University of Groningen
Case assignment by prepositions in Russian aphasiaTrofimova, Maria
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.
Document VersionPublisher's PDF, also known as Version of record
Publication date:2009
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):Trofimova, M. (2009). Case assignment by prepositions in Russian aphasia. Groningen: s.n.
CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.
Download date: 15-01-2020
133
CHAPTER VI
SENTENCE COMPLETION EXPERIMENT
A brief review of previous findings on case-morphology in aphasia
To introduce the main topic of the study presented in this chapter, a short survey
of the relevant previous findings is provided here. A fuller review of the previous
research, which provides the background for this thesis, can be found in chapter III.
Broca’s aphasia is typically considered to be a grammatical disorder causing impairment
of the morphosyntactic component of speech. Speech production in a ‘telegraphic style’,
with omissions and substitutions of free and bound grammatical morphemes, is a typical
characteristic of Broca’s aphasia. The distribution patterns of omissions and substitutions
have been shown to be language specific; Broca’s aphasic speakers have been reported to
omit functional morphemes only if they do not form an obligatory part of a word, and
when their omissions do not result in non-words in a particular language (Avrutin, 2001;
Grodzinsky, 1984; Kehayia et al., 1990; Kertesz & Osmán-Sági, 2001; Lehečková, 1988,
Miceli et al., 1984; Miceli et al., 1989; Tsapkini et al., 2001). In Broca’s aphasia, free
grammatical morphemes are believed to be more liable to omissions, while bound
morphemes are more prone to substitutions. Case-marking morphemes in many synthetic
and inflecting languages with rich morphology are usually bound morphemes. Previous
studies have reported that the production of correct case-marking morphemes is
problematic for Broca’s aphasic speakers, and these morphemes are subject to
substitution rather than omission errors within the same grammatical category
(Lehečková, 1988, 2001; MacWhinney & Osmán-Sági, 1991; Ulatowska et al., 2001).
Production of correct case-marking morphemes and other elements with an overt
manifestation of case, such as determiners, has been found to depend to a considerable
extent on the realization of the case-assigning category of verbs (Ruigendijk et al., 1999;
Ruigendijk & Bastiaanse, 2002). Prepositions belong to the free-standing morphemes
CHAPTER VI
134
known to be vulnerable in aphasia, and were reported to be misused (Beyn et al., 1979;
Leikin, 1996; Lehečková, 2001; Tesak & Hummer, 1994; Tsvetkova & Glozman, 1975)
and misunderstood in Broca’s aphasia (Friederici, 1985; Friederici et al., 1982; Mack,
1981; Schwartz et al., 1980).
Wernicke’s aphasia is generally characterized by paragrammatic speech with
substitutions rather than omissions of function words and grammatical morphemes.
Previously, this type of aphasia was considered to affect the lexical-semantic domain and
cause lexical-retrieval difficulties. That is why the (morpho-)syntactic problems of
Wernicke’s aphasic speakers were neglected for so long. Indeed, the extensive lexical
difficulties of Wernicke’s aphasic speakers are more widespread and obvious than their
syntactic problems; however, they are no longer disclaimed. Wernicke’s aphasic speakers
were found to make errors in verb inflections (Butterworth & Howard, 1987), and to
produce inaccurate syntactic structures (Edwards, 2002), particularly with respect to
complex sentences (Bastiaanse et al., 1996; Bastiaanse & Edwards, 2004; Edwards, 1995;
Edwards & Bastiaanse, 1998; Edwards et al., 1994; Martin & Blossom-Stach, 1986).
Several studies describe the problems that Wernicke’s aphasic speakers have with
production of grammatical morphemes and, specifically, with case-marking morphemes
(Bates et al., 1991; Edwards, 2002; Martin & Blossom-Stach, 1986). Wernicke’s aphasic
speakers have been reported to produce pronouns incorrectly marked for case (Edwards,
2002; Martin & Blossom-Stach, 1986), to use incorrect determiners, to produce incorrect
prepositions in adverbial phrases of time and place, and to substitute prepositions in verb
argument structures (Martin & Blossom-Stach, 1986). Moreover, similar to Broca’s
aphasic speakers, Wernicke’s aphasic speakers were found to attach incorrect case-
inflections to nouns, which resulted in case-substitution errors (Bhatnagar, 1980;
Ruigendijk, 2002) and to produce incorrect case-marking morphemes of the complements
of verbs (Bates & Wulfeck, 1989; Bhatnagar, 1980). With respect to prepositions in
Wernicke’s aphasia, it has been stated that they are subject to substitution errors within
the same category (Friederici, 1981). Moreover, Russian Wernicke’s aphasic speakers
have been found to substitute prepositions and produce case morphemes appropriate for
the produced prepositions (Beyn et al., 1979).
SENTENCE COMPLETION EXPERIMENT
135
The main goal of the study
One of the previous studies reported here (described in chapter V) analyzed production of
case-assigning prepositions and their complement nouns in the narrative speech of fluent
and non-fluent aphasic speakers. It was found that, unlike fluent aphasic speakers, non-
fluent aphasic speakers produced fewer prepositions and complete correct prepositional
phrases in their spontaneous speech and narrative samples compared to the control group.
Both groups of aphasic speakers were found to be able to assign the correct case to
complements of prepositions when these prepositions were present in their speech.
Within Case theory (Chomsky, 1981, 1986) prepositions, on par with verbs, are regarded
as principle case assigners; therefore, this finding is in line with earlier aphasiological
cross-linguistics research which, particularly for Russian, states that realization of a
correct case-assigning verb is essential for production of nouns with correct case-marking
morphemes (Ruigendijk, 2002). However, it is known that the language deficits of
aphasic speakers can be manifested to a different degree in different tasks and
circumstances (Bastiaanse, 1995; Caplan et al., 2006; Lehečková, 2001). Thus, the main
aim of the study presented here is to describe production of prepositions and their
dependent case-marking morphemes on nouns by aphasic speakers in experimental
conditions, and to give an account for the errors observed in the performance of the
aphasic speakers and the distribution patterns of these errors. This study investigates
whether the presence of a correct case-assigning preposition, similar to case-assigning
verbs (Ruigendijk, 2002), leads to production of correct case-marking morphemes on a
complement noun of this case-assigning preposition. The experiment needs to elucidate
several issues addressed in the research questions that are formulated in the next section.
The research questions
The data acquisition for the study of narrative speech (chapter V) was administered in an
informal conversational style, which could have allowed aphasic speakers to avoid using
constructions that are problematic for them. The non-fluent aphasic speakers were found
CHAPTER VI
136
to produce fewer prepositions than the control group; between the fluent aphasic speakers
and the control speakers no difference emerged. Next, the performance of the aphasic
speakers on production of prepositions in experimental tasks was examined. A sentence
completion test with pictures was designed, and included two subtests to target the
questions posed in the study. In the first subtest of the sentence completion test, in order
to complete a sentence the participants were required to produce a preposition and to
case-mark its complement noun, which was provided in the nominative case. This subtest
is hereafter referred to as Subtest-1, ‘noun provided’. In the second subtest, the
participants had to produce a noun with a correct case-marking morpheme while its case-
assigning preposition was provided. This subtest is hereafter referred to as Subtest-2,
‘case assigner provided’. Both subtests, the testing procedures, requirements, and test
materials will be described in more detail below.
The first question to be addressed concerns production of prepositional phrases in
quantitative terms. Namely, this work considers whether aphasic speakers perform worse
than the control speakers when they have to produce a preposition and case-mark its
complement noun provided in the nominative case form. Subtest-1, ‘noun provided’
targets this question. It facilitates the examination not only of the ability of participants to
produce correct prepositional phrases, but also of the incorrect items, and an investigation
of which errors aphasic speakers make with respect to production of prepositions in
obligatory contexts; in other words, it enables the study of whether prepositions were
omitted or produced incorrectly.
The second question addressed in this experiment considers case-marking
morphemes produced by aphasic speakers on nouns complements of prepositions. It
investigates whether aphasic speakers are able to produce a noun with the correct case-
marking morpheme when its case-assigning preposition is provided. Subtest-2, ‘case
assigner provided’ also investigates another issue concerning case marking, namely, what
happens to case-marking morphemes when obligatory case assigners are omitted or
produced incorrectly.
Finally, it is questioned whether there is direct relationship between the presence
and absence of case-assigning prepositions in experimental sentences and the production
of correct case-marking morphemes of nouns. In other words, this experiment examines
SENTENCE COMPLETION EXPERIMENT
137
whether the presence of (correct) case-assigning prepositions assists correct case
marking. The two subtests have been designed with minimal differences, namely, the
presence (in subtest-2, ‘case assigner provided’) or absence (in subtest-1, ‘noun
provided’) of case assigners in the experimental sentences, which permits direct
comparisons between them to address the issue at hand.
Hypotheses
The control group is expected to perform well on the task; they are not expected to omit
obligatory prepositions or to produce incomplete or ungrammatical sentences. With
respect to case assignment and case marking, no errors are anticipated. With regard to
performance of aphasic speakers, prepositions in Russian are free-standing grammatical
morphemes, which are known to be vulnerable in non-fluent aphasia. This was shown in
the analysis of narrative speech described in chapter II of this thesis: the non-fluent
aphasic speakers produced fewer prepositions than the control speakers, whereas the
performance of the fluent aphasic speakers did not differ from the control group. It is
therefore expected that when required to produce prepositions in the experimental
sentences, the non-fluent aphasic speakers will omit obligatory prepositions from their
responses. However, taking into account the results of Ruigendijk’s (2002) study into
case assignment by Russian verbs, and the results of the analysis of case assignment with
respect to prepositions in the narrative speech of the non-fluent aphasic speakers
described in this thesis, the non-fluent aphasic speakers are hypothesized to be able to
produce correct case marking on nouns when they manage to produce correct case-
assigning prepositions, and especially when these case assigners are already provided in
the experimental sentences. Moreover, presence of correct case-assigning prepositions is
expected to assist case marking by the non-fluent aphasic speakers; therefore, in the
presence of case-assigning prepositions, the non-fluent aphasic speakers are expected to
be able to deal with case-morphology on nouns, although not faultlessly.
Prepositions have been reported to be substituted by fluent aphasic speakers
(Friederici, 1981); in addition, complement nouns of such prepositions have been found
CHAPTER VI
138
to be case-marked in accordance with these prepositions by Russian fluent aphasic
speakers (Beyn et al., 1979). A similar observation was made in the analysis of narrative
speech discussed in chapter V. In narrative speech, fluent aphasic speakers produced as
many prepositions as the neurologically intact control group. In this experiment, the
fluent aphasic speakers were not expected to omit obligatory prepositions; rather, they
might tend to substitute these prepositions and produce other incorrect prepositions that
do not match the context. Therefore, it is hypothesized that when fluent aphasic speakers
substitute prepositions they will produce the complement nouns with case-marking
morphemes suitable for these prepositions. It is not expected from either group of aphasic
speakers that they will produce noun stems deprived of case-marking morphology, which
would lead to the production of nonsensical words.
To recapitulate, there are two hypotheses for the present study:
° When non-fluent aphasic speakers manage to produce a correct case-assigning
preposition, they will also be able to produce a correct case-marking morpheme of
the complement noun of this preposition; therefore, the presence of a correct case-
assigning preposition will facilitate case assignment.
° When fluent aphasic speakers substitute case-assigning prepositions, they will
produce their complement nouns with the case-marking morphemes appropriate
for the produced prepositions.
Methods
Participants
The experiment was administered to a group of 22 Russian-native aphasic speakers and
to a control group of 22 neurologically intact speakers. In the control group all speakers
were volunteers; their age ranged in years from 28 to 54, with mean age of 41.3 years; the
group included 10 male and 12 female participants. None of the participants in the
control group had any previous history of neurological disease or any other major illness;
their vision was normal or corrected to normal. All participants were right-handed native
SENTENCE COMPLETION EXPERIMENT
139
speakers of the Russian language, originating from the central part of the Russian
Federation; none of them displayed any accent. The control group of neurologically intact
speakers is, for the sake of brevity, further referred to as the control speakers. The
individual data of the control speakers are presented in Appendix III, table I; their group
data are summarized in table 6.1 below.
Table 6.1 Group data of all participants who participated in the sentence completion
experiment
Number of participants
Group mean age in years
Gender
Control speakers 22 41.3 10 male 12 females
Fluent aphasic speakers 8 43.25 3 males 5 females
Non-fluent aphasic speakers
12 38.83 7 males 5 females
All aphasic speakers were tested in rehabilitation centers in Russia,27
27 The neurological department of the Rehabilitation center “Zelenyj gorod”, Nizhnij Novgorod; Institute of the Human Brain, Russian Academy of Science, Saint-Petersburg; Federal Center of Speech Pathology and Neurorehabilitation, Moscow, Russia.
where they
participated in a course of neurological rehabilitation and a course of speech and
language therapy. They were diagnosed in the rehabilitation centers by speech therapists
on the basis of Luria’s classification (Luria, 1973). The diagnosis and speech
characteristics of the aphasic speakers were taken into account when they were judged as
being fluent or non-fluent. In the group of 8 fluent aphasic speakers, the age ranged in
years from 17 to 69, with a mean age of 43.25 years; 3 males and 5 females were in the
group. The data of aphasic speaker Fl 6 were excluded from the analysis due to failure to
complete the experiment. All fluent aphasic speakers were aphasic as a result of
cerebrovascular accidents in the left hemisphere, except speaker Fl 20, whose aphasia
etiology was traumatic as the result of a gunshot wound and consequent surgical removal
of a subdural hemorrhage in the left hemisphere. The aphasia types, established in
CHAPTER VI
140
accordance with Luria’s classification, differed. Aphasic speaker Fl1 was diagnosed as
having sensory aphasia with elements of acoustic-gnostic aphasia; Fl 10 was diagnosed
with amnestic aphasia; Fl19 with afferent motor aphasia with elements of acoustic-
gnostic aphasia, Fl7, Fl14, Fl17, Fl20, and Fl24 were diagnosed with afferent motor
aphasia; these are all known to be fluent aphasias. Neurological data and the linguistic
characteristics of the fluent aphasic speakers are shown in Appendix VI, table I. In the
group of 12 non-fluent aphasic speakers the age ranged in years from 22 to 58; the mean
age in the group was 38.83 years; there were 7 male and 5 female participants in the
group. Aphasia etiology in the group was diverse; aphasic speakers Nf 3, Nf 11, Nf 13,
Nf 15, Nf 21, and Nf 23 were aphasic due to cerebrovascular accidents in the left
hemisphere; aphasic speakers Nf 2, Nf 4, Nf 9, Nf 16, Nf 18, and Nf 22 suffered from
traumatic brain injury as the result of car accidents. The non-aphasic speakers were
diagnosed with efferent-motor aphasia. All aphasic speakers were at least six months
post-onset.
Materials
To assess the case-assigning abilities of the aphasic speakers a sentence completion
experiment was designed. For each item a black-and-white line drawing consisting of an
incomplete sentence with a gap marked by dots was created. The experiment consisted of
two subtests, with 98 items each. For both subtests, in 50 experimental sentences the case
assigners were prepositions, and in 48 control sentences the case assigners were verbs.
The results of all sentences later entered data analysis. All prepositions used in the
experimental sentences were non-derived primary prepositions used in their lexical
function (see chapter II). As discussed previously, Russian prepositions assign five out of
six cases (excluding the nominative), thus, five cases were tested in combination with
prepositions in this experiment (genitive, dative, accusative, instrumental, and
prepositional). Russian verbs assign five out of six cases; these are the nominative case of
subjects of sentences and another four oblique cases excluding the prepositional. In this
experiment four oblique cases were tested in combination with verbs (genitive, dative,
SENTENCE COMPLETION EXPERIMENT
141
accusative, and instrumental). Examples of case assignment by prepositions and verbs are
provided in table 6.2.
Table 6.2 Examples of case assignment by prepositions and verbs in Russian
Case assignment by prepositions Case assignment by verbs Nominative - Devochka spit
“A girl sleeps” Genitive Sasha stoit okolo stula Katja boitsja sobaki
“Sasha stand next to the chair” “Katja is afraid of the dog” Dative Korabl’ plyvet po okeanu Muzhchina zvonit synu
“A ship sails in the ocean” “A man calls his son” Accusative Sobaka prygaet cherez koshku Petja gladit rubashku
“A dog jumps over the cat” “Petja irons a shirt” Instrumental Kot lezhit pod stolom Sobaka viljaet khvostom
“A cat is lying under the table” “A dog wags its tail” Prepositional Kot sidit na divane -
“A cat sits on the sofa”
In subtest 1, ‘noun provided’, a case assigner was missing from the sentences,
while a noun was provided in the pictures in the nominative (default) case. For this
subtest, the participants were instructed to complete the sentences and were required to
use the given noun in their answers. The participants were also explicitly reminded to
change the word form of the noun if necessary. Examples of test materials used in
subtest-1, ‘noun provided’ are presented in figure 6.1.
CHAPTER VI
142
Sobaka prygaet … (koshka) Muzhchina … (zapiska)
The dog jumps … (cat) The man … (a note)
The answers expected from a participant:
Sobaka prygaet cherez koshku [ACC.case] Muzhchina pishet zapisku [ACC.case]
The dog jumps over the cat The man writes a note
Figure 6.1 Examples of test materials for Subtest-1, ‘noun provided’: a case assigner is
missing; a noun is provided in the nominative (default) case (left: case assignment by a
preposition; right: case-assignment by a verb)
In subtest 2, in all sentences a case assigner was provided, but a noun phrase that
had to be case-marked was missing. Participants were required to complete the sentences
with an appropriate noun correctly marked for case. This subtest addressed the issue of
whether there is a direct relationship between the presence of a preposition and correct
case marking of a noun. It allowed for the investigation of whether the presence of case-
assigning prepositions assists correct case marking. Furthermore, subtest 2 is labeled as
subtest-2, ‘case assigner provided’, referring to the element present in the sentences – a
case assigner. Examples of the test materials of subtest-2, ‘case assigner provided’ are
presented in figure 6.2.
SENTENCE COMPLETION EXPERIMENT
143
Devochka nyrjaet v … Devochka gladit …
The girl dives into … The girl strokes …
The answers expected from a participant:
Devochka nyrjaet v vodu [PREP.case] Devochka gladit koshku [ACC.case]
The girl dives into the water The girl strokes a cat
Figure 6.2 Examples of test materials for Subtest-2, ‘case assigner provided’: a case
assigner is provided, a noun is missing (left: case assignment by a preposition; right:
case assignment by a verb)
When test materials for the experiment were designed, all nouns were checked for
case-homonymous forms, and these were not used in the test in order to avoid confusion
in scoring. To denote objects shown in the pictures, nouns of feminine and masculine
genders of the first and second declension classes were used, which are typical and
representative of their class.
CHAPTER VI
144
Procedure
Prior to testing, the participants were instructed by the experimenter. They were asked to
look at a picture on the computer screen, to read aloud the sentence underneath it, and to
fill in the gap. To familiarize the participants with the task, a trial session of 4 items was
administered, from which feedback was provided. During the actual task the examiner
did not give participants any feedback on their performance. During testing, a participant
and the experimenter would typically sit in a quiet room facing the computer. Testing
started with a slide with instructions asking participants to complete the sentences
provided with each picture. The participants were allowed to take as much time as they
needed to complete the task. Both subtests made use of the same stimuli; that is why each
subtest was subdivided into two blocks: for subtest-1, ‘noun provided’, block a-1
included items from 1 to 49, and block a-2 included items from 50 to 98. For subtest-2,
‘case assigner provided’, block b-1 included items 1 to 49 and block b-2 included items
50 to 98, which can be represented visually as in figure 6.3. Each block was preceded by
a trial session and instructions, as described above.
Figure 6.3 Schematic representation of the subtests administered to the participants;
arrows indicate possible combinations of blocks within one testing session
Two sessions were administered on two different days with at least one day in between.
The order of the administration of testing blocks was semi-random per participant. Three
SENTENCE COMPLETION EXPERIMENT
145
principles of the administration of the experimental materials were respected: (1) two
blocks of the same subtest were never presented within one session; (2) two blocks with
the same items were never presented within one session; (3) the order of block
presentation varied across participants. The order of administration of the experimental
materials is shown in Appendix VI, table II.
Scoring and data analysis
The test performance of the participants was scored on the paper score form and was
audio taped for further detailed transcription and analysis. Data scoring obtained in this
experiment went through the general scoring procedure developed in this thesis.
Therefore, three issues were taken into account during data analysis:
1) which case case assigners were produced by aphasic speakers – correct or
incorrect case assigners, or which case assigners were omitted in the responses of
the aphasic speakers;
2) how aphasic speakers dealt with case assignment when
a. they produced correct case assigners,
b. they produced incorrect case assigners and substituted them with other
elements,
c. they omitted case assigners from the responses;
3) whether patterns of performance with case-assigning prepositions were the same
as patterns with case-assigning verbs.
A complete prepositional phrase includes two elements – a preposition and its
complement noun. An error incurred in one of the elements results in an incorrect
prepositional phrase. Therefore, scoring included several levels of analysis.
At the first level, a simple correct–incorrect scoring was performed. At the second
level, the responses were further examined with respect to case assigners, and were
grouped into items where case assigners were (1) produced correctly, (2) omitted, or (3)
substituted by other members of the same category. At the third level, all responses were
CHAPTER VI
146
grouped according to the case-marking morphemes of their complement nouns into (1)
correct case-marking morphemes (hereafter termed correct case marking) and (2)
incorrect case-marking morphemes. At the forth level, incorrect case-marking
morphemes were examined and subdivided into (a) nominative case-marking
morphemes, which were erroneous, since case assigners cannot assign the nominative
case in the contexts provided; (b) noun stems, which did not bare case-marking
morphemes and thus resulted in non-words; (c) incorrect case-marking morphemes,
which denoted cases that could not be used in combination with particular case assigners;
(d) substituted case-marking morphemes, which were used with possible case assigners
but denoted cases other than the ones required by the context of the sentences.
Substituted case-marking morphemes mostly occurred in sentences with case-assigning
prepositions. The difference between (c) and (d) error types can be explained by the
asymmetry between Russian prepositions and case systems. One preposition in Russian
can assign several cases, all manifested with different case-marking morphemes; that is
why there are several possible combinations of a preposition and case morphemes that
occur in different contexts. The four possibilities for case marking, outlined above, are
referred to as nominative case marking, noun stems, incorrcet case marking and
substituted case marking, respectively.
As discussed earlier in the Materials section, the conditions of the two subtests
were balanced and matched as precisely as possible. However, their requirements
differed slightly, which was inevitable. To illustrate this, in subtest-2, ‘case assigner
provided’, a case assigner (a preposition or a verb) was provided, which made the phrase
frame rather transparent. The only possible completion of the sentences in subtest-2,
‘case assigner provided’ was a noun. Whereas in subtest-1, ‘noun provided’, a case
assigner was missing from the experimental stimuli, while a noun was provided in the
default nominative case. This allowed for more possible sentence completion schemes;
therefore, the participants had more freedom in this subtest to complete the sentence. This
was impossible to overcome and resulted in occasional category shifts (when participants,
in response to an experimental sentence that had to be completed, for example, with a
verb phrase, produced a prepositional phrase instead: Mal’chik ukazyvaet na vazu: “The
boy points at the vase”, instead of Mal’chik trogaet vazu: “The boy touches the vase”).
SENTENCE COMPLETION EXPERIMENT
147
As mentioned above, one of the reasons for the occurrence of such responses was the
combination of the test design and the characteristics of the Russian language. To avoid
confusion in scoring and data analysis several experimental stimuli were omitted. From
subtest-1, ‘noun provided’, fourteen experimental sentences that were required to be
completed with VPs were excluded from the main analysis for all participants. From
subtest-2, ‘case assigner provided’, two similar experimental sentences were also
excluded from the data of all participants. In subtest-1, ‘noun provided’, this resulted in
50 items that required completion with PPs, and 34 items that required completion with
VPs; in subtest-2, ‘case assigner provided’, 50 items required completion with PPs and
46 with VPs.
Statistical tools
Responses obtained from all participants were counted. For a statistical analysis the raw
numbers of correct responses produced by each participant were converted to proportions
from the total number of responses of each participant. To preclude possible deviations
from the normal distribution of the data, an arcsine transformation was applied to the
square root of all proportions before any statistical analysis was run. However, during the
provisional analysis Shapiro-Wilk normality tests showed that the data of all groups of
participants remained non-normally distributed. Moreover, Levene’s test showed that the
groups had heterogeneous variances. In other words, two main assumptions of parametric
testing were violated: normality of data distribution and homogeneity of variances.
Therefore, it was decided to apply non-parametric statistical tests. For comparisons of the
three groups of participants – the control speakers, the fluent aphasic speakers and the
non-fluent aphasic speakers – a non-parametric equivalent of ANOVA, the Kruskal-
Wallis test, was used. For post-hoc comparisons between the groups of participants, an
alternative to the Mann-Whitney-U test – the Kolmogorov-Smirnov Z test – was applied.
This test is considered to have more power when the sample sizes tested have less than 25
participants per group (Field, 2005). For comparisons of the data within each group of
participants, a non-parametric alternative of the paired t-test, the Wilcoxon signed-rank
CHAPTER VI
148
test, was used. To reduce the possibility of Type I errors, and to decrease the risk of
overclaiming, a Holm’s sequential Bonferroni correction was applied where appropriate.
This procedure tends to be less strict than Bonferroni adjustment and, unlike Bonferroni,
does not run the risk of underclaiming. To avoid the erroneous conclusions over
significance that may occur when significance is computed for small datasets the Exact
significance test option was applied throughout, which is more suitable for smaller
sample sizes than the Monte Carlo significance estimate (Field, 2005).
Results
In this section, the results of the experiment are presented. Initially, the results of subtest-
1, ‘noun provided’ are given. In this subtest, case assigners were missing from the
experimental sentences; participants had to complete a sentence, producing an
appropriate preposition or a verb. In addition, they also had to use the noun provided in
the experimental picture, and produce this noun with correct case-marking morphemes
appropriate for the provided preposition or the verb. Description of the results starts with
a report on the overall results of the groups of aphasic speakers. It is followed by an
account of the first analysis regarding production of case assigners, irrespective of the
case-marking morphemes of their complement nouns. It was analyzed whether case
assigners were produced correctly, produced incorrectly, or omitted. Subsequently, an
account of the second analysis of case-assigned nouns, examining their case-marking
morphemes, is presented. Case-marking morphemes are examined separately for items
with correct, incorrect and omitted case assigners (prepositions and verbs). The
description of the results of subtest-1, ‘noun provided’ is rounded up with an interim
general summary. The same order of presentation is maintained in relation to the results
obtained in subtest-2, ‘case assigner provided’, where case assigners were provided in the
experimental sentences, but their complement nouns were missing. Thereafter, the results
of the two subtests are compared.
SENTENCE COMPLETION EXPERIMENT
149
Results: Subtest-1, ‘noun provided’
The first research question in this experiment concerned the quantitative differences
between the group of aphasic speakers and the control group in terms of the numbers of
complete correct prepositional phrases produced in the restricted tasks, which demanded
usage of particular constructions. Subtest-1, ‘noun provided’ was designed to address this
issue. The number of complete correct prepositional phrases was counted in the data
corpus acquired from each participant. These raw numbers were converted to percentages
from the total number of items to enable direct comparisons between the groups of
participants. The same was done with respect to control sentences with verb phrases. The
group of control speakers performed at ceiling on sentences with prepositional (PPs) and
verb phrases (VPs), and hereafter their data are discarded from the error analysis. Table
6.3 shows correct and incorrect responses, as percentages, produced by groups of fluent
and non-fluent aphasic speakers. Individual results for the fluent and non-fluent aphasic
speakers on items with prepositional and verb phases in subtest-1, ‘noun provided’ of the
sentence completion experiment are shown in Appendix VI, table III.
Table 6.3 Overall test results of the fluent and non-fluent aphasic speakers on subtest-1,
‘noun provided’ (a case assigner is omitted from experimental sentences; a noun is
provided in the nominative case)
Fluent aphasic speakers (N=8)
Non-fluent aphasic speakers (N=12)
Items with Items with PPs
(Nitems=400) VPs
(Nitems=272) PPs
(Nitems=600) VPs
(Nitems=408) Correct responses with assigners 83.50 84.56 72.17 81.62 Incorrect responses with assigners 14.50 3.31 25.33 4.42 Other errors 2 12.13 2.5 13.97
CHAPTER VI
150
The category ‘correct responses’ implies items with complete correct prepositional or
verb phrases depending on the context in which case assigners were produced correctly;
they were not omitted or substituted. Moreover, their complement nouns were case-
marked correctly in accordance with these case assigners. ‘Incorrect responses’ are items
in which errors occurred either with respect to case assigners (where they were either
omitted or produced incorrectly), or which affected case markings of their complement
nouns. These two error categories are discussed in more detail in the next sections of the
error analysis. The category ‘other errors’ comprises non-responses, un-analyzable
utterances, and responses in which aphasic speakers produced structures other than those
required; these responses will not be further analyzed.
The Kruskal-Wallis tests revealed significant differences among the groups of
participants in sentences with PPs (H(2)=31.12, p<.001) and in sentences with VPs
(H(2)=28.50, p<.001). Further paired group comparisons were administered with the
Kolmogorov-Smirnov Z test, which showed that the group of fluent aphasic speakers
produced significantly fewer correct responses with PPs (K-S Z=2.12; p<.001) and with
VPs (K-S Z=2.42; p<.001) than did the group of control speakers. The group of non-
fluent aphasic speakers also performed significantly worse than the group of control
speakers. They produced fewer correct responses with PPs (K-S Z=2.55; p<.001) and
with VPs (K-S Z=2.32; p<.001) than the control speakers. The two groups of aphasic
speakers were also compared; however, no significant differences were found between
these two groups, either in sentences with PPs (K-S Z=31.00; p<.1) or in sentences with
VPs (K-S Z=39.00; p<.253). Hereafter, comparisons within each group of aphasic
speakers were performed with the Wilcoxon signed-rank test across correct items with
prepositional and verb phrases. No significant differences were found either in the
performance of the fluent aphasic speakers in sentences with PPs and with VPs (Z=-.280;
p=.422, r=-.07), or in the performance of non-fluent aphasic speakers in sentence with
PPs and with VPs (Z=-1.412; p=.088, r=-.29).
SENTENCE COMPLETION EXPERIMENT
151
Error analyses: Subtest-1, ‘noun provided’
The error analysis was conducted on the total number of analyzable items only.
Analyzable items are sentences completed by aphasic speakers in accordance with the
sentence structure required by the task – which was a preposition and a noun, or a verb
and a noun, depending on the sentence context. Analyzable items are also sentences that
were intended to be completed as required; therefore, their structures are transparent, but
one or the other element necessary for the sentence completion is missing but can be
reconstructed from the context. For example, if aphasic speakers were required to
produce a sentence as shown in example (1), but instead produced a sentence as shown in
examples (2), (3), (4) or (5), these sentences were judged as analyzable and entered into
error analyses:
(1) Mal’chik idet s zontom [INSTR]
“The boy walks with an umbrella”
(2) Mal’chik idet s …
“The boy walks with …”
(3) Mal’chik idet … zontom [INSTR]
“The boy walks … an umbrella”
(4) Mal’chik idet … zont [NOM]
“The boy walks … an umbrella”
(5) Mal’chik idet … zonte [PREP]
“The boy walks … an umbrella”
Also, if aphasic speakers produced sentences in which one or the other element was
incorrect, these responses were entered into the error analysis. For example, a sentence in
(6) is complete, but the case-marking morpheme of the complement noun is incorrect; it
CHAPTER VI
152
represents the dative case, instead of the required instrumental case. In a sentence type as
exemplified in (7), an incorrect preposition na: “on”, in place of s: “with”, is used.
Responses similar to the ones described above were analyzed. The same is true for the
items requiring completion with a verb and a noun.
(4) Mal’chik idet s zontu [DAT]
“The boy walks with umbrella [DAT]”
(7) Mal’chik idet na zonte
“The boy walks on an umbrella”
Non-responses, un-analyzable utterances and sentences with structures other than those
required, (for example, the production of a sentence like this: A man goes home instead of
A man goes towards the house) were excluded from the error analysis.
The main objective of the error analysis was to provide an answer to the second
and third research questions of this experiment, namely, what happens to case marking on
noun phrases when case assigners are produced correctly, when they are omitted, or when
they are produced incorrectly. To answer these questions it was necessary to perform two
analyses: the first regards case assigners and the second regards case marking.
Error analysis of case assigners
After correct–incorrect scoring and analysis, the second level of scoring procedure
concerned case assigners. It was examined here whether aphasic speakers produced
correct case assigners, or whether they omitted obligatory case assigners, or produced
them incorrectly. All analyzable sentences were examined with regard to case assigners
(prepositions and verbs) produced, irrespective of the case marking that appeared on their
complement nouns. Patterns in the production of case-marking morphemes were
analyzed later, and they are discussed separately in the following sections of error
analysis. For the analysis of case assigners, omitted and incorrectly produced case
SENTENCE COMPLETION EXPERIMENT
153
assigners were counted in the groups of aphasic speakers. These numbers were converted
into percentages from the total number of errors made in each group. Patterns of case-
marking morphemes were analyzed thereafter and will be discussed separately in the
following sections of the error analysis. It was observed that in the group of eight fluent
aphasic speakers, three participants omitted prepositions, and seven of them substituted
prepositions. As for the non-fluent aphasic speakers, seven omitted the required
prepositions and eleven substituted prepositions. With respect to verbs, only one non-
fluent aphasic speaker omitted a verb and one fluent aphasic speaker substituted a verb.
These data are not included in table 6.4, which reviews the error patterns of the fluent and
non-fluent aphasic speakers with respect to case-assigning prepositions.
Table 6.4 Performance of the fluent and non-fluent aphasic speakers with respect to case-
assigning prepositions
No significant association was found between the category of error made and the
fluency of the aphasic speakers (χ2 =.07; df=1 p=.987). Obligatory case-assigning
prepositions were omitted in 18.53 percent of incorrect responses of the fluent aphasic
speakers, and in 21.26 percent of incorrect responses by the non-fluent aphasic speakers.
With regard to substitutions of prepositions, they occurred in 81.47 percent of incorrect
responses of the fluent aphasic speakers, and in 78.74 percent of incorrect responses of
the non-fluent aphasic speakers.
Fluent aphasic speakers Non-fluent aphasic speakers Items with Items with PPs (nitems=42) PPs (nitems=143) Case assigner omitted 18.53 21.26 Case assigner incorrect 81.47 78.74 χ2 =.07; df=1 p=.987
CHAPTER VI
154
Error analysis of case-assigned nouns and their case marking
The next stage of the experiment undertook an error analysis of the case-marking
morphemes of the complement nouns of prepositions and verbs. Patterns of case-marking
morphemes were looked into independently in items with two case assigners –
prepositions and verbs. Moreover, all responses were also examined separately, as
follows, (1) responses with case assigners produced correctly, (2) responses where case
assigners were omitted by aphasic speakers, (3) responses in which case assigners were
produced incorrectly and substituted by the members of the same category. All errors
made on case-marking morphemes within each category where (1) ‘case assigner is
produced correctly’, (2) ‘case assigner is omitted’, and (3) ‘case assigner is substituted’
were counted and converted to percentages from all errors within this particular category.
The main question for this error analysis was What happens to case marking of a
complement noun when its case assigner is produced correctly, when it is omitted or
when it is substituted by another case assigner?
Case-marking morphemes of complement nouns in phrases with correct case
assigners
Fr this error analysis, all items with correct case-assigning prepositions produced in each
group of aphasic speakers are considered to constitute 100 percent; errors in the
production of case-marking morphemes in sentences with correct case-assigning
prepositions are computed in percentages from this total. The same is done with respect
to items with correct case-assigning verbs produced by the fluent and non-fluent aphasic
speakers.
When both groups of aphasic speakers produced correct case-assigning
prepositions in their responses, they virtually always (except once) produced their
complement nouns with case-morphology. In the group of fluent aphasic speakers, in
responses with correct prepositions, correct case-marking morphemes were produced in
SENTENCE COMPLETION EXPERIMENT
155
92.40 percent of complement nouns; in responses with correct verb, correct case-marking
morphemes were produced in 95.89 percent of these responses.
Similarly, in the group of non-fluent aphasic speakers, in responses with correct
prepositions, case-marking morphemes were produced correctly in 96.51 percent of these
phrases; in responses with correct verbs, correct case-marking morphemes were produced
in 94.30 percent of complement nouns. It was obvious that only a few occasional errors
occurred in case marking of complement nouns in items with prepositions and in items
with verbs in both groups of aphasic speakers; however, no clear error pattern emerged in
either group. Patterns of case marking of complement nouns in the presence of correct
case-assigning prepositions and verbs by aphasic speakers are shown in table 6.5.
Table 6.5 Case-marking patterns of nouns in the presence of correct case-assigning
prepositions and verbs produced by aphasic speakers
Fluent aphasic speakers
Non-fluent aphasic speakers
Items with Items with PPs
(nitems=350) VPs
(nitems=238) PPs
(nitems=442) VPs
(nitems=362) Correct case marking produced
92.40
95.89
96.51
94.30
Incorrect case marking produced
1.63
1.13
2.24
3.75
Nominative case marking produced
4.63
2.98
0.19
1.95
Case marking other than required, but matching a preposition
1.34
-
0.71
-
Noun stem produced
-
-
0.35
-
CHAPTER VI
156
Case-marking morphemes of complement nouns in phrases with omitted case-
assigning prepositions and verbs
As was described in the previous section, omission errors were encountered in the data of
both groups of aphasic speakers. However, omissions were made only with respect to
prepositions; with respect to verbs, only one non-fluent aphasic speaker omitted a verb
and produced a correct case-marking morpheme of its complement noun; one fluent
aphasic speaker substituted a verb and produced a nominative case-marking morpheme of
its complement noun. These data are not displayed in table 6.6.
In the group of fluent aphasic speakers, only three participants omitted
prepositions. When lexical prepositions were omitted, the fluent aphasic speakers
produced correct case-marking morphemes in 66.67 percent of complement nouns, and in
33.33 percent they produced nominative case-marking morphemes. In the group of non-
fluent aphasic speakers, seven participants omitted prepositions, and in 62.34 percent of
their complement nouns produced correct case-marking morphemes; in 24.23 percent,
case-marking morphemes were incorrect, and in another 13.53 percent, case-marking
morphemes represented nominative case marking. Patters in the production of case-
marking morphemes of complement nouns of omitted prepositions are shown in table 6.6.
SENTENCE COMPLETION EXPERIMENT
157
Table 6.6 Error patterns in case marking when case assigners were omitted by aphasic
speakers
As shown in table 6.6, both groups always produced case-marking morphemes on
the complement nouns. Even when case assigners were omitted, their complement nouns
were never deprived of case morphemes and they were never produced as bare noun
stems. On the obtained data, no strong tendency concerning production of case-marking
morphemes of complement nouns in responses with omitted case-assigning prepositions
can be discerned. However, the general pattern is fairly clear: even when obligatory case-
assigning prepositions were omitted from the responses of both groups of aphasic
speakers, they were more likely to produce correct case-marking morphemes on the
complement nouns.
Case-marking morphemes of complement nouns in phrases with incorrect case
assigners
The last error category concerns items in which case-assigning prepositions and verbs
were produced incorrectly by aphasic speakers. As mentioned above, only one verb was
Fluent aphasic speakers
Non-fluent aphasic speakers
Items with Items with PPs (nitems=9) PPs (nitems=32) Correct case marking produced
66.67
62.34
Incorrect case marking produced
-
24.13
Nominative case marking produced
33.33
13.53
Noun stem produced
-
-
CHAPTER VI
158
substituted in this subtest; this was done by a fluent aphasic speaker; items with
substituted prepositions are discussed below.
In the group of fluent aphasic speakers, substitution errors of prepositions were
made by seven participants. In these responses, fluent aphasic speakers tended to produce
case-marking morphemes matching the substituted prepositions in 80 percent of
complement nouns, and they also produced correct case-marking morphemes that would
match the required prepositions in 14.29 percent of responses. In the group of non-fluent
aphasic speakers, prepositions were produced incorrectly and substituted by other
prepositions by eleven participants. When prepositions were substituted, the non-fluent
aphasic speakers in 95.75 percent of their complement nouns also substituted case-
marking morphemes of their complement noun, and produced case-marking morphemes
appropriate for the substituted prepositions. Unlike the group of fluent aphasic speakers,
correct case-marking morphemes matching the required prepositions were produced in
only 1.91 percent. Patterns of case-marking morphemes of nouns in sentences with case-
assigning prepositions produced incorrectly are shown in table 6.7.
Table 6.7 Error patterns in case marking when case assigners were substituted by the
aphasic speakers
Fluent aphasic speakers
Non-fluent aphasic speakers
Items with Items with PPs (nitems=33) PPs (nitems=111) Correct case marking produced
14.29
1.91
Incorrect case marking produced
0.71
1.44
Nominative case marking produced
4.29
0.90
Substituted case marking produced
80
95.97
Noun stem produced
0.71
-
SENTENCE COMPLETION EXPERIMENT
159
It was observed that both groups of aphasic speakers tended to produce case-
marking morphemes of complement nouns with substituted prepositions matching these
prepositions. For example, instead of producing preposition na: “on”, which assigns a
prepositional case in the provided context of the sentence (in 8, below), aphasic speakers
produced the preposition pod: “under”, which assigns the instrumental case. Furthermore,
they also case-marked the complement noun vetk-a: “branch nom.case” of preposition
pod for instrumental case, which is manifested with a morpheme oj/ej28
vetk-oj’ (as in 9,
below).
(8) Golub’ sidit na vetke
“A pigeon sits on a branch [PREP.]”
(9) Golub’ sidit pod vetkoj
“A pigeon sits under a branch [INSTR.]”
Summary of the results: Subtest-1, ‘noun provided’
Subtest-1, ‘noun provided’ of the sentence completion experiment targeted the
first question of the study concerning the quantitative differences between the numbers of
complete correct prepositional phrases produced by the aphasic speakers and the control
speakers. Namely, it was investigated whether aphasic speakers performed worse than the
group of control speakers when they had to produce a preposition and case-mark its
complement noun provided in the nominative case form.
The group of control speakers performed at ceiling on experimental sentences
with PPs and control sentences with VPs. They always produced correct case-assigning
prepositions and case-assigning verbs and case-marked their complement nouns
correctly.
28 Some cases can be manifested with a morpheme and its allomorphs, depending on the declension class of a noun, here, and further a morpheme of a particular case is followed by its allomorph after the slash, here; for the instrumental case the morpheme is oj and its allomorph is ej.
CHAPTER VI
160
Both the fluent and non-fluent aphasic speakers performed significantly worse on
sentences with PPs and VPs than did the group of control speakers. Between the groups
of aphasic speakers no differences were found either in their performance in sentences
with PPs or in sentences with VPs. Within each group of aphasic speakers, performance
in sentences with PPs and VPs did not differ either.
Examination of complete PPs and VPs showed that when both groups of aphasic
speakers produced correct case-assigning prepositions and verbs, they also produced
correct case-marking morphemes of their complement nouns more than 90 percent of the
time. With regard to production of case-assigning prepositions and verbs, it was observed
that fluent and non-fluent aphasic speakers occasionally omitted and substituted case-
assigning prepositions, but not verbs. Both groups of aphasic speakers were more likely
to substitute prepositions than omit them. When both groups omitted case-assigning
prepositions, the general tendency was the production of correct case-marking
morphemes of complement nouns. When case-assigning prepositions were substituted by
fluent and non-fluent aphasic speakers, they tended to produce case-marking morphemes
in accordance with the substituted prepositions. In addition, it was observed that both
groups of aphasic speakers hardly ever produced complement nouns deprived of case-
marking morphology, either in items with correct case assigners produced, or with
substituted case assigners, or even with omitted case assigners.
To recapitulate, then, the results of subtest-1, ‘noun provided’ of the sentence
completion experiment provided some data to support the hypotheses put forward at the
beginning of this chapter. The non-fluent aphasic speakers were indeed more likely to
produce correct case-marking morphemes of the complement nouns of prepositions when
they managed to produce correct prepositions. The fluent aphasic speakers showed a
tendency to produce case-marking morphemes of complement nouns of substituted
prepositions in accordance with these prepositions. However, so far, on the basis of the
results of subtest-1, ‘noun provided’, these two statements did not prove to be unique to
either group of aphasic speakers; it could be applied to both.
SENTENCE COMPLETION EXPERIMENT
161
Results: Subtest-2, ‘case assigner provided’
The second question of this experiment regarded case-marking morphemes of the
complement nouns of prepositions. Specifically, it was questioned whether aphasic
speakers are able to produce a noun with a correct case-marking morpheme when its
case-assigning preposition is provided. The results of subtest-2, ‘case assigner provided’,
which was performed to target this question, are reviewed here.
Data scoring and data analysis in the subtest was carried out along the same lines
developed for the sentence completion experiment (previously followed in subtest-1,
‘noun provided’). The number of complete correct prepositional phrases was counted for
each participant, and converted to percentages from the total number of items. The same
was done with respect to control sentences with verb phrases. The group of control
speakers performed at ceiling on sentences with prepositional (PPs) and verb phrases
(VPs); their data are not considered further. The overall results of the groups of aphasic
speakers are shown in table 6.8. Individual results of the fluent and non-fluent aphasic
speakers in items with prepositional and verb phases in subtest-2, ‘case assigner
provided’ of the sentence completion experiment are shown in Appendix VII, table I.
Table 6.8 Overall results of the fluent and non-fluent aphasic speakers in subtest-2, ‘case
assigner provided’ (a case assigner is provided in the experimental sentences; a noun is
missing).
Fluent aphasic speakers (N=8)
Non-fluent aphasic speakers (N=12)
Items with Items with PPs
(Nitems=400) VPs
(Nitems=368) PPs
(Nitems=600) VPs
(Nitems=552) Correct responses with prepositions
92
85.59
91.67
88.04
Erroneous responses 7.75 4.89 7.5 4.17 Other errors 0.25 9.52 0.85 7.79
CHAPTER VI
162
Similar to the error analysis done for the previous subtest, in subtest-2, ‘case assigner
provided’, all responses were subdivided into three categories: ‘correct responses’,
‘erroneous responses’, and ‘other errors’. The latter category includes non-responses, un-
analyzable utterances, and responses in which aphasic speakers produced structures other
than those required; these responses will not be analyzed further.
The Kruskal-Wallis tests revealed significant differences among the groups of
participants in sentences with prepositions (H(2)=22.31, p<.001) and in sentences with
verbs (H(2)=29.69, p<.001). Further paired group comparisons were made using the
Kolmogorov-Smirnov Z test. Although both groups of aphasic speakers performed very
well on the task, it was found that the group of fluent aphasic speakers produced
significantly fewer correct responses with PPs (K-S Z=1.51; p<.001) and with VPs (K-S
Z=2.42; p<.001) than did the control speakers. The group of non-fluent aphasic speakers
also performed significantly worse on sentences with PPs (K-S Z=2.09 p<.001) and on
sentences with VPs (K-S Z=2.32; p<.001) than the group of control speakers. The results
from the two groups of aphasic speakers were also compared. The comparisons revealed
no significant difference between the two groups of aphasic speakers, either in sentences
with VPs (K-S Z=.639; p<.611) or in sentences with PPs (K-S Z=1.00; p<.122).
Further comparisons within each group of aphasic speakers were performed with
the Wilcoxon signed-rank test across correct items with prepositional and verb phrases. A
significant difference was found in the performance of the fluent aphasic speakers in
sentences with PPs and sentences with VPs at α-level of .0125 (Z=-2.521; p=.004, r=-
.063). A trend was observed in the performance of the non-fluent aphasic speakers on
sentences with PPs and sentences with VPs (Z=-1.647; p=.055, r=-.35).
Error analyses: Subtest-2, ‘case assigner provided’
As with subtest-1, ‘noun provided’, only analyzable items were considered in this error
analysis. The same criteria described in the error analysis section of the previous subtest
were applied to determine which items were analyzable. In subtest-2, ‘case assigner
provided’, a case assigner was provided in the experimental sentences, and only a noun
SENTENCE COMPLETION EXPERIMENT
163
was missing. Hence, the sentence structure was fairly clear; the only possible completion
of the sentences in subtest-2, ‘case assigner provided’ was a noun. Nevertheless, both
case assigners and case-marking morphemes of their complement nouns were examined.
Error analysis of case assigners
After correct–incorrect scoring, the second level of scoring procedure concerned case
assigners. Because case assigners were already present in the experimental sentences the
participants had only to read them correctly and complete a sentence with a noun;
therefore, hardly any errors were expected in relation to case assigners. However, it was
still possible that aphasic speakers either would not realize these case assigners
phonetically, in other words, they would not pronounce them and would omit these case
assigners from the responses; or aphasic speakers would read case assigners incorrectly –
reading preposition pod: “under” as nad: “above”, which looks rather similar in Cyrillic
transcription: над: “under”, and под: “above”, for example.
In fact, very few errors occurred. Two non-fluent aphasic speakers did not read
five case-assigning prepositions, and omitted them from their responses. Correct case-
marking morphemes were produced on two complement nouns; incorrect case-marking
morphemes appeared on two other complement nouns; and one complement noun had a
nominative case-marking morpheme. For five non-fluent aphasic speakers, nine
prepositions were read incorrectly and were substituted by others; seven complement
nouns of these prepositions were produced with case-marking morphemes matching the
new prepositions, and case-marking morphemes of two other prepositions were incorrect.
In the group of fluent aphasic speakers, only one participant failed to read a
preposition and produced correct case-marking morphemes on its complement noun.
Also, one fluent aphasic speaker read a preposition incorrectly and produced a
nominative case-marking morpheme of its complement noun.
As for the production of case-assigning verbs by the fluent and non-fluent aphasic
speakers, fewer errors occurred in both groups. A verb was omitted by a non-fluent
CHAPTER VI
164
aphasic speaker, and a verb was substituted by a fluent aphasic speaker; on their
complement nouns, nominative case-marking morphemes were produced.
Error analysis of case-assigned nouns and their case marking
Subsequently, as in the previous subtest, error analysis regarding case marking on nouns
was performed. However, because hardly any errors occurred that would result in
omissions or substitutions of case-assigning prepositions and verbs, and moreover, these
errors were already described in the previous section, only a detailed error analysis of the
case-marking morphemes of complement nouns of correct case assigners is presented
here.
Case-marking morphemes of complement nouns in phrases with correct case
assigners
When both groups of aphasic speakers read case-assigning prepositions and verbs
correctly, they virtually always (except once) produced their complement nouns with
case-morphology. Only one fluent aphasic speaker produced a noun stem in the response
with a case-assigning preposition. Patterns of case-marking morphemes were looked at
separately for items with case-assigning prepositions and items with case-assigning
verbs. The group of errors with case-marking morphemes being considered was restricted
to items with correct case-assigning prepositions. Errors in case marking in sentences
with correct case-assigning prepositions were computed in percentages from this total.
The same was done with items with correct case-assigning verbs.
In the group of fluent aphasic speakers, in responses with correct prepositions,
correct case-marking morphemes were produced in 92.49 percent of these responses; and
in responses with correct verbs, correct case-marking morphemes were produced in 94.06
percent. Similarly, in the group of non-fluent aphasic speakers, in responses with correct
prepositions, case-marking morphemes were produced correctly in 94.59 percent of the
SENTENCE COMPLETION EXPERIMENT
165
time; in responses with correct verbs, correct case-marking morphemes were produced in
95.41 percent of complement nouns. Just a few occasional errors occurred in the case
marking of complement nouns in both groups of aphasic speakers; no clear error pattern
emerged in either group. Patterns of case marking of complement nouns of case-
assigning prepositions and case-assigning verbs read correctly by aphasic speakers are
shown in table 6.9.
Table 6.9 Case-marking patterns of nouns in responses with case-assigning prepositions
and verbs read correctly by aphasic speakers
Summary of the results: Subtest-2, ‘case assigner provided’
The second question of the sentence completion experiment considered case-marking
morphemes produced by aphasic speakers on the noun complements of prepositions.
Namely, it examined whether aphasic speakers are able to produce a noun with correct
case-marking morphemes when its case-assigning preposition is provided in a sentence.
Subtest-2, ‘case assigner provided’ was designed to answer this question, and also to
Fluent aphasic Speakers
Non-fluent aphasic speakers
Items with Items with PPs
(nitems=398) VPs
(nitems=335) PPs
(nitems=581) VPs
(nitems=509) Correct case marking produced
92.49
94.06
94.59
95.41
Incorrect case marking produced
1.02
2.15
4.05
2.69
Nominative case marking produced
4.17
3.79
0.86
1.90
Case marking other than required, but matching a preposition
1.54
-
0.51
-
Noun stem produced
0.78
-
-
-
CHAPTER VI
166
investigate patterns of case-marking morphemes with omitted and substituted case
assigners.
The performance of the group of control speakers on experimental sentences with
PPs and control sentences with VPs was found to be at ceiling. They always read
experimental sentences correctly, and did not omit case assigners or read them
incorrectly. In addition, no errors were made by the group of control speakers with
respect to the production of correct case-marking morphemes of complement nouns.
Although both groups of aphasic speakers performed very well on the task,
statistically their performance in sentences with PPs and VPs was significantly worse
than the performance of the control speakers. Between the groups of aphasic speakers no
differences were found between responses with either type of case assigner. When
performance in sentences with PPs and VPs was compared within each group of aphasic
speakers, the fluent aphasic speakers were found to produce more correct responses with
PPs than with VPs; non-fluent aphasic speakers showed a similar trend. Examination of
case assigners revealed hardly any errors since case-assigning prepositions and verbs
were already present in the sentences. Examination of complete PPs and VPs showed that
the fluent and non-fluent aphasic speakers virtually always produced nouns with case-
morphology. Furthermore, when both groups of aphasic speakers read case assigners
correctly, in more than 90 percent of the time they produced complement nouns of these
case assigners with correct case-marking morphemes.
To summarize, the results of subtest-2, ‘case assigner provided’ of the sentence
completion experiment provided additional evidence in favor of the hypotheses of this
study. The non-fluent aphasic speakers proved to be able to produce correct case-marking
morphemes of the complement nouns of prepositions when these prepositions were
provided in the sentences. They made occasional errors but these errors hardly ever
resulted in the production of bare noun stems. The fluent aphasic speakers showed a
performance very similar to the non-fluent aphasic speakers. Similar to the results of
subtest-1 ‘noun provided’, the results of subtest-2, ‘case assigner provided’ do not
provide sufficient grounds for differentiating between the two aphasic impairments.
SENTENCE COMPLETION EXPERIMENT
167
Comparison of the results of the two subtests
The third research question of this study investigates the relationship between the
presence and absence of case-assigning prepositions in experimental sentences and in the
production of correct case-marking morphemes of their complement nouns. It was
questioned whether the presence of (correct) case-assigning prepositions facilitates
production of correct case-marking morphemes of the complement nouns of these
prepositions. The results of subtest-1, ‘noun provided’ showed that when aphasic
speakers managed to produce correct case-assigning prepositions, both groups of aphasic
speakers were also able to produce correct case-marking morphemes of complement
nouns of these case assigners. In subtest-2, ‘case assigner provided’, fluent and non-
fluent aphasic speakers produced correct case-marking morphemes of more than 90
percent of complement nouns of the case-assigning prepositions provided in the
experimental sentences. The two subtests were designed with minimal differences – the
presence of case assigners (in subtest-2, ‘case assigner provided’) and absence of case
assigners (in subtest-1, ‘noun provided’) in the experimental sentences – which permitted
direct comparisons between the subtests to address this issue. When comparing the
subtests, the hypothesis was investigated claiming that the presence of a correct case
assigner in a sentence assists in production of correct case-marking morphemes of its
complement noun. Below, the results of this comparison are presented for each group of
aphasic speakers.
Fluent aphasic speakers: comparison of the subtests
Performance of the group of fluent aphasic speakers on items with complete correct
prepositional phrases in subtest-1, ‘noun provided’ was compared to their performance on
similar items in subtest-2, ‘case assigner provided’. A significant difference was found in
their performance in sentences with PPs in the two subtests at α-level of .017 (Z=-2.383;
p=.008, r=.59); no difference emerged in the performance on sentences with VPs (Z=-
.840 p=.230, r=.21). The group of fluent aphasic speakers produced more items with
CHAPTER VI
168
complete correct PPs in subtest-2, ‘case assigner provided’ (92.00%) than in subtest-1,
‘noun provided’ (83.50%). They produced equal numbers of correct responses with
complete correct VPs in the two subtests: 84.56 percent of responses in subtest-2, ‘case
assigner provided’ and 85.59 percent of responses in subtest-1, ‘noun provided’. The
performance of the fluent aphasic speakers in the two subtests is shown in figure 6.4 in
terms of percentages of correct items.
Figure 6.4 Performance of the group of fluent aphasic speakers on the two subtests; data
presented in percentage of correct items derived from the total number of analyzable
sentences produced in this group
Non-fluent aphasic speakers: comparison of the subtests
With the group of non-fluent aphasic speakers, a significant difference emerged in their
performance on sentences with PPs in the two subtests (Z=-3.059; p<.001, r=.63),
whereas no significant difference was found in the performance in sentences with VPs
(Z=-1.334; p=.102, r=.27). The group of non-fluent aphasic speakers produced more
items with complete correct PPs in subtest-2, ‘case assigner provided’ (91.67%) than in
subtest-1, ‘noun provided’ (71.17%), and they produced equal numbers of correct
responses with complete correct VPs in the two subtests: 88.04 percent of correct
SENTENCE COMPLETION EXPERIMENT
169
responses in subtest-2, ‘case assigner provided’ and 81.86 percent of correct responses in
subtest-1, ‘noun provided’. The performance of the non-fluent aphasic speakers in the
two subtests is shown in figure 6.5.
Figure 6.5 Performance of the group of non-fluent aphasic speakers in the two subtests
Comparisons of the results of the aphasic speakers in the two subtests were
administered within each group of aphasic speakers to confirm the hypothesis claiming
that presence of (correct) case-assigning prepositions facilitates case assignment and
leads to production of correct case-marking morphemes of the complement nouns of
these prepositions. Because the principle difference in the test designs of the two subtests
was the presence or absence of case assigners, the numbers of complete correct PPs and
VPs produced in each group of aphasic speakers were compared across the two subtests.
It was found that for both groups of aphasic speakers it was easier to produce complete
correct PPs in subtest-2, ‘case assigner provided’, where case-assigning prepositions were
provided in the experimental sentences, than in subtest-1, ‘noun provided’, where case
assigners had to be produced by aphasic speakers.
CHAPTER VI
170
Discussion
In the narrative speech analysis reported in chapter V, it was found that non-fluent
aphasic speakers produced fewer prepositions and complete correct prepositional phrases
in their narrative speech when compared to the neurologically intact control speakers.
The fluent aphasic speakers did not differ from the group of control speakers. With
respect to case marking, it was found that both groups of aphasic speakers were able to
assign correct case to the complement nouns of case-assigning prepositions when they
produced these prepositions in their narrative speech, and they were able to produce the
correct case-marking morphemes to denote these particular cases. From previous cross-
linguistic research, is known that language deficits of aphasic speakers can be manifested
to a different degree in different tasks (Bastiaanse, 1995; Caplan et al., 2006; Lehečková,
2001).
The main goal of the present research was to describe production of prepositions
and production of case-marking morphemes of their complement nouns by aphasic
speakers in experimental conditions. Performances of two aphasic populations were
examined in the study, namely, fluent aphasic speakers and non-fluent aphasic speakers.
Therefore, two hypotheses were put forward.
Taking into account the results of Ruigendijk’s 2002 study of case assignment by
Russian verbs, and the results of the analysis of case assignment with respect to
prepositions in the narrative speech of non-fluent aphasic speakers described in this
thesis, the non-fluent aphasic speakers were hypothesized to be able to produce a correct
case-assigning preposition; they would manage to produce a correct case-marking
morpheme of the complement noun of this preposition. Therefore, the presence of correct
case-assigning prepositions would facilitate case assignment.
With respect to the fluent aphasic speakers, taking into consideration results
(Friederici, 1981) and observations (Beyn et al., 1979) of previous studies, it was further
hypothesized that when fluent aphasic speakers substitute case-assigning prepositions,
they will produce their complement nouns with case-marking morphemes appropriate for
the produced prepositions.
SENTENCE COMPLETION EXPERIMENT
171
The first question of the study concerned production of prepositional phrases in
quantitative terms, namely, it was examined whether aphasic speakers perform worse
than the control speakers when they have to produce a preposition and case-mark its
complement noun provided in the nominative case form.
The second question considered case-marking morphemes of complement nouns
of prepositions produced by aphasic speakers. It was investigated whether aphasic
speakers are able to produce a noun with a correct case-marking morpheme when its
case-assigning preposition is provided.
The third research question regarded relations between presence of a correct case-
assigning preposition and production of a correct case-marking morpheme of its
complement noun. It was examined whether the presence of (correct) case-assigning
prepositions assists correct case marking.
The results of the experiment reported here are in line with the findings of the
previous studies outlined above. Both groups of aphasic speakers experienced difficulties
in the production of prepositions; they produced significantly fewer complete correct
prepositional phrases than the group of control speakers. Moreover, both groups of
aphasic speakers showed similar performance patterns; when fluent and non-fluent
aphasic speakers managed to produce correct prepositions, they also tended to manage to
produce their complement nouns with correct case-marking morphemes. The same is true
for those sentences where case-assigning prepositions were already provided: more than
90 percent of their complement nouns had correct case-marking morphemes. These
findings support the results of the previous study of Russian aphasia, which found that
when Russian non-fluent aphasic speakers produced correct case-assigning verbs, they
also produced correct case-marking morphemes of their complement nouns (Ruigendijk,
2002). Moreover, the analysis of control sentences with verb phrases in the sentence
completion experiment replicated these findings.
When required to produce prepositions, both groups of aphasic speakers made
occasional errors. The fluent aphasic speakers substituted prepositions more often than
they omitted them, which is in line with the earlier studies reporting that the difficulties
aphasic speakers experience with prepositions are reflected in their substitutions rather
than in their omissions (Butterworth & Howard, 1987; Edwards et al., 1994; Martin &
CHAPTER VI
172
Blossom-Stach, 1986). When the fluent aphasic speakers substituted case-assigning
prepositions, quite often they produced case-marking morphemes of complement nouns
of these prepositions in accordance with the prepositions. This performance has also been
observed before in Russian fluent aphasia (Beyn et al, 1979), and it supports the proposed
hypothesis. The non-fluent aphasic speakers are known to tend to omit prepositions
(Beyn et al., 1979; Tsvetkova & Glozman, 1975). Although this was also observed in this
study, the Russian non-fluent aphasic speakers, similarly to the fluent aphasic speakers,
were found to substitute prepositions more often than to omit them. Substitutions of
prepositions by Russian non-fluent aphasic speakers have been reported only once before
(Leikin, 2002). In the present study, when non-fluent aphasic speakers substituted
prepositions, in analogy with the fluent aphasic speakers, they also produced case-
marking morphemes in accordance with these prepositions. When non-fluent aphasic
speakers omitted case-assigning prepositions, they still tended to produce correct case-
marking morphemes of the complement nouns, whereas fluent aphasic speakers were
more likely to produce nominative case-marking morphemes. Previously, it was reported
that when Russian non-fluent aphasic speakers omitted case-assigning verbs they did not
show any preference in case-marking patterns (Ruigendijk, 2002). With regard to patterns
in production of case-marking morphemes by Russian aphasic speakers in this
experiment, it should be noted that both groups of aphasic speakers virtually always
produced nouns with case-marking morphemes and hardly produced any noun stems.
This observation is consistent with previous research, which claimed that patterns of
omissions and substitutions largely depend on the characteristics of a particular language
(Avrutin, 2001; Grodzinsky, 1984; Kehayia et al., 1990; Lehečková, 1988; MacWhinney
& Osmán-Sági, 1991; Miceli et al., 1984; Ulatowska et al., 2001, and others). If
omissions are not authorized in particular contexts (for example, omissions of case-
marking morphemes in Russian, which result in non-words), aphasic speakers are not
prone to making omissions, and tend instead to produce substitutions of these
morphemes.
To recapitulate, both groups of aphasic speakers showed that, although not
faultlessly, they were able to produce correct case-marking morphemes on nouns either
when they managed to produce correct case assigners, or when these case assigners were
SENTENCE COMPLETION EXPERIMENT
173
already provided in the sentences. Therefore, comparisons of their performances on
sentences with provided case assigners (subtest-2, ‘case assigner provided’), and on
sentences in which they had to produce case assigners (subtest-1, ‘noun provided’), were
performed, which allowed for exploration of whether there is a relationship between the
presence and absence of case-assigning prepositions in experimental sentences and
production of correct case-marking morphemes of their complement nouns. Both groups
of aphasic speakers were found to produce significantly more responses with correct
complete prepositional phrases when case-assigning prepositions were provided in the
experimental sentences (subtest-2, ‘case assigner provided’) than when aphasic speakers
had to produce case-assigning prepositions themselves (subtest-1, ‘noun provided’); in
relation to items with verb phrases, no differences were found. Hence, the conclusion can
indeed be drawn that the presence of correct case-assigning prepositions in experimental
items has a positive influence on the production of the correct case-marking morphemes
of the complement nouns of these prepositions. And as expected, the presence of correct
case-assigning prepositions facilitates case assignment and assists in production of
correct case-marking morphemes. This replicates the earlier findings with respect to
Russian verbs (Ruigendijk, 2002), and supports the main hypothesis proposed for non-
fluent aphasic speakers and put forward in relation to this experiment at the beginning of
this chapter.
It is worth noting that although there were two hypotheses proposed for the
performance patterns of the groups of fluent and non-fluent aphasic speakers
respectively, both hypotheses found some support in the performance of each group of
aphasic speakers. Previously, similar performances of fluent and non-fluent aphasic
speakers had been observed (Bastiaanse & Edwards, 2004; Edwards, 2002; Edwards &
Bastiaanse, 1998), which were accounted for in terms of the general difficulties aphasic
speakers have when required to integrate “lexical-semantic and grammatical processes at
sentence level” (Bastiaanse & Edwards, 2004: 101). Following this logic, in both subtests
of the sentence completion experiment, aphasic speakers had to perform two operations.
In subtest-1, ‘noun provided’, they had to produce a case assigner and modify its
complement noun provided in the nominative case and to produce it with other correct
case marking. Production of a case assigner is a lexical-semantic process, whereas case
CHAPTER VI
174
marking of the nominative form of its complement noun is a grammatical process. In
subtest-2, ‘case assigner provided’, the participants had to produce the complement noun
of the case assigner, which was provided, and to case-mark this complement noun in
accordance with the case assigner provided. Similar to the previous subtest, then,
production of a noun activates lexical-semantic abilities, and case marking as required by
the provided case assigner is a grammatical process.
The next chapter will investigate whether the hypotheses supported by the results
of the present study are valid for all prepositions irrespective of the function of a
preposition and the context in which a particular preposition occurs, since from previous
cross-linguistic research it is known that prepositions are liable to differential impairment
in aphasia, which is predetermined by the context and by the function of this preposition
(Bennis et al., 1983; Friederici, 1981, 1982; Friederici, 1985; Friederici et al., 1982;
Grodzinsky, 1988).