g., types of sentence structures trained; genre of

discourse sample), and training characteristics (e.g., number, length, and intensity

of TUF sessions). Therefore, future research must delineate which patient and

stimulus criteria as well as therapy schedules are most likely to foster generalisationof TUF’s morphosyntactic benefits to discourse.

DM’s relatively greater pragmatic than morphosyntactic improvement might

reflect the influence of our Discourse Training Module. This module built-in ‘‘loose’’

training procedures (e.g., reinforcing patient-initiated output, encouraging patient-

elaboration of earlier utterances) similar to those used in Response Elaboration

Training (RET; Kearns, 1985; Kearns & Scher, 1989). It differed, however, in that

RET emphasises utterance content only, whereas our module underscored both

content and form. Despite this difference, our discourse findings corresponded wellwith those of RET: Adults with nonfluent aphasia profiles produce an increased

TUF IN A DISCOURSE CONTEXT 159

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 23: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

number and variety of open-class words (especially nouns) and relatedly, longer and

more informative verbal output (Kearns & Scher, 1989; Wambaugh & Martinez,

2000; Wambaugh, Martinez, & Alegre, 2001). However, our study design does not

allow certain treatment outcomes to be ascribed to specific components of our

training protocol, and thus further research is required to disentangle the relative

contributions of TUF, the Discourse Training Module, and homework activities to

DM’s improvement pattern across the various discourse measures.

When the discourse data are broken down by modality, an interesting trend isobserved: Although TUF, the Discourse Module, and homework activities focused

on DM’s writing abilities, his verbal output gains across all discourse measures

approximated or exceeded those made in his writing. These findings might indicate

that DM’s spoken discourse improvements were a further manifestation of the cross-

modal generalisation pattern observed in the sentence production probe data. Such

an account, however, would not explain why DM’s spoken discourse gains were

greater rather than comparable to those in his written discourse. Instead, differences

between the discourse modalities may reflect sample characteristics. For example,DM’s spoken samples were, both pre- and post-treatment, longer and less efficient

than his written samples; therefore, in his speech there was more room for variables

like %CIUs, a measure of communicative efficiency, to improve. Alternately, greater

gains in spoken discourse may reflect the natural communicative priority assigned by

most adults, including those with aphasia, to speech, and thus the proportionally

greater rehearsal that priority engenders—that DM automatically rehearsed aloud

sentences trained in treatment supports this contention. Finally, spoken discourse

improvements might also, at least in part, be related to DM’s concurrent, aphasiasupport group participation: Group activities often concentrated on pragmatic

communication skills, and relatedly, spoken discourse variables registering the

greatest positive change were pragmatic in nature. However, a primary goal of group

sessions was to encourage use of multi-modality communication (e.g., use of

gestures, drawing, sound effects, etc.) rather than solely verbal communication.

Likewise, DM had been attending this support group for several semesters prior to

this study and had never previously demonstrated the degree of change in his verbal

abilities that was observed following our experimental treatment. Given that noprevious TUF studies have evaluated for cross-modality generalisation at the

discourse level, there remains the need to investigate further whether DM’s

substantial spoken discourse gains are idiosyncratic or can be induced in other

individuals with similar or different symptom profiles.

In summary, the current study serves not only as a confirmation and extension of

the potential clinical efficacy of TUF, but also as an invitation for its reassessment.

Augmenting a written version of TUF with a Discourse Training Module and

homework activities resulted in treatment outcomes comparable to those in previousTUF studies (e.g., Ballard & Thompson, 1999; Thompson et al., 2003): acquisition of

trained structures with concomitant improvements in untrained, syntactically related

sentence structures. Our findings broadened TUF’s application by showing that

training written sentence production can evoke substantial cross-modal general-

isation to spoken production of trained and untrained sentence types. Questions

persist, however, regarding how to maximise generalisation of TUF’s treatment

effects to discourse production. Despite pairing structured TUF training with a less

constrained and more contextual Discourse Training Module, pragmatic versusmorphosyntactic aspects of DM’s written and spoken discourse showed most

160 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 24: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

improvement, an outcome pattern similar to that observed in prior TUF research

(Ballard & Thompson, 1999). Therefore, although research consistently supports

that TUF can efficiently enhance sentence processing in constrained contexts,

continued investigation is recommended to determine (a) whether it is additionally

an efficient means by which to treat sentence processing in discourse, or alternately

(b) what modifications will ensure generalisation to discourse contexts in a broader

spectrum of aphasic patients.Manuscript received 24 July 2006

Manuscript accepted 19 September 2006

REFERENCES

Ballard, K. J., & Thompson, C. K. (1999). Treatment and generalisation of complex sentence production

in agrammatism. Journal of Speech, Language and Hearing Research, 42, 690–707.

Bastiaanse, R., Edwards, S., & Rispens, J. (2002). Verb and Sentence Test. Bury St. Edmunds, UK:

Thames Valley Test Company.

Beeson, P., Rising, K., & Volk, J. (2003). Writing treatment for severe aphasia: Who benefits? Journal of

Speech, Language, and Hearing Research, 46, 1038–1060.

Berndt, R. S., Haendiges, A. N., Mitchum, C. C., & Sandson, J. (1997). Verb retrieval in aphasia. 2.

Relationship to sentence processing. Brain and Language, 56, 107–137.

Brookshire, R. H., & Nicholas, L. E. (1997). The Discourse Comprehension Test. (rev. ed.). Minneapolis,

MN: BRK Publishers.

Cherney, L. R. (1998). Pragmatics and discourse: An introduction. In L. R. Cherney, B. Shadden, &

C. A. Coelho (Eds.), Analysing discourse in communicatively impaired adults (pp. 1–7). Gaithersburg,

MD: Aspen Publishers.

Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research.

Journal of Verbal Learning and Verbal Behavior, 11, 671–684.

Dabul, B. (2000). Apraxia Battery for Adults – 2. Austin, TX: Pro-Ed.

Davis, G. A., & Wilcox, J. M. (1985). Adult aphasia rehabilitation: Applied pragmatics. San Diego, CA:

College-Hill.

Freedman, J. C., & Steinbruner, J. D. (1964). Perceived choice and resistance to persuasion. Journal of

Abnormal and Social Psychology, 68(6), 262–265.

Gibson, E. A. (1998). Linguistic complexity: Locality of syntactic dependencies. Cognition, 68, 1–76.

Gierut, J. A. (2001). Complexity in phonological treatment: Clinical factors. Language, Speech and

Hearing Services in Schools, 32, 229–241.

Grodzinsky, Y. (1995). Trace deletion, theta-roles and cognitive strategies. Brain and Language, 51,

467–497.

Grodzinsky, Y. (2000). The neurology of syntax: Language use without Broca’s area. Behavioural and

Brain Sciences, 23, 1–71.

Hammill, D. D., Brown, V. L., Larsen, S. C., & Wiederholt, J. L. (1994). Test of Adolescent and Adult

Language. (3rd ed.). Austin, TX: Pro-Ed.

Jacobs, B. J. (2001). Social validity of changes in informativeness and efficiency of aphasic discourse

following Linguistic Specific Treatment (LST). Brain and Language, 78, 115–127.

Jacobs, B. J., & Thompson, C. K. (2000). Cross-modal generalisation effects of training non-canonical

sentence comprehension and production in agrammatic aphasia. Journal of Speech, Language, and

Hearing Research, 43, 5–20.

Kearns, K. P. (1985). Response elaboration training for patient initiated utterances. Clinical Aphasiology,

14, 196–204.

Kearns, K. P., & Scher, G. P. (1989). The generalisation of response elaboration training effects. Clinical

Aphasiology, 18, 223–245.

Kertesz, A. (1982). Western Aphasia Battery. New York: Grune & Stratton.

Kiesler, C. A., & Sakamura, J. (1966). A test of a model for commitment. Journal of Personality and Social

Psychology, 3, 349–353.

Kim, M., & Thompson, C. K. (2004). Verb deficits in Alzheimer’s disease and agrammatism: Implications

for lexical organization. Brain & Language, 88, 1–20.

TUF IN A DISCOURSE CONTEXT 161

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 25: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

Kiran, S., & Thompson, C. K. (2003). Effects of exemplar typicality on naming deficits in fluent aphasia.

Journal of Speech, Language, and Hearing Research, 46, 602–622.

Lewis, R. L., & Vasishth, S. (2005). An activation-based model of sentence processing as skilled memory

retrieval. Cognitive Science, 29, 375–419.

Li, E., Ritterman, S., Della Volpe, A., & Williams, S. (1996). Variation in grammatic complexity across

three types of discourse. Journal of Speech-Language Pathology and Audiology, 20, 180–186.

Maas, E., Barlow, J., Robin, D., & Shapiro, L. (2002). Treatment of phonological errors in aphasia and

apraxia of speech: Effects of phonological complexity. Aphasiology, 16, 609–622.

Mauner, G., Fromkin, V., & Cornell, T. (1993). Comprehension and acceptability judgements in

agrammatism: Disruptions in the syntax of referential dependency. Brain and Language, 45, 340–370.

Mitchum, C. C., Haendiges, A. N., & Berndt, R. S. (1993). Model-guided treatment to improve written

sentence production: A case study. Aphasiology, 7, 71–109.

Murray, L. L., Ballard, K. J., & Karcher, L. (2004). Linguistic Specific Treatment: Just for Broca’s

aphasia? Aphasiology, 18(9), 785–809.

Murray, L. L., & Karcher, L. (2000). Treating written verb retrieval and sentence construction skills: A

case study. Aphasiology, 14, 585–602.

Nicholas, L., & Brookshire, R. (1993). A system for quantifying the informativeness and efficiency of the

connected speech of adults with aphasia. Journal of Speech and Hearing Research, 36, 338–350.

Peach, R. K., & Wong, P. C. M. (2004). Integrating the message level into treatment for agrammatism

using story retelling. Aphasiology, 18, 429–441.

Richards, K., Singletary, F., Gonzalez-Rothi, L. J., Koehler, S., & Crosson, B. (2002). Activation of

intentional mechanisms through utilisation of nonsymbolic movements in aphasia rehabilitation.

Journal of Rehabilitation Research and Development, 39, 445–454.

Richards, S. B., Taylor, R. L., Ramasamy, R., & Richards, R. Y. (1999). Single subject research:

Applications in educational and clinical settings. San Diego, CA: Singular Publishing.

Robertson, I. H., Ward, T., Ridgeway, V., & Nimmo-Smith, I. (1994). The Test of Everyday Attention.

Gaylord, MI: Northern Speech Services.

Robson, J., Marshall, J., Chiat, S., & Pring, T. (2001). Enhancing communication in jargon aphasia: A

small group study of writing therapy. International Journal of Language and Communication Disorders,

36, 471–488.

Rose, M. (2006). The utility of arm and hand gestures in the treatment of aphasia. Advances in Speech-

Language Pathology, 8, 92–109.

Rose, M., Douglas, J., & Matyas, T. (2002). The comparative effectiveness of gesture and verbal

treatments for a specific phonologic naming impairment. Aphasiology, 16, 1001–1030.

Ross, K. B., & Wertz, R. T. (1999). Comparison of impairment and disability measures for assessing

severity of, and improvement in, aphasia. Aphasiology, 13, 113–124.

Saffran, E. M., Berndt, R. S., & Schwartz, M. R. (1989). The quantitative analysis of agrammatic

production: Procedure and data. Brain and Language, 37, 440–479.

Shadden, B. B., Burnette, R. B., Eikenberry, B. R., & DiBrezzo, R. (1991). All discourse tasks are not

created equal. Clinical Aphasiology, 20, 327–342.

Thompson, C. K. (1996). Northwestern Sentence Comprehension Test. Unpublished.

Thompson, C. K. (2001). Treatment of underlying forms: A linguistic specific approach for sentence

production deficits in aphasia. In R. Chapey (Ed.), Language intervention strategies in aphasia and

related neurogenic communication disorders. (4th ed., pp. 605–674). New York: Lippincott, Williams, &

Wilkins.

Thompson, C. K., Ballard, K. J., & Shapiro, L. P. (1998). The role of syntactic complexity in training wh-

movement structures in agrammatic aphasia: Optimal order for promoting generalisation. Journal of

the International Neuropsychological Society, 4, 661–674.

Thompson, C. K., & Shapiro, L. P. (2005). Treating agrammatic aphasia within a linguistic framework:

Treatment of Underlying Forms. Aphasiology, 19, 1021–1036.

Thompson, C. K., Shapiro, L. P., Ballard, K. J., Jacobs, B. J., Schneider, S. L., & Tait, M. E. (1997).

Training and generalised production of wh- and NP-movement structures in agrammatic speakers.

Journal of Speech, Language, and Hearing Research, 40, 228–244.

Thompson, C. K., Shapiro, L. P., Kiran, S., & Sobecks, J. (2003). The role of syntactic complexity in

treatment of sentence deficits in agrammatic aphasia: The complexity account of treatment efficacy

(CATE). Journal of Speech, Language, and Hearing Research, 46, 591–607.

Thompson, C. K., Shapiro, L. P., Tait, M., Jacobs, B., Schneider, S., & Ballard, K. (1995). A system for

systematic analysis of agrammatic language production. Brain and Language, 51, 124–129.

162 MURRAY, TIMBERLAKE, EBERLE

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6

Page 26: context Treatment of Underlying Forms in a discourse · abilities in discourse post-treatment. Methods & Procedures: A single subject, multiple baseline across-behaviours design was

Thompson, C. K., Shapiro, L. P., Tait, M. E., Jacobs, B. J., & Schneider, S. L. (1996). Training wh-

question production in agrammatic aphasia: Analysis of argument and adjunct movement. Brain and

Language, 52, 175–228.

Tompkins, C. A., Bloise, C. G., Timko, M. L., & Baumgaertner, A. (1994). Working memory and

inference revision in brain-damaged and normally aging adults. Journal of Speech and Hearing

Research, 37(4), 896–912.

Wambaugh, J. L., & Martinez, A. L. (2000). Effects of modified response elaboration training with

apraxia and aphasic speakers. Aphasiology, 14, 603–617.

Wambaugh, J. L., Martinez, A. L., & Alegre, M. N. (2001). Qualitative changes following application of

modified response elaboration training with apraxic-aphasic speakers. Aphasiology, 15, 965–976.

TUF IN A DISCOURSE CONTEXT 163

Dow

nloa

ded

by [

Uni

vers

ity o

f O

rego

n] a

t 17:

05 2

5 M

arch

201

6