This article was downloaded by: [University Of Maryland]
On: 04 April 2013, At: 08:42
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Aphasiology
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/paph20
Selective treatment of regular
versus irregular verbs in agrammatic
aphasia: Efficacy data
Yasmeen Faroqi-Shah
a
a
Department of Hearing and Speech Sciences, University of
Maryland, College Park, MD, USA
Version of record first published: 07 Mar 2013.
To cite this article: Yasmeen Faroqi-Shah (2013): Selective treatment of regular versus irregular
verbs in agrammatic aphasia: Efficacy data, Aphasiology, DOI:10.1080/02687038.2012.751577
To link to this article: http://dx.doi.org/10.1080/02687038.2012.751577
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-andconditions
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any
representation that the contents will be complete or accurate or up to date. The
accuracy of any instructions, formulae, and drug doses should be independently
verified with primary sources. The publisher shall not be liable for any loss, actions,
claims, proceedings, demand, or costs or damages whatsoever or howsoever caused
arising directly or indirectly in connection with or arising out of the use of this
material.
Aphasiology, 2013
http://dx.doi.org/10.1080/02687038.2012.751577
Selective treatment of regular versus irregular verbs
in agrammatic aphasia: Efficacy data
Yasmeen Faroqi-Shah
Downloaded by [University Of Maryland] at 08:42 04 April 2013
Department of Hearing and Speech Sciences, University of Maryland, College
Park, MD, USA
Background: Production of verb morphology, especially tense marking, is frequently
impaired in persons with agrammatic aphasia. Very little research has examined theoretically driven treatments for verb morphology deficits in aphasia.
Aims: This study examined the relative efficacy of using regular (wash-washed,
rob-robbed) versus irregular (drink-drank, keep-kept) verbs as stimuli to treat morphological impairments in individuals with aphasia. This comparison was motivated
by differences in the lexical organisation of regular and irregular verbs proposed in
psycholinguistic theory.
Methods & Procedures: A single-participant multiple-baseline design was used to examine treatment outcomes in six individuals with agrammatic aphasia. Participants received
training to produce tense morphology using only either regular or irregular verbs, and
the crucial outcome measure was generalisation to untrained past tense forms (regular to
irregular and vice versa).
Outcomes & Results: All participants improved in the trained tenses and generalised to the
production of regular tense morphology on untrained verbs. Generalisation to untrained
irregular past tense was relatively modest, irrespective of whether regular or irregular
verbs were trained.
Conclusions: The results replicate previous findings that verb morphology deficits respond
to intervention, and extend the findings by suggesting that choice of stimuli may have
consequences for generalisation effects. The implication for aphasia rehabilitation is that
tense training using irregularly inflected verbs generalises to a greater variety of untrained
verb inflections (including regular past) than does the use of regular verbs.
Keywords: Aphasia; Brain damage; Inflection; Morphology; Language production;
Treatment efficacy; Verb.
Address correspondence to: Yasmeen Faroqi-Shah, Department of Hearing and Speech Sciences,
University of Maryland, College Park, 0100, Lefrak Hall, College Park, MD-20742, USA. E-mail:
[email protected]
This project was supported by General Research Board award from the University of Maryland to YF-S.
The author wishes to thank all the participants and their families for their participation and support for this
research. Special thanks to students who helped with various stages of data collection and analyses (listed
alphabetically): Heather Alexander, Sarah Camponeschi, Isabelle Dunn, Xiomy Esteban, Lauren Graham,
Melissa Graham, Jessica Greenberg, Jen Maultasch, Rachel Mont, Monica Sampson, Mohan Singh, and
Pascale St. Victor. Partial findings of this study were presented at the Clinical Aphasiology Conference
(2008) and the Annual Convention of the American Speech Language and Hearing Association (2008).
© 2013 Taylor & Francis
Downloaded by [University Of Maryland] at 08:42 04 April 2013
2
FAROQI-SHAH
Agrammatic Broca’s aphasia is characterised by syntactic and morphological difficulties that frequently encompass verbs. For example, verb morphology is severely
impaired while noun morphology is relatively spared (Goodglass, Christiansen, &
Gallagher, 1993; Tsapkini, Jarema, & Kehayia, 2002). A cross-linguistic pattern that
has generated considerable interest among aphasiologists is that verb morphology is
selectively more impaired when expressing semantic notions such as tense rather than
when marking syntactic constraints such as subject–verb agreement. For instance,
completing sentences that demand temporal referencing, such as Tomorrow, the boys
_____ [will swim], is significantly more difficult than completing sentences in which
local syntactic constraints cue verb form, as in The boy(s) ___ [swim(s)] or The boys
will______ [swim] (Arabatzi & Edwards, 2002; Benedet, Christiansen, & Goodglass,
1998; Faroqi-Shah & Dickey, 2009; Faroqi-Shah & Thompson, 2007; Friedmann &
Grodzinsky, 1997; Kok, van Doorn, & Kolk, 2007; Varlakosta et al., 2006; Wenzlaff
& Clahsen, 2004). This paper presents efficacy data for a treatment approach for verb
morphology deficits that focused on temporal reference. The treatment paradigm is a
replication and extension of an earlier study (Faroqi-Shah, 2008). This study extends
previous work by examining whether the type of stimuli used for treatment (verb with
regular versus irregular past morphology) can impact treatment outcomes. A background on relevant theoretical frameworks and empirical evidence is presented in the
next section. This is followed by research questions and an overview of prior treatment
research of verb morphology in aphasia.
THEORETICAL AND EMPIRICAL BACKGROUND
This selective impairment of tense morphology has been attributed either to a
representational deficit for tense or to a processing limitation. Proponents of the
representational deficit use the syntactic tree framework of generative grammarians
to claim that the tense node in the mental syntactic tree is erased (Friedmann &
Grodzinsky, 1997). The processing argument attributes the greater difficulty with tense
morphology to the need to integrate information between two levels of computation:
conceptual intent and syntactic formulation (Avrutin, 2000, Goodglass et al., 1993).
Several recent proposals have focused on the semantic functions of verb inflections
because at least two syntactic functions are relatively less affected: agrammatic individuals perform better on verb forms that can be selected solely by syntactic rules
(Faroqi-Shah & Thompson, 2007; Goodglass et al., 1993), and judgement of syntactic violations is relatively spared (Linebarger et al., 1983). That is, the process of
translating temporal information into verb morphology has been identified by multiple authors as a potential source of difficulty in agrammatic production (Bastiaanse,
2008; Diacritical encoding and retrieval hypothesis of Faroqi-Shah & Thompson,
2007; Tense Underspecification Hypothesis of Wenzlaff & Clahsen, 2004; Tense and
Agreement Underspecification Hypothesis of Burchert, Swoboda-Moll, & De Bleser,
2005; see also Penke, 2003, for a similar explanation based on accessing inflectional
morphology).
Although there is a rather productive mapping between semantic notions (such
as tense/mood) and morphological structure, it is somewhat inconsistent in several
Germanic languages due to the existence of irregular verbs such as go-went, sleep-slept,
and drink-drank. Patterns of performance on regular and irregular verbs have been
variable in aphasia: some studies reported worse performance on irregulars (Balaguer,
Costa, Sebastian-Galles, Juncadella, & Caramazza, 2004; Inglis, 2005; Kok et al.,
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
3
2007; Miozzo, 2003; O’Conner, Obler, & Goral, 2007; Penke & Janssen, 1999), others reported worse performance on regulars (Ullman et al., 1997), or no differences
between regular and irregular past (Bird, Lambon Ralph, Seidenberg, McClelland, &
Patterson, 2003; Druks, 2006; Faroqi-Shah & Thompson, 2007; Wenzlaff & Clahsen,
2004). A meta-analysis of studies that compared regular and irregular verb performance revealed that more than half of the published 110 datasets showed no
differences (Faroqi-Shah, 2007). The meta-analysis also revealed that data sets that
found worse performance on regulars were from English speakers and the stimuli
were confounded by phonological complexity of regulars compared to irregulars. The
general trend is for irregular verb morphology to be more difficult than regular morphology for agrammatic individuals in languages where irregulars have a subregular
affix (Penke & Janssen, 1999 in German; Balaguer et al., 2004, and O’Connor et al.,
2007 in Spanish; Kok et al., 2007 for Dutch). Subregular refers to semi-productive patterns within the class of irregulars, such as ring-rang, sing-sang, drink-drank in English,
and -e(n) suffix for feminine nouns in German irregular plurals. The crucial point is
that the putative performance differences between regulars and irregulars are over and
above the difficulty with tense morphology. That is, the tense information conveyed by
both regular and irregular verb forms (past tense in English) is essentially the same, but
what differs is whether there is an additional influence of morphological complexity.
There is considerable debate about the extent to which regular and irregular words
differ in lexical representation, and providing a comprehensive review is beyond the
scope of this paper (Bybee, 1995; Clahsen, 1999; Halle & Marantz, 1993; McClelland
& Patterson, 2002; Pinker & Prince, 1991; Rumelhart & McClelland, 1986; Ullman &
Pinker, 2002). Although it is not the intention of this paper to validate any one theory
of morphological processing, a review of experimental results from a variety of sources
such as frequency effects, priming, and child language acquisition indicates that morphologically regular forms are generally generated by a default mechanism of applying
affixation rules, such as Verb (+PAST) = Verbed, while irregular forms are generally retrieved from the mental lexicon1 (see also Okrent, 2004; Zimmermann, 2001).
Experimental evidence from tasks such as inflectional priming and sentence matching
indicates that lexical entries are morphologically underspecified for redundant features that can be represented by a parsimonious rule, such as Verb (+PAST) = Verbed
(Clahsen, Eisenbeiss, Hadler, & Sonnenstuhl, 2001; Janssen & Penke, 2002; Penke,
Janssen, & Eisenbeiss, 2004). In contrast, the lexical entries of irregular verbs include
additional specification for word structure. For instance, the lexical entry of give, will
specify the perfect form (give(+PERFECT) = given) and for the simple past form
(give(+PAST) = gave). The Unique Entry Principle assumes that a general rule such
as Verb(+PAST) = Verbed is replaced by specific rules such as give(+PAST) = gave
for irregular verbs (Pinker, 1984; Pinker & Prince, 1988). In order to explain why (or
how) default rules are not universally applied to irregular verbs, thus preventing words
such as singed or sanged in adult speakers, a blocking mechanism is proposed (Aronoff,
1976; Glass & Lau, 2003; Kiparsky, 1982; Pinker, 1999). The Blocking Principle allows
the affixation rule to apply to irregular verbs, however, the existence of lexically stored
irregular (past tense) forms blocks an over-regularised form from being produced.
1 Exceptional situations where regular forms may be stored include very high frequency forms (Alegre
& Gordon, 1999). Conversely, nonword past tense generation tasks suggest the existence of subregular
schemas for irregular verbs that are partially based on phonological form (Bybee & Slobin, 1982;
Stemberger, 2002).
Downloaded by [University Of Maryland] at 08:42 04 April 2013
4
FAROQI-SHAH
Figure 1. Illustration contrasting the lexical entries and morphological encoding mechanisms of regular (rake) and irregular (drink) verbs. The rectangular boxes representing lexical entries illustrate the
under-specification of regular verb entries and lexical representation of irregular verb forms (unique entry
principle). The shaded arrows represent the default affixation rule, which is proposed to be activated for
both regular and irregular verb production. The figure illustrates that production of an irregular verb entails
default affixation (arrows 1 & 2 for drink), and selection of the lexical entry (arrow 3 for drank) blocks (arrow
4) the regularised form from entering phonological encoding (represented in ovals). The blocking mechanism is denoted by . Affixation rule activation during both regular and irregular verb retrieval can be used
to predict generalisation effects for aphasia.
Figure 1 illustrates the more complex nature of irregular verb lexical entries and the
application of the blocking mechanism on an over-regularised verb after successful
retrieval of an irregular past.
The above principles can be used to make predictions about directionality effects
upon treatment of verb inflections in agrammatic aphasia. It can be hypothesised
that, when irregular verb production is practiced by an agrammatic individual, this
not only strengthens the practised irregular verb, but is also likely to re-activate and
strengthen the past tense affixation rule (the shaded arrows in Figure 1). Hence, the
effects of treatment with irregular inflection, via strengthening of the affixation rule,
could trickle down to the production of regular past tense forms. This predicts that
treatment of irregular verb inflections could potentially improve both irregular and
regular verb production. However, practice with regular verb inflections is unlikely to
automatically benefit irregular verb production. Specifically, there is a theoretical possibility of asymmetrical generalisation effects, that is, treatment of tense marking using
irregular verbs may result in generalised improvement of tense marking on regulars,
but not vice versa.
Another aspect that has the potential of influencing treatment effects is the presence of subregular structural patterns within the class of irregulars (Bybee & Slobin,
1982). One notable subregular group in English is irregulars that undergo a vowel
change (drank, sank, rang). Another subregular group is one that includes both
vowel change and addition of an alveolar consonant (henceforth referred to as suffixed irregulars: kept, slept, felt, and sold, told, lost). According to theories such as
Distributed Morphology (Halle & Marantz, 1993), suffixed irregulars pattern both
with vowel-change-only irregulars as well as regulars (due to alveolar consonant addition). In support of this view, a priming study with unimpaired participants found that
suffixed irregulars patterned with regulars rather than with vowel-change irregulars
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
5
(Kierlar, Joanisse, & Hare, 2008). By this logic, treatment using suffixed irregulars
would be more likely to generalise to production of regulars than treatment that used
vowel-change irregular stimuli.
This treatment study addresses the above predictions by comparing the relative
efficacy of using regular versus irregular verb stimuli in achieving within- and betweenverb type generalisation. Of course, with English-speaking aphasic participants as is
the case in this study, the question is only valid for past tense (simple and perfective) since present and future tense morphology is highly regular. The first question
addressed in the present study pertained to within-verb type generalisation; that is,
if tense training for any one verb type (e.g., regulars) would generalise to untrained
exemplars of the same verb type (regulars). Based on principles of affixation rules
versus storage for regular versus irregular morphology, it was predicted that training regular inflections would generalise to other verbs using regular inflections (past,
present, and future). However, the uniqueness of lexical entries of irregular verbs
would preclude generalisation of irregular past tense training to untrained irregular past verbs. The second question was whether tense training would result in
between-verb type generalisation (regular to irregular morphology and vice versa).
The blocking and unique entry principles were used to predict that training of irregular past tense would inadvertently improve regular past tense production. However,
between verb type generalisation would be less likely with regular verb training due to
the absence of mapping between the underspecified entries of regular verbs and those
of irregular verbs. In other words, it was hypothesised that tense training with regular verbs would promote within verb type generalisation, while tense training with
irregular verbs would result in between verb type generalisation.
These two questions were addressed by training participants with either regular
verbs (three participants) or with irregular verbs (three participants), while generalisation to untrained regular and irregular verbs was tested. Given the subregularities
among irregulars, participants who received training with irregulars were exposed to
a homogeneous set of irregulars, either vowel-change only or suffixed irregulars only.
Given the proposal that suffixed irregulars actually encompass features of both vowelchange irregulars and rule based regular affixation (addition of t/d) (Halle & Marantz,
1993), we decided to train vowel-change irregulars to ensure no “contamination” by
affixation rules. Hence, two out of three participants received irregular verb training with vowel-change irregulars. We trained suffixed irregulars in one participant in
order to verify if the generalisation patterns obtained with vowel change irregulars
were replicable with suffixed irregulars.
PRIOR TREATMENTS OF INFLECTIONAL MORPHOLOGY
In spite of the pervasive occurrence of verb inflection impairments in nonfluent
aphasia, and the widespread empirical and theoretical interest in this impairment,
there has been a relative dearth of treatment studies for inflectional deficits. The earliest treatment study for verb inflections used Garrett’s (1988) framework of language
production (Mitchum & Berndt, 1994). As per Garrett’s proposal that verb inflections are embedded in sentence frames, Mitchum and Berndt trained present, past,
and future tenses such as The noun is verbing, The noun has verbed, and The noun
will verb in one aphasic individual. It is noteworthy that the trained sentences were
Downloaded by [University Of Maryland] at 08:42 04 April 2013
6
FAROQI-SHAH
morphosyntactically similar in including an auxiliary and verb, but were semantically varied in the type of aspectual markers, with progressive aspect for present tense,
and perfective aspect for past tense, but no aspect trained for future tense. One half
of the trained verbs had a regular perfective maker (has closed) while the other half
were irregular (has eaten). After treatment the patient improved in the production of
both untrained regular and irregular past perfect tense. Another tense treatment study
used computerised icons and trained a combination of regular and irregular verbs in
three nonfluent aphasic individuals (Weinrich, Shelton, Cox, & McCall, 1997; see also
Boser, Weinrich, & McCall, 2000). The stimuli differed in two aspects from those of
Mitchum and Berndt. First, simple past tense was trained, and second, irregular verbs
with a variety of transformations were used. All participants improved in the production of untrained regular past tense, while improvements for untrained irregular past
tense were limited. Weinrich and colleagues attributed the limited generalisability of
irregulars to the use of heterogeneous irregulars in treatment. The present study differs
from these two past studies by training only regular or irregular verbs in any one aphasic participant. Also, all training was strictly for simple tense forms and no aspectual
marking was included.
In a prior study, the relative efficacy of training verb form production
(morphophonological treatment) versus semantic aspects of verb forms
(morphosemantic treatment) was compared in four individuals with agrammatic
aphasia (Faroqi-Shah, 2008). Only the morphosemantic treatment resulted in
increased accuracy of verb inflections, although both treatments improved the number and variety of inflected verbs produced. Hence agrammatic individuals benefited
maximally from practice in mapping tense features (e.g., +PAST, +PRESENT,
+FUTURE) onto semantically matched verb forms (folded, folds, will fold). Practice
with verb form production in the absence of semantic context was less successful.
Hence the present study used the morphosemantic procedure of tense training
proposed by Faroqi-Shah (2008).
METHOD
Participants
Six individuals with aphasia were recruited for the current study (age range:
44–65 years). All participants gave written informed consent using procedures
approved by the institutional review board of the University of Maryland, College
Park, and hence this study was performed in accordance with the ethical standards laid
down in the 1964 Declaration of Helsinki. All participants had a single left hemisphere
lesion in the region of the middle cerebral artery resulting from a cerebrovascular accident (CVA) and were at least 1 year post-onset of the CVA. One of the participants
(P6) had a history of grand mal seizures, and none of the other participants had any
complicating neurological conditions such as prior neurosurgery, degenerative or psychiatric conditions, or a history of substance abuse. They passed puretone audiometric
screening at 500, 1000, 2000, and 4000 Hz at 40 dBHL ANSI: 1969 in both ears and
passed a vision screen, defined as at least 20/40 corrected or uncorrected vision on
a Snellen’s chart and the absence of spatial neglect and visual field deficits (as per
self-report). All participants were native speakers of English, right-handed before the
CVA, and had at least a high school diploma. Demographic and neurological details
of all participants are given in Table 1.
REGULAR AND IRREGULAR VERB TREATMENT
7
TABLE 1
Demographic and neurological information of participants, sorted by treatment verb type
Participant
Age/gender
Downloaded by [University Of Maryland] at 08:42 04 April 2013
Regular verb treatment
P1
37/M
Years
education
Occupation
18
Police officer
P2
44/M
19
Software
engineer
P3
55/M
18
Admiral in
Navy
17
Insurance
adjuster
Homemaker
Irregular verb treatment
P4
56/M
P5
55/F
18
P6
39/M
14
Financial
analyst
Lesion
Months
post-onset
Left MCA infarct involving the
frontal and parietal regions
Left MCA infarct, involving
frontal, subcortical, and basal
ganglia
Left MCA involving the
temporo-parietal and
extending into the occipital
region
84
Left MCA infarct
19
Left MCA territory infarct and
occlusion of the left ICA
Left MCA territory infarct
resulting in a paucity of distal
branches. No major branch
occlusion
26
16
29
25
MCA = middle cerebral artery; ICA = internal carotid artery.
Language testing
In order to establish their diagnosis of non-fluent aphasia with agrammatic speech
and morphological impairments, all aphasic participants completed a variety of tests
including: (1) Western Aphasia Battery (WAB, Kertesz, 1982), (2) narrative and
oral expression sections of the Boston Diagnostic Aphasia Examination (BDAE,
Goodglass, Kaplan, & Barresi, 2001), (3) six sequences from the Narrative Story
Cards (Helm-Estabrooks & Nicholas, 2003), and (4) Verb Inflection Test (FaroqiShah, unpublished). The narratives obtained from the tests described above were
transcribed and coded using the Systematic Analysis of Language Transcripts (SALT)
(Miller, 2004) for lexical and syntactic measures as described in Faroqi-Shah (2008).
Using SALT the narratives were analysed to determine rate of speech and mean length
of utterance. The first 50 utterances were used to compute the proportion of sentences,
proportion of grammatical sentences, noun–verb ratio, open–closed class ratio, proportion of suffixed verbs (the ratio of suffixed verbs to the total number of verbs that
require affixation in contexts where affixation was possible), diversity of tense marking
(type–token ratio of tense computed as the ratio of unique tense-aspect markings to
the total number of verbs), and accuracy of tense marking. The first 50 utterances were
used to compute these values so that comparisons across participants and pre- versus
post-treatment measures could be made over a consistent number of utterances. These
measures were used not only to determine the presence of agrammatic speech, but
also as general outcome measures that could be compared across studies in the future.
The Verb Inflection Test (Faroqi-Shah, unpublished) elicited 20 sentences in past,
present, and future tenses using a picture description with adverbial cues (Tomorrow,
Downloaded by [University Of Maryland] at 08:42 04 April 2013
8
FAROQI-SHAH
Right now, Yesterday, and Every day/Nowadays).2 In addition single word repetition of
20 regular and irregular past tense verbs was tested using stimuli matched for phonological structure (taken from Bird et al., 2003). Individual participant scores are given
in Table 2.
Inclusionary criteria from these tests were: (1) a profile of non-fluent aphasia with
relatively spared comprehension (all participants were classified as having Broca’s
aphasia as per the WAB), (2) agrammatic sentence production gleaned from narratives and defined as the presence of fragmented utterances, decreased proportion
of sentences, paucity of grammatical morphemes and verbs, (3) difficulty with verb
morphology, defined as an accuracy below 60% on the Verb Inflection Test, (4) no
significant differences in the production of regular and irregular past tense verbs,
determined from tests of sentence production and single word repetition, (5) absence
of significant apraxia or dysarthria as per the oral expression subtests of the BDAE
and the checklist of apraxic symptoms listed in the Apraxia Battery for Adults-2nd
edition (Dabul, 2000), and (6) functional reading abilities as per the single word reading subtests of the WAB. The inclusionary criterion of comparable performance for
regular and irregular past verbs prior to the onset of treatment was necessary to
examine between-verb type generalisation effects after treatment and also because of
discussions in prior literature about whether regular and irregular verbs are dissociated
in aphasia (Bird et al., 2003; Faroqi-Shah, 2007; Faroqi-Shah & Thompson, 2007).
Upon determination of eligibility, participants were assigned to receive treatment
with either irregular or regular verbs. Although it was not possible to make exact
pairwise matches, we used a pseudo-random assignment method so that participants
receiving regular versus irregular treatment did not vary dramatically in three respects:
severity of the aphasia quotient (AQ) calculated from the WAB, accuracy of verb tense
production calculated from the Verb Inflection Test, and age at the time of the study.
This is evident from the language scores in Table 2: participants who received treatment with regular past tense verbs (P1, P2, P3) did not differ in the above criteria from
the participants who received treatment with irregular past tense verbs (P4, P5, P6) in
terms of overall aphasia severity.
Stimuli
Black and white line drawings in three tenses were made by an artist for 40 regular and
40 irregular imageable verbs (an example of the verb steal is shown in Figure 2 and the
verb list is given in Appendix A). A total of 20 regular verbs were used as treatment
stimuli and the remaining 20 regular verbs were used to test generalisation. The irregular verbs included an equal number of verbs with vowel change and suffixed verbs.
Verbs with two acceptable past tense forms (such as dove-dived) were avoided to the
extent possible. As explained earlier, one research question pertained to between verbtype generalisation: whether training of irregulars would generalise to regulars, and
so we wanted to train irregulars that had little overlap with regulars. Hence two participants (P4, P5) received training on vowel-change irregulars. The third participant
(P6) received training on suffixed irregulars in order to replicate the generalisation
patterns obtained with vowel change irregulars. Treatment and generalisation verbs
were matched in logarithmic lemma frequency as per CELEX word frequency counts
(Baayen, Piepenbrock, & van Rijn, 1993).
2
All participants were pre-tested for comprehension of temporal adverbs using a calendar pointing
task, on which they had to identify yesterday, today, tomorrow, next month, etc.
AQ (max. = 100)
inflection Test (%)
Word repetition (max=10)
Regular past
Irregular past
c Narrative Speech
Rate (words per minute)
Mean length of utterance
Proportion sentences
Open: Closed Class
Proportion verbs
Proportion suffixed verbs
Variety of tense
Accuracy of tense
21.6
3.74
0.48
1.14
0.16
0.63
0.82
0.22
8
9
77.4
55
49.2
3.12
0.46
1.3
0.19
0.66
0.77
0.2
92.1#
Post
58.8
1.92
0.22
4.33
0.19
0.83
0.08
1
9
10
45.6
25
Pre
P2
44.4
3.61
0.61∗
1.95∗
0.2
0.77
0.18∗
1
75.8#
Post
27.6
1.68
0.1
3.53
0.15
0.54
0
0
5
7
61.8
30
Pre
P3
32.4
1.91
0.39
1.59∗
0.14
0.83∗
0.89∗
0.38∗
67.8#
Post
197.4
2.84
0.4
1.28
0.09
0.85
0.86
0.17
10
10
80.5
25
Pre
P4
265.8
3.4
0.44
0.95∗
0.08
1∗
0.95∗
0.1
93.7#
Post
55.2
1.54
0.1
15
0.09
0.14
0
0
5
10
22.7
5
Pre
P5
19.8
2.8
0.62∗
1.47∗
0.11
0.6∗
0.6∗
0.11
61.9#
Post
Irregular verbs trained
55.8
2.94
0.36
0.84
0.07
0.8
0.79
0.27
9
10
71.8
5
Pre
P6
33
4.28
0.62∗
0.9
0.11∗
0.87
0.67
0.14
74.1
Post
Aphasia Battery (Kertesz, 1982); AQ-Aphasia Quotient; b Faroqi-Shah (unpublished); c Normative values are given in Appendix B. # McNemar’s change test,
p < .05;
greater than 2 standard deviations of normative sample.
∗ change
a Western
b Verb
a WAB
Pre
P1
Regular verbs trained
TABLE 2
Pre- and post-treatment scores on language outcome measures
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
9
Downloaded by [University Of Maryland] at 08:42 04 April 2013
10
FAROQI-SHAH
Figure 2. Example of stimuli. The verb illustrated here is steal. Lexical items were provided by the tester to
elicit sentences. For example, the past tense was elicited using the following verbatim instructions: “In these
three pictures, the action is steal and here is a man and a thief . Here the action is about to happen, here it is
going on, and here it is done. Now use a sentence to describe this picture [while pointing to the third picture
in the sequence].”
Design and procedures
This study primarily used a single-participant design, although descriptive betweengroup comparisons were made to assess generalisation patterns. Data collection
consisted of three phases: baseline, treatment, and follow-up. During each baseline probe, a picture description task was used to elicit 30 sentences distributed
across present, past, and future tenses with the treatment and generalisation verbs.
Participants received between four and eight baseline probes (ranging from 120 to
240 sentences total) in order to demonstrate that treatment effects were unrelated
to repetitive exposure to baseline probes. The number of baseline sessions was randomly assigned to each participant. Participants received individual morphosemantic
treatment 4 days a week in 1–2-hour sessions. Criteria for cessation of treatment were
treatment probe accuracies exceeding 80% over at least two sessions and a minimum
of 12 sessions. Post-treatment testing was conducted immediately following the cessation of treatment and consisted of a re-administration of treatment and generalisation
probes, WAB, and narrative measures. Follow-up testing was done twice, at 2 weeks
(F1) and 2 months (F2) after the cessation of treatment, during which treatment and
generalisation probes were administered. The typical duration of involvement in the
study (excluding follow-up testing) was 6 weeks.
Experimental control was achieved using varying numbers of baseline measures
across participants, thus documenting the absence of spontaneous improvement and
demonstrating treatment-induced improvement regardless of the amount of exposure
to baseline measures.
Treatment. Each treatment session began with administration of the treatment
probes, in which the 20 sentences with training verbs were elicited in various
tenses. Generalisation probes were administered every fourth session and consisted
of regular and irregular past tense as well as present and future tenses of these
verbs. The procedure for morphosemantic treatment focused on learning the semantic implications of various verb forms and has been described in detail elsewhere
(Faroqi-Shah, 2008). The following steps were completed for each verb using three
pictures, which depicted past, present, and future actions. (1) Confrontation naming:
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
11
In this step the participant was asked to name the action depicted in the three picture sequences (e.g., steal or stealing for Figure 2). (2) Grammaticality judgement:
The experimenter read sentences that contained the action elicited in step (1) and
the participant responded whether or not the sentences were grammatically correct
(e.g., Yesterday the woman rang the bell, Tomorrow the woman rang the bell). The primary emphasis was on learning to match the temporal adverb with the verb inflection.
(3) Comprehension. In this step the experimenter randomly staggered the three pictures and read a sentence. The participant was asked to match the sentence with its
corresponding picture (e.g., The man steals the wallet). (4) In the written sentence
completion task the participant was given a printed sentence with the inflected verb
omitted (e.g., Yesterday the lion ________) and the participant was asked to write the
appropriate verb inflection that matched a given picture. Assistance with spelling was
provided. (5) Sentence anagram: In this step the participant was asked to select word
anagrams (The man/ stole/ steals/ will steal/ the wallet) that matched a given picture
and arrange these to form a sentence. Each step was followed by feedback regarding
accuracy and an accompanying explanation. The above steps were repeated for each
of the 20 treatment verbs. After the first few treatment sessions all participants were
able to complete two iterations of 20 verbs during each session. Consistent with the
procedure of Faroqi-Shah (2008), none of the steps listed above required extensive
oral production of verb morphology.3 All three tenses were trained in the following
sequence: past, present (third person singular, Verb+s), and future (will Verb). Only
simple tenses were trained without introducing aspectual markers. Participants had to
achieve an accuracy of 90% in the written sentence completion step (see (iv) above) to
proceed to the next tense.
Outcome measures and scoring
The measure that determined cessation of treatment was tense accuracy of treatment
sentences. Generalisation probes—that is, tense accuracy of both regular and irregular generalisation verbs (in all three tenses)—were the primary dependent variable
that specifically addressed the research questions. A significant generalisation effect
was operationally defined as a statistically significant difference between baseline and
post-treatment probes (McNemar’s change test, p < .05). Generalisation probes were
administered every fourth session for two reasons: to limit improvement in accuracy
as a result of repetitive daily exposure, and because of the time required to administer the very large number of generalisation items (N = 120). The method of eliciting
sentences for these treatment and generalisation probes was identical and involved
describing one of three pictures in a sequence (see Figure 2). The nouns and verbs in
the pictures were provided by the experimenter in order to avoid confounds created
by lexical retrieval difficulties. Responses that minimally contained a noun and a verb
were scored, and the target verb was scored as correct if the verb morphology matched
the tense being elicited even if the entire sentence was not grammatically well formed.
That is, “Woman peeled” was considered for scoring and scored as correct because the
3 Given that the primary purpose of the Faroqi-Shah (2008) study was to compare the relative efficacy of
morphosemantic and morphophonological treatments, oral production was minimised for morphosemantic
treatment to preclude any morphophonological practice. The same protocol was repeated for the current
study although the lack of oral practice has no direct implications for the research questions in the current
study. Fidelity with the treatment protocol between studies serves to replicate and extend the findings of
Faroqi-Shah (2008).
12
FAROQI-SHAH
Downloaded by [University Of Maryland] at 08:42 04 April 2013
target past tense morphology was produced even though the determiner for “woman”
and the noun argument (the potatoes) were not produced.
Participants were given the choice of writing their responses if they experienced difficulty with oral production, and both oral and written responses were accepted. All
participants preferred orally producing their responses. However, three participants
were limited by intermittent oral production difficulties specifically for verb endings
(P2, P5, and P6) and would orally perseverate on a single ending. All three participants
would self-correct after oral production by writing out their intended verb morphology (this happened for a total of 5%, 20%, and 17% of all probes for P2, P5, and
P6 respectively). No feedback regarding accuracy was provided. Additionally, narrative measures and the WAB were repeated at the end of treatment as general outcome
measures that would enable future comparison with the outcomes of other treatment
studies.
Reliability
Inter-rater reliability was obtained for 30% of all selected dependent measures, including scoring of treatment and generalisation probes, as well as transcription and
coding of narrative samples. Additionally, reliability was obtained for 30% of treatment sessions for accuracy of administration of the treatment procedure, which is
the independent variable. Reliability measures were obtained using audio recordings,
video recordings and/or on-line scoring. Point to point inter-rater agreement for all
measures exceeded 90% (Cohen’s Kappa statistic, 1960, K > 0.95).
Data analysis
Each participant’s percent accuracy scores were calculated for treatment and generalisation probes for each of the three phases of the study (i.e., baseline, treatment, and
follow-up). Our primary research question pertained to generalisation to untrained
verb morphology and hence the percent accuracy of past tense sentences among the
generalisation probes was also separately calculated. McNemar’s change test was used
to determine statistically significant change in level for each participant by comparing
mean baseline accuracy with post-treatment accuracy (see Boo & Rose, 2011; Conroy,
Sage, & Lambon Ralph, 2009; Edmonds & Babb, 2011; for a similar use of McNemar’s
test). Two effect size measures were computed, effect size for change in level (Busk
& Serlin, 1992) and effect size including slope (Olive & Smith, 2005). The effect
size for change in level (SMDpre-post ) was computed by subtracting each participant’s
mean baseline accuracy from the final treatment session accuracy. This difference was
divided by that participant’s baseline standard deviation. To compute the effect size
for slope, mean accuracy was computed over all treatment sessions (hence SMDall ).
This effect size (SMDall ) was calculated by subtracting mean baseline accuracy from
mean treatment accuracy and dividing this number by the baseline standard deviation
(Olive & Smith, 2005). SMDall values are typically smaller than SMDpre-post because
probe accuracies during the treatment phase are lower than post-treatment accuracy.
One participant (P4) had zero percent accuracy for baseline probes and hence effect
sizes could not be calculated.4
4 In cases of zero-standard deviation during baseline, Busk and Serlin (1992) proposed calculation of
standard deviation by combining baseline and follow-up scores if both phases have similar performance.
Since P4 improved significantly following treatment, Busk and Serlin’s condition for calculation of pooled
standard deviation was not met.
REGULAR AND IRREGULAR VERB TREATMENT
13
RESULTS
Downloaded by [University Of Maryland] at 08:42 04 April 2013
The mean accuracy of baseline sessions was 3.4% (range: 0–35), and all participants
showed relatively stable baselines with a small standard deviation. During baseline
testing, regular and irregular past tense probes were of comparable accuracy for all
participants with the exception of P1 who showed better performance in irregular past
compared to regular past (14/20 versus 6/20); χ 2 (N = 40) = 4.9, p < .05. The numbers of baseline sessions, treatment sessions, and total hours of treatment are given
for each participant in Table 3. The average number of treatment sessions was 15. The
average final treatment accuracy for trained verbs was 93.4% (range: 87–100) showing
a significant treatment effect, Paired t-test, t(5) = 9.9, = p < .001. In the following sections results of each treatment verb type will be presented separately. This is followed
by scores on general outcome measures, namely narratives and WAB.
Regular verb treatment
The treatment trajectories of P1, P2, and P3 are presented in Figure 3. The duration
of treatment and effect sizes are given in Table 3. All three participants showed significant gains in trained verbs (McNemar’s change test, p < .001) with an average final
treatment accuracy of 95%. As for the first research question (within-verb type generalisation), Figure 3 shows that all three participants generalised treatment effects to
untrained regular past tense, and the scores for untrained regular past in the final treatment session were near ceiling (Mean = 96.7%). Effect sizes did not differ from that of
trained regular past tense (Fisher’s exact test, p >1 for all three participants). The data
for other trained tenses (which includes combined scores for simple present and future
tenses) were also near ceiling (Mean = 96.6%), and effect sizes did not differ from
that of trained regular past tense (Fisher’s exact test, p > 1 for all participants). The
second research question regarding between-verb type generalisation was examined
with accuracy of irregular past tense. The effects are mixed: P1 and P2 improved by
30% points in the production of irregular past tense (pre- to final-treatment accuracy
for P1 is 70% to 100%; McNemar’s change test, p < .05; pre- to final-treatment accuracy for P2 is 0% to 30%; McNemar’s change test, p < .05). P3 remained unchanged
for irregular verbs (pre- and final-treatment accuracy = 0%). To summarise, participants were able to apply trained morphology (regular past, simple present, future) to
untrained verbs, but did not improve consistently in untrained morphology (irregular
past).
All participants maintained the effects of treatment in follow-up testing (data points
F1 and F2 in Figure 3), with a mean follow up accuracy score of 87.1% (range =
75–100). This accuracy was significantly greater than baseline accuracy (McNemar’s
change test, p < .05). While there was a slight decline from final treatment session
accuracy, this difference was not statistically significant (McNemar’s change test,
p > 1).
Irregular verb treatment
Figure 4 and Table 3 show the treatment trajectories of P4, P5, P6, and other relevant
data respectively. All three participants achieved 80% accuracy criterion for treatment
probes with an average final treatment accuracy of 95% (McNemar’s change test, p
< .01). Data pertaining to the first research question (within-verb type generalisation), show that all three participants improved significantly in accuracy of untrained
17
15
15
15.6
Irregular verb treatment
P4
8
P5
4
P6
6
Group Mean
26.25
12.5
19
19.2
23.75
24
25.75
24.5
Treatment hours
n/a
6.6
7
6.8
6.4
25.6
75
35.6
Tx
n/a
7.4
5.4
6.4
7.8
48
80
45.2
GenReg
n/a
0.5
1.9
1.2
0.8
17.4
−0.6
5.8
GenIrreg
n/a
9.2
9.7
9.4
6.8
44.8
88.4
46.6
GenOther
n/a
12.6
10.1
11.3
7.8
71.4
193
90.7
Tx
n/a
12.6
7.5
10
7.8
71.4
193
90.7
GenReg
n/a
3.6
4.9
4.2
3.15
21
−0.6
7.8
GenIrreg
n/a
12.5
12.6
12.5
8.8
71.4
91.7
57.3
GenOther
Effect sizes for level (SMDpre-post )
Notes: The effect sizes in the left panel reflect slope (SMDall ), which includes the accuracy of every treatment session. The values on the right reflect change in level
(SMDpre-post ). Tx = trained verbs, all tenses; GenReg = Untrained verbs, regular past tense; GenIrreg = Untrained verbs, irregular past tense; GenOther = Untrained
verbs, third person singular present and future tenses. As described in the text, effect sizes could not be calculated for P4 because he had zero percent accuracy for all
baseline probes.
8
18
17
14.37
Treatment sessions
Regular verb treatment
P1
4
P2
8
P3
6
Group Mean
Baseline sessions
Effect sizes for slope (SMDall )
TABLE 3
Treatment data for each participant, sorted by type of treatment verb
Downloaded by [University Of Maryland] at 08:42 04 April 2013
14
FAROQI-SHAH
15
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
Figure 3. Percent accuracy for participants who received tense training with regular verbs (P1, P2, P3) during baseline (B), treatment (T), and follow-up (F). During baseline, combined scores for both verb types
(regular, irregular) and all three tenses (simple past, simple present, future) are plotted for the sake of visual
simplicity. The treatment data (Tx Reg All tense) reflect combined accuracies for all trained verbs (three
tenses). The generalisation data are separately plotted for regular and irregular past (Gen Reg Past, Gen Irr
Past) and for other tenses, which include simple present and future tense (Gen Other Tense).
FAROQI-SHAH
Downloaded by [University Of Maryland] at 08:42 04 April 2013
16
Figure 4. Percent accuracy for participants who received tense training with irregular verbs (P4, P5, P6) during baseline (B), treatment (T), and follow-up (F). During baseline, combined scores for both verb types
(regular, irregular) and all three tenses (simple past, simple present, future) are plotted for the sake of visual
simplicity. The treatment data (Tx Reg All tense) reflect combined accuracies for all trained verbs (three
tenses). The generalisation data are separately plotted for regular and irregular past (Gen Reg Past, Gen Irr
Past) and for other tenses, which include simple present and future tense (Gen Other Tense).
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
17
irregular past forms (overall pre- to final-treatment accuracy: 0–48.3%; P4: 0–80%;
P5: 0–30%; P6: 0–35%; McNemar’s change test, p < .01 for P4, p < .05 for P5 and
P6). Table 3 shows that the effect sizes for untrained irregulars were relatively small
when compared to the effect sizes for trained irregulars (0.5 versus 6.6 for P5 and
1.9 versus 7 for P6). For the second research question (between-verb type generalisation), accuracy of regular past tense showed a statistically significant gain for all
three participants, with a mean final treatment accuracy of 93.3% compared to 7.7%
at baseline testing (McNemar’s change test, p < .01). The effect sizes for regular past
are larger in magnitude than the corresponding trained irregular and untrained irregular effect sizes (see Table 3). Production of future and present tense morphology on
untrained verbs also improved significantly for all three participants (Mean pre- and
final treatment scores are: 12% and 87.6%; McNemar’s change test, p < .01 for all
three participants). The participant who received treatment with suffixed irregulars
(P6) showed no significant advantage over the participants who received treatment on
vowel-change irregulars (P4, P5). To summarise, treatment using irregular verbs produced generalisations in untrained regular and irregular verb morphology. However,
the gain in untrained irregular past verb forms was lower in magnitude than that for
regular morphology.
All participants maintained the effects of treatment in follow-up testing, with
an average accuracy score of 80.8% for trained irregulars (data points F1 and
F2 in Figure 4; range = 66–100). While mean follow-up accuracy was significantly
greater than baseline accuracy (McNemar’s change test, p < .05), the decline from
final-treatment session accuracy was significant (McNemar’s change test, p < .05).
Regular versus irregular verb treatment
Table 3 shows that quantitative aspects such as the mean number treatment sessions required to reach criterion (14.4 versus 15.6), treatment effect sizes (SMDpre-post :
11.1 versus 11.8), and the number of treatment hours (24.5 versus 19.2) showed no
consistent advantage for any group. However, there were differences in the pattern
and magnitude of generalisation to untrained verb types, which can be compared in
the accuracy scores of final treatment and follow-up sessions across Figures 3 and 4.
As for the first research question, the mean within-verb type generalisation, computed
as the difference between average baseline and final treatment session accuracy, was
99.5% for the regular-trained group (gains of 100%, 99%, and 99.5% for P1, P2, and
P3 respectively) and 48.3% for the irregular-trained group (gains of 80%, 30%, and
35% for P4, P5, and P6 respectively). Mean between-verb type generalisation (regular trained → irregular tested, and vice versa) was 20% for the regular trained group
(gains of 30%, 30%, and 0% for P1, P2, and P3 respectively) and 79% for the irregular trained group (gains of 60%, 98%, and 79% for P4, P5, and P6 respectively).
That is, while within-verb type generalisation was of higher magnitude for the regular
trained group, between-verb type generalisation was larger for the irregular trained
group. In other words, irrespective of whether regular or irregular verbs are used in
training, generalisation to untrained regular morphology was consistently found and
was larger in magnitude than for irregular verb morphology. It is also noteworthy
that there were no remarkable differences in generalisation patterns between the two
participants who received treatment with vowel change irregulars and the participant
who received treatment on suffixed irregulars.
18
FAROQI-SHAH
Downloaded by [University Of Maryland] at 08:42 04 April 2013
Other outcome measures
Post-treatment scores of the WAB and narratives are given in Table 2. Five out of
six participants (with the exception of P6) significantly improved in their WAB AQ
(McNemar’s change test, p < .05). For the most part, these changes were in the fluency score, and additionally in a variety of subtests. Normative values of narrative
measures are given in Appendix B for comparison. Given that typical statistical tests
cannot be used for most narrative measures, an a priori criterion of change greater
than two standard deviations of the normative sample was used as threshold for significant change (Rochon, Laird, Bose, & Scofield, 2005). The post-treatment narrative
measures in Table 2 show that rate of speech was higher than the pre-treatment sample
for three participants (P1, P3, P4) and was decreased for the other three participants
(P2, P5, P6). All participants except one (P1) showed a quantitative increase in mean
length of utterance (MLU) although this was not a significant change (< 2 SD of
normative sample). All except one participant (P1) had a higher proportion of sentences in the post-treatment sample, and this was significant for P2, P5, and P6. The
ratio of open-class to closed-class words showed a significant post-treatment decrease
(agrammatic participants have abnormally high ratios, hence a decrease is a favourable
change) in four participants (all except P1 and P6). All participants except P4 showed
a quantitative increase in the proportion of verbs, and this change was significant for
one participant (P6). Three participants (P3, P4, P5) increased significantly in the proportion of suffixed verbs. Four participants improved significantly in the variety of
tense (all except P1 and P6). Only one participant (P3) improved significantly on the
accuracy of tense measure. The pre-treatment accuracy of tense is relatively high in
some participants (P2) because of the over-use of the present progressive morphology
(is Ving).
DISCUSSION
The primary aim of this study was to investigate the relative extent of within- and
between-verb generalisation that could be achieved with morphosemantic treatment
of tense using regular or irregular verb stimuli. This is a first study that directly
compares the relative efficacy of using regular and irregular verbs in verb tense training, and is a phase II investigation of morphosemantic treatment (Robey, Schultz,
Crawford, & Sinner, 1999; WHO). The data show that morphosemantic training of
both verb types results in significant gains in trained items for all participants. The
morphosemantic treatment also promotes generalised improvement in tenses that are
conveyed by predictable and regular morphological markers (such as V+d, V+s, will
V ). On the other hand, generalisation to untrained irregular past forms is partial,
irrespective of whether regular or irregular verbs are trained. For most participants
there were also changes in related narrative measures such as variety of tense. In the
following sections we discuss the implications of these findings for aphasia rehabilitation, the understanding of verb tense impairments in agrammatism, and the lexical
representation of verb morphology.
Implications for aphasia therapy
First, this study adds to the body of research that demonstrates the efficacy of targeted
aphasia treatment in individuals with chronic aphasia (Holland, Fromm, DeRuyter,
& Stein, 1996). Morphosemantic treatment, as described here and in previous work
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
19
(Faroqi-Shah, 2008), results in significantly improved production of verb tense morphology in elicited speech and modest improvements in narrative speech. The data on
regular rule recovery and generalisation show that all six aphasic individuals were able
to re-learn rules or patterns and generalise the use of V+d, V+s, will V in the correct
temporal context to untrained items.
Given that three tenses for each verb were sequentially trained, it is worth questioning if it is even necessary to train all tenses (or verb forms). In other words, if it is
sufficient to train only one tense and obtain generalisation to untrained tenses (or verb
forms). Two prior studies provide answers to this question. Bastiaanse, Hurkmans,
and Links, (2006) trained only non-finite or finite verbs in two Dutch-speaking participants. The key treatment step was sentence completion and training of temporal
reference was not a part of this treatment. The authors used the Complexity Account
of Treatment Efficacy (Thompson, Shapiro, Kiran, & Sobecks, 2003) to predict that
finite verb treatment would generalise to non-finite verbs since the former are syntactically more complex than the latter. Both participants showed limited improvement
of trained verb types and failed to improve in the untrained verb type. Thompson
et al. (2006) trained either past tense morphology (V+ed) or agreement morphology
(V+s) in a group of 12 participants, and examined generalisation to the untrained
morphology. Temporal context was utilised in this treatment with adverbs Yesterday
and Nowadays. Thompson et al. found mixed patterns of generalisation, with an equal
number of participants generalising or failing to generalise to the untrained morphology. To summarise, the results so far reveal weak generalisation across verb forms,
suggesting that each verb tense and its corresponding form (V+ed, V+s, will V )
might need to be trained before participants will apply the morphological pattern
to untrained verbs in the appropriate temporal context. The exception in generalisation to untrained verb forms are the three participants (P4, P5, P6) who improved
significantly in the regular past form even though regular past was not trained and
one participant (P1) who improved on irregular past when only regulars were trained.
This finding of generalisation across past tense verb forms is further discussed in the
following sections.
This study differed from previous verb morphology treatment research in several
respects (Bastiaanse et al., 2006; Mitchum & Berndt, 1994; Thompson et al., 2006;
Weinrich et al., 1997). First, past tense training involved only one verb type for any
participant, making it relatively straightforward to examine production of past tense
across untrained regular and irregular verbs. Another previous study examined generalisation across regular-irregular morphology (Weinrich et al., 1997), and differed
from the present study in three ways: training used computerised icons, both regular and irregular verbs, and trained six irregular verbs which had different subpatterns
(while the present study trained a homogeneous set of irregulars: vowel change only or
suffixed only). The findings were consistent with the present study: improved production of regular past morphology on untrained verbs, but limited improvements with
untrained irregular past (see also participants TP, PP and SK in Faroqi-Shah, 2008,
for similar findings). While rule-based generalisation for regular morphology was discussed earlier, there are at least two possible explanations for irregular verb outcomes.
From a psycholinguistic perspective, irregular verbs are unlikely to have strong interlinkages with each other in the mental lexicon (Pinker, 1999) and hence exert limited
influence on other irregulars. Second, composition of stimuli used for irregular verb
training presents a caveat: while the present study tested generalisation for irregular
verb types that were not used for training (that is, suffixed irregulars were tested when
Downloaded by [University Of Maryland] at 08:42 04 April 2013
20
FAROQI-SHAH
vowel change irregulars were trained and vice versa), Weinrich et al. (1997) trained
six heterogeneous irregulars. No study has yet trained and tested generalisation within
the same subfamily of irregulars, and therefore this possibility for stronger irregular
verb generalisation remains to be explored.
A second difference with prior treatment studies is the de-emphasis on oral production in the treatment steps. The outcomes showed that there was a general
improvement in oral production of verb morphology even though this was not an
explicit part of the treatment (as mentioned in the section on scoring, P2, P5, and
P6 provided written responses for 5–20% of probes). This finding is consistent with
Faroqi-Shah (2008), who found that accuracy of tense production failed to improve
when participants received extensive oral practice (morphophonological training) but
training on temporal context was not provided. A third difference from prior studies was the inclusion of treatment steps that tapped input processing (grammaticality
judgement and comprehension). The motivation for including these steps was drawn
from prior research showing poor performance in such tasks (Dickey, Milman, &
Thompson, 2008; Faroqi-Shah & Dickey, 2009). Given that Bastiaanse et al. (2006)
failed to find significant treatment gains, and their treatment included oral sentence
completion without any input processing steps, one can infer that the input steps
used in the present study may be augmenting the treatment outcomes. However, it is
preliminary to make definitive statements about the benefit of using input processing
steps without systematically comparing outcomes of morphosemantic treatment with
and without input steps. A final difference with prior treatment studies is the robust
improvement in narrative measures across most participants (except P1).
The implications for aphasia therapy based on the current study and previous
findings are that: (1) morphosemantic training of verb morphology is a promising
treatment approach for tense deficits for participants whose aphasia profile is similar
to those reported in this study and Faroqi-Shah (2008), that is, those with agrammatic
aphasia and minimal apraxia of speech, (2) contextually accurate use of regular verb
morphology can be trained quite successfully. Further, use of regular past ending is
highly likely to improve even when it is not explicitly trained. Findings to date are
less conclusive about generalisation to untrained irregular verb stimuli, irrespective of
whether regular or irregular verbs are used for training, (3) oral practice may not be
necessary for improved outcomes, (4) use of input processing steps may be responsible for the positive outcomes (although this final point needs further verification),
and (5) each tense may need to be trained for generalisation of that tense to untrained
verbs.
The findings of this study also have implications for the Complexity Account of
Treatment Efficacy, which states that the “linguistic complexity of materials selected
for treatment will influence learning and generalisation patterns” (Thompson et al.,
2003, p. 602). In other words, treatment of complex linguistic processes is expected
to produce generalised improvement of linguistically related less complex processes.
Most evidence for the complexity account comes from syntactic (Thompson &
Shapiro, 2007) and semantic treatments (Kiran, 2007). The present study adds morphological treatment to the body of research that supports the complexity account;
namely, the mental representation of irregular verbs is more complex (as per the
evidence presented in the Introduction) and treatment of irregular past tense generalises to the production of regular past tense. Given that P1, the patient with least
severe pre-treatment verb inflection scores, showed improvement for the more complex
irregulars following regular training, further research is necessary to examine the
applicability of CATE to intervention of verb morphology.
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
21
Obviously much more research is warranted to validate the above clinical recommendations. First, although we found improvement in regular past tense for patients
who were trained on irregulars, a caveat is that all patients were also trained on
future and present tenses, which are morphologically regular. In future work, we
intend to examine the potential for generalisation to regular past following exposure to irregular past and no training of other regular morphology. It also remains
to be seen if morphosemantic treatment is a viable treatment approach for individuals
who speak inflectionally rich languages (such as Italian, French, Spanish, Kannada)
where verb morphology represents additional information such as mood, number, and
person. The patterns of generalisation across regular and irregular verbs and verbs
with different conjugations in other languages are also worth investigating. Finally,
morphosemantic treatment in its current form does not emphasise oral production.
It remains to be seen if incorporating oral practice (such as oral sentence completion
or sentence repetition) would enhance treatment outcomes.
Implications for lexical organisation and theories of morphological
processing
The significant improvement in regular past tense production in the irregular trained
group provides neuropsychological support for activation of the regularisation rule
during irregular production and is consistent with theories that invoke either the
Blocking mechanism (Aronoff, 1976; Kiparsky, 1982; Pinker, 1999) or parallel
access of rule-based and storage mechanisms (Augmented Addressed Morphology:
Caramazza, Laudanna, & Romani, 1988; Frauenfelder & Schreuder, 1992; Schreuder
& Baayen, 1995). However, a confound that limits the interpretation of automatic
activation of regularisation rules in the presence of irregulars is that the irregular
treatment group was exposed to other regularisation rules (V+s). In future research
exclusively training irregular past without accompanying simple present or future
tense training will better elucidate if regularisation rules are automatically strengthened with exposure to irregulars. A second point from these results is that the
effectiveness of generalisation to regulars did not vary with the type of irregular
trained (vowel change for P4 and P5 versus suffixed irregulars for P6). This is inconsistent with models that postulate a gradation between regulars and irregulars on the
basis of phonological shape (such as the Distributed Morphology of Halle & Marantz,
1993; and the single route account of Eddington, 2002). This finding is also inconsistent with priming data on unimpaired individuals where suffixed irregulars patterned
with regulars while vowel change irregulars did not (Kielar et al., 2008). It is likely
that there are multiple principles of organisation in the mental lexicon and the various experimental tasks tapped different organisational principles. The third theoretical
implication comes from the finding that untrained irregulars only partially improved
with irregular treatment. The Unique Entry Principle and its recent variants (Glass &
Lau, 2003) propose the relative independence of each irregular verb entry and hence
can accommodate this finding related to limited irregular verb generalisation.
Implications for theories of tense deficits in agrammatism
The patterns of acquisition and generalisation obtained in this study and prior treatment studies provide some insight into the nature of tense deficits in agrammatic
aphasia. Given that training on associating temporal context with verb morphology
Downloaded by [University Of Maryland] at 08:42 04 April 2013
22
FAROQI-SHAH
resulted in improved production of tense morphology, the findings are generally consistent with theoretical accounts of agrammatic aphasia that attribute difficulty with
semantic or extrasentential aspects of tense morphology (Faroqi-Shah & Thompson,
2007; Wenzlaff & Clahsen, 2004). The pattern of tense deficits in this study is
inconsistent with Bastiaanse’s (2008) proposal that only computation of past tense
morphology is impaired in agrammatic aphasia. Our findings of impaired present,
future, and past tense morphology are instead consistent with findings of a generalised
tense impairment (Burchert et al., 2005; Thompson et al., 2006; Wenzlaff & Clahsen,
2004).
The occurrence of over-regularisation errors for irregular targets during the course
of treatment and generalisation testing needs to be accounted for. After producing an
over-regularised verb, participants frequently commented on their unfamiliarity with
the uttered word, but were still unable to generate the correct form. Less frequently,
production of an over-regularised form was followed by production of an accurate
irregular target. Although this over-regularisation pattern has not been reported for
English-speaking agrammatic participants, it is consistent with data from German and
Dutch agrammatic participants described by Penke and Westermann (2006). Their
data, obtained using a sentence completion task (for example in German: Ich schreibe.
Ich habe __ [geschrieben] [I am writing. I have ___ [written]]), revealed 81% and 98%
over-regularisation errors for German and Dutch participles respectively. It is also
noteworthy that the difficulty with irregulars was reported for participles, which are
non-finite (untensed), with the tense information having been conveyed by an already
existing auxiliary (habe or hebben) in the sentence completion task. This predominant
pattern of over-regularisation errors cannot be explained without invoking difficulties
with word form access. A further indication of difficulty with word form access comes
from the limited generalisation to untrained irregulars found in our study. That is,
production of irregular morphology was not facilitated by tense training alone. A third
pointer to word form access is the influence of lexical frequency on the production of
irregular (Penke & Westermann, 2006) and regular verb inflections (Faroqi-Shah &
Thompson, 2004). It is widely reported in psycholinguistic literature that the language
production process most robustly influenced by lexical frequency is word form access
(e.g., Levelt, Roelofs, & Meyer, 1999).
In their present form, two accounts can accommodate a verb form retrieval deficit.
Penke (2003), using the framework of lexicalist approaches to inflectional morphology, proposed that problems in accessing inflectional affixes were at the core of
agrammatic speech. Penke cited two patterns of data in German agrammatic patients
to support this argument: inflectional errors were more likely with lower-frequency
verb forms and in inflectional paradigms with fewer members. Agrammatic errors in
German, such as clause-final verb placement and inflectional errors in the second verb
were argued to be accommodated by this account (Penke, 2001). Additionally, the
Diacritical Encoding and Retrieval (DER) hypothesis can accommodate a word form
retrieval problem that is likely to underlie the over-regularisation errors with irregular
targets. The DER identifies two potential deficits: selection of diacritical features that
correspond to the intended message (+PAST, +PLURAL, etc.) and/or selection of
verb forms that correspond to these diacritics (Faroqi-Shah & Thompson, 2007). The
latter deficit can explain why, despite training with (and knowledge of) past tense, our
participants were unable to retrieve the irregular past and instead applied the default
past tense affixation rule. Faroqi-Shah and Thompson (2004) argued that agrammatic
participants are unable to retrieve verb forms that match the intended tense, and hence
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
23
resort to producing the most easily accessible verb form. In most cases verb forms
with higher word frequencies or higher affix frequencies are thus produced (see also
Centeno, Obler, Cairns, Garro, & Merrifield, 1996 for consistent data from Spanish,
and Luzzatti, Mondini, & Semenza, 2001 for data from Italian). The data with irregular verbs suggest that representational differences between verbs add another level of
complexity that can impact the production of inflections.
Accounts of verb inflection deficits are incomplete if they fail to address another
facet: that of morphophonological complexity, which is often implicated to undermine the production of regular inflections in English (Bird et al., 2003; Braber,
Patterson, Ellis, & Lambon Ralph, 2005; Druks, 2006; Ullman et al., 1997). Druks
(2006) suggested that morphosyntactic (tense and agreement encoding) difficulties
might interact with morphophonological encoding such that only those inflections
that are morphosyntactically impaired experience additional deficits at the level of
morphophonology. This was proposed to account for the relatively spared production
of plural and progressive markers in comparison to that of past tense inflections, and
a greater deficit for regular past versus irregular past tense inflections observed for
Druks’ patient.
In sum, agrammatic participants described here and in previous studies evince multiple levels of difficulties with verb inflections: morphosemantic (diacritical feature
specification), morphosyntactic (as in subject–verb agreement), morpholexical (as in
verb form access/encoding), and morphophonological (Bird et al., 2003; Burchert
et al., 2005; Druks, 2006; Faroqi-Shah & Thompson, 2004, 2007; Friedmann &
Grodzinsky, 1997; Luzzatti & De Blaser, 1996; Wenzlaff & Clahsen, 2004). The emerging pattern is that, while morphosemantic deficits are pervasive (with tense diacritics
most frequently impaired), these may or may not co-occur with other deficits in varying combinations. There is a cascading complexity effect, where deficits at any level
get magnified by deficits at ensuing levels. The recent computational load account
of agrammatic production (Kok et al., 2007) reveals that task demands add another
dimension to this complexity cascade and is not incompatible with the data presented
here.
Conclusions
Psycholinguistic, neuroimaging, and neuropsychological differences in the processing of regularly and irregularly inflected verbs in Germanic languages have played
a central role in theories of linguistic processing. However, there have been hardly
any attempts to apply these findings to rehabilitation of morphological deficits
in aphasia (but see Thompson et al., 2006, for application of linguistic theory).
This study presents one such attempt in which representational differences between
verbs were used as a framework to guide treatment. The primary finding is that
re-learning of morphological rules is quite successful in persons with agrammatic
aphasia with morphosemantic treatment (within-verb type generalisation). The potential for improvement in rule-based morphology following training with irregular verbs
(between-verb type generalisation) is promising, but needs further research before
conclusive statements can be made.
Manuscript received 16 September 2011
Manuscript accepted 12 November 2012
First published online 7 March 2013
24
FAROQI-SHAH
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REFERENCES
Alegre, M., & Gordon, P. (1999). Frequency effects and the representational status of regular inflections.
Journal of Memory and Language, 40, 41–61.
Arabatzi, M., & Edwards, S. (2002). Tense and syntactic processes in agrammatic speech. Brain and
Language, 80, 314–327.
Aronoff, M. (1976). Word formation in generative grammar. Cambridge, MA: MIT Press.
Avrutin, S. (2000). Comprehension of discourse-linked and non-discourse-linked questions by children and
Broca’s aphasics. In Y. Grodzinsky, L. Shapiro, & D. Swinney (Eds.), Language and the brain (pp.
295–315). San Diego, CA: Academic Press.
Baayen, R., Piepenbrock, R., & van Rijn, H. (1993). The celex lexical database (release 1) [CD-rom].
Philadelphia: Linguistic Data Consortium; University of Pennsylvania [distributor].
Balaguer, R. D. D., Costa, A., Sebastian-Galles, N., Juncadella, M., & Caramazza, A. (2004). Regular
and irregular morphology and its relationship with agrammatism: Evidence from two Spanish-Catalan
bilinguals. Brain and Language, 91, 212–222.
Bastiaanse, R. (2008). Production of verbs in base position by Dutch agrammatic speakers: Inflection versus
finiteness. Journal of Neurolinguistics, 21, 104–119.
Bastiaanse, R., Hurkmans, J., & Links, P. (2006). The training of verb production in Broca’s aphasia: A
multiple baseline across-behaviors study. Aphasiology, 20, 298–311.
Benedet, M. J., Christiansen, J. A., & Goodglass, H. (1998). A cross-linguistic study of grammatical
morphology in Spanish- and English-speaking agrammatic patients. Cortex, 34, 309–336.
Bird, H., Lambon Ralph, M. A., Seidenberg, M. S., McClelland, J. L., & Patterson, K. (2003). Deficits in
phonology and past-tense morphology: What’s the connection? Journal of Memory and Language, 48,
502–526.
Boser, K. I., Weinrich, M., & McCall, D. (2000). Maintenance of oral production in agrammatic aphasia:
Verb tense morphology training. Neurorehabilitation & Neural Repair, 14, 105–118.
Boo, M., & Rose, M. L. (2011). The efficacy of repetition, semantic, and gesture treatments for verb retrieval
and use in Broca’s aphasia. Aphasiology, 25, 154–175.
Braber, N., Patterson, K., Ellis, K., & Lambon Ralph, M. A. (2005). The relationship between phonological
and morphological deficits in Broca’s aphasia: Further evidence from errors in verb inflection. Brain and
Language, 92, 278–287.
Burchert, F., Swoboda-Moll., M., & De Bleser, R. (2005). Tense and agreement dissociations in German
agrammatic speakers: Underspecification vs. hierarchy. Brain and Language, 94, 188–199.
Busk, P., & Serlin, R. (1992). Meta-analysis for single case research. In T. Kratochwill & J. Levin (Eds.),
Single-case research design and analysis (pp. 197–198). Hillsdale, NJ: Lawrence Erlbaum Associates.
Bybee, J. (1995). Diachronic and typological properties of morphology and their implications for representation. In L. B. Feldman (Ed.), Morphological aspects of language processing (pp. 225–246). Hillsdale,
NJ: Lawrence Erlbaum Associates.
Bybee, J., & Slobin, D. (1982). Rules and schemas in the development and use of the English past tense.
Language, 58, 265–289.
Caramazza, A., Laudanna, A., & Romani, C. (1988). Lexical access and inflectional morphology. Cognition,
28, 297–332.
Centeno, J., Obler, L. K., Cairns, H., Garro, L., & Merrifield, P. (1996). Frequency effects on verb inflection
use by agrammatic Spanish-speaking subjects. Brain and Language, 57, 47–50.
Clahsen, H. (1999). Lexical entries and rules of language: A multidisciplinary study of German inflection.
Behavioral and Brain Sciences, 22, 991–1060.
Clahsen, H., Eisenbeiss, S., Hadler, M., & Sonnenstuhl, I. (2001). The mental representation of inflected
words: An experimental study of adjectives and verbs in German. Language, 77, 510–543.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement,
20, 37–46.
Conroy, P., Sage, K., & Lambon Ralph, M. A. (2009). A comparison of word versus sentence cues as therapy
for verb naming in aphasia. Aphasiology, 23, 462–482.
Dabul, B. (2000). Apraxia Battery for Adults – Second edition (ABA-2). Austin, TX: Pro-Ed.
Dickey, M. W., Milman, L. H., & Thompson, C. K. (2008). Judgment of functional morphology in
agrammatic aphasia. Journal of Neurolinguistics, 21, 35–65.
Druks, J. (2006). Morphosyntactic and morphophonological deficits in the production of regularly and
irregularly inflected verbs. Aphasiology, 20, 993–1017.
Eddington, D. (2002). Dissociation in Italian conjugations: A single-route account. Brain and Language, 81,
291–302.
Downloaded by [University Of Maryland] at 08:42 04 April 2013
REGULAR AND IRREGULAR VERB TREATMENT
25
Edmonds, L. A., & Babb, M. (2011). Effect of verb network strengthening treatment in moderate-to-severe
aphasia. American Journal of Speech-Language Pathology, 20, 131–145.
Faroqi-Shah, Y. (2007). Are regular and irregular verbs dissociated in nonfluent aphasia? A meta-analysis.
Brain Research Bulletin, 74, 1–13.
Faroqi-Shah, Y. (2008). A comparison of two theoretically driven treatments of verb inflection deficits in
aphasia, Neuropsychologia, 46, 3088–3100.
Faroqi-Shah, Y., & Dickey, M. W. (2009). On-line processing of tense and temporality in agrammatic
aphasia. Brain and Language, 108, 97–111.
Faroqi-Shah, Y., & Thompson, C. K. (2004). Semantic, lexical, and phonological influences on the
production of verb inflections in agrammatic aphasia. Brain and Language, 89, 484–498.
Faroqi-Shah, Y., & Thompson, C. K. (2007). Verb inflections in agrammatic aphasia: Encoding of tense
features. Journal and Memory and Language, 56, 129–151.
Frauenfelder, U. H., & Schreuder, R. (1992). Constraining psycholinguistic models for morphological processing and representation: The role of productivity. In G. E. Booij & J. van Marle (Eds.), Yearbook of
morphology (pp. 165–183). Dordrecht: Kluwer.
Friedmann, N., & Grodzinsky, Y. (1997). Tense and agreement in agrammatic production: Pruning the
syntactic tree. Brain & Language, 56, 397–425.
Garrett, M. F. (1988). Processes in language production. In N. Frederick (Ed.), Linguistics: The Cambridge
survey (Vol. 3, pp. 69–96). Cambridge, MA: Cambridge University Press.
Glass, A., & Lau, J. (2003). Grammatical intuitions about irregular verb inflections. American Journal of
Psychology, 116, 51–70.
Goodglass, H., Christiansen, J. A., & Gallagher, R. (1993). Comparison of morphology and syntax in free
narrative and structured tests: Fluent vs. nonfluent aphasics. Cortex, 29, 377–407.
Goodglass, H., Kaplan, E., & Barresi, B. (2001). Boston diagnostic aphasia examination, 3rd edition.
Philadelphia, PA: Lippincott Williams & Wilkins.
Halle, M., & Marantz, A. (1993). Distributed morphology and the pieces of inflection. In K. Halle & S.
J. Keyser (Eds.), The view from building 29: Essays in linguistics in honor of Sylvian Bromberger (pp.
111–175). Cambridge, MA: MIT Press.
Helm-Estabrooks, N., & Nicholas, M. (2003). Narrative story cards (6th ed.). Austin, TX: PRO-ED.
Holland, A. L., Fromm. D. S., DeRuyter, F., & Stein, M. (1996). Treatment efficacy: Aphasia. Journal of
Speech and Hearing Research, 39, S27–S36.
Inglis, A. L. (2005). Is ‘time’ up for the irregularity of tense in aphasia, in relation to semantics and the
temporal lobe? Brain and Language, 95, 189–191.
Janssen, U., & Penke, M. (2002). How are inflectional affixes organised in the mental lexicon? Evidence
from the investigation of agreement errors in agrammatic aphasics. Brain and Language, 81, 180–191.
Kertesz, A. (1982). Western aphasia battery. New York, NY: Grune & Stratton.
Kielar, A., Joanisse, M., & Hare, M. L. (2008). Priming English past tense verbs: Rules or statistics? Journal
of Memory and Language, 58, 327–346.
Kiparsky, P. (1982). From cyclic phonology to lexical phonology. In H. van der Hulst & N. Smith (Eds.),
The structure of phonological representations (Part 1). Dordrecht: Foris.
Kiran, S. (2007). Complexity in the treatment of naming deficits. American Journal of Speech Language
Pathology, 16, 18–29.
Kok, P., van Doorn, A., & Kolk, H. (2007). Inflection and computational load in agrammatic speech. Brain
and Language, 102, 273–283.
Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production.
Behavioral and Brain Sciences, 22, 1–75.
Linebarger, M. C., Schwartz, M. F., & Saffran, E. M. (1983). Sensitivity to grammatical structure in socalled agrammatic aphasics. Cognition, 13, 361–392.
Luzzatti, C., & De Blaser, R. D. (1996). Morphological processing in Italian agrammatic speakers: Eight
experiments in lexical morphology. Brain and Language, 54, 26–74.
Luzzatti, C., Mondini, S., & Semenza, C. (2001). Lexical representation and processing of morphologically
complex words: Evidence from the reading performance of an Italian agrammatic patient. Brain and
Language, 79, 345–359.
McClelland, J. L., & Patterson, K. (2002). Rules or connections in past-tense inflections: What does the
evidence rule out? Trends in Cognitive Sciences, 6, 465–472.
Miller, J. (2004). Systematic analysis of language samples, Language Analysis Lab., Waisman Center,
University of Wisconsin-Madison.
Downloaded by [University Of Maryland] at 08:42 04 April 2013
26
FAROQI-SHAH
Miozzo, M. (2003). On the processing of regular and irregular forms of verbs and nouns: Evidence from
neuropsychology. Cognition, 87, 101–127.
Mitchum, C., & Berndt, R. (1994). Verb retrieval and sentence construction: Effects of targeted intervention. In M. J. Riddoch & G. Humphreys (Eds.), Cognitive neuropsychology and cognitive rehabilitation.
Hove, UK: Lawrence Erlbaum Associates.
O’Connor, B. A., Obler, L. K., & Goral, M. (2007). The importance of verb form-regularity in agrammatism. Brain and Language, 103, 29–30.
Okrent, A. (2004). From meaning to words: An investigation of past tense verb inflection in English comparing
a form to form mapping task with a meaning to form mapping task (Unpublished doctoral dissertation).
University of Chicago, Chicago, IL.
Olive, M. L., & Smith, B. (2005). Effect size calculations and single subject designs. Educational Psychology,
25, 313–324.
Penke, M. (2001). Controversies about CP: A comparison of language acquisition and language impairments in Broca’s aphasia. Brain and Language, 77, 351–363.
Penke, M. (2003). On the morphological basis of syntactic deficits. Brain and Language, 87, 50–51.
Penke, M., & Janssen, U. (1999). The representation of inflectional morphology: Evidence from Broca’s
aphasia. Brain and Language, 68, 225–232.
Penke, M., Janssen, U., & Eisenbeiss, S. (2004). Psycholinguistic evidence for the underspecification of
morphosyntactic features. Brain and Language, 90, 423–433.
Penke, M., & Westermann, G. (2006). Broca’s area and inflectional morphology: Evidence from Broca’s
aphasia and computer modeling. Cortex, 42, 563–576.
Pinker, S. (1984). Language learnability and language development. Cambridge, MA: Harvard University
Press.
Pinker, S. (1999). Words and rules: The ingredients of language. New York, NY: HarperCollins.
Pinker, S., & Prince, A. (1988). On language and connectionism: Analysis of a parallel distributed processing model of language acquisition. In S. Pinker & J. Mehler (Eds.), Connections and symbols
(pp. 73–193). Cambridge, MA: MIT Press.
Pinker, S., & Prince, A. (1991). Regular and irregular morphology and the psychological status of rules of
grammar. Berkeley Linguistics Society, 17, 230–251.
Robey, R., Schultz, M., Crawford, A., & Sinner, C. (1999). Single-subject clinical-outcome research:
Designs, data, effect sizes, and analyses. Aphasiology, 13, 445–473.
Rochon, E., Laird, L., Bose, A., & Scofield, J. (2005). Mapping therapy for sentence production impairments in nonfluent aphasia. Neuropsychological Rehabilitation, 15, 1–36.
Rumelhart, D., & McClelland, J. (1986). On learning the past tenses of English verbs: Implicit rules or
parallel distributed processing. In D. Rumelhart & J. McClelland (Eds.), Parallel distributed processing:
Explorations in the microstructure of cognition (pp. 216–271). Cambridge, MA: MIT Press.
Schreuder, R., & Baayen, R. H. (1995). Modeling morphological processing. In L. B. Feldman
(Ed.), Morphological aspects of language processing (pp. 131–154). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Stemberger, J. P. (2002). Overtensing and the effect of regularity. Cognitive Science, 26, 737–766.
Thompson, C. K., Milman, L. H., Dickey, M. W., O’Connor, J. E., Bonakdarpour, B., Fix, S. C., . . .
Arcuri, D. F. (2006). Functional category production in agrammatism: Treatment and generalisation
effects. Brain and Language, 99, 79–81.
Thompson, C. K., & Shapiro, K. (2007). Complexity in treatment of syntactic deficits. American Journal of
Speech Language Pathology, 16, 30–42.
Thompson, C. K., Shapiro, K., Kiran, S., & Sobecks, J. (2003). The role of syntactic complexity in treatment
of syntactic deficits in agrammatic aphasia: The Complexity Account of Treatment Efficacy (CATE).
Journal of Speech, Language and Hearing Research, 46, 591–607.
Tsapkini, K., Jarema, G., & Kehayia, E. (2002). A morphological processing deficit in verbs but not in
nouns: A case study in a highly inflected language. Journal of Neurolinguistics, 15, 265–288.
Ullman, M. T., Corkin, S., Coppola, M., Hickok, G., Growdon, J. H., Koroshetz, W. J., & Pinker, S.
(1997). A neural dissociation within language: Evidence that the mental dictionary is part of declarative memory, and that grammatical rules are processed by the procedural system. Journal of Cognitive
Neuroscience, 9, 266–276.
Ullman, M. T., & Pinker, S. (2002). The past and future of the past tense. Trends in Cognitive Sciences, 6,
456–463.
REGULAR AND IRREGULAR VERB TREATMENT
27
Varlakosta, S., Valeonti, N., Kakavoulia, M., Lazaridou, M., Economou, A., & Protopapas, A. (2006).
The breakdown of functional categories in Greek aphasia: Evidence from agreement, tense and aspect.
Aphasiology, 20, 723–743.
Weinrich, M., Shelton, J. R., Cox, D. M., & McCall, D. (1997). Remediating production of tense
morphology improves verb retrieval in chronic aphasia. Brain and Language, 58, 23–45.
Wenzlaff, M., & Clahsen, H. (2004). Tense and agreement in German agrammatism. Brain and Language,
89, 57–68.
Zimmermann, K. (2001). Full-form storage and blocking in morphology and syntax (Unpublished doctoral
dissertation). The State University of New Jersey, New Brunswick.
APPENDIX A
Downloaded by [University Of Maryland] at 08:42 04 April 2013
Verbs used for treatment and generalisation testing, listed by verb type and participant
conditions.
Regular verbs
Treatment for P1,
P2, P3
rake
peel
wrap
shave
pull
wash
kick
smoke
hug
rob
play
fold
juggle
sew
ski
paint
carry
dry
tie
skate
Irregular verbs
Generalisation for all
participants
iron
water
plant
climb
mow
comb
push
empty
kiss
vacuum
jump
open
bake
clean
cross
pour
sail
wipe
save
scrub
Treatment for P4,
P6Generalisation for
P1, P2, P3, P5
ring
sing
swim
sink
break
wear
tear
steal
drink
wake
sit
eat
drive
light
dig
fly
write
throw
lead
bite
Treatment for P5
Generalisation for all other
participants
sleep
keep
sell
sweep
deal
weep
leave
buy
hear
kneel
creep
lose
leap
flee
catch
tell
dream
feel
teach
pay
28
FAROQI-SHAH
APPENDIX B
Normative values for narrative speech obtained from 13 age-matched healthy normal
adults.
Downloaded by [University Of Maryland] at 08:42 04 April 2013
Rate (words per minute)
MLU
Proportion sentences
Open:closed class ratio
Proportion verbs
Proportion suffixed verbs
Variety of tense
Accuracy of tense
Mean
SD
136.06
10.80
0.97
0.79
0.15
0.63
0.08
0.94
26.53
1.63
0.03
0.07
0.02
0.07
0.02
0.08
Age range = 46 to 66 years, Mean age: 56.6; seven females, Mean years of education = 18.1.