K. R. C.

2001 1

Language Impairment,
Neurology and Linguistic
Theory
M.A. Thesis

Ken Ramshøj Christensen
Department of Linguistics, University of Aarhus
ken@aarhus.mail.telia.com

Supervisors:
Peter Bakker, Department of Linguistics
Sten Vikner, Department of English






K. R. C. 2001 2
Contents:


1 Preliminaries ........................................................................................................... 4
1.1 Introduction........................................................................................................... 4
1.2 Acknowledgements............................................................................................... 4
1.3 Abbreviations........................................................................................................ 5
2 Linguistic Theory.................................................................................................... 6
2.1 Modular Organization of the Grammar ................................................................ 7
2.1.1 X-bar Theory................................................................................................. 8
2.1.2 The Lexicon ................................................................................................ 10
2.1.3 The Projection Principle ............................................................................. 11
2.1.4 Extended Projection Principle..................................................................... 11
2.1.5 Move-α & Trace ......................................................................................... 12
2.1.6 Case Theory................................................................................................ 12
2.1.7 Theta Theory............................................................................................... 13
2.1.8 The VP-Internal Subject Hypothesis .......................................................... 14
2.1.9 Parameters................................................................................................... 15
3 Normal Brains: The Language Area................................................................... 17
3.1 Evolution of Language........................................................................................ 17
3.2 Localization, Distribution and Lateralization ..................................................... 21
3.3 The Language Zone ............................................................................................ 22
3.4 Language Acquisition......................................................................................... 24
4 Language Breakdown........................................................................................... 27
4.1 Mental Retardation and Impaired Language: Autism......................................... 27
4.2 Mental Retardation and Spared Language.......................................................... 28
4.2.1 Down’s Syndrome ...................................................................................... 28
4.2.2 Williams Syndrome .................................................................................... 30
4.2.3 Christopher – a Linguistic Idiot Savant ...................................................... 31
4.3 Impaired Language and Spared Cognition ......................................................... 32
4.3.1 Aphasia ....................................................................................................... 32
4.3.1.1 Broca’s Aphasia...................................................................................... 33
4.3.1.2 Wernicke’s Aphasia................................................................................ 34
4.3.1.3 Global Aphasia ....................................................................................... 35
4.3.1.4 Conduction Aphasia................................................................................ 35
4.3.1.5 Transcortical Aphasia ............................................................................. 36
4.3.1.6 Anomia.................................................................................................... 36
4.3.2 Specific Language Impairment ................................................................... 37
4.4 Discussion........................................................................................................... 39
5 Cerebral Area and Function................................................................................ 43
6 A Syntactic Approach to Broca’s Aphasia ......................................................... 49
K. R. C. 2001 3
6.1 Comprehension: The Trace Deletion Hypothesis............................................... 50
6.2 Production: The Tree-Pruning Hypothesis ......................................................... 56
7 Is Grodzinsky’s thesis applicable to Danish?..................................................... 61
7.1 The Danish Language ......................................................................................... 61
7.2 Predictions About Agrammatism in Danish ....................................................... 67
7.2.1 Comprehension ........................................................................................... 68
7.2.2 Production................................................................................................... 71
7.2.3 Summary..................................................................................................... 76
7.3 Empirical Tests ................................................................................................... 77
7.3.1 Comprehension: the Sentence-to-Picture Matching Test ........................... 77
7.3.2 Production: Repetition Test ........................................................................ 79
7.3.3 Patient TJ .................................................................................................... 82
7.3.3.1 Medical history ....................................................................................... 82
7.3.3.2 Comprehension Test Results................................................................... 83
7.3.3.3 Production Test Results .......................................................................... 85
7.3.3.4 Conclusions............................................................................................. 93
8 Discussion: Cerebral Area and Function Revisited........................................... 94
9 Conclusions.......................................................................................................... 103
10 References............................................................................................................ 107
11 Appendix A: Sentence-to-Picture Matching Test ............................................ 113
12 Appendix B: Sentence Types and Tokens......................................................... 118
13 Appendix C: Repetition Test: Structure Distribution..................................... 121
14 Appendix D: TJ's Comprehension Test Data................................................... 124
15 Appendix E: TJ's Production Test Data........................................................... 126


K. R. C. 2001 4
1 Preliminaries
This chapter contains the preliminaries: first a general introduction to the structure of
the paper, followed by my acknowledgements and a list of abbreviations.

1.1 Introduction
This study is about the neural predisposition of the brain underlying human species-
specific linguistic competence. It is about the connection between neurology and
linguistic theory. I discuss how findings from language impairments can play a crucial
part in determining the proper linguistic theory, as the latter must be compatible with
the patterns found in language deficits.
The structure of the study is as follows: First, I give a general introduction to the
grammatical framework of generative grammar in chapter 2. I describe the modular
structure of the grammar and introduce the modules (sub-theories) relevant for this
paper. Chapter 3 is about the biology of language - the evolution of language, the
localization of language in the language zone on the left hemisphere, and language
acquisition as viewed from the brain. I then turn to language impairment in chapter 4,
discussing the dissociation between language and intelligence as reflected in
developmental deficits such as Down's syndrome and in acquired deficits, i.e. different
types of aphasia. The survey of language breakdown will also provide evidence for the
internal modular structure of the human language competence. Chapter 5 is a discussion
on the correlation between lesion site and language function. In chapter 6 I present a
syntactic approach to Broca's aphasia based on the general framework of generative
grammar, which I use in chapter 7 to make some predictions about agrammatism in
Danish. These predictions are tested empirically on a Danish subject with some aphasic
symptoms due to lesions in the language area. Based on the previous chapter and the
test results I return to the discussion on brain area and function in chapter 8. Finally, In
chapter 9 I draw my conclusions, one of which is that language is reflected in the
architecture of the brain and therefore linguistic theory can be used to predict specific
symptoms of language breakdown in any language.

1.2 Acknowledgements
Thanks to TJ, who volunteered to help me by taking the language tests, and to Gerald
Fischer (speech therapist at Kommunehospitalet, Aarhus) for establishing contact with
K. R. C. 2001 5
TJ, and for being very helpful, for example on the interpretation of TJ's medical file.
Thanks to my supervisors: Peter Bakker (Department of Linguistics, University of
Aarhus) for being supportive and helpful beyond duty, and Sten Vikner (Department of
English, University of Aarhus) for his invaluable help on the specifics of grammatical
structure.

1.3 Abbreviations

Adject Adjectival
AdjP Adjective Phrase
Adv Adverb
AdvP Adverbial Phrase
AgrP Agreement Phrase
AUX Auxiliary verb
C Complementizer
Compl Complement
Conj Conjunction
CP Complementizer Phrase
DP Determiner Phrase
e Empty X
0
FEM Feminine
I Inflection
INFL Inflection
INF Infinitive
IP Inflection Phrase
LF Logical Form
MASC Masculine
N Noun
NegP Negation Phrase
NP Noun Phrase
Obj Object
P Preposition
PASS Passive
PERF Perfect
PERF+ Double perfect
PERS Person
PF Phonetic Form
PL. Plural
PP Preposition Phrase
PRES Present
Psych Psychological
PT Past
PTCP Participle
SING Singular
Spec Specifier
Sub Subject
t trace
TnsP Tense Phrase
V Verb
VP Verb Phrase
X
0
Head of an XP

K. R. C. 2001 6
2 Linguistic Theory
The linguistic framework in which I shall be working is generative grammar. I shall try
to avoid being too loyal to any specific version of generative grammar while trying to
retain the principles underlying most recent versions of the theory within the principles
and parameters frame, and therefore my grammatical analyses will be very much
influenced by Government and Binding Theory (Chomsky 1981, Haegeman 1994,
Vikner 1999). At least in principle, it will also be compatible with more recent
linguistic developments such as the Minimalist Program (Chomsky 1993, 1995) and
Optimality Theory (e.g. Archangeli & Langendoen 1997, Grimshaw 1997, Kager 1999,
Vikner 2000), as these frameworks all share some of the same core principles.
Generative linguistics takes as its goal to give an account of the linguistic
competence of humans, and as language is a universal human trait this competence is
conceived as a universal grammar (UG). This grammar consists of two parts: First, the
basic underlying universal principles that are shared by all languages of the world. They
are motivated by the simple fact that language is universal. Second, there is a set of
parameters whose settings determine the specifics of any language of the world, such as
the basic word order, for example Subject–Verb–Object in Danish. This is motivated by
the fact that languages are different in various but not arbitrary ways. Language
acquisition is viewed as the setting of these parameters on the universal grammar by
exposure to the ambient language. From this follows that the goal of generative
linguistics is not to describe the details of one specific language but rather to formulate
what principles determine the grammar of any language. In my description I only
include the principles relevant for the task at hand and only the necessary elaboration.
For general overviews I refer the reader to Grodzinsky 1990 (chapter 2 – an excellent
introduction), Haegeman 1994 and Vikner 1995 (chapter 2). The following is only a
brief and very general overview, but I shall give more specific definitions and introduce
further principles when required throughout the text.

K. R. C. 2001 7
2.1 Modular Organization of the Grammar
The grammar generates several levels of representation for every sentence. First, there is
an underlying structure called the D-structure. This is posited to account for the
structural (and semantic) connection between corresponding actives and passives. For
example, there is a strong connection between “the boy kisses the girl” and “the girl is
kissed by the boy”. They are not quite synonymous, but the relation is clearly felt. The
scenario (the kissing), and the thematic relations (the kisser and the kissed) between the
participants (the boy and the girl) are the same. The active and passive sentences are
two instantiations of the second level of representation: S-structure. As will be
explained below, some of the elements of the sentence are moved to other positions in
order to reach the S-structure from D-structure, cf. the difference in structure between
the active and the passive. Third, there is a phonological representation, or Phonological
Form (abbreviated PF), which is the form used for phonetic interpretation. Fourth, there
is the logico-semantic representation called Logical Form (LF). The levels are
organized as follows:

D-structure

S-structure

PF LF
Figure 1: The organization of the grammar

The principles of grammar are modularly organized. A module is a mechanism, which
is responsible for a designated domain. Therefore, the principles do not necessarily
apply at all levels, as they may or may not share the same domain or even parts of the
same domains. Anticipating the following descriptions of the principles involved, I just
list the principles here without going into details. In the figure below the principles are
placed in boxes, and the arrows point to the levels at which they apply, i.e. their domain
(I leave out principles not presented in this paper):

K. R. C. 2001 8

X-bar Theory D-structure
Theta Theory
Case Theory S-structure (Extended) Projection Principle
Move-α
PF LF
Figure 2: The modular organization and the domains of application of the
grammatical principles (adapted from Grodzinsky 1990: 25)

Consider the following example, in which I have left out some syntactic elaboration that
will be introduced later (I exclude PF and LF throughout the text, as they are not crucial
for my account):


“The boy kissed the girl” “The girl was kissed by the boy”
D-structure [ – ed [ [the boy] kiss [the girl] ]]
S-structure [[the boy] – [ – kiss-ed [the girl]] [[the girl] was [ – kissed – ]] by the boy
Table 1: syntactic representations.

This is an oversimplification, but the concepts will become clear as the principles are
presented. The important thing to notice in this example is the common underlying
structure. The lines indicate the movement of the verbal inflection “-ed” and the noun
phrases “the boy” and “the girl”, while the “–“ indicate the empty space left by the
movement.

2.1.1 X-bar Theory
Important to the theory are the notions of head, complement, specifier, and projection.
To explain this and the following principles it may be useful to consider an example:

(1) The boy kissed the girl from Sweden

K. R. C. 2001 9
A head is an element that gives a larger unit its characteristics. In the sentence above
“the girl” constitutes a noun phrase. The head of the phrase is the noun “girl”
1
, and
hence the phrase it heads becomes a noun phrase, abbreviated NP. The determiner
specifies the head, i.e. it is not just any girl, but the one “pointed out” by the specifier.
The complement of the head “girl” is the preposition phrase PP “from Sweden”, which
in turn is headed by the preposition “from” taking the NP “Sweden” as complement
(abbreviated as a triangle in (2) below). The elements are hierarchically ordered in the
following way:

(2) NP

spec N’

N
0
compl

The girl from Sweden

In the representation spec is the term used for specifier, and compl is the term for
complement. The head of the noun phrase is called N
0
, the zero level of the noun phrase,
and the N’ is the next higher level called N-bar (which is the level that has become the
name of the theory: this level is present in all phrases so in an XP (e.g. a noun phrase or
a preposition phrase this level is called X-bar). Finally, the highest level is called NP
(for noun phrase). This is also called the maximal projection of the head.
The X-bar schema is a constraint on all syntactic categories, which means that
all phrases must abide by it. Hence, depending on the head, the phrases are: NP (noun
phrase), VP (verb phrase), PP (preposition phrase), AdvP (adverbial phrase), and AdjP
(adjective phrase). In addition to these lexical categories (they are lexical because the
heads are words) there are functional heads, for example IP (inflectional phrase), which
is headed by the verbal inflection. All the phrases, regardless of the type of the head,
which is therefore abbreviated X, share same structure: the X-bar structure, which is
represented in the following way (a sort of ‘flattened tree’):



1
I leave out the notion of determiner phrases DPs where the head is the determiner (e.g. “the”), which
takes a NP as complement ([
DP
the [
NP
girl]]) and use the classic term NP instead.

K. R. C. 2001 10
(3) [
XP
spec [
X’
X
0
compl]]

For example:

(4) [
NP
the [
N’
girl [
COMPL
from Sweden]]]
(5) [
NP
boring [
N’
books [
COMPL
about linguistics]]]

I shall use a slightly abbreviated version of this notation leaving out the X’ branching
orthographically as shown in (6) and exemplified in (7):

(6) [
XP
spec X compl]

(7) [
NP
boring books [
PP
about linguistics]]

In this NP, “boring” is the specifier of N
0
, “books”, and the preposition phrase (PP) is
the complement. Henceforth I shall refer to the specifier position of an XP as [spec,
XP], which can be read as “specifier of XP”.
The structure of the clause includes functional as well as lexical categories. For
example an embedded clause like “(they said) that the boy kissed the girl” is introduced
by the complementizer phrase CP headed be the complementizer “that”, which takes as
its complement an inflectional phrase IP headed by the verbal inflection. This in turn
takes a VP as its complement. This is illustrated in:

(8) [
CP
that [
IP
[
NP
the boy] - [
VP
kissed [
NP
the girl]]]]

2.1.2 The Lexicon
Words are stored in a mental dictionary called the lexicon. A word’s lexical entry
specifies its syntactic category and its semantic meaning. Consider the verb, or rather
the predicate “kiss” in the example in (1) above. The lexical entry specifies that it is a
verb (V), and that takes a nominal complement (NP) – the verb subcategorizes for an
NP, which is the entity receiving the kiss (THEME). In addition it selects another NP,
which is the entity doing the kissing (AGENT). In other words, the predicate “kiss” has
K. R. C. 2001 11
two arguments. The lexical entry specifies their syntactic categories and their semantic
roles, and it may be represented like this:

(9) kiss: V <AGENT (NP), THEME (NP)>

2.1.3 The Projection Principle
The semantics of the predicate (the verb) determines to a large extent the structure of
the sentence. First, the specifics of the predicate (its lexical entry) determine the number
of arguments and their categories and semantic roles. So, lexical information is also
reflected in the structure of the clause:

(10) [
NP
The boy] [
VP
kissed [
NP
the girl]]

So, to obey the Projection Principle lexical information must be syntactically
represented. Leaving out, for example one of the arguments would result in
ungrammatical sentence, as indicated with an asterisk *:

(11) *[
VP
kissed [
NP
the girl]]
2.1.4 Extended Projection Principle
According to the Extended Projection Principle clauses are projected by the verb. The
VP contains the core thematic information: the predicate and its arguments (under the
VP-internal Subject Hypothesis, defined below). In short, the functional categories (CP
and IP, the latter to be further split up in section 6.2 below) do not contribute to the
thematic information as such, but tense inflection locates the event in time and the
[spec, IP] position assigns NOMINATIVE case to the subject. From a semantic point of
view, the clause is an Extended Projection of the verb, hence the Extended Projection
Principle (cf. Grimshaw 1991).

K. R. C. 2001 12
2.1.5 Move-α αα α & Trace
As shown in Table 1 above, the connection between D-structure and S-structure
involves certain changes to the positions of some of the constituents. These changes are
known as transformations or movements. The principle (or sub-component) of grammar
responsible for these transformations is the rule of Move-Alpha or Move-α:

(12) Move-α αα α
“Move anything anywhere”

These movements are constrained by the other modules of grammar in order to rule out
ungrammatical forms, such as “*the the kissed boy girl” and “*was kissed the girl by the
boy”. It is the interaction of Move-α with the other syntactic principles that results in
grammatical strings. It is responsible for the mapping between active and passive, for
example.
Movement of constituents results in a phonologically silent but structurally
represented position, which is filled with a construct, called a trace (t). The moved
constituent is the antecedent of the trace and together they form a chain. The two are
further linked together by a shared index. Consider again the example from (8) (without
the CP):

(13) [
IP
[
NP
The boy]
1
t
2
[
VP
t
1
kissed
2
[
NP
the girl]]]

The verbal inflection that headed the IP has moved to the verb “kiss” in the VP, and has
left a co-indexed trace (t
2
) behind in I
0
. The subject “the boy” has also undergone
movement and left a trace, and this movement is due to the Case Filter, which I shall
now explain.

2.1.6 Case Theory
This is the component of grammar specifying the structural position in which lexical
NPs may appear. According to this theory all NPs must be assigned abstract case. In
some languages this case is overtly realized as case endings, e.g. German, while in
others it is invisible, e.g. English. Certain elements are case assignors, such as tense
K. R. C. 2001 13
inflection, verbs, and prepositions. Thus, syntactic structures containing NPs unmarked
for abstract case are labeled ungrammatical and filtered out by what is called the Case
Filter at the level of S-structure, cf. Figure 1:

(14) Case Filter:
*NP
[+lexical, -case]

In other words: all lexical NPs must have case.

In the example in (13) above, the subject was forced to move to [spec, IP] (specifier of
the inflection phrase IP) in order abide the Case Filter. If it remains in its base position
inside VP (I shall return to the position of the subject shortly; so far, suffice it to say that
the subject is moved) it will not be case marked and the structure will be filtered out.
The subject case NOMINATIVE is assigned to [spec, IP], and object case
ACCUSATIVE is assigned by the verb to its complement “the girl”. Thus, the
interaction of the Case Filter and Move-α determines the position of the subject.

2.1.7 Theta Theory
Arguments are assigned semantic / thematic roles, known as theta-roles or θ-roles, such
as AGENT (the ‘doer’ of the action specified by the predicate, e.g. “the boy” in (1)) or
THEME (the entity affected by the action). A principle called the Theta Criterion
(Chomsky 1981) ensures that all arguments are assigned one and only one θ-role and
that all θ-roles of a predicate are assigned to appropriate structures in a one-to-one
relation. This can be illustrated as follows:

(15) Argument
1
– θ-role
α

Argument
2
– θ-role
β


θ-Roles constitute a universal set including the following:

(16) AGENT: The one intentionally doing or initiating the action expressed by
the predicate.
K. R. C. 2001 14
THEME: The entity affected by the action or state expressed by the
predicate.
EXPERIENCER: The entity experiencing the psychological state expressed by the
predicate.

There are other θ-roles, but the ones defined above are the only ones needed for my
purposes in this study. For further information, see e.g. Haegeman (1994: 49ff) and
Jackendoff (1972).

2.1.8 The VP-Internal Subject Hypothesis
According to the VP-Internal Subject Hypothesis, the subject is base generated inside
the verb phrase in [spec, VP] (specifier of VP), cf. Burton & Grimshaw (1992). As
explained above, to abide the Case Filter the subject moves to [spec, IP] to be assigned
NOMINATIVE case leaving behind a trace, with which it shares a common index (i):

(17) IP

NP
i
I’

I
0
VP

t
i
V’

V
0
compl

The idea is that in the underlying structure (the D-structure) the subject (NP
i
in the
diagram) is generated in [spec, VP], where V
0
assigns a θ-role to the NP (in other
words, the verb theta-marks the subject). In English NOMINATIVE is assigned to
[spec, IP] by I
0
(cf. Haegeman 1994):

NOMINATIVE
(18) [
CP
spec C
0
[
IP
spec I
0
[
VP
spec V
0
compl]]]
[
IP
he
1
t
2
[
VP
t
1
loves
2
her]]]
“he-NOM loves her-ACC”

K. R. C. 2001 15
In Danish, NOMINATIVE is assigned to [spec, IP] by C
0
(e.g. Vikner 1995: 54ff.):

NOMINATIVE
(19) [
CP
spec C
0
[
IP
spec I
0
[
VP
spec V
0
compl]]]
[
CP
han
1
elsker
2
[
IP
t
1
t
2
[
VP
t
1
t
2
hende]]]
“he-NOM loves her-ACC”

2.1.9 Parameters
As mentioned above, the grammar consists of principles and parameters. An example of
the latter is reflected in the difference between (18) and (19) above. The parameter
involved ‘decides’ what assigns NOMINATIVE case: in Danish C
0
is the
NOMINATIVE-assignor and in English it is I
0
.
Another parameter concerns word order: in e.g. Danish and Swedish the finite
verb is always the second constituent in the sentence (preceded by the subject or a
topicalized element), while this is not the case in English and French. All four language
shave Subject-Verb-Object (SVO) word order, but in Danish and Swedish the verb is
always second, so if e.g. the object is topicalized the subject follows the verb: Topic-
Verb-Subject. This is not the case in English and French, where the verb must follow
the subject: Topic-Subject-Verb. This parameter is the V2 (verb second) parameter:
Danish and Swedish are V2 languages, English and French are not. Briefly, in the
grammar of a child acquiring e.g. Danish parameters will be set to SVO and V2, while
an English child will have the same parameters set to SVO and non-V2 (this is a
simplification but it will suffice).
Still another example of a parameter is the type of case system a language has.
For example English, German, Hebrew and Danish all have a NOMINATIVE-
ACCUSATIVE case system while for instance Chukchi (a language of the north-east
Siberia, cf. Comrie 1989: 104) and Bandjalang (spoken in the northern New South
Wales in Australia, cf. Crowley 1997: 137) has ABSOLUTIVE-ERGATIVE. Other
languages have both types, such Diyrbal (a language of the northeastern Queensland in
Australia, cf. Comrie 1989: 113).



K. R. C. 2001 16
INTRANSITIVE TRANSITIVE
“she pushed” “she pushed him”
Grammatical Roles Sub Sub Obj
Accustive System NOM NOM ACC
Ergative System ERG ABS ERG
Table 2: (NOMINATIVE-) ACCUSATIVE and
(ABSOLUTIVE-) ERGATIVE cases systems.

In the ACCUSATIVE system the subject in both transitive and intransitive clauses is
always in NOMINATIVE case. In the ERGATIVE system the subject of an intransitive
clause has the same case marking as the object of a transitive clause, i.e. ERGATIVE
case. The English gloss of the ERGATIVE version of “she pushed him” would be “her
pushed he”.

In summary, the principles and theories listed here are sub-components of a
modularly structured grammar. In interaction they generate the grammatical structures
of a language. In chapter 6 I shall argue that the modular structure of the grammar is
relevant and even crucial in the definition and diagnostics of aphasia.


K. R. C. 2001 17
3 Normal Brains: The Language Area
In this chapter I first discuss the evolution of language and the brain, which most likely
started with a cognitive ‘arms race’ leading to selection for bigger brains. In time
language has become represented in the architecture of the human brain: species-
specific, universal, and innate. Then I move on to discuss where language functions are
located in the brain: nowhere, everywhere, in one place or in many places. This leads to
the next section in which I zoom in and look at an area in the left hemisphere. This area,
I shall argue, is the language organ and I support this claim with a brief survey of
language acquisition from a neural perspective. In short, this chapter is about the
biology of language.

3.1 Evolution of Language
As mentioned in the preceding chapter, the goal of generative linguistics is to provide
an account of the human linguistic competence. This competence is both unique to
mankind and universal among the peoples of the world. This species-specificity
naturally entails that language is different from any other kind of communication in the
animal kingdom (cf. Deacon 1997; Donald 1991; Pinker & Bloom 1990). Some aspects
of animal communication and the rudimentary symbolic communication (laboriously)
taught to apes may have some analogies to human language in the same way that bat
wings are analogous to the ‘wings’ of flying squirrels – both are used for flying. It is
clear though that the two kinds of wings are not homologous, which means sharing the
same basic structural design stemming from a common ancestor. Bat wings are
modifications of the hands of the common mammalian ancestor and the wings of the
flying squirrel are modifications of the rib cage (cf. Pinker 1994, 1995). Universality
covers the observation of the fact that all peoples of the world have a language – no
mute and non-signing tribe has ever been discovered. Together these two facts, species-
specificity and universality, point to some kind of specific human capacity or faculty
represented in the brain as some sort of neural circuitry that is unique and universal to
humans and dedicated to language processing. In other words, the universal grammar
described above is innate. Such a claim, I think, has to be supported by a feasible
account of the origins of such a competence. The following is a brief discussion of how
this competence could possibly as well as probably have come to be innate.
K. R. C. 2001 18
It is generally believed (cf. Deacon 1997; Donald 1991; Dunbar 1993; Pinker &
Bloom 1990; Pinker 1994, 1995) that the onset of human evolution and the break off
from the chimps was the emergence of bipedalism in the Australopithecus Afarensis
approximately 4 million years before present, which so to speak freed the hands for tool
use. However, the evidence for a social structure capable of supporting language
evolution is better with the later Homo Habilines, who cooperated in hunting and
nurturing.
According to one theory (Deacon 1997) what was special about the early
hominid social groups was that they were monogamous couples living together in larger
groups. Following his argument, monogamy is restricted to animals living in deserted
and preferably vast areas where there is little contact with other members of the species.
The early humans on the other hand, had much contact as they were living in groups.
This lead to the evolution of symbolic communication that would enable them to
establish and maintain symbolic social relations, or rather marital contracts, and
furthermore ensure that all members of the group honor them.
Pinker (1984) proposes a different theory, in which the primary initiator was not
monogamy. What initiated the evolution of symbolic communication (besides the
obvious benefit of sharing information) was the need to outwit potential social cheaters:
those that reap the benefits of this cooperative hunting and nurturing, while not paying
the costs of participation. See also Bickerton (1990), Donald (1991), and Foley (1997)
for discussions on the evolution of language.
In either case there was selection pressure for organisms with communicative
skills. This led to a cognitive ‘arms race’ which in turn lead to the enlargement of the
human brain. The human brain is large in relation to the body – a phenomenon known
as encephalization. This trait is not unique to mankind (sparrows and mice for example
have even bigger brains in relation to their bodies than humans), but in this respect we
clearly differ from the other apes. Actually our brain is three times larger than expected
on the basis of our body size (cf. Deacon 1997). Of course, a larger brain does not
inevitably produce support for language. In other words, language is not the result of a
larger brain – whales for example have quite larger brains than humans, but they do not
have a language. However, as Dunbar (1993) points out, there is a strong connection
between group size, brain size and language. As primate groups grow in size, so does
the structural complexity: there are more social relations to keep track of, so to speak
(unlike large flocks of cattle, which has very little structure). This calls for greater
K. R. C. 2001 19
cognitive abilities, which in turn gives rise to the enlargement of the brain. However, it
is unresolved whether this results from the complexity of living in groups of
monogamous couples as Deacon claims or from the cognitive demands of outsmarting
uncooperative social parasites as Pinker claims, i.e. sexual versus natural selection
respectively (Darwin 1859). I suspect the truth lies somewhere in between.
The human brain is not just an ‘inflated’ chimp brain. Not all of it is larger than
would be expected for a primate of human size. Due to the encephalization a relatively
smaller portion of the cortex of the human brain is devoted to controlling bodily
functions, the body being relatively small in relation to the brain. The size of the motor
areas is thus actually reduced in relation to the rest of the brain. However, the auditory
areas in the superior temporal gyrus including Wernicke’s area (see Figure 3 below) is
enlarged, and most important the prefrontal cortex including Broca’s area (both areas
are further discussed below) is twice its expected size (cf. Deacon 1997:217). This
means natural selection has given the prefrontal cortex dominance over the rest of the
brain. In development the different parts of the brain connect to all other parts of the
brain. Through a process named neural Darwinism (Edelman 1998) the largest part,
which thus has the largest number of connections to a given area, wins the control over
it – in other words, this is ‘survival of the fittest’ played out on a developmental
neuronal level.


Figure 3: Idealized map of cerebral areas in the left hemisphere. The black line
surrounds the so-called perisylvian region. The names in white indicate functional
areas, while the black names merely indicate location. In the text the term
posterior refers to areas (in the figure) to the right of the motor area.
K. R. C. 2001 20

To Deacon (1997) all this indicates that the prefrontal cortex houses a ‘symbolic
processor’ (and (some) support comes from Donald 1991), which is also responsible for
language. Language acquisition is thus based on a specialized learning capacity for
symbolic systems (below I shall argue that this is not altogether plausible). In other
words, humans are predisposed to language in the sense that they have a special
learning algorithm for language. Deacon argues that the cortical areas in the left
perisylvian region (see Figure 3) are not specialized for language but that they are
computational centers also utilized for linguistic purposes.
According to Pinker (1994) on the other hand the left perisylvian region is the
language organ, i.e. these areas are specifically dedicated to language. To Pinker
language acquisition is not dependent on a general symbolic learning capacity, but
rather a matter of instinctive and automatic acquisition of the ambient language,
consisting of setting the parameters of the innate universal grammar to match the
surrounding language. This explains why language is universal and species-specific and
independent on general learning capacities.
Any way we slice it, language and the brain have co-evolved: language has
shaped the brain and vice versa (cf. Christensen 2000 for a discussion on the co-
evolution of language and the brain and a comparison of Pinker 1994 and Deacon
1997). It therefore seems natural to say that language has been internalized, i.e.
represented in the neural architecture of the brain, and therefore the brain is already
prepared for linguistic input. Furthermore, since neither typological language variation
(Comrie 1989) nor diachronic language change (Crowley 1997) is completely arbitrary
and unconstrained, it seems natural and plausible to claim that there is such a thing as a
universal grammar (cf. Pinker 1994). Both cross-linguistic and cross-time variation is
constrained by the specialized structures of the human brain. Below I shall argue that
the areas in the left perisylvian region indeed are specialized for language.
Physiologically, the most obvious adaptation of the human body to language is
the (again) species-specific position of the larynx deep in the throat, which gives us two
cavities (mouth and throat), which together define the large array of possible human
vowel sounds. This obvious benefit for our phoneme inventory comes with a severe
disadvantage for breathing and swallowing, as there is a risk of food falling into our
lungs (cf. Lieberman 1984).

K. R. C. 2001 21
3.2 Localization, Distribution and Lateralization
Where is language represented in the brain? There are four logically possible answers:
1: nowhere, 2: everywhere, 3: in one place, or 4: in several places. The first answer is
not an option, as damage to the brain can lead to selective impairment of language, as I
shall discuss in more detail in chapter 4. The second possibility, ‘everywhere’, is not
plausible either. The implication is that language acquisition is done by a general
learning mechanism and hence dependent on general intelligence and vice versa.
Language is independent and distinct from ‘general cognition’, and impaired general
intelligence need not have an impact on language, as I shall show below. This is of
course closely related to the notion of modularity (Fodor 1983, 1985). Language is a
cognitive module, a somewhat self-contained subsystem of the human mind (see also
Pinker 1994, 1997). I shall return to the matters of modularity and dissociation between
language and intelligence in the following sections.
This leaves us with the last two possibilities: 3: language is localized in one area,
and 4: language is distributed over several areas. A logical place to start is to find out if
language is represented in the right, the left, or in both hemispheres, i.e. left-, right-, or
not lateralized. Studies of brain damage leading to language disorders have shown that
language normally is localized in the left hemisphere, as the correlation of aphasia and
right-hemisphere damage is very rare, and encountered only if the patient suffers from
early left hemisphere damage (cf. Bishop 1988). Right hemidecortication (removal of
the entire cortex in one hemisphere) and lesions in the right hemisphere cause some
semantic and/or pragmatic disorders, such as problems with understanding narratives
and jokes (cf. Damasio 1992: 537; Deacon 1997), and aprosodia, a syndrome where the
patient’s speech is flat and lacks prosody (cf. Donald 1991: 80 and Calvin & Ojemann
1994: chapter 4). These deficits are nonlinguistic, as language per se is not damaged. It
would be very strange to diagnose a person as language impaired if he or she has
problems with getting the point of jokes and stories or just speaks in a flat and
monotonous manner. Grodzinsky (2000: 19) concludes:

Thus, the evidence is that this side of the brain [i.e. the right side. K. R. C.] has
an important role in communication but makes no syntactic contribution to
language use.

K. R. C. 2001 22
The discussion so far points to a localization of language in the left hemisphere,
and this is supported by results from tests where one hemisphere is anesthetized prior to
brain surgery (this is called the Wada test, cf. Bishop 1988: 206). This is done in order
to discover whether language functions are located in the left (as is almost always the
case) or in the right, which is sometimes though rarely the case. Anesthetizing the
‘linguistic’ hemisphere causes language impairment (cf. Bishop 1988, Calvin &
Ojemann 1994: chapter 3 and Donald 1991:80). Deacon (1997) views the difference
between the hemispheres not as linguistic versus non-linguistic computation, but as a
difference in speed of computation. The right hemisphere is specialized in large time
domains (linguistic and non-linguistic), which should account for the above mentioned
problems with narratives and prosody for patients with right-hemisphere injuries.
Prosody is a feature of the inter-sentential domain, as it spans the entire utterance
regardless of the number of clauses. The left side is specialized in short time domains.
In other words, the left hemisphere is speed optimized, which should account for the
breakdown of syntax and morphology in left-hemisphere injuries (cf. Deacon 1997:
316).
Much in the same line as Deacon, based on evidence from bilingual aphasics
Paradis (1998) places what he calls implicit linguistic competence in the left
hemisphere. Regarding the right hemisphere, he states that “one can safely assume that
the RH [right hemisphere, K. R. C.] is crucially involved in the processing of pragmatic
aspects of language use” (Paradis 1998: 422).

3.3 The Language Zone
Within the left hemisphere language functions are localized in an area around sulcus
lateralis, also known as the Sylvian fissure (see Figure 4 below), which is the long
‘gorge’ that constitutes the upper ‘border’ of the temporal lobe. In the vast majority of
cases where brain damage causes language deficits it is located the perisylvian region,
i.e. in the region surrounding the Sylvian fissure (cf. Damasio 1992, Damasio &
Damasio 1992, Calvin & Ojemann 1984, Gazzaniga 1989). PET studies have shown
that there is increased blood flow, hence cerebral activation, in this area during
linguistic processing. Broca’s area (the left inferior frontal gyrus) is especially active
during tasks involving syntactic analysis (Stromswold et al. 1996).

K. R. C. 2001 23

Figure 4: Language areas (from Pinker 1994:308).

This area, the left perisylvian region, is often discussed as if it consisted of a collection
of separate parts or areas. This is apparent in Figure 4, where the different areas around
the Sylvian fissure seem to be delimited in a nice clear-cut fashion. The brain does not
have a flat surface; its surface is full of convolutions, much like if you take a large piece
of cloth and crumble it up into a ball. Therefore, much of the surface is not ‘visible’ and
the topmost parts seem adjacent, but sometimes they may appear to be separate.
However, Gazzaniga (1989) has shown what this region of the brain would look like if
it had been ironed flat. All the areas that are associated with language (and language
breakdown, see below) form a contiguous area, which he labels the language zone
(Gazzaniga 1989: 950). For the same reason Pinker (1994) and Grodzinsky (2000) call
this region the language organ.
Further evidence comes from studies of bilinguals. Dehaene et al. (1997)
investigated a group of French-English bilinguals. The subjects were scanned in a fMRI
(functional magnetic resonance imaging) while listening to stories in the two languages.
The results of the tests showed a consistent and significant increase in activity in the left
perisylvian region during comprehension of the subjects' first language (L1, English),
whereas the second language (L2, French) showed a weaker but similar pattern.
Dehaene et al. suspect that the latter may be due to different teaching strategies in the
schools, where the subjects had learned their L2. Whatever the reasons for the less
consistent pattern of L2 in their investigation, the subjects' native language (L1) was
located in the perisylvian region.
Others have found more consistent results showing that both L1 and L2 are
located in the language zone. According to Paradis (1998: 422) (see also Menn et al.
K. R. C. 2001 24
1995) both languages are represented in the language areas of the left hemisphere in
bilingual as well as unilingual individuals.

3.4 Language Acquisition
The maturation of the human brain is the driving force in language acquisition. In this
section I briefly discuss the processes in the brain during language acquisition.
Before birth all the nerve cells of the brain, the neurons, are formed and migrate
from the place where they are generated into their proper places. At birth, the brain is
already biased towards a left-hemisphere specialization for language (cf. Bishop 1988:
212). After birth the size of the brain and the thickness of the cerebral cortex increase in
the first year. The white matter (underlying the cortex or gray matter), which houses the
long-distance connections between the major regions is not complete until around nine
months of age. The synapses, which are the connections between the neurons in the
cortex, continue to develop and peak in number sometime between nine months and two
years, depending on the region of the brain. At approximately two years of age the brain
has 50% more synaptic connections than the adult brain. The metabolic activity reaches
an adult level at around nine months and rapidly exceeds it, peaking at around 4 years of
age. Huge numbers of neurons die from birth to around the age of seven. The synaptic
connections decrease in number until adolescence, when the brain falls back to the adult
level of metabolism (for elaboration see Deacon 1997: chapter 6; Elman et al. 1996:
chapter 5; Müller 1996: section 3.2.1; Pinker 1994: chapter 9, 1995).
I have summed up the correlation between developmental neural and linguistic
events in the following table:

Age Neural development Linguistic development
birth Completed cell formation and migration left hemisphere specialization

around 9 months Adult distribution of metabolism,
long-distance connections established
suppression of non-native sounds - fine-tuning
the phonetic inventory
around 12 months one word stage
around 18 months Peak in number of connections within and
between cortical regions
vocabulary spurt, two word stage - primitive
syntax
around 36 months grammar explosion - rapid increase in number
of syntactic constructions and complexity
around 48 months peak in overall brain metabolism successful language acquisition
Table 3: Developmental neural and linguistic events.

K. R. C. 2001 25
Evidence for the early left-hemisphere bias of language at birth comes from
cases of brain damage. According to Bishop (1988) the majority of children suffering
left hemidecortication or brain damage within the first couple of years of life do not
develop aphasia. The ability to recover rapidly decreases with age and chances are best
before the age of ten. Furthermore, there is a strong correlation between early damage to
the language area and right-hemisphere language specialization. This is due to the fact
that the child brain is very plastic, i.e. functional reorganization is possible in the very
early stages. As Bishop states:

Provided that the traditional language areas were spared, even extensive early
lesions of the left hemisphere did not result in right-hemisphere language
representation. However, lesions that encroached upon Broca's area or
Wernicke's area were likely to bring about a functional reorganization of the
brain, with right-hemisphere specialization for language. (Bishop 1988: 207)

It appears then that there is some sort of 'default brain plan'. The brain does not
consist of a vast number of identical neurons – it is not an equipotential mass (Müller
1996: section 3.2.1) or a meatloaf (Pinker 1994). As mentioned above, the neurons
migrate from the place where they are generated to their proper locations. The neurons
themselves are thus in place at around seven months after conception (cf. Elman et al.
1996:288), but the number of neurons and the number and strength of the connections
between them need to be adjusted to suit the environment. The important point is that
the brain seems to be 'prepared' for language but still dependent on stimuli from the
environment. The default brain develops following the bioprogram (Bickerton 1988),
while adjusting / reorganizing itself to match the ambient language. This reorganization
is dependent on the neural plasticity, which is highest in infancy and rapidly decreases
during the first years of life, reaching the adult low at around puberty. This period may
be called the critical period (Lenneberg 1967) or the sensitive period (e.g. Elman et al
1996). This period is the time window in which the brain is open to successful language
acquisition and recovery from lesions. After this period successful first language
acquisition is rare. Also, bilingual speakers’ first language will usually permanently
influence the acquisition of their second language (for example through accent) after
this sensitive period, which is not the case if both L1 and L2 are acquired within the
period.
K. R. C. 2001 26
Recall from chapter 2 that in generative linguistics a language is viewed as
parameter setting on a universal grammar. Language acquisition is viewed as the
process in which these parameter settings become fixed. This fits very well with the
facts of brain development. Both nature and nurture, or genetics and environment must
contribute in order to secure successful language acquisition. Thus, innateness does not
mean that everything is pre-wired in the brain of the infant:

No nativist has ever supposed that innate capacities are unaffected or unformed
by environmental interactions. […] How many times has this point been made in
the last 30 years? How many times in the last three centuries? How many times
is one going to have to make it again?” (Fodor 1985: 36)

So far, I have argued that evolution has lead to a neural predisposition to
language learning / acquisition. This predisposition is reflected in the architecture of the
human brain in that linguistic functions are located in the left hemisphere, i.e. language
is left lateralized. Within the left hemisphere, language is localized in the region
surrounding the Sylvian fissure – the language zone. Furthermore, language functions
are distributed over the language zone. This neural specialization is the basis of
language acquisition. The infant brain develops according to some kind of a default plan
/ bioprogram, a process to which the environment provides crucial input. This
interaction between (genetic) neural predisposition and environment has to take place
within a limited time frame, the critical period, to be successful.
In the next chapter, I discuss what happens when the language zone is damaged
– either by developmental deficits, which leads to neural malformation, or by damage to
previously normal adult brains.
K. R. C. 2001 27
4 Language Breakdown
This chapter is about the kinds of impairment or trauma that lead to the breakdown of
language. It is a brief survey of language deficits, some of which are developmental
deficits – autism, Down’s syndrome, Williams syndrome, and Specific Language
Impairment – and other are acquired deficits due to brain damage – aphasia and
anomia. From this survey I shall conclude like others before me that there is a double
dissociation between language and intelligence. Hence, language acquisition is not a
general learning problem for some central processor. Language is innate, which is also
evident from the very robustness of language, which is (sometimes) acquired or spared
in spite of brain damage. This also points to modularity – externally, in that it is
independent of e.g. hearing and intelligence, and internally, in that there exist deficits
specific to certain aspects of grammar. The dissociation between language and
intelligence is used to establish a typology of language impairment and mental
retardation, which again supports the modularity hypothesis. Finally, I link the different
kinds of acquired language impairments to different areas of the brain – or rather to a
certain region in the left hemisphere: the language organ.

4.1 Mental Retardation and Impaired Language: Autism
Autism is a severe developmental deficit that typically appears some time during the
first three years of life. People suffering from an autistic disorder (Bishop 1989) tend to
be (but are not always) mentally retarded with an I.Q. below 60. They lack what is
known as a theory of mind; that is, they lack empathy and the awareness of other people
as conscious beings with feelings and emotion. They tend to be socially withdrawn.
According to Fay (1988) mutism is a frequent characteristic symptom associated with
autism. Furthermore, autistic people tend to lack communicative intent. Interestingly,
some appear to be deaf as they do not respond when spoken to, but this deafness is
restricted to linguistic sounds, while they may show signs of intolerance to other
environmental sounds, such as vacuum cleaners. Of the minority that does develop
speech approximately 75% go through a prolonged period of echolalia, the repetition of
words or fragments of sentences just uttered by others. Their speech is often
monotonous and mechanical.
K. R. C. 2001 28
As is evident from this short description of autism, it is a mentally retarding
deficit causing low intelligence and a blocking or mal-acquisition (abnormal acquisition
leading to abnormal language) of language per se – not just of speech. In short, autism
is a case of mental retardation and impaired language.
Yet another strange characteristic of the syndrome is the so-called idiot savants
(see Frith 1989: 84), who seem to have an ‘island’ of spared skill on a background of
generally poor learning abilities. Often this involves very good long-term memory,
which is used to memorize long lists of names, animals, telephone numbers, bus routes,
etc. Other kinds include mathematical geniuses who can mentally multiply large
numbers at a rapid speed, and musical prodigies.
Neurologically, the basis of autism is still not completely resolved (cf. Frith
1989: 68-81). According to Deacon (1997: 275) (based on Damasio 1994), autistic
people tend to have smaller cerebellums and brain stems, but the only direct evidence of
cortical involvement in the deficit seems to focus on the reduced blood flow of the
prefrontal lobes, which includes Broca’s area.

4.2 Mental Retardation and Spared Language
In the following three subsections I first discuss two developmental deficits, Down’s
syndrome and Williams syndrome, and then discuss a patient called Christopher, who is
a so-called idiot savant. All three are instances of mental retardation and spared
language. They are not isolated cases; actually, this dissociation between language and
mental retardation or alternative intelligence is abundant: dementia, schizophrenia,
psychopathy, and insanity – all examples of non-normal intelligence/cognition and
undamaged language, the latter sometimes quite clearly intact.

4.2.1 Down’s Syndrome
As in autism, people suffering from Down’s syndrome (also known as mongolism) are
mentally retarded with an upper limit of a mental age comparable to that of a normal 4
to 5 year-old child. The intellectual maturation tends to be complete at around 12 to 15
years of physical age.
Unlike autistic patients, subjects suffering from Down’s syndrome do not lack
communicative intent and / or ability. Most children with Down’s syndrome begin to
speak between the age of two and four years in spite of many physical abnormalities.
K. R. C. 2001 29
Among the numerous factors working against them are pathologies such as an
undersized mouth cavity, a protruding and edematous tongue, a larynx often positioned
too high in the throat, hypotonia (reduced sensitivity to stimulation) of the speech
muscles, abnormal lips, hearing loss due to malformations of the inner ear, etc. As
Rondal puts it:

Given the number and severity of factors that militate against them, it is
surprising that individuals with Down’s syndrome develop language at all.
Should we need another proof of the robustness of language in the face of
biological and psychological hazards, Down’s syndrome subjects amply supply
this proof. (Rondal 1988: 166)

Even though there is inter-individual variability within the syndrome, language
acquisition is generally delayed, but it goes through the same stages as normal subjects.
However, due to the physiological deficiencies, speech production is impaired. They
have a reduced lexicon but it is used and understood correctly. According to Rondal
(1988: 171) their utterances tend to be short and telegraphic with somewhat limited
morpho-syntactic elaboration, and they show limited use of ‘function words’, such as
prepositions, copulas, auxiliaries, pronouns and conjunctions. These claims are
supported by measures of MLU (mean length of utterance), which however say nothing
about morpho-syntactic elaboration, use of function words, or about telegraphic speech
for that matter. In fact, the only example in the text (Rondal 1988: 170) includes all of
these ‘function words’.
As opposed to autistic patients, Down’s syndrome subjects tend to be very social
and generally happy. Furthermore, they are very close to normal people regarding the
theory of mind. Baron-Cohen et al. (1985) tested 20 autistic children (mean I.Q. 82), 14
Down’s syndrome children (mean I.Q. 64), and 27 clinically normal children. The test
consisted of a marble and two dolls, named Sally and Anne, hence the test is called the
‘Sally-Anne’ Test. Sally placed a marble in her basket and left the scene. Then Anne
took the marble out of the basket and hid it in her box, after which Sally returned. The
experimenter then asked the subjects where Sally would look for the marble. The
correct response was to point at the basket and the wrong to point at the box. 86% of the
normal subjects and 85% of the Down’s syndrome children were correct, while only
20% of the autistic subjects were correct.
K. R. C. 2001 30
4.2.2 Williams Syndrome
Williams patients are characterized by a peculiar facial appearance, which includes a
star-like pattern in the iris, eyebrow hair growth towards the nose, a narrow face, a
broad forehead, a flat nasal bridge, sharp chins, and thick lips (cf. Bellugi et al. 1988:
178, Pinker 1994: 52). For this reason they are sometimes called “elfin-like” or “elfin-
people” (or as Pinker points out, they look like Mick Jagger). They are severely
mentally retarded with an I.Q. at approximately 50 and are unable to live independent
lives even as adults. They have difficulties with dressing themselves, remembering
routines, tying their shoes, finding their way, etc. Their visuospatial capacity is also
severely impaired. For example, when they draw an elephant or a bicycle it is not a
cohesive, recognizable drawing resembling what was intended, but a collection of parts
(though correct parts). In spite of this visuospatial impairment their facial recognition is
unimpaired, and they are friendly and highly social people, and notably very loquacious.
A striking dissociation exists between their inability to draw e.g. an elephant and
their ability to describe it – which they do eloquently. Another characteristic of
Williams syndrome is the unusual vocabulary for the mental age of the patients.
Whereas a normal child would mention a dog, cat, horse, cow, etc. when asked to
mention some animals, a Williams patient would mention such (to Europeans unusual)
animals as hippopotamus, chihuahua, antelope, condor, vulture, sabretooth, etc. As
opposed to Down’s syndrome patients, their range of vocabulary is above their mental
age.
Language acquisition is delayed, but apart from that their language (both
comprehension and production) appears to be intact and normal. Their expressive
language is complex in terms of morphological and syntactic structures, such as
inflection, tense and aspect markers, passives and various types of embeddings, and
generally grammatically correct, and they are able to use these structures productively
and appropriately. “Complex structures in the spontaneous speech of the Williams
syndrome children are abundant” (Bellugi et al. 1988: 183). And in sharp contrast to the
autistic patients, their language use is clearly not echolalic or formulaic.
Williams syndrome subjects suffer from supravalvular aortic stenosis (a
narrowing of the aorta) and from abnormalities in the metabolism of both calcium and
calcitonin (Bellugi et al. 1988: 178). This rare and genetically based metabolic disorder
is called hypercalcemia and has various effects on the organism:
K. R. C. 2001 31

The syndrome seems to be associated with a defective gene on chromosome 11
involved in the regulation of calcium, and it acts in complex ways on the brain,
skull, and internal organs during development, though no one knows why it has
the effects it does. (Pinker 1994: 52)

As Deacon (1997:269) explains, postmortem and MRI analyses of Williams patients’
brains have revealed reduction of the entire posterior lobes, while the prefrontal lobes
are spared. The brains of Williams syndrome patients are even more ‘front heavy’ than
normal brains, which already have relatively large frontal lobes (cf. section 3.1 above).
Interestingly, this is the direct opposite neural pattern of the autistic subjects.

4.2.3 Christopher – a Linguistic Idiot Savant
Smith & Tsimpli (1995) describe a very special and quite unique person called
Christopher. He is institutionalized because he cannot look after himself. He is mentally
retarded and has an I.Q. around 60 (42-76 depending on the specific test, ibid. p. 4). As
Smith & Tsimpli put it, his medical and neurological history is rather opaque, but
several tests and scans have shown that he has suffered hydrocephalic brain damage and
severe neural impairment of his motor co-ordination, some cerebral atrophy with wide
sulci over both hemispheres. An EEG (electroencephalogram) scan showed slow waves
in the frontal lobes, i.e. reduced activity there. He has not been diagnosed as autistic, but
some evidence points in this direction. First, he failed on an equivalent to the ‘Sally-
Anne’ test (see section 4.2.1 above) showing an apparent lack of a theory of mind,
which is characteristic of (and perhaps specific to, cf. Baron-Cohen et al. 1985) autistic
subjects. Second, the mentioned reduced activity in the frontal lobes is consistent with
the reduced cortex characteristic of (at least some types of) autism. Third, and most
interesting, he has a single spared (or overly developed) mastery or skill in the face of
otherwise retarded intelligence: language. Christopher is a linguistic ‘idiot savant’ (cf.
Frith 1989): he cannot button his shirt or tie his shoes but he can read, write and
communicate (though with some differences in degree ranging from fluency to bare
elements) in fifteen to twenty different languages for which he has had no formal
training. This is very surprising because autistic subjects tend to lack any form of
communicative intent and ability, as mentioned above, and if they have any it is
K. R. C. 2001 32
severely impaired. This single linguistic skill is characteristic of Williams syndrome,
which also fits his very social nature but not his lack of theory of mind.
Christopher is a clear example of a person whose dissociation between language
and intelligence is very clear. I mention Christopher in this separate section because he
is not a prototypical case of autism but represents a ‘borderline’ case with elements
characteristic of both Williams syndrome and autism. The former is characterized by
spared linguistic abilities in the face of severe mental retardation, and he clearly has the
will to communicate (as do Williams patients), which is not characteristic of autism.
The latter is however a disorder with much interpersonal variance, and as he lacks the
theory of mind and has a single ‘super skill’, he certainly has to be considered autistic to
a certain degree, as these two traits are among the hallmarks of autism.


4.3 Impaired Language and Spared Cognition
In the next subsections I discuss the kinds of deficits that selectively damage language
while leaving general cognition (relatively) intact. First, I give a survey of the different
types of aphasia, which are all caused by brain damage and then I discuss the
developmental deficit called Specific Language Impairment, or SLI.

4.3.1 Aphasia
The term aphasia actually covers a number of disorders that have as common ground of
intact non-verbal cognition and impaired language. Aphasia is “the loss of or
impairment of language abilities following brain damage” (Pinker 1994: 473). In other
words, it is an acquired disorder of linguistic processing. Typically, six subtypes are
distinguished (cf. Damasio 1992: 536, Bishop 1988:204): Broca’s -, Wernicke’s -,
global -, conduction -, and transcortical aphasia, and anomia. I shall briefly discuss the
different types in the same order in the following subsections.




K. R. C. 2001 33
4.3.1.1 Broca’s Aphasia
Damage to Broca’s area, i.e. Brodmann’s areas 44 and 45 (also called the left inferior
frontal gyrus), see Figure 5 below, and its vicinity
2
results in Broca’s aphasia, which is
a drastic loss of speech fluency making speech effortful and telegraphic. The hallmark
of Broca’s aphasia is agrammatism, an impairment of the subject’s grammar which
manifests itself as an inability to organize words into grammatical sentences and an
improper use or non-use of grammatical words and morphemes, such as conjunctions,
prepositions, auxiliary verbs and tense-inflection. Furthermore, their ability to assemble
phonemes into words is also defective, a feature shared with Wernicke’s aphasics (cf.
below).


Figure 5: Brodmann’s areas. The left arrow points to area 44, Broca’s area. The
right arrow points to the posterior (rightmost) part of area 22 – this posterior part
is Wernicke’s area. The symbols indicate different cortical cell structures, which
roughly coincide with functional areas.
From Sobotta & Becher (1975: 6, Fig 7)


2
The vicinity refers to three areas: the operculum, which is the lower part of the motor strip, just to the
right of Broca’s area); The insula, which is a group of convolutions at the base of the Sylvian fissure to
the below to the right of Broca’s area. Finally, it refers to the subjacent white matter, (beneath the cortex
towards the inner parts of the brain).

K. R. C. 2001 34
Comprehension is also impaired. Broca’s aphasics have difficulties with the
interpretation of semantically reversible passives, which are sentences like “John kissed
Jill”, where both “John” and “Jill” can be “kissers” AGENTS (cf. chapter 2) as well as
“be kissed” (THEME). An example of a non-reversible is “John kicked the ball”, because
only “John” can be a “kicker”. Broca’s aphasics have little or no difficulties with non-
reversible passives where only one noun phrase can be the AGENT, such as (20).
Example (21) is an active semantically reversible sentence. It does not cause any
problems for the aphasic. The passive counterpart in (22), however, does.

(20) The car was driven by the woman.
(21) The man touched the woman.
(22) The woman was touched by the man.

The mechanism underlying this selective impairment has been the focus of research
done by Yosef Grodzinsky (e.g. 1986, 1990, 1995a+b, 2000) and it is the topic of
chapter 6 below so I will not go into further detail here.
Another aspect of this disorder is the aphasic’s inability to repeat sentences they
hear. They are even unable to repeat the types of sentences that they fully understand,
much to their own surprise and dismay. Broca’s aphasics are well aware of their
impairment and are often depressed (Damasio 1992).

4.3.1.2 Wernicke’s Aphasia
Wernicke’s aphasia is caused by damage to the posterior region of the left auditory-
association cortex (Brodmann’s Area 22, see Figure 5 above) and its adjacent areas.
This type of aphasia is characterized by the fluency of the speech of the subjects and is
therefore sometimes called fluent aphasia. Speech is laborious and produced with
normal (or even above normal) speed and intonation. In this sense it can be said to be
the complement of Broca’s aphasia, which is effortful and slow. The utterances are
fluent and more or less grammatical, but they are characterized by their lack of sense
and frequent neologisms and word substitutions.
Wernicke’s aphasics suffer from verbal or semantic paraphasia – they have
great difficulty selecting the appropriate words that accurately match the intended
meaning. For example they might say “chair” instead of “table”, “knee” for “elbow”, or
“dog” for “queen”. They also suffer from phonemic paraphasia, which cause them to
K. R. C. 2001 35
make phonemic substitutions or distortions such as “tubber” for “butter” or “leasing”
for “ceiling”. Furthermore, they use many (unintelligible) neologisms such as “robbli”
for “queen” (cf. Harley 1995: 280; Pinker 1994: 310-311). As their utterances tend to be
grammatical with ‘distorted’ words, and it has often been assumed that Wernicke’s
aphasia is a semantic deficit.
Comprehension in Wernicke’s aphasics is also impaired, and sometimes their
inability to understand what others are saying makes them anxious and agitated and
perhaps even paranoid (Damasio 1992: 534). However, Wernicke’s aphasics are less apt
to become depressed and frustrated than Broca’s aphasics are. They seem to be less
aware of their impairment, which perhaps relates to general intelligence. How can they
not be aware of their impairment? Indeed, studies have shown that there are some
differences in intelligence in aphasics: Broca’s and conduction aphasics (see section
4.3.1.4 below) and anomics (see section 4.3.1.6 below) are not intellectually impaired,
whereas Wernicke’s and global aphasics (see below) have been shown to have below
normal intelligence (cf. Kertesz & McCabe 1975).

4.3.1.3 Global Aphasia
The combination of Broca’s and Wernicke’s aphasia (i.e. damage to both Broca’s area
and Wernicke’s area) is called Global aphasia and is the almost complete loss of
language, both comprehension and speech. On the production side global aphasics’
deliberate speech is limited to a few words and sentences. These structures may be used
repeatedly and inappropriately in a vague attempt to communicate. Their non-deliberate
(automatic) speech is preserved. They are able to appropriately use an inventory of
expletives such as “god damn it” with correct inflection and articulation. Their auditory
comprehension is limited to a small set of nouns, verbs, and idioms, and they seem to
have no understanding of grammatical words or complex sentences.

4.3.1.4 Conduction Aphasia
This kind of aphasia shares three features with Broca’s and Wernicke’s aphasia:
phonemic paraphasia (word substitution), naming problems, and the lack of capacity for
verbatim repetition. Conduction aphasia is distinguished by the relative preservation of
auditory comprehension (unlike Wernicke’s aphasia) and speech production (unlike
Broca’s aphasia). It is caused by damage to one of two loci: (1) The supramarginal
gyrus (Brodmann’s area 40, located in the posterior region of the temporal lobe, just
K. R. C. 2001 36
above Wernicke’s area, see Figure 5. See also Figure 4). (2) The left primary auditory
area (Brodmann’s areas 41 and 42 just left of Wernicke’s area, cf. Figure 5), the insula
and the underlying white matter (see footnote 2, page 33), cf. Damasio (1992: 534).

4.3.1.5 Transcortical Aphasia
So far the types of aphasia discussed all involved an impairment of the capacity for
repetition. Transcortical aphasia is distinguished by the (relative) preservation of
verbatim repetition (cf. Harley 1995: 275). It is divided into two variants: a sensory and
a motor variant. Transcortical sensory aphasia is fluent and involves impaired
comprehension. It is caused by damage to temporal or parietal cortex in the vicinity of
Wernicke’s area (i.e. parts of Brodmann’s areas 21, 22, 37, 39, and 40). Transcortical
motor aphasia is somewhat the mirror image of the sensory variant. Speech is non-
fluent and comprehension is largely intact. It is caused by damage to the left frontal
cortex above and in front of (and sometimes involving) Broca’s area (i.e. parts of
Brodmann’s areas 9, 44, 45, and 46) (cf. Damasio 1992: 535).

4.3.1.6 Anomia
Anomia is a naming disorder, which can be found in isolation or accompanying
Wernicke’s and Broca’s aphasia. People with anomia are impaired in their ability to
name objects or pictures of objects, as well as in their ability to recognize spoken or
written names of things (cf. Harley 1995: 272). As Pinker (1994: 311) states, the name
of the deficit speaks for itself. Literally, anomia means “no-name-ia”.
Pure anomics do not suffer from the symptoms of the other kinds of aphasia.
They are fluent patients with normal comprehension and no severe substitutions of
words or inflections. They have problems with naming and finding the right words (cf.
Bates et al. 1991: 144). Anomia can be described as a chronical state of “it’s just on the
tip of my tongue”.
Neurologically, anomia is caused by trauma to an area to the right of Wernicke’s
area called the angular gyrus (cf. Grodzinsky 2000: 20), see Figure 4 in section 3.3
(page 23), also known as Brodmann’s area 39, see Figure 5 above.

All types of aphasia have distinct symptoms and are all associated with distinct
areas in the perisylvian region. This is summarized in Figure 6 on page 41 below.

K. R. C. 2001 37
4.3.2 Specific Language Impairment
Specific language impairment (SLI) is characterized by severe impairment in the
development of language comprehension and production without any mental
retardation, motor, hearing, social, or emotional disorders that could account for the
impairment. The child is otherwise normal, only language is impaired. A most
interesting aspect of this disorder is that it is not caused by brain damage:

Few children with specific language impairment have any history of brain
disease or any hard neurological signs, and techniques such as CT scan or
electroencephalography (EEG) reveal abnormalities only in a minority of
children with particularly severe problems. Overall the evidence for an acquired
brain lesion as the cause of specific language impairment is slim. (Bishop &
Mogford 1988: 258)

However, SLI tends to run in families, which points to a genetic explanation for the
impairment. Of course, the mere fact that a form of behavior runs in families does not
prove a genetic foundation. Consider for instance recipes, lullabies, stories, etc. which
are transmitted in families, but friends, neighbors, colleagues and such may also “be
contaminated” or “inherit” the use of them. SLI can reasonably be said to have a genetic
cause as it runs in the family much in the same way as psoriasis, which affects only
some (not all) descendants of a common ancestor and does not afflict close age-mates,
friends or other family members. For example, in one single large family of 30
members, the KE family (Gopnik & Crago 1991), half of the members were language
impaired, which is quite many as SLI affects only about 7% of children in general (van
der Lely et al 1998: 1253).
SLI appears to be a rather heterogeneous disorder. Subjects may be with or
without articulatory, phonological, or comprehension impairments (cf. van der Lely &
Stollwerck 1996), and according to Bishop & Mogford (1988: 259) it is largely defined
by exclusion – i.e. if the impairment is not caused by any of the other syndromes it is
labeled SLI. Future research may answer the question whether SLI is a cover term for
several distinct disorders or a single deficit. In fact, several homogeneous subgroups
have been identified, cf. van der Lely & Stollwerck (1996: 486). Other findings point to
K. R. C. 2001 38
a single disorder with a common underlying deficit with several manifestations (as I
will argue below that Broca’s aphasia is).
However, SLI subjects are mostly characterized by varying degrees of
grammatical and morphological impairment in comprehension and/or expression, i.e. a
grammar deficit. Van der Lely et al. 1998 have investigated a case of pure grammatical
SLI. The subject named AZ has a non-verbal I.Q. of 119, i.e. with intelligence above
average. His grammatical skills (morpho-syntax), on the other hand, are severely
impaired. AZ has been tested on a battery of tests in three domains: grammatical
abilities, non-grammatical language abilities, and non-verbal cognitive abilities.
Analysis of AZ’s speech has revealed that he uses only short sentences with frequent
(70-80%) omission of inflection (‘-s’ for person and tense) and omission of phrases
(“the dog was poking [his head] in [-to the jar]”). He uses very few embedded sentences
(2 of 26) and only simple phrases (nothing like “the small black dog”), and he has
severe problems in producing wh-questions. In fact, 83% of the wh-sentences contain
errors, for example omission or ungrammatical use of auxiliaries, and he has severe
problems with morphology and inflection. Furthermore, he is impaired in his ability to
assign reference to pronouns and reflexives, when syntactic grammatical knowledge is
crucial for the judgement. On the other hand, when non-grammatical knowledge is
sufficient for the assignment AZ’s performance is normal (96% correct). For example,
in “Grandpa says Grandma is pinching him” the pronoun “him” can only refer to
“grandpa” because it refers to a male and therefore only semantic knowledge is
necessary for the assignment. On the other hand, in a sentence like “John says that Jack
is pinching him/himself” the pronoun “him” can only refer to “John” and the reflexive
“himself” only to “Jack”. Here grammatical knowledge is crucially required for the right
interpretation as both “John” and “Jack” are males and hence match the semantics of
both “him” and “himself”. Here, AZ’s performance did not differ significantly from
chance, cf. Van der Lely et al. (1998: 1255).
AZ has achieved an overall non-verbal I.Q. ranging from 119 to 131, clearly
above average. He shows no deficits in any non-grammatical domain. His impairment is
restricted to his grammar.
This resembles the diagnosis for Broca’s aphasia, which is also associated with
grammatical impairment and normal intelligence – even though Broca’s aphasia is most
clearly characterized by reduced speech fluency. However, it is important to distinguish
between acquired deficits like aphasia, and developmental deficits like SLI, autism and
K. R. C. 2001 39
Down’s syndrome. SLI is not caused by cerebral trauma. According to Bishop &
Mogford (1988: 259), “it is possible that specific language impairment in children
reflects an underlying immaturity of neurological development of those brain areas
concerned with language development”.

4.4 Discussion
We have seen that autistic subjects suffer from a developmental deficit that, in most
cases, renders them mentally retarded with an I.Q. below 60. They are socially
withdrawn and have no theory of mind. They lack the ability and intent to communicate
and mostly they are mute but those who do acquire language are echolalic and speak
monotonously and mechanically. Autism is associated with reduced activity and/or size
of the frontal lobes, which result in the blocking or mal-acquisition of language. Down’s
syndrome is also a developmental deficit that causes mental retardation resulting in an
I.Q. of about 60. Unlike the autistic subjects, Down’s syndrome subjects are very social
and possess a theory of mind. Their lexicon is reduced compared to normal subjects, but
they are able to use and understand it correctly. Furthermore, Down’s syndrome is clear
evidence for the robustness of language, as the subjects acquire and use language in
spite of massive physical handicaps in the speech production and perception systems.
In comparing autism and Down’s syndrome it is clear that language is
independent of intelligence. Both syndromes show severe mental retardation, but they
differ on a very important point: one blocks out language while the other does not.
Hence, language is not dependent on normal or high intelligence and it is not blocked
by low intelligence. Therefore, language acquisition is not based on general learning,
as these syndromes clearly limit learning abilities in almost every other field and the
subjects are generally poor learners. This is an (indirect) argument in favor of the
innateness hypothesis of language for two reasons: one, some sort of neural
specialization must be a prerequisite for language learning since general learning cannot
apply; and two, there is dissociation between intelligence and linguistic competence.
Another conclusion that might (incorrectly) be drawn from this comparison is
that language acquisition is dependent on the theory of mind. Perhaps, one might argue,
if one does not acknowledge other people as agents and speakers, the required
parameter setting would not take place, and hence leave the child without a language.
However intuitive and reasonable this may seem the hypothesis does not hold.
K. R. C. 2001 40
Christopher is a special case of autism (though he is not a typical case and has not been
clinically diagnosed as autistic, his test results point to autism, cf. Smith & Tsimpli
(1995) and section 4.2.3 above), because he has language as his special ‘super skill’. He
lacks the theory of mind, but speaks several languages. In this way he is a counter
example to the hypothesis. Thus, language is independent of the theory of mind.
Williams syndrome is also a developmental deficit. The subjects have an I.Q. of
around 50, and they are very social and loquacious. They have an unusual vocabulary,
clearly above their mental age. Their language is intact, both in terms of comprehension
and production. Their posterior lobes are reduced, while the frontal lobes are spared (i.e.
of normal size) – quite the opposite pattern of autism. This pattern leads Deacon (1997)
to propose a symbolic processor in the frontal lobes that controls language. But again,
the case of the autistic Christopher provides counter evidence. He has reduced activity
in the frontal lobes, but he is clearly not deprived of language.
The last of the developmental deficits I discuss is SLI, which is characterized by
normal or even high intelligence and a selectively impaired language. Grammatical
impairment seems to be a defining characteristic of SLI, though the deficit is rather non-
homogenous and has several manifestations. The study by Van der Lely et al. 1998 of
pure grammatical SLI offers further support for the above mentioned dissociation
between language and intelligence: The subject AZ had an I.Q. above normal but a
severe grammar deficit. This also points to a modular internal structure of language, as
the deficit was confined to morpho-syntactic aspects of language. SLI has no neuro-
pathological basis, as it is not caused by deformity of the brain. It appears to be caused
by an immaturity of the development of the language areas.
We turn now to the different kinds of aphasia, language impairments caused by
brain damage. Broca’s aphasia, caused by damage to Broca’s area and its vicinity, is
characterized by agrammatism and a drastic loss of speech fluency, which result in
effortful and telegraphic speech. Comprehension and repetition is also impaired. In
Wernicke’s aphasia, caused by damage to Wernicke’s area and its vicinity, speech is
effortless and fluent and it is produced with normal or above normal speed, but is
nonsensical and full of semantic and phonemic substitutions and neologisms. Their
comprehension is also impaired. The combination of damage to both Broca’s and
Wernicke’s area results in global aphasia, which has features from both associated
types of aphasia: both production and comprehension are severely limited. Conduction
aphasia, caused by damage to either supramarginal gyrus or to the left primary auditory
K. R. C. 2001 41
area, is associated with phonemic paraphasia (substitutions), naming problems, an
inability of verbatim repetition, and interestingly a sparing of production and
comprehension. The two variants of transcortical aphasia are characterized by the
ability to repeat spoken sentences in contrast to the other types of aphasia. Finally,
anomia, caused by damage to the angular gyrus, is a naming disorder. Anomics
understand nouns and names correctly but have problems finding the right ones for
expression. I have summarized the loci of the different types of aphasia in Figure 6
below:


Figure 6: Cerebral regions and their associated types of aphasia.
Based on Sobotta & Becher (1975: 4, Fig. 3)

It is now possible to establish a typology of language and general intelligence, or rather
mental retardation, which clearly shows the double dissociation between the two. It also
offers some support for the modularity hypothesis. This typology is summarized in
Table 4 below. The categories ‘mentally normal’ and ‘mentally retarded’ are not
K. R. C. 2001 42
intended to be interpreted too literally, as for example Wernicke’s aphasia is also
associated with reduced I.Q. as a consequence of the trauma.

MENTALLY NORMAL MENTALLY RETARDED

NORMAL / GOOD
LANGUAGE
nerologically
normal
Normal subjects e.g. insanity,
psychopathy
nerologically
abnormal
e.g. blindness due to
brain damage
Down’s syndrome
Williams syndrome
IMPAIRED / POOR
LANGUAGE
nerologically
normal
S.L.I. ?
nerologically
abnormal
Aphasia Autism
Table 4: A typology of language impairment and mental retardation. Some areas
are shaded, because they refer to cases not covered in this study.
3



3
For a similar classification see Vikner’s web-site [http://www.hum.au.dk/engelsk/engsv/medfoedt.htm].
K. R. C. 2001 43
5 Cerebral Area and Function
As is clear from the survey of the different types of aphasia, the classic clinical
description of language is based on the processes of comprehension and production. For
this reason this kind of clinical description and definition is insufficient in that it seems
to presuppose that language is a unitary skill (underlying the processes of
comprehension and production). However, none of the above mentioned language
impairments involve only a complete and undifferentiated loss of only linguistic
comprehension or only production. For similar reasons a description in terms of syntax
versus semantics is also insufficient, as for example aspects of syntax are impaired in
e.g. Broca’s aphasia, not all of it. Broca’s aphasia is not a loss of syntax and likewise
Wernicke’s aphasia is not a loss of semantics. A description of the function of a cerebral
area in terms of language processes, such as production or comprehension, or in terms
of traditional linguistic distinctions, such as syntax and semantics, is misconceived.
What, then, are the functions of Broca’s and Wernicke’s areas? Let us consider
some proposals. According to Damasio (1992: 534), Broca’s area (and its vicinity, cf.
footnote 2, page 33) is part of the network concerned with the relational aspects of
language – that is, grammatical structure (cf. the characteristic agrammatism of Broca’s
aphasia), morphemes and verbs (Broca’s aphasics have difficulties especially with
verbs, cf. Bates et al 1991). In other words, Broca’s area is concerned with syntax and
morphology. On the other hand, Wernicke’s area is

[…] a processor of speech that allows sounds to be mapped as words and to be
used subsequently to evoke conceptual meanings [i.e. the coupling of
phonological representations with semantic concepts. K. R. C.]. […] Wernicke’s
area is no longer seen as a center for word selection. Rather, it appears that
once a word is selected for possible use in an utterance, Wernicke’s area is part
of the network required to implement its constituent speech sounds, in the form
of an internal auditory representation or of vocalization. (Damasio 1992: 534)

At a first sight, Pinker (1994) seems to imply a classic localizationist approach:
grammar in Broca’s area and lexicon in Wernicke’s area. But, as mentioned in section
3.3, Pinker labels the left perisylvian region the language organ, and hence he cannot at
the same time lump all of grammar into one distinct small area like Broca’s area and the
K. R. C. 2001 44
entire lexicon into Wernicke’s area. If he did, then what would be the function of the
rest of the language area (the perisylvian region)? He clearly does not make this
mistake:

So, is Broca’s area the grammar organ? Not really. Damage to Broca’s area
alone usually does not produce long-lasting severe aphasia. […] And, most
surprisingly of all, some kinds of grammatical abilities seem to survive damage
to Broca’s area. When asked to distinguish grammatical from ungrammatical
sentences, some Broca’s aphasics can detect even subtle violations of the rules
of syntax […]. (Pinker 1994: 309) Still, aphasics do not detect all
ungrammaticalities, nor do all aphasics detect them, so the role of Broca’s area
in language is maddeningly unclear (ibid. p.310).

Wernicke’s area he assigns a role in looking up words in the lexicon and sending them
on to other areas of the grammar. Thus, the lexicon itself can be distributed over a much
larger area; Wernicke’s area is just the “librarian”, to put it crudely.
This model, I think, is quite compatible with Damasio’s (1992) model described
above. Damasio suspects that Wernicke’s area has something to do with assigning a
chosen word its proper phonological form but does not mention lexical semantics,
which then must be handled by something else. In Pinker’s view, a word’s lexical entry
includes both semantic content and phonological form. Both authors seemingly agree
that this phonological form is implemented by the grammar in some sort of an internal
representation of the utterance (which, at least for Pinker, would correspond to PF in
Figure 1 in chapter 2). This is at least partly done in Broca’s area, which they both state
is part of the network responsible for ‘relational aspects’ of language, such as sequential
order, sentence structure and morphology. Damasio (1992) is typical of clinical papers
on these matters in that no particular linguistic theory is mentioned and no linguistic
examples are included in the text. For this reason it is not possible to determine exactly
what Damasio has in mind.
Deacon (1997), based on work by Roman Jakobson (1956), has a different
approach. To him, there is no language-specific cerebral region. Rather, Wernicke’s and
Broca’s areas are responsible for the computation of paradigmatic and syntagmatic
relations respectively and are as such bottlenecks in larger computational chains,
K. R. C. 2001 45
linguistic as well as non-linguistic. He explains the two different kinds of relationships
as follows:

In the most general sense, all words of the same part of speech are paradigmatic
of each other to some degree since they can substitute for one another. […]
Syntagmatic operations are reflected in the complementary relationships
between words from different parts of speech (e.g., nouns, verbs, adjectives,
adverbs, or articles) and the way these different classes of words alternate in
sequence in a sentence. (Deacon 1997: 305-306)

To Deacon, the frontal lobes house a symbolic processor. The frontal lobes are
very large in proportion to the rest of the brain in humans (cf. section 3.1), and Deacon
correlates this fact to the species-specificity of language – or symbolic communication
and thinking. Coupled with the fact that trauma to the perisylvian region causes
language deficits, this leads him to postulate that the frontal lobes house the central
symbolic processor and that Broca’s and Wernicke’s areas are computational centers for
information from the frontal lobes. Deacon’s proposal fits quite well with the finding
that there is a severe reduction in blood flow in the frontal lobes of autistic subjects with
severe communicative impairments and with the spared frontal lobes of William’s
syndrome patients (cf. sections 4.1 and 4.2.2 respectively). But the mere facts, that
normal subjects with normal frontal lobes have normal language, that Williams patients
have normal frontal lobes and (near) normal language, and that autistic patients have
abnormal frontal lobes and abnormal language do not place language in the frontal
lobes. Again, it should be kept in mind that there is a difference between acquired
impairment and developmental deficits. The facts just stated cannot account for the
aphasic effects of brain damage to the language area. Furthermore, a lack of drive to
communicate (mutism, not autism), which Deacon correlates with the diminished frontal
lobes in autism, can be caused by damage to the internal cerebral surface of the left
hemisphere in the area including the anterior cingulate gyrus and supplementary motor
area, cf. Damasio (1992:537).
As Broca’s area is part of the region affected in autism one would predict that
the capacity for syntagmatic computation per se should be severely impaired as well. I
find this at least dubious. Cases of idiot savants (Fay 1988, Frith 1989) clearly cast a
shadow of doubt on this. How can one explain autistic geniuses of e.g. art, music, or
K. R. C. 2001 46
mathematics? These abilities clearly involve syntagmatic computations, as they cannot
be described as mere skills of substitution, or in other words, as paradigmatic
operations. It is rather superfluous to state that music and mathematical equations are
sequentially ordered and dependent on syntagmatic computation. Furthermore, and
clearly contrary to the facts, Christopher (who I argued is autistic in section 4.2.3
above), should not be able to learn language, and if so (as he actually has) he should
only be able to compile lists of words in paradigmatic relation to each other (due to the
sparing of Wernicke’s area). This prediction is not borne out. Christopher’s linguistic
“super skill” is clearly not limited to recitation of huge lists of remembered words, since
his abilities involve grammatical skills as well, cf. Smith and Tsimpli (1995).
Deacon states that “Syntactic operations and grammatical judgements can
involve many different syntagmatic and paradigmatic processes, and these can differ
from language to language” (1997: 306). As support for his hypothesis he cites findings
by Bates et al. (1991)
4
who show that there is cross-linguistic variation in the correlation
between lesion site and aphasia type. The important point is that a lesion in Broca’s area
has different effects on e.g. English and Italian. English (and German) uses fixed word
and phrase position to signal such grammatical relationships as possession,
subordination, question vs. statement, active vs. passive. The same relationships are
signal by inflectional affixes in highly inflected languages like Italian (and Latin).
Hence, the claim is, damage to Broca’s area, which supposedly is responsible for
syntagmatic relations, will have great effect on English, but a much less severe effect on
Italian, which in turn would be more effected than English (regarding syntax) by a
damage to Wernicke’s area. Deacon interprets this as evidence for his claim that there is
no language-specific module:

So if there is a grammar module, then the parts of this module map in very
different ways to different grammatical operations, depending on the relative
importance of positional or inflectional tricks for cuing [sic] grammatical
decisions in different languages. This sort of module is a will-o’-the-wisp.
(Deacon 1997: 307)


4
Bates et al. (1991) is actually a summary of findings by a number of researchers published as articles in
a special issue of Brain and Language, issue 41.
K. R. C. 2001 47
But this variation is only strange if one assumes that language competence is divided
into only two modules: grammar and lexicon. As already mentioned, lumping together
all aspects of grammar into one is a misconception. Grammatical abilities are not lost in
an either/or manner, such that a lesion in Broca’s area would lead to a complete loss of
all grammatical competence. No one, at least in linguistic circles (as far as I know),
believes in this kind of gross localizationist approach anymore. In fact, one of the
cornerstones of generative grammar is that language is modular, internally (cf. section
2.1 above) as well as externally (cf. section 4.4).
Actually, I don’t think that there is much in favor of Deacon’s hypothesis that
Broca’s area and Wernicke’s area are but computational centers that also compute
linguistic signals from the frontal lobes. The mere distinction between syntagmatic and
paradigmatic relations within a semiotic system has little explanatory or descriptive
value. His claim that the frontal lobes house the central processor that is also
responsible for language is also rather dubious, as language is clearly located in the
language zone of the left hemisphere (the frontal lobes may, however, still be crucially
involved in symbolic cognition as such). Furthermore, he claims that language is
acquired so easily and fast because humans have a special kind of learning ability to do
so, due to the co-evolution of language and brain. Deacon claims that general symbolic
cognition, most clearly reflected in the theory of mind, is a prerequisite for language, i.e.
general intelligence. I have already argued that this dependency is false (cf. chapter 4).
Damasio, Pinker, and Deacon are all too unspecific in their proposals on the
functions of the areas inside the language zone. Nevertheless, they share a common
feature: they all place a responsibility for syntagmatic, relational, or sequential
computation of grammatical structure on Broca’s area (and vicinity). In the following
chapter I shall return to this aspect with a refined theory of grammar.
What is lacking in the three proposals is a grounding in a specific linguistic
theory, which is capable of describing the linguistic competence of normal language
users and which accounts for the observed breakdown patterns – a theory that is
breakdown compatible (Grodzinsky 1990: 111). That is, a framework that provides the
proper descriptive tools to give an account beyond the inadequate dichotomies, such as
syntax vs. lexicon or syntagmatic vs. paradigmatic. Though Pinker adheres to the
Principles and Parameters framework (see chapter 2) he does not go sufficiently into
detail in his 1994 account. Damasio is a clinician and it is unclear to which linguistic
framework he would adhere. Deacon’s proposal, as I have shown, has little to offer but
K. R. C. 2001 48
the old distinction between syntagmatic and paradigmatic relations. The fact that
language deficits vary cross-linguistically is, I think, the only linguistic phenomenon he
does account for and that is not covered by the others. However, as I shall show below,
this phenomenon is accountable for by a generative framework, which also provides a
description of normal speakers’ competence in terms of a specific grammatical
framework.

K. R. C. 2001 49
6 A Syntactic Approach to Broca’s Aphasia

The distinction between syntax and lexicon, and the conception of the two being
localized in two distinct areas of the left hemisphere, was argued in the preceding
chapters to be insufficient and wrong on empirical grounds, both cross-linguistically (as
shown in e.g. Bates et al 1991) as well as pathologically. Language functions are
distributed over the language zone. It appears that the anterior part is responsible for the
syntagmatic aspects of linguistic computation, which clearly includes syntax. Damage
to this area, however, does not lead to the loss of all of syntax. As Grodzinsky (2000: 4)
states:

“On testing, Broca’s aphasics showed near-normal abilities in comprehension
and grammaticality judgement on many syntactic constructions, and thus did not
appear to have “asyntactic comprehension”. There was a disruption, but it was
restricted to certain aspects of syntax. It was becoming clear, then, that a
distinction between different levels of linguistic analysis would not suffice, and
that distinctions within syntax were needed to account for the comprehension
deficit, just as they were for speech production. (Emphasis added.)

(The meaning of the loose terms “many syntactic constructions” and “certain aspects of
syntax” will become clear below.)
Let me once again point out the important distinction between developmental and
acquired impairment. The former is the result of a brain that is different from a normal
brain, while the latter is the result of damage to a hitherto normal brain. As Broca’s
aphasia is an acquired impairment, it is obvious to consider Broca aphasics’ linguistic
competence as a normal grammar with some damaged part or parts – not as a
completely different grammar. Therefore, the theory that accounts for the linguistic
competence in normal people has to able to account for the competence in aphasics as
well. The theory must meet the Criterion of Breakdown-Compatibility:




K. R. C. 2001 50
(23) The Criterion of Breakdown-Compatibility:
Every pattern of impairment and sparing of linguistic ability must be accounted
for in a natural, non-ad hoc fashion. (Grodzinsky 1990: 111; see also
Grodzinsky 1986: 137)

The accounts in the preceding chapters have primarily been clinical accounts. The
scientific field concerned with the linguistic abilities of the human species is, of course,
linguistics. Therefore, the framework of linguistics should also be applied to language
impairments such as aphasia, in order to properly describe what impact trauma to
different regions of the brain have on peoples linguistic competence, and in turn to aid
in establishing proper diagnostics. Granted that this is true, it is natural to approach the
domain of aphasia from a syntactic point of view.
In the next sections agrammatic comprehension and production will be viewed
from a syntactic point of view, and I shall argue that the deficits to both of these
domains are due to selective impairments to aspects of grammar. First, in section 6.1 I
present an account of the comprehension deficit. Briefly, the hypothesis is that the
comprehension problems associated with Broca’s aphasia are due to a loss or deletion of
traces in the syntactic representation. Hence, the hypothesis is called the Trace Deletion
Hypothesis. This deletion of traces is what causes the subsequent misinterpretation of
thematic roles. Then, in section 6.2 I present the Tree-pruning Hypothesis, which
accounts for the production impairment in Broca’s aphasia. According to the
hypothesis, the production errors are caused be a disruption to the syntactic tree, in
which one or more nodes (maximal projections, cf. chapter 2) are missing from the
representation.

6.1 Comprehension: The Trace Deletion Hypothesis
As Broca’s aphasics have some linguistic competence, their entire grammar cannot be
lost. In fact, as already mentioned, their understanding is not entirely impaired, and their
speech production also present intact features of grammar, such as word order. Based on
research both by the author himself as well as others in a diversity of languages
including Japanese, Chinese, Hebrew, Dutch, Russian, Italian and English, Grodzinsky
(2000; see also 1990) argues that in fact most of the Broca aphasics’ grammar is intact.
The following domains of grammar are supposedly left unimpaired in processes of
K. R. C. 2001 51
comprehension in Broca’s aphasia (cf. Grodzinsky 1995: 31, 2000: 4) (I shall deal with
production below):

i. In comprehension, Broca’s aphasics can construct basic syntactic trees for simple
sentences not containing intra-sentential dependencies (such as trace-antecedent
relations), for example active sentences. They are also able to detect violations of
phrase structure rules (cf. Grodzinsky 1995a), such as ii-iv.

ii. They seem to have no impairment to the part of the lexicon that interfaces with
sentence grammar. They are able to detect violations of subcategorization, e.g. the
verb "to eat" subcategorizes for a NP such as "a cake"; and argument structure, i.e.
how many and what kinds of arguments (AGENT, THEME etc.).

iii. The module called the Theta-theory is also intact. Broca’s aphasics know the theta-
roles of predicates and are able to assign them directly to positions, which is
evident from comprehension tasks on simple structures such as active sentences.

iv. They are able to compute (interpret) some intra-sentential dependencies, such as
grammatical case (for example ACCUSATIVE or DATIVE) which is typically
assigned to canonical positions, such as the object position. They are also able to
handle binding relations (cf. Grodzinsky et al. 1993), which are constraints on
anaphoric relations between pronouns and reflexives and their antecedents
(Chomsky 1981, Haegeman 1994).
However, certain aspects of pronominal reference are impaired, but they have to
do with discourse (pragmatic skills) and not with intra-sentential binding relations
(cf. Balogh & Grodzinsky 2000, Grodzinsky & Reinhart 1993).

In contrast to these spared abilities, Broca’s aphasics suffer severe difficulties with
constructions involving syntactic movement, such as passives, where a noun phrase is
moved from object position to subject position leaving a trace:

(24) [The boy]
1
was pushed t
1
by the girl.

K. R. C. 2001 52
In (24) the NP "the boy" is the THEME and "the girl" is the AGENT, the one doing the
pushing. Canonically, the role of THEME is assigned by the verb to the position to its
right – the typical object position of actives. The passive morphology absorbs (cf.
Haegeman 1994) the argument to the left (in this case "the girl") which is optionally
adjoined in a PP. In order to abide the Case Filter, the NP the boy is moved to [spec, IP]
to be assigned NOMINATIVE, subject case (see chapter 2 for more information).
Numerous experiments (see for example Grodzinsky 1990 and 1995a) on
semantically reversible sentences have supported the claim that transformation (i.e.
movement) implies comprehension problems. The experiments typically involve the
sentence-to-picture-matching test, in which the patient hears a sentence and then he/she
has to point out a picture, which depicts the semantic content of the sentence. For
example, the patient hears the sentence "the boy pushed the girl" and has to choose
between a picture where a boy is pushing a girl and a picture where a girl is pushing a
boy. The experiment is supposed to probe the patient’s abilities on theta-assignment to
structural positions. The data categorizes sentential structures into three sets: Those
where the patients perform above chance, at chance level, and those where the patients
perform below chance (for statistical details, see e.g. Grodzinsky 1995b: 491 or Balogh
& Grodzinsky 2000: 17). I have summarized these findings in the following table:

Construction Example Theta-Structure Performance
Active the girl pushed the boy AGENT-THEME above chance
Subject-subject relative
Object-subject relative
the girl, who pushed the boy, is angry
show me the girl, who pushed the boy
AGENT-THEME
AGENT-THEME
above chance
above chance
Subject cleft it is the girl who pushed the boy AGENT-THEME above chance
Adjectival passive the boy was interested in the girl EXP.- THEME above chance
Psychological verb the boy admires the girl EXP.- THEME above chance
Passive [the boy]
1
was pushed t
1
by the girl THEME-AGENT Chance
Subject-object relative
Object-object relative
[the boy]
1
who the girl pushed t
1
was tall
show me [the boy]
1
who the girl pushed t
1
THEME-AGENT
THEME-AGENT
Chance
Chance
Object cleft it is [the boy]
1
who the girl pushed t
1
THEME-AGENT Chance
Psychological passive [the girl]
1
is admired t
1
by the boy THEME-EXP. below chance
Table 5. For expository reasons the total syntactic complexity of the examples is
left out, and will be described in due course below. Roughly, chance refers to 30%-
70% correct performance. Note also that the θ θθ θ-role EXP. is short for EXPERIENCER.

K. R. C. 2001 53
The term psychological verb refers to non-agentive verbs, such as feel, love, hate, and
admire, which denote experience rather that action. You can deliberately kick someone
(you are the AGENT), so kick is an agentive verb. You can not deliberately love
someone, but you can experience (feel) love for someone (you are the EXPERIENCER).
From the table above it is clear that the distinguishing factor dividing the syntactic
constructions is movement. In the types on which agrammatic aphasics perform above
chance (near normal) there are no traces in the examples given in the table. The other
types all involve movement of the ‘underlying’ object to subject position and hence the
representations have traces. This finding gives rise to the Trace-Deletion Hypothesis (I
refer the reader to Grodzinsky 1986, 1990, 1995a+b and 2000 for elaboration):

(25) The Trace-Deletion Hypothesis:
In agrammatism, traces in θ-positions (structural positions to which θ-roles are
assigned) are deleted from the syntactic representations.

So, in the representations of a sentence in Broca’s aphasics, the moved NP will not have
a theta-role assigned to them because the trace that would normally transmit the role is
deleted. Perhaps a note on the term ‘deleted’ is in order. In comprehension the hearer
constructs a mental grammatical representation of the utterance, which can be
represented orthographically as a syntactic tree. The agrammatic tree either lacks the
traces of the moved NPs or the links between the trace and its antecedent is broken. In
either case, the theta-transmission fails. I will not go into further discussion on the two
possibilities (no traces or broken chains - note that chains are instances of the relational
aspects of language discussed in the previous chapter). The point is that the theta-
transmission fails and this can be accounted for by saying the traces of the moved NPs
are deleted. Keep in mind that traces are constructs (cf. chapter 2), and deletion does not
refer to some sort of mental eraser removing ts from trees in the patient’s head.
The Trace-Deletion Hypothesis alone, however, is not enough to account for all the
constructions in the table, for three reasons. First, the hypothesis does not predict
chance performance in itself in tests on e.g. passives, only that they are problematic and
that comprehension is impaired. Second, it does not predict that the performance on
psychological passives would be below chance. Third, under the VP-internal Subject
Hypothesis (see chapter 2) all constructions involve movement of the subject, and hence
K. R. C. 2001 54
the prediction is that all constructions would be problematic, which is not borne out.
What remedies these problems is the application of a heuristic non-linguistic strategy
that assigns NPs theta-roles according to their linear position. This strategy is based on
knowledge of the world, such as frequency of occurrence / canonical order of roles,
which means that the initial (argument) NP will be assigned the role of AGENT:

(26) The Default Strategy:
Phrasal constituents with no theta-role are assigned one by default, by linear
considerations (NP
1
=AGENT).

This strategy applies after the grammar has built the syntactic representation with the
missing theta-role(s). In normal as well as in aphasic speakers, the information the
grammar uses in the construction of representations is not available to higher cognition,
such as world knowledge – this is called the informational encapsulation of the
grammar (cf. Fodor 1983, Grodzinsky 1990). Therefore, in Broca’s aphasia the Default
Strategy is applied blindly resulting in counterintuitive assigning:

(27) John was killed _ by Bill
THEME AGENT (normal assignment)
AGENT AGENT (agrammatic assignment)

In this example Bill is not moved and is therefore assigned the θ-role AGENT by the
grammar. The NP John, on the other hand, is moved and its trace is deleted (illustrated
with _) and therefore theta-transmission is impossible leaving the NP without a θ-role.
Then the Default Strategy kicks in and assigns it the role of AGENT. The representation
now has two AGENTS, which are in competition. This leads to chance performance as the
agrammatic subject is forced to guess which is correct.
The second problem was with psychological passives. Let’s consider an
example:

(28) [The girl] is admired _ by the boy
THEME EXPERIENCER (normal assignment)
AGENT EXPERIENCER (agrammatic assignment)
K. R. C. 2001 55
Here the result of the application of the strategy no longer leads to competition between
two identical roles, as was the case in (27) above. The two θ-roles in the agrammatic
representation are not identical like they were in (27), hence only one interpretation is
made in (28), even though it may seem a bit odd to consider admiration as something
one deliberately does (I deliberately admire her / I admire her on purpose). The
agrammatic interprets such sentences directly complementary of normal language users:
In (28) the girl is doing the admiring, not the boy. To account for this we have to
consider how the different theta-roles are related. Jackendoff (1972) ranks the θ-roles in
a hierarchy according to thematic salience:

(29) The Thematic Hierarchy:
AGENT > EXPERIENCER > LOCATION, SOURCE, GOAL > THEME

Thus, the NP assigned the highest ranking is ‘the doer’, in (28) the girl is doing the
admiring because of the role as AGENT, not the NP the boy even though it is assigned the
role of EXPERIENCER by the grammar as this role is lower ranked. Another such
hierarchy is the Animacy Hierarchy. A NP is higher in animacy (more animate) the
more to the left it is on the following hierarchy (cf. Comrie 1989: 128):

(30) 1
ST
/ 2
ND
PERS. PRONOUNS > OTHER HUMAN NPS > ANIMAL NPS > INANIMATE NPS

For an account of experiments involving non-agentive animate and inanimate NPs and
therefore θ-roles other than AGENT, THEME, and EXPERIENCER, see Grodzinsky 1995b. In
my account I restrict myself to NPs that are animate.
This consistent reversal of the roles in sentences with psychological verbs is the
reason why an account merely based on canonical order fails. According to such an
account, when the aphasic encounters a structure that does not have the AGENT first, he
or she will guess, and this would result in chance performance. However, the data
shows performance below chance and a reversal of the roles, which is not predicted by
such an account. This supports the transformational account outlined here (see also
Grodzinsky 1989).
The third problem was of a theoretical nature. According to the VP-internal
subject hypothesis the subject moves out of the verb phrase (in which it is base
K. R. C. 2001 56
generated) to receive NOMINATIVE. Without the Default Strategy the subject would
not be assigned a θ-role. Through the application of the Default Strategy the subject is
assigned AGENT in non-problematic actives:

(31) [John [(t) killed Bill]]
AGENT THEME (normal assignment: θ-transmission from t)
AGENT THEME (agrammatic assignment: Default Strategy)

Under normal circumstances the subject is assigned its θ-role by transmission from its
trace. In agrammatic comprehension, however, this is not possible as the trace is deleted
(or the trace-antecedent chain is broken). The strategy assigns it AGENT, and thus
compensates for the lost grammatical θ-transmission and leading to the right
interpretation. The same θ-role is assigned, but for a different reason.
In short, the account of agrammatic comprehension consists of two interacting
parts: the Trace Deletion Hypothesis and the Default Strategy. According to the Trace
Deletion Hypothesis, traces in θ-positions are deleted from the syntactic representation,
which leads to separation of θ-assignor and θ-assignee, which in turn results in NPs
without θ-roles. The Default Strategy states that constituents without grammatically
assigned θ-roles are assigned one by linear considerations, such that the first NP (in a θ-
position) is assigned AGENT. The interaction of the two parts lead to chance
performance on movement derived clauses (except psychological passive, which is
associated with below chance performance).

6.2 Production: The Tree-Pruning Hypothesis
The production side of the deficit does not mirror the comprehension side. After all, it
was long believed that Broca’s aphasia was an impairment of speech production with a
sparing of comprehension. Mostly, it is characterized as improper use or omission of
grammatical words and morphemes (cf. section 4.3.1 above). However, a number of
researchers have pointed out that this is not altogether true, as there is cross-linguistic
variation within this deficit (e.g. Bates et. al 1991 and Grodzinsky 1990, 2000). Bates et
al. (1991) relate this variation to differences in competition between two emergent
connectionist neural networks, somewhat reminiscent of the Jakobsonian syntagmatic
vs. paradigmatic computation of information that Deacon (1997) supports. I have
K. R. C. 2001 57
already explained why this approach, in my opinion, is wrong. Grodzinsky’s approach
to this is, I think, more elegant. It obeys the important Criterion of Breakdown-
Compatibility.
The differences between languages are constrained by the typological space
defined by the possible parametric variation of the universal grammar. Therefore, it
should not be surprising to find that at least some aspects of agrammatism reflect this
parametric variation. Grodzinsky’s studies (1984, 1990, and 2000) have shown a
correlation between zero-morphology and omission of inflection in agrammatics (i.e.
people suffering from agrammatism) in a variety of languages. In both nominal and
verbal inflected elements morphemic omission is observed if the bare stem is a real and
well-formed word in the current language, i.e. if the language has +zero-morphology.
Otherwise, if the omission of inflection results in nonwords, i.e. if the language has –
zero-morphology, substitution errors occur instead:

(32) +Zero-morphology omission (e.g. English and Japanese)
-Zero-morphology substitution (e.g. Russian, Italian, Hebrew)

Agrammatics do not produce non-words (cf. Grodzinsky 1990: 52). There is thus no
violation of constraints on lexical well-formedness, as omission of inflection only
occurs if the resulting word is a real word. Otherwise the inflectional morpheme is
substituted, giving rise to grammatical aberration. This supports the proposed
connection between the typological parameter (+/- zero-morphology) and observed
agrammatic behavior.
Here are some examples of omission and substitution respectively (taken from
Grodzinsky 1990: 52):

(33) Omission:
English: Uh, oh, I guess six month… my mother pass away
(Omission of number and tense inflection)

Japanese: inorimasu (correct: inorimasushita)
I-pray I-prayed
(Omission of tense inflection)

K. R. C. 2001 58
Substitution:
Russian: grustnaja malchik
sad (FEM.) boy (MASC.)
(Substitution of agr. inflection: grustn-iy > grustn-aja)

Italian: Cappuceto rossa andava
Little Ridinghood (MASC.) Red (FEM.) went
(Substitution of agreement inflection: ross-o > ross-a)

Hebrew: tiylu anaxnu ba’ali ve’ani
‘take-a-walk’.PAST.3
rd
PERS.PL. we husband.my and-I
(“They took a walk, we my husband and I”)
(Substitution of agreement inflection: tiyl-nu > tiyl-u)

The deficit is, however, more restricted than to the correlation between
omission/substitution and +/-zero-morphology. The omission/substitution is determined
by the elements’ structural position. According to Grodzinsky and Friedmann (1997,
2000), there is dissociation between tense and agreement inflection. This runs contrary
to the common belief that Broca’s aphasics have equal problems with all functional
categories. Evidence from Hebrew, Arabic, Italian, French, Dutch, German, Icelandic,
Swedish, Finnish, Japanese has shown that agrammatics may be impaired in tense,
while not having problems with agreement (based on both new tests as well as
retrospective literature reviews). On the other hand, the opposite, impaired agreement
and spared tense, is never found, cf. Friedmann & Grodzinsky (1997, 2000).
Impaired agreement implies impaired tense, not vice versa. These findings offer
support for the Split-Inflection Hypothesis (Pollock 1989), according to which the INFL
node IP (inflection phrase) in (34) is split up into an agreement phrase AgrP and a tense
phrase TnsP in (35) (NegP is short for negation phrase):

(34) [
CP
spec C
0
[
IP
spec I
0
[
NegP
spec N
0
[
VP
spec V
0
compl ]]]]
(35) [
CP
spec C
0
[
TnsP
spec Tns
0
[
NegP
spec N
0
[
AgrP
spec Agr
0
[
VP
spec V
0
compl ]]]]]

K. R. C. 2001 59
The ordering or ranking of the functional categories is not uncontroversial, something
which I will come back to in the next chapter. This dissociation between tense and
agreement suggests a possible deficit that involves the tense node but not the agreement
node. It also allows for varying degrees of severity. If the AgrP node is impaired, the
syntactic tree is disrupted from AgrP and up, and errors of agreement will occur
accompanied by errors of tense. If the tense node is impaired, TnsP and CP are
disrupted, and hence only tense errors will occur (no agreement errors). Based on these
findings Grodzinsky (2000:16) states the following hypothesis:

(36) The Tree-Pruning Hypothesis:
Agrammatic aphasic patients produce trees that are intact up to the Tense node
and “pruned”
5
from this node up

Or in more formal terms (from Friedmann & Grodzinsky 2000:25 [pdf version]):

a) T is impaired in agrammatic production
b) An impaired node cannot project any higher.

The hypothesis is further supported by the finding that the verb stays in situ (in its base
position) instead of moving up the tree
6
:

In Germanic languages such as Dutch, German, and Icelandic, patients frequently
use the infinitive instead of the inflected verb. Crucially, a non-finite form always
appears in a sentence-final position, indicating that the verb has not moved up the

5
The term “pruned” refers to the way trees are trimmed. A gardener may trim a tree to control its growth
by cutting off branches, and thus prevent he tree from growing higher. This is called pruning. In the same
way the syntactic tree in agrammatic production is cut or rather destroyed from the tense node up, which
means that the TnsP and CP ‘branches’ are missing.

6
This phenomenon is also known from studies of language acquisition (cf. Wexler 1994). Children also
produce sentences with uninflected verbs in situ; such verbs are also called root infinitives. Wexler (1994)
argues that children go through an “optional infinitive” stage, in which they utter both sentences with
finite verbs and sentences with non-finite verbs (infinitives). The interesting point is the position of the
verbs: e.g. English and French speaking children put the finite verb before the negation and the non-finite
(infinitive) after the negation; Dutch and German children place the finite verb in V2 position, while the
non-finite forms are sentence final. The same results are predicted in agrammatism: if the verb cannot
move due to impaired nodes under the Tree-pruning Hypothesis it stays in situ in the infinitive form in
V
0
; if the verb moves it will be finite (inflected for tense).

K. R. C. 2001 60
tree to C, where tensed verbs in matrix clauses of V2 languages should move.
(Friedmann & Grodzinsky 2000: 8 [pdf version])

Furthermore, Grodzinsky & Friedmann (1997, 2000) show that agrammatics make no
Wh-questions and no embedded sentences involving a complementizer in CP. Both
structures depend on the presence of CP, since wh-elements (such as English who and
which) normally move to [spec, CP], and complementizers like that are generated in C
0
.
Correlated with the disruption of tense and the sparing of agreement, the absence of
embeddings supports the Tree-Pruning Hypothesis. As mentioned, their model also
allows for varying degrees of impairment. In this (rather simplified) tree structure the
double slashes (//) indicate possible points of impairment:

(37) [
CP
// [
TnsP
// [
NegP
[
AgrP
// [
VP …

Agrammatics may be impaired from agreement up (deficit to agreement, tense, and CP),
from the tense node up (deficit to tense and CP), or just to the topmost node CP (see
Grodzinsky & Friedmann 1997).
In the next chapter I want to apply the Trace-Deletion Hypothesis, the Default
Strategy and the Tree-Pruning Hypothesis to Danish, which is a verb-second (V2)
language. Because of its V2 status certain predictions can be made about Danish
agrammatic speech production. Furthermore, I present an alternative order of the
functional categories TnsP and AgrP. I also show how differences in degrees of severity
are predicted to be reflected in the speech production of Danish Broca aphasics.

K. R. C. 2001 61
7 Is Grodzinsky’s thesis applicable to Danish?
Linguistic competence is universal to humans and only humans (the hypotheses of
Universality and Species-Specificity) and based on common cerebral predisposition (the
Innateness Hypothesis). Therefore a theory assumed to be able to account for language
breakdown has to be applicable to all languages of the world. For this reason it is
important to test whether Grodzinsky’s hypotheses are applicable to Danish (as well as
to every other language).

7.1 The Danish Language
Danish is the official language of Denmark. It belongs to the East Scandinavian
language group of the Germanic family. It has a fairly strict word order, SVO (Subject-
Verb-Object), in both main and embedded clauses. Moreover, Danish is a verb-second
(V2) language, which means that the finite verb is always the second constituent of the
main clause. Yes/no questions constructions constitute the only exception, where the
verb is sentence initial (see (42) in Table 6 below). Furthermore, if another constituent
is fronted (topicalized), it triggers inversion of the subject and the finite verb, which
places the finite verb in second position (V2) and the subject in third ((40) in Table 6
below). In embedded clauses fronting and inversion are prohibited. In brief, the linear
word order in Danish is as follows
7
:



7
I leave out structures that contain both a direct and an indirect object. Either the indirect object is
realized as a preposition phrase or as part of a double object construction (cf. Larson 1988, Vikner 1989),
in which the VP has the following structure:
For example:
[
CP
han
1
har
2
[
IP
t
1
t
2
[
VP
t
1
t
2
[
VP
t
1
købt
3
[
VP
hende t
3
[
NP
en gave]]]]]]
he has bought her a present

K. R. C. 2001 62
Conj Sub/Topic Verb-finite (Sub) Adv Neg Verb-nonfinite Obj
(38)
Jeg
I (Sub)
spiser
eat-PRES
endda
even
ikke
not
bananer
bananas
(39)
Men
But
jeg
I (Sub)
har
have-PRES
tit
often
smagt
taste-PT.PTCP
ferskener
peaches
(40)
Pærer
Pears (Obj+Top)
spiser
eat-PRES
jeg
I
også
also

(41)
Og
And
Hvad
What
køber
buy-PRES
du?
you?
(42)
køber
Buy-PRES
du
you
også
also
æbler?
apples?
Table 6: Topological analysis of Danish main clauses.


Conj Sub Adv Neg Verb-finite Verb-nonfinite Obj
(43)
du ved
you know
at
that
jeg
I
endda
even
ikke
not
spiser
eat-PRES
bananer
bananas
(44)
Piger
Girls
der
who
tit
often
har
have-PRES
smagt
taste-PT.PTCP
ferskener
peaches
(45)
… hvis
if
du
you
aldrig
never
prøver
try-PRES
Table 7: Topological analysis of Danish embedded clauses.


The structural analysis of main clauses and embedded clauses are given below in (46)
and (47) respectively, with the split-IP order proposed by Pollock’s (1989) (the symbol
‘–‘ indicates an unfilled specifier position. The word “ikke” is the Danish negation
equivalent to the English “not”.) For expository reasons I have left out the structure of
the complement (compl) of the verb.

(46) Main Clause (preliminary version):
[
CP
Conj [
CP
Sub
1
Verb
2
[
TnsP
t
1
t
2
[
NegP
[
AdvP
Adv]
[
NegP
(ikke) t
2
[
AgrP
- t
2
[
VP
t
1
t
2
Compl ]]]]]]]

For example:
[
CP
og [
CP
jeg
1
spiser
2
[
TnsP
t
1
t
2
[
NegP
[
AdvP
faktisk]
[
NegP
ikke t
2
[
AgrP
- t
2
[
VP
t
1
t
2
bananer]]]]]]]
“And I actually do not eat bananas” (lit.: and I eat actually not bananas)
K. R. C. 2001 63

(47) Embedded Clause (preliminary version):
[
CP
Conj [
TnsP
Sub
1
t
2
[
NegP
[
AdvP
Adv] [
NegP
(ikke) t
2
[
AgrP
- t
2
[
VP
t
1
Verb
2
Compl ]]]]]]

For example:
[
CP
at [
TnsP
jeg
1
t
2
[
NegP
[
AdvP
faktisk] [
NegP
ikke t
2
[
AgrP
- t
2
[
VP
t
1
spiser
2
bananer]]]]]]
“That I actually do not eat bananas.” (lit.: that I actually not eat bananas)

Note the CP-recursion in the main clause (46): In V2 languages the verb always moves
from V
0
through Agr
0
, Neg
0
and Tns
0
to C
0
in main clauses. SO, in the presence of a
coordinating conjunction (it must be coordinating as it occurs in a main clause) another
CP is projected, which is headed by the conjunction (the complementizer), cf. Vikner
(1995), because the first CP is ‘occupied’ by the verb. Note also that in embedded
clauses the tense and agreement inflection is lowered to the verb in V
0
instead of being
attached to the verb by movement out of V
0
.
Perhaps a description of the derivation of a standard main clause is in order.
Consider the following example:

(48) [
CP
hunden
1
bed
2
[
TnsP
t
1
t
2
[
NegP
[
AdvP
faktisk]
[
NegP
ikke t
2
[
AgrP
- t
2
[
VP
t
1
t
2
katten]]]]]]
“The dog actually didn’t bite the cat.”
(Literally: the dog bit actually not the cat.)

First of all, following the Projection Principle, the verb “bide” projects a verb phrase
VP and subcategorizes for two NPs, one for each argument “hunden” og “katten” (their
structure is left out for expository reasons). The former is placed left of V
0
receiving the
AGENT role and the latter to the right receiving the role of THEME, thus satisfying the
K. R. C. 2001 64
Theta Criterion. The sentential adverbial phrase (AdvP) is adjoined to the NegP
8
. Due
to the Extended Projection Principle the clause structure is projected by the verb, and
the Case Filter forces the subject to move from its base position in [spec, VP] to [spec,
TnsP] where it is assigned NOMINATIVE case. The object is assigned ACCUSATIVE
by the verb in V
0
. The verb is moved up to Agr
0
and via Neg
0
on to Tns
0
to receive
tense inflection, and finally assuming V2 position in C
0
. The subject moves from [spec,
TnsP] to [spec, CP], which is vacant as no other topic is fronted.
As mentioned in the section 6.2, the above ordering of the functional categories
TnsP, NegP, and AgrP is not uncontroversial. Belletti (1990), Haegeman (1994), and
Vikner (1995) have a different order of tense, negation and agreement, i.e. [
AgrP
[
TnsP

[
NegP
]]] rather than [
TnsP
[
Neg
[
AgrP
]]] as Pollock 1989 proposes. This re-ordering
(Belletti 1990) has serious consequences:

i. Under the Breakdown Hypothesis, in order to maintain neurological support,
Pollock’s order somehow has to be possible, as it is applicable to agrammatic
production in the languages studied by Grodzinsky (e.g. 2000).
ii. However, it is widely accepted that Belletti’s order is correct. The argument is
based on the ordering of the inflection on the verb. This order should reflect the
movement of the verb. This is known as the Mirror Principle. If the verb moves
to Tns
0
before Agr
0
then the tense inflection should be closer to the verb stem
than the agreement inflection. This is the case in for example German: “Du
glaub-t-est” ‘you think-PAST-2
nd
PERS.SING’ (cf. Haegeman 1994 for discussion).
This may imply that the ranking of the functional categories is not universal, but
open to parametric variation.

8
In order not to complicate matters unnecessarily, I only include sentence adverbials and leave out
structures with VP-adverbials, which are adjoined to VP:

… [
NegP
[
AdvP
Adv ] [
NegP
spec Neg
0
[
VP
[
AdvP
Adv ] [
VP
spec V
0

/ \
Sentence adverbial VP-adverbial:

An example of a VP-adverb is “langsomt” (slowly), which can only appears after the negation:

jeg spiser ikke langsomt
“I don’t eat slowly” (literally: I eat not slowly)

*jeg spiser langsomt ikke


K. R. C. 2001 65
iii. It follows from the possibility of parametric variation in the ordering of the
functional categories that the Tree-Pruning Hypothesis will give rise to language
specific variation in agrammatism. Hence, impairment to the Agreement phrase
alone with a sparing of Tense is possible, but as mentioned in section 6.2 above
this has never been found (Friedman & Grodzinsky 1997, 2000). This may be an
artifact of the languages studied. Perhaps there are agrammatic speakers of some
language that is only poorly studied. On the other hand, if no such language can
be found then we will be forced to reevaluate the internal structure of IP.

Movement of the verb V
0
is cyclic and is done step-wise and cannot skip a
position (this is known as the Head Movement Constraint). For this reason, it is
necessary to assume that negation in Danish (“ikke”) is in [spec, Neg
0
] like English
“not” and French “pas”. Otherwise it would block verb movement. The English
negation “n’t” (and the French “ne”) is a Neg
0
element, i.e. a head. It is attached to
(incorporated into) the verb, resulting in e.g. “didn’t” (thanks to Sten Vikner for
pointing this out). This gives us the following structure, main clause structure version 2,
in which the order of TnsP and AgrP is reversed in the line of Belleti:

(49) Structure of Danish Main Clauses , final version (replaces (46)):
[
CP
Conj [
CP
Sub
1
Verb
2
[
AgrP
t
1
t
2
[
TnsP
- t
2
[
NegP
[
AdvP
Adv] [
NegP
(ikke) t
2
[
VP
t
1
t
2
Compl ]]]]]]]

For example: [
CP
hunden
1
bed
2
[
AgrP
t
1
t
2
[
TnsP
t
1
t
2
[
NegP
[
AdvP
faktisk]
[
NegP
ikke t
2
[
VP
t
1
t
2
katten]]]]]]
“The dog actually didn’t bite the cat” (Literally: the dog bit actually not the cat)

(50) Structure of Danish Embedded Clauses, final version (replaces (47)):
[
CP
Conj [
AgrP
Sub
1
t
2
[
TnsP
- t
2
[
NegP
[
AdvP
Adv]
[
NegP
(ikke) t
2
[
VP
t
1
Verb
2
Compl ]]]]]]
For example: [
CP
at [
AgrP
hunden
1
t
2
[
TnsP
t
1
t
2
[
NegP
[
AdvP
faktisk]
[
NegP
ikke t
2
[
VP
t
1
bed
2
katten]]]]]]
“That the dog actually didn’t bite the cat” (Literally: that the dog actually not bit the cat)


K. R. C. 2001 66
The figures below show the structure of main clauses and embedded clauses
respectively. In Figure 7 shows the movement of the subject and the verb in main
clauses from their base positions inside VP to their surface positions in CP, which result
in the obligatory V2 word order. Figure 8 shows the movement in embedded clauses,
where the subject moves from [spec, VP] to [spec, AgrP], and the verbal inflection
moves downwards to the verb, which remains in its base position in V
0
. For expository
reasons, I have left out the structure of the complement (compl) and the adverbial
phrase AdvP:


Figure 7: Structure of Danish main clauses.


K. R. C. 2001 67

Figure 8: Structure of Danish embedded clauses.

Note that NOMINATIVE is assigned to [spec, Agr
0
] (in Danish by C
0
and in English by
Agr
0
for reasons not discussed here, cf. Vikner 1995; see also section 2.1.8).

7.2 Predictions About Agrammatism in Danish
Based on the interaction between the Trace-Deletion Hypothesis and the Default
Strategy and on the Tree-Pruning Hypothesis, certain predictions can be made about the
performance of agrammatic aphasics in Danish (as well as in any other language).
Depending on the specifics of the Danish typological parameter-settings, the aphasic’s
comprehension of certain types of sentences will be impaired, while other types will
turn out normal. In production, the structure of the sentence will be the crucial factor in
predicting impairment, as the hypothesis states that certain parts of the grammatical
representation (syntactic tree) are disrupted. Furthermore, predictions can be made with
regards to omission or substitution of inflection based on the presence or absence of
zero-morphology.



K. R. C. 2001 68
7.2.1 Comprehension
First of all, let us take a look at the structure of the relevant clauses in Danish in the
following two tables:

Type Structure
Simple
active
[
CP
pigen
1
skubber
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
v
2
[
NP
drengen]]]]]]]]

pigen skubber drengen
the girl pushes the boy
AGENT THEME

[
CP
pigen
1
peger
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
v
2
[
PP
på [
NP
drengen]]]]]

pigen peger på drengen
the girl points at the boy
AGENT THEME
Subject
-subject
relative
pigen
1
[
CP
som
1
e [
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
skubber
2
[
NP
drengen]]]]]] er høj

pigen
1
[ som
1
skubber drengen ] er høj
the girl who pushes the boy is tall
AGENT THEME
Object-
subject
relative
vis mig pigen
1
[
CP
som
1
e [
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
skubber
2
[
NP
drengen]]]]]]

vis mig pigen
1
[ som
1
skubber drengen ]
show me the girl who pushes the boy
AGENT THEME
Subject
cleft
det er pigen
1
[
CP
som
1
e [
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
skubber
2
[
NP
drengen]]]]]]

det er pigen
1
[ som
1
skubber drengen ]
it is the girl who pushes the boy
AGENT THEME

(Continued on next page)
K. R. C. 2001 69

Type Structure (continued from above)
Lexical
passive
[
CP
drengen
1
synes
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
v
2
[
PP
om [
NP
pigen]]]]]]]

drengen synes om pigen
the boy thinks about the girl (“the boy likes the girl”)
EXPERIENCER THEME
Adject.
Passive
[
CP
drengen
1
er
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
v
2
[
AdvP
t
1
interesseret [
PP
i [
NP
pigen]]]]]]]]

drengen er interesseret i pigen
the boy is interested in the girl
EXPERIENCER THEME
Psych.-
verb
[
CP
drengen
1
beundrer
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
t
1
v
2
[
NP
pigen]]]]]]

drengen beundrer pigen
the boy admires the girl
EXPERIENCER THEME
Table 8: Danish clause structure 1.


Type Structure
Verbal
passive
[
CP
drengen
1
skubbes
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
[
VP
t
1
v
2
t
1
] [
PP
af [
NP
pigen]]]]]]

drengen
1
skubbe-s t
1
af pigen
the boy push-PRES.PASS by the girl (“the boy is pushed by the girl”)
THEME AGENT
“Blive”
passive
[
CP
drengen
1
bliver
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
[
VP
t
1
v
2
[
VP
t
1
skubbet t
1
]] [
PP
af [
NP
pigen]]]]]]]

drengen
1
bliver skubbet t
1
af pigen
the boy is pushed by the girl
THEME AGENT

[
CP
drengen
1
bliver
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
[
VP
t
1
v
2
[
VP
t
1
peget [
PP
på t
1
]]] [
PP
af [
NP
pigen]]]]]]]

drengen
1
bliver peget på t
1
af pigen
the boy is pointed at by the girl
THEME AGENT

(Continued on next page)
K. R. C. 2001 70

Type Structure (continued from above)
Subject
-object
relative
drengen
1
[
CP
som
1
e [
AgrP
pigen
2
v
3
[
TnsP
- v
3
[
NegP
- v
3
[
VP
t
2
skubber
3
t
1
]]]]] er høj

drengen
1
[ som
1
pigen skubber t
1
] er høj
the boy who the girl pushes is tall
THEME AGENT
Object-
object
relative
vis mig drengen
1
[
CP
som
1
e [
AgrP
pigen
2
v
3
[
TnsP
- v
3
[
NegP
- v
3
[
VP
t
2
skubber
3
t
1
]]]]]]]]

vis mig drengen
1
[ som
1
pigen skubber t
1
]
show me the boy who the girl pushes
THEME AGENT
Object
cleft
det er drengen
1
[
CP
som
1
e [
AgrP
pigen
2
v
3
[
TnsP
- v
3
[
NegP
- v
3
[
VP
t
2
skubber t
1
]]]]]]]]

det er drengen
1
[ som
1
pigen skubber t
1
]
it is the boy who the girl pushes
THEME AGENT
Psych.
Passive
[
CP
pigen
1
beundres
2
[
AgrP
t
1
v
2
[
TnsP
- v
2
[
NegP
- v
2
[
VP
[
VP
t
1
v
2
t
1
] [
PP
af [
NP
drengen]]]]]]]

pigen
1
beundre-s t
1
af drengen
the girl admire-PRES.PASS by the boy
THEME EXPERIENCER
Table 9: Danish clause structure 2.

The structures in the two tables above are all semantically reversible transitives. The
distribution of performance rates between above chance, at chance, and below chance is
expected to be the same as in English. Table 8 contains the structures on which
agrammatics are expected to perform above chance, as all the clauses have AGENT-
THEME theta-structure. The subject, which is moved out of [spec, VP] under the VP-
internal Subject Hypothesis, is assigned the AGENT-role by the Default Strategy and no
comprehension problem arises. The object has not undergone movement and is
therefore assigned its role of THEME grammatically.
Table 9 contains the structures expected to be problematic. All the clauses
involve movement of the object. The prediction is that due to the Trace-Deletion
Hypothesis the θ-role can no longer be transferred through the link to the trace and the
result is a wrong assignment by the Default Strategy. The double occurrence of the
AGENT-role leads to chance performance. The exception to this is, as in English, the
K. R. C. 2001 71
psychological passive in which the result of θ-role assignment (in interaction with the
Thematic Hierarchy) gives rise to the reversal of the roles (or rather their salience)
resulting in below chance performance.
These predictions about Danish aphasic comprehension (which in fact is
identical to the pattern observed in English) are summarized in the following table:

Structure Normal
Assignment
Aphasic
Assignment
Predicted
Performance
Simple active
Subject -subject relative
Object- subject relative
Subject cleft
Lexical passive
Adject. Passive


AGENT-THEME



AGENT-THEME

Psychological verb EXPERIENCER-THEME AGENT-THEME


above chance
Verbal passive
“Blive” passive
Subject-object relative
Object-object relative
Object cleft


THEME-AGENT



AGENT-AGENT



chance

Psychological passive THEME-EXPERIENCER AGENT-EXPERIENCER below chance
Table 10: Predicted Danish aphasic performance on comprehension tests.

7.2.2 Production
The Tree-Pruning Hypothesis predicts impairment to the top-most nodes of the
syntactic tree – more precisely, from the tense (TnsP) node up, from the agreement
(AgrP) node up, or to the CP alone depending on the degree of severity of the deficit. In
the illustration below the points of breakdown is indicated with double lines:
K. R. C. 2001 72

Figure 9: Possible points of breakdown in the syntactic representation in Danish
(the numbers refer to examples below).

However, as Danish is a V2 language, the verb always (in normal language users)
moves to C
0
(see Table 8 and Table 9), and as there is no agreement inflection in
Danish, it is not possible to tell whether the agreement node is intact if the tense node is
impaired. For this reason it is not possible to decide on the relative ordering of AgrP and
TnsP. So in practice it will only be possible to place the site of breakdown in TnsP or
CP (or actually perhaps only in CP; I return to this problem shortly). This will reveal
itself in the following way: If the impairment is restricted to CP (51) (cf. Figure 9
above) the verb will only be able to move as far as to Tns
0
, where it is inflected for tense
(perhaps the verb can move to Agr
0
(51) but there would be no way of telling. If the
tense node is disrupted (53) (cf. Figure 9) in the entire TnsP is missing (and so are AgrP
and CP) and the verb cannot move farther than to Neg
0
and thus appears uninflected for
tense – i.e. in the infinitive.

(51) *[
AgrP
– Verb
1
[
TnsP
– t
1
[
NegP
[
AdvP
Adv] [
NegP
- t
1
[
VP
{sub} t
1

(52) *[
TnsP
– Verb
1
[
NegP
[
AdvP
Adv] [
NegP
- t
1
[
VP
{sub} t
1

(53) *[
NegP
[
AdvP
Adv] [
NegP
- Verb
1
[
VP
{sub} t
1

(54) *[
VP
{sub} Verb…

(51)

(52)

(53)
K. R. C. 2001 73
There is of course another logical possibility, which is not shown in Figure 9.
The impairment may theoretically be from the negation phrase (NegP) up, in which case
the verb remains in its base position inside VP, as shown in (54). This would be
distinguished from (53) by the position of the verb in relation to the negation in negative
clauses and to the medial adverb (adjoined to VP) in positive clauses. If the NegP is
intact the verb precedes the adverb or negation, if NegP is impaired the verb follows it.
An impaired NegP would result in non-normal use or non-use of negation, for the
simple reason that there would be no base position for the negation.
There are thus four possibilities, where the first two are phonetically
indistinguishable (for the sake of brevity I have left out the adjoined AdvP), listed here
in order of degree of severity of impairment (from least to most severe):

(55) Normal: Unimpaired CP
[
CP
han
1
skubber
2
[
AgrP
t
1
t
2
[
TnsP
- t
2
[
NegP
ikke t
2
[
VP
t
1
t
2
pigen ]]]]]
he push-PRES not the girl

(56) Impaired CP
*[
AgrP
– skubber
1
[
TnsP
– t
1
[
NegP
ikke t
1
[
VP
{han} t
1
pigen ]]]]
push-PRES not {he} the girl

(57) Impaired AgrP
*[
TnsP
– skubber
1
[
NegP
ikke t
1
[
VP
{han} t
1
pigen ]]]
push-PRES not {he } the girl

(58) Impaired TnsP
*[
NegP
ikke skubbe
1
[
VP
{han} t
1
pigen ]]
not push-INF {he} the girl

(59) Impaired NegP
*[
VP
{han} skubbe hende …
{he} push-INF her

The consequences of disruptions of the syntactic tree not only affects verb-
movement; it also affects other aspects of syntax related to the node in question. The
subject is dependent on C
0
to assign NOMINATIVE case in Danish. Being an
argument, the subject normally undergoes A-movement (cf. Haegeman 1994 and Vikner
1995), which is movement from one argument position to another, i.e. directly from
K. R. C. 2001 74
[spec, VP] to [spec, AgrP] not via the intermediate specifier positions. This movement
is only possible in (51)/(56) where the [spec, AgrP] position is intact. However, even in
this case the subject will not be overt (phonetically realized), because it has not been
assigned NOMINATIVE case. Arguments depend on case in order to pass the Case
Filter, and if CP is missing C
0
will not be present to assign case. This means that
theoretically the subject will be dropped. Therefore, I have put the subjects in (51)-(54)
and (56)-(59) above in “curly” brackets ({}).
According to Friedmann and Grodzinsky (2000) subject pronouns are only
dropped when verbal inflection is substituted, i.e. in languages without zero-
morphology, cf. (32) above. As Danish has +zero-morphology, Danish speaking
aphasics should not drop subject pronouns, if the predictions hold. Lexical subjects as
well as pronouns should not be overt without case, so Friedmann & Grodzinsky (2000:
14-15 [pdf version]) proposes that “the agrammatics may use subjects as topicalized
elements, and assign them a default case […]”, which in Danish is NOMINATIVE.
This means that the subject is moved to the topmost available specifier. The fact that it
is overt (phonetically realized) may be due to an impaired Case Filter (the same would
be the case if the subject was adjoined to the topmost node). Here is an example of a
clause pruned at the tense node:

(60) *[
NegP
SUB
1
VERB
2
[
VP
t
1
t
2
compl…
the boy push (inf.) the girl

Another problem with this proposal is that only [spec, VP], [spec, AgrP], and
[spec, CP] are argument positions. If none of these positions are available the subject
will have nowhere to go, so to speak, without breaking a general rule of movement:
arguments move to and from argument positions only. An admittedly a bit ad hoc
solution to this problem might be that the subject is placed in front of the sentence after
the representation has left the grammar (which is informationally encapsulated, cf.
section 6.1). In other words, the subject is fronted and overt due to knowledge other
than grammatical (such as pragmatics and meta-linguistic knowledge), i.e. in the same
non-linguistic (non-grammatical) way as with the Default Strategy. Perhaps the subject
is ‘named’ or stated first and then the rest of the sentence. The subject and the rest of the
sentence may be uttered in two distinct but related parts instead of in one construction.
Presently, I see no apparent solution to this problem.
K. R. C. 2001 75
Above I briefly mentioned that perhaps it would only be possible to place the
locus of breakdown in CP. The problem with subject movement just discussed is related
to a problem of verb movement. Danish is, as mentioned, a V2 language, which means
that the verb always moves to C
0
in main clauses. Any disruption of the syntactic tree
will affect CP, and therefore all main clauses in Danish will also be affected. If the CP
is missing the verb will not be able to move out of VP unless one assumes that the verb
always moves if possible, which in fact is the strategy I have used so far in this section.
However, in generative grammar it is generally assumed that movement does not take
place unless it is necessary (for example, in the Minimalist Program this is called
movement as last resort). As the C
0
position is not present to motivate movement, the
verb will most like remain uninflected in its base position in V
0
. Danish is thus quite
vulnerable to agrammatism, as in theory any structural degree of severity will have the
same effect on main clauses:

(61) Normal main clause: intact tree
[
CP
han
1
skubber
2
[
AgrP
t
1
t
2
[
TnsP
- t
2
[
NegP
ikke t
2
[
VP
t
1
t
2
pigen ]]]]]
he push-PRES not the girl

(62) Agrammatic main clause: impaired tree
*[han] ([
AgrP
- e [
TnsP
- e) [
NegP
ikke e [
VP
{han} skubbe pigen ]]
he not {he} push-INF the girl

(I have placed the subject of (62) in front of the sentence in square brackets to indicate
the indeterminacy whether it is cognitively or grammatically fronted; either it is a free-
standing constituent or else it occupies the top-most available specifier position (or it is
adjoined).) It is up to empirical research to choose between the two different models of
verb movement outlined here: either the verb always moves if possible or it only moves
when necessary. Studying agrammatism may provide the evidence needed to verify one
and falsify the other.
Another effect of pruned trees is that “Wh-questions and embedded clauses are
nonexistent or completely ill-formed in the speech of the patients” (Grodzinsky 2000:
16). This is due to (again) the missing CP, where the complementizer or wh-element
should be. That means that relatives (with wh-elements or complementizers) and clefts
(see Table 8 above for subject relatives and clefts and Table 9 for object relatives and
clefts) will be either nonexistent or ill-formed. However, some embeddings are
K. R. C. 2001 76
constructed without complementizers or wh-elements, for example the Danish
complementizer “at” (that) is optional, as it sometimes is in English:

(63) han ved [
CP
(at) [
AgrP
hun
1
t
2
[
TnsP
- t
2
[
NegP
ikke t
2
[
VP
t
1
kommer
2
]]]]]
he know-PRES (that) she not come-PRES
“He knows she’s not coming”

Embedded clauses with optional complementizers should not be as vulnerable to
impairment as other sentence types, because the CP is not involved in verb movement
and therefore the verb should theoretically be inflected for tense and agreement
depending on which node is impaired. However, the subject depends on C
0
to assign
NOMINATIVE. So if C
0
is missing so may the subject – unless it may occupy a non-
argument position, such as [spec, TnsP], as it cannot be fronted (the main clause is
already there). A Danish agrammatic person with (only) an impaired CP node would
probably utter (63) as:

(64) *[han][AgrP - e [TnsP - e [VP {han} vide[AgrP hun1 t2 [TnsP - t2 [NegP ikke t2 [VP t1 kommer2
he know-INF she not come-PRES
“He know she’s not coming”

7.2.3 Summary
All of these predictions of the previous section can be summed up in a series of
questions that need to be answered through empirical tests in order to either validate or
falsify the hypotheses of trace deletion and tree-pruning underlying agrammatism in
Broca’s aphasia.

i. In comprehension, do Danish agrammatics perform according to the distribution
outlined in Table 10 above:

No syntactic movement No comprehension problem above chance performance
Syntactic movement comprehension problem chance performance
Psychological passive comprehension problem below chance performance

The following questions deal with production:
K. R. C. 2001 77

ii. Do Danish aphasics produce constructions without movement of the finite verb?
If so, are these instead produced as infinitives?
iii. If the verb has moved, is it inflected for tense?
iv. What is the position of the verb relative to the medial adverb or negation?
v. What is the degree of severity?
vi. Do speakers drop the subject? If not, is it fronted?
vii. Danish has zero-morphology, implying omission of inflection. True or false?


7.3 Empirical Tests
Agrammatism manifests itself differently with regards to comprehension and
production. Therefore, the linguistic tests used to examine agrammatism are also
different: one for comprehension and one for production. First I describe the tests I have
used, and then I discuss the data I have obtained from applying the test to a patient with
agrammatism.
I am very much aware that a single case is not much to base any statistics on,
and that the value and weight of my empirical data would have been significantly
increased had I had several test subjects. For various reasons, my efforts to find more
than one patient for this study were in vain. First of all, the institutions and speech
therapists responsible for the treatment and rehabilitation of aphasics were not open to
outside research by linguists. Second, of the two helpful places one institution had no
patients with Broca’s aphasia and the other currently treated only one agrammatic
aphasic patient – the patient tested here.

7.3.1 Comprehension: the Sentence-to-Picture Matching Test
The comprehension test is a sentence-to-picture matching test. As described in section
4.3.1 above, this test proceeds as follows. First the patient is presented with two pictures
that semantically mirror each other. The patient then hears a semantically reversible
sentence and the task is then to point out the picture that depicts the meaning of the
sentence. Consider the following example. The patient is presented with the pictures
below depicting a boy loving a girl (A) and a girl loving a boy (B):

K. R. C. 2001 78

Figure 10: Picture of “the boy loves the girl” and “the girl loves the boy”.

Next, the patient hears the sentence "The girl loves the boy" and has to point out the
corresponding picture, in this case the correct answer is of course (B). I have devised 15
sets of pictures that correspond to the two interpretations of semantically reversible
predications. I have also constructed 144 semantically reversible sentences that can be
matched with the pictures, such that each of the two pictures of the same set is
represented equally frequent. The pictures and the sentences that have the semantic
meaning corresponding to the pictures can be found in appendix A, left and right
column respectively. I have constructed the set of sentences in such a way that that each
type of construction is represented by ten token sentences, see appendix B. Lexical
passive is an exception with only four tokens, as it is quite rare in Danish - rare in the
sense of distribution. The few lexical passives are actually quite frequently used, such
as:

(65) Drengen væmme-s ved pigen
The boy disgust-PRES.PASS with the girl
“The boy is disgusted at the girl”

Note that the lexical passive does not involve movement of the object (see Table 8)
unlike for example the verbal passive (see Table 9).
Another important thing is that I have included ten additional tokens of simple
actives and ten "blive" passives that have a verb that subcategorizes for a preposition
K. R. C. 2001 79
phrase (cf. Section 2.1.2). In the following example the first preposition "på" is
subcategorized for by the verb and it is therefore inside the VP that is projected by the
verb, while the second preposition "af" heads the PP adjoined to the VP (structural
information not central to this point is left out for expository reasons):

(66) Pigen
1
bliver [
VP
[
VP
t
1
peget på t
1
] [
PP
af drengen]]
the girl is pointed at by the boy

The VP-internal preposition is governed by the verb (I shall not go into further formal
detail of government relations. The terms VP-internal and adjoined to VP will suffice.
For elaboration see Haegeman (1994)). It will thus be possible to see whether the
presence or non-presence of a governed proposition affects comprehension.
The sentences are presented to the patient in a random order, such that the
correct answer, either (A) or (B), of the presented sentences appear in series of no more
than three. For example, the correct answers may come in sequences such as A-A-B-A-
B-B-A-A-A-B, but not for example A-B-A-A-A-A-B-B. The latter has a sequence of
more than three consecutive identical correct answers: 4 As in a row. This is done to
secure that the intact comprehension of certain structures that happens to be presented
consecutively does not affect performance, which may otherwise become biased
towards either (A) or (B) if longer series were allowed. Furthermore, sentences that
relate to the same pair of pictures (such as the pair in Figure 10) do not occur in series
of more than two, also to avoid any bias or other unpredicted influence on the results.

7.3.2 Production: Repetition Test
The production test is a repetition test. The patient hears a sentence, which he or she
then has to repeat. This is done to ensure that all types of structure are present in the
corpus. Moreover, it can be difficult sometimes to determine what was intended when
the utterance is severely malformed. By doing a repetition test, it is ensured that both
intended and produced are known.
I have constructed a set of 100 sentences listed in appendix C. The sentences are
distributed over the 13 structural types (see Table 10 above). Each type is represented
by seven tokens, one for each of the tense/aspects:

K. R. C. 2001 80
(67) Tense/Aspects:

Past double perfect
9
:
Drengen hav-de haf-t skubb-et pigen
The boy have-PAST have-PAST.PTCP push-PAST.PTCP the girl
“The boy had pushed the girl”

Past perfect
Drengen hav-de skubb-et pigen
The boy have-PAST push-PAST.PTCP the girl
“The boy had pushed the girl”

Past
Drengen skubb-ede pigen
The boy push-PAST the girl
“The boy pushed the girl”

Present double perfect
Drengen har-Ø haft skubb-et pigen
The boy have-PRES have-PAST.PTCP push-PAST.PTCP the girl
“The boy has pushed the girl”

Present perfect
Drengen har-Ø skubb-et pigen
The boy have-PRES push-PAST.PTCP the girl
“The boy has pushed the girl”

9
The double perfect constructions are not used by all Danish speakers and some even find them
ungrammatical. However, most speakers would not doubt accept the double perfect in sentences about
phone calls, such as “jeg har haft ringet [men du var ikke hjemme]” (‘I have phoned you [but you weren’t
home]’). I include the construction knowing it may be controversial.

K. R. C. 2001 81

Present
Drengen skubb-er pigen
The boy push-PRES the girl
“The boy pushes the girl”

Future
Drengen vil-Ø skubb-e pigen
The boy will-PRES push-INF the girl
“The boy will push the girl”


In addition I have added a set of simple actives and "blive" passives with the negation
"ikke" in order to be able to test whether the use of negation is intact.
However, eleven of the 105 sentences generated in the set had to be excluded
due to ungrammaticality (see appendix C). Most of the excluded sentences are in the
passive voice and constructed with the double perfect. For example this verbal passive,
which is ungrammatical for two independent reasons: first, the verbal passive is only
possible in the present tense (and in past constructions involving the modality verbs
“skulle” (should) or “ville” (would)); second the passive cannot be constructed on a
participle (i.e. in the perfect aspect):

(68) **Hunden hav-de vist haf-t bid-es af katten
The dog have-PAST probably have-PAST.PTCT bite-PASS by the cat

To reach a total of 100 sentences I have added six simple passives. The set includes
sentences both with and without governed prepositions.
In order to detect movement of the verb all the sentences include sentence
medial adverbs or negation.
Finally, I have included sentences both with and without VP-internal (governed)
prepositions, in order to see whether one or the other is dropped. According to
Grodzinsky (1988) in agrammatics only the VP-internal prepositions are deleted
(phonetically silent), while the adjoined prepositions that head phrases like “[the boy
was hit] by the girl” are not.
K. R. C. 2001 82
The sentences are presented in random order, where the same structural type
never occurs in sequences above two to avoid any bias.
Both tests were performed on two normal subjects with no prior history of brain
damage or neural pathology. One was a 28-year-old male fellow student of linguistics at
the University of Aarhus. The other was a 26-year-old woman studying to become an
architect at the Aarhus School of Architecture. Both subjects performed 100% correct
on both the comprehension test and the repetition test.

7.3.3 Patient TJ
In order to test the hypotheses and predictions of section 7.2 I have tested a patient
diagnosed as showing signs of Broca’s aphasia. I first provide his medical history and
then the results of the test described in the previous sections.

7.3.3.1 Medical history
TJ is a 31-year-old (right-handed) male with a master's degree in art history, who was
hit by a truck while riding his bicycle. According to his medical report
10
, the accident
caused massive trauma to his left hemisphere. He suffered massive hematomas (blood-
filled swellings) to the areas in the perisylvian region, stretching into the frontal-,
parietal-, and temporal lobes. He had infarction (cell death) in most of his left temporal
lobe. In addition he had a small hematoma in the right parietal lobe but, as shown in
section 3.2 above, the right hemisphere makes no grammatical contribution to language
use. I have summarized the zone within which the hematomas in the left hemisphere
have occurred in the following illustration:


10
The medical report is not listed as a reference due to the fact that it is confidential and not publicly
accessible.
K. R. C. 2001 83

Figure 11: Left-hemisphere damage in patient TJ. The medical report is rather
vague with regards to the precise location of the hematomas. The shaded area is a
gross summary of hematoma sites. The arrows point to major hematoma centers.
Based on Sobotta & Becher (1975: 4, Fig. 3)

According to the medical report, the neurologists diagnosed him as suffering from
severe expressive aphasia, while his impressive functions (comprehension) were
relatively spared. I tested TJ eight months after the accident, and at that stage he had
improved significantly, according to his speech therapist.

7.3.3.2 Comprehension Test Results
TJ was tested on the comprehension test described above. Due to fairly successful
rehabilitation most of his linguistic abilities were restored to a near-normal level at the
time of testing. This had the disadvantage that he was no longer a clear example of
Broca’s aphasia and therefore his test performance could not be expected to follow the
predictions completely. On the other hand, it had the advantage that he was easily
understandable. The following table shows a summary of his performance on the
comprehension test, which was applied twice with a week’s interval, each take with a
duration of about half an hour.



K. R. C. 2001 84
Type Correct % Performance
Simple active 20/20, 20/20 100
Subject–subject relative 10/10, 10/10 100
Object-subject relative 10/10, 10/10 100
Subject cleft 10/10, 09/10 95
Lexical passive 03/04, 03/04 75
Adjectival passive 10/10, 18/10 90
Psychological verb 08/10, 10/10 90
Verbal passive 10/10, 10/10 100
“Blive” passive 20/20, 20/20 100



Above chance
Subject-object relative 07/10, 07/10 70 Chance
Object-object relative 10/10, 10/10 100
Object cleft 10/10, 08/10 90
Psychological passive 09/10, 09/10 90

Above chance
Table 11: TJ's performance on the Sentence-to-Picture Matching Test. I define
‘chance performance’ as between 30% and 70% knowing that this is a gross
simplification of the mathematics of statistics. However, the important point is that
the lowest performance (70%) is on a predicted type of construction.

As can be seen in the table, TJ performed above chance on all but one sentence type, i.e.
subject-object relative. The fact that he performs normal on subject-subject relatives
shows that he has no difficulties with center-embedding as such. Neither does it matter
whether the AGENT has the same or different grammatical 'roles' (subject / object) in the
matrix clause and the embedded clause – only when the subject of the matrix clause is
the object of the embedded clause. Also evident from the table is that he does not
perform according to a canonical/non-canonical word order or role order distinction.
There is also a clear reduction in his performance on lexical passives, if one only
considers the percentage. However, there are only four tokens of lexical passive, which
is not much of a base for conclusions. Furthermore, he failed on the same token both
times:

(69) Drengen væmme-s ved pigen
The boy disgust-PRES.PASS at the girl
“The boy is disgusted at the girl”

This seems to imply a problem with the lexical entry of the verb rather than the
construction; but then again this is based on one of only four tokens.
K. R. C. 2001 85
The presence or absence of governed (VP-internal) prepositions did not affect
comprehension.

7.3.3.3 Production Test Results
On the production side, TJ was tested once on the Repetition Test, which took about
twenty minutes. The results are given in the table below (the ‘Changes’ column for
example ‘1 = Simple passive’ should to read as ‘one sentence was produced as a simple
active’):

Type Error Rate Changes Correct
Simple active 5,3% (1/19) 1 = Simple passive 94,7%
Subject–subject relative 14,3% (1/7) 1 = Subject-object relative 85,7%
Object-subject relative 14,3% (1/7) 1 = ? O 85,7&
Subject cleft 0,0% (0/7) 100,0%
Lexical passive 71,4% (5/7) 5 = ? OOOOO 28,6%
Adjectival passive 0,0% (0/5) 100,0%
Psychological verb 14,3% (1/7) 1 = Simple active 85,7%
Verbal passive 25,0% (1/4) 1 = Simple active 75,0%
“Blive” passive 0,0% (0/7) 100,0%
Subject-object relative 85,7% (6/7) 2 = Subject–subject relative
1 = Object-subject relative
1 = Adjectival passive
2 = ? OO
14,3%
Object-object relative 57,1% (4/7) 4 = Object-subject relative
1 = [
NP
N [object relative]]O
42,9%
Object cleft 28,6% (2/7) 2 = Object-subject relative 71,4%
Psychological passive 20,0% (1/5) 1 = Psychological active 80,0%
Table 12: TJ's performance on the Repetition Test.

In the table and henceforth the numbers in black circles refer to a set of sentences in the
test data results in appendix E (they are also marked in the appendix):

(70) O#93 O#14 O#35 O#52 O#43 O#15 O#55 O#71 O#44

K. R. C. 2001 86
This is done in order to keep the results and my treatment of them as transparent as
possible and to be able to use the numbers as short hand references.
TJ has a tendency to avoid the passive morphology “–s”. Of 13 intended
instances of passive morphology
11
, eight (61,5%) are not produced correctly. The
passive morphology was intended in three passive construction types: 7 lexical passives,
2 psychological passives, and 4 verbal passives:

(71) Passive “-s” omission
Type: Error:
Lexical passive: 3 drop main V O O O
2 drop passive morphology O O
Psych. passive: 1 drop passive morphology (#50 in appendix E)
Verbal passive: 2 drop passive morphology (#81 and #85)

For example this verbal passive:

(72) Intended (#85):
drengen skubb-ede-s jo af pigen
the boy push-PAST-PASS after-all
12
by the girl

Produced:
*drengen skubb-ede jo af pigen
the boy push-PAST after-all by the girl

The entire verb was dropped in three instances; the passive morphology was omitted in
five cases, whereas no errors were made on the "blive" passive, which involves an
auxiliary verb and no passive morphology. The omission of the passive ending is
consistent with the prediction from above (page 57) that +zero-morphology leads to

11
It should be noted that when talking about passive morphology, I am only referring to the Danish
passive “-s” on the main verb, NOT to the passive “-et” (participle) inflected on the auxiliary, such as
“pigen blev skubb-et af drengen” (the girl was push-ed by the boy).


12
The Danish “jo” is equivalent to the German “ja” and is hard to translate without any context. Consider
for example this German example: Er ist ja ein Berliner. Depending on the context, it can be translated as
for example “after all, he is a Berliner” or “but he is a Berliner”.
K. R. C. 2001 87
omission and not substitution in agrammatism. There are no examples of morphemic
substitution in the test data, but there are examples of word substitutions (semantic
paraphasia), for example "kigget på" (looked at) instead of "inspireret af" (inspired by)
(see #31 in appendix E).
TJ has a strong tendency to use forms without movement of the (underlying)
object as is evident from his performance on object relatives: Subject-object relative:
14,3% correct, object-object relative: 42,9% correct. Furthermore, he performs only
71,4% correct on object clefts, which is clearly a reduction. Again, a reference to center-
versus right-embedding is insufficient. Subject-subject relatives are center-embedded
and he performs near normal on them, while his performance is only 14,3% correct on
the subject-object relatives, which are also center-embedded. He performs near normal
on object-subject relatives but almost consistently wrong on subject-object relatives,
both of which are right-embedded.
The performance on tense inflected forms is central to the Tree-pruning
Hypothesis, cf. sections 6.2 and 7.2.2 above. The following table shows TJ's
performance on the different tense/aspect constructions:

Tense & AUX Correct
Performance
Produced Verb Reduction
1: PAST PERF+
[AUX AUX V]
0,0% 0X = PAST PERF+
1X = PAST PERF
3X =PAST
1X = PRES PERF+
4X = PRES PERF
1X = ? O

-1 AUX
-2 AUX

-1 AUX
-1 AUX, -main V
2: PAST PERF
[AUX V]
42,9% 6X = PAST PERF
6X = PAST
1X = PRES PERF
1X = ? O

-1 AUX

-main V
3: PAST
[V]
90,9% 20X = PAST
2X = PRES


(Continued on next page)
K. R. C. 2001 88

(Continued from above)
Tense & AUX Correct
Performance
Produced Verb Reduction
4: PRES PERF+
[AUX AUX V]

0,0%
2X = PAST
0X = PRES PERF+
6X = PRES PERF
2X = ? O O
-2 AUX

-1 AUX
-1 AUX, -main V
5: PRES PERF
[AUX V]

42,9%
5X = PAST
6X = PRES PERF
1X = PRES
1X = ? O
1X = ? O
-1 AUX

-1 AUX
-1 AUX
-1 AUX, -main V
6: PRES
[V]

68,8%
3X = PAST
11X = PRES
2X = ? O O


-main V
7: FUTURE (PRES)
[AUX V]

64,3%
5X = PAST
9X = FUTURE
-1 AUX
Table 13: TJ's tense/aspect performance.

TJ performs 0% correct on the double perfect constructions (PERF+) in both past and
present tense. Only one is produced as a double perfect, but in the wrong tense. In
addition, he performs only 42,9% correct on the perfect constructions (PAST.PERF and
PRES.PERF) and 64,3% on the future constructions. All of these constructions involve an
auxiliary verb. Clearly there is a tendency to reduce the number of auxiliary verbs. This
is summarized in the following table:

Verbs Intended Produced %
AUX AUX V 20 1 5,0%
AUX V 52 33 63,53%
V 38 58 152,6%
? 8
TOTAL 100 100
Table 14: Auxiliary verb reduction. Whatever one’s opinion on the double perfect
(see footnote 9 on page 80), the reduction of the number of VPs is still clear from
the comparison of the remaining structures.

K. R. C. 2001 89
TJ has a tendency to use constructions without auxiliary verbs, i.e. to minimize the
structure – recall from section 2 that each verb projects a VP. It is not the case that
tense/aspect per se is omitted, only the auxiliary verbs. All the verbs are inflected for
tense, i.e. past or present, and therefore it is a case of structure reduction not
tense/aspect omission.
Next consider the distribution of inflection for past and present. As mentioned
all verbs are inflected. In some (5/8) of the clauses where TJ omitted the main verb
(OOOOO) he produced the auxiliary verb, which was inflected for tense. In three
sentences (OOO) he omitted the entire embedded clause. In summary:

Tense Inflection Intended Produced %
PAST 46 53 115,2%
PRESENT 54 44 81,5%
? 3
TOTAL 100 100
Table 15: The distribution of inflection for past and present in TJ's performance
data. The question mark indicates that the verb as well as the rest of the clause has
been omitted.

The following changes in tense inflection are made:

(73) 10x past = present
17x present = past
3x present = ? (O O O)

Note that present is changed more frequently than past: of 30 changes 20 are made to
the present tense (66,7%). There is no clear ’default’ tense, but a slight preference for
past. This is evident from Table 13 above, where it is shown that TJ performs 90,9%
correct on the constructions with only one verb and inflected for past tense, while he
performs only 68,8% correct on the same construction inflected for present.
Importantly, he produces no unmoved uninflected verbs (i.e. root infinitives), which
suggests that the tense node TnsP is intact.
TJ's use of negation is normal. There are twelve clauses with the negation "ikke"
(indicated with a P in the third column in appendix E), and TJ performed 100% correct.
This shows that NegP is intact.
K. R. C. 2001 90
All the 100 sentences in the test set has a sentence medial adverb, which is
adjoined to NegP, cf. Figure 7 and Figure 8 in section 7.1. In all but one sentence
(99/100) the position of the adverb in relation the verb is correct. In matrix clauses the
adverbial follows the finite verb and in embedded clauses the adverb precedes the finite
verb. The only example of an incorrect adverb position is O, which should have had the
matrix word order but instead has the word order of an embedded clause. The correct
order of verb and adverb (V
FINITE
-Adv) shows that the verb has moved out of VP (as the
AdvP is adjoined to a node (NegP) higher than VP the base-generated order is Adv-
V
FINITE
, see Figure 7 on page 66). This of course means that there must be a position to
which the verb can move. Therefore TnsP must be intact (recall from section 7.2.2 that
it is not possible to tell whether AgrP is intact). There is no example of a non-moved
verb in the infinitive in the data.
TJ correctly produces embedded clauses. This shows that the CP node is intact,
because it contains the relative pronoun “som” (who) in [spec, CP] in the relative
clauses. This, of course, also shows that both AgrP, TnsP, and NegP are intact because
“an impaired node cannot project any higher”, as stated in the Tree-pruning Hypothesis
(cf. (36) in section 7.2.2 above). This is also supported by the fact that the subject is
always present and in its correct position – [spec, CP] – preceding the finite verb in C
0
.
As mentioned, the fact that he produces embeddings with a wh-element shows
that the CP node is intact. In his main clauses, the verb has undergone movement and is
in V2 position. On the basis of the evidence from TJ, it is not possible to decide on the
problem (from section 7.2.2) of deciding between verb movement as ‘movement when
possible’ or ‘movement when necessary’.
Anyway, as the entire tree structure is intact, TJ has a very mild form of
agrammatism. That is, he has evidently undergone successful rehabilitation.
In section 4.3.1 I mentioned that Broca’s aphasics have an improper use (or non-
use) of prepositions. According to Grodzinsky (1988) agrammatics frequently omit only
one kind of preposition, i.e. the prepositions that are inside the VP, i.e. those that are
governed by the VP, while they produce those that are adjoined to the VP, i.e.
ungoverned by the VP (cf. section 7.3.2 above). This structural difference between the
two types of prepositions is also found in the speech of TJ:
K. R. C. 2001 91

Preposition Intended Omitted Error Rate
Governed 27 4 14,8%
Ungoverned 19 1 5,3%
Table 16: Preposition drop in TJ’s production. In appendix E the sentences with
governed prepositions are marked with a P in the second column on the left. The
omission of a governed or ungoverned preposition is indicated with –G and –U
respectively in the second column on the right.

Few prepositions are dropped but there is, however, a marked difference between
governed and ungoverned prepositions. The governed prepositions are omitted almost
three times as often as the ungoverned. Still, only 14,8% of the governed prepositions
are deleted, which is close to normal production. Further evidence for TJ’s relatively
restored ability to use prepositions correctly, comes from the six cases where he
substitutes a verb that does not subcategorize for a PP with one that does. In each of
these cases he correctly produces the governed preposition. Consider for example this
simplified version of #31 from appendix E (slightly modified for expository reasons):

(74) Target:
pigen var [
VP
inspireret] [
PP
af [
NP
drengen]]
the girl was inspired by the boy

(75) Produced:
pigen har [
VP
kigget [
PP
på [
NP
drengen]]]
the girl has looked at the boy

In (74) the verb “inspirere” (inspire) does not subcategorize for a preposition phrase;
the optional PP is adjoined to VP. In (75) the verb “kigge (på)” (look (at)) optionally
subcategorizes for a PP, i.e. if the object is present it is governed by the VP.
In appendix E such cases of substitution ([
VP
] [
PP
] = [
VP
[
PP
]]) are marked with
+G in the second column on the right. There is no ungrammatical use of governed or
ungoverned prepositions.
There is also evidence of a single impaired lexical entry in TJ’s production. The
Danish verb “slå” (hit) is a very common word, which has irregular inflection. Compare
K. R. C. 2001 92
the two common verbs “ligne” (resemble, look like) and “slå” (hit), which have regular
and irregular inflection respectively:

(76) Tense: Regular: Irregular:
Infinitive lign-e slå-Ø
Present lign-er slå-r
Past lign-ede slog
Past Ptcp. lign-et slå-et
Present Ptcp. lign-ende slå-ende

What is interesting is TJ’s use of the past tense form of “slå”. The correct form is
“slog”, but TJ only produced it once out of three intended and once before self-
correcting it. In cases where he produced the wrong tense form and substituted it with
the past form (4/8), he produced the verb stem with the regular inflection, i.e. “slå-ede”
instead of the correct “slog”:

Intended Produced Appendix E
slået slået Correct (#17, #18, #37, #69)
slået slåede Wrong (#23, #65)
slået slog slåede Wrong, Self-correction, Wrong (#38)
slået slåede slået Self-correction, Correct (#95)
slå slå Correct (#28)
slog slog Correct (#04)
Slog slår Wrong (#29)
Slog kiggede efter Wrong (#68)
Table 17: TJ’s use of the verb “slå” (hit).

No other verb or other type of word was used in a similar manner. It seems that the
lexical entry for this verb is selectively impaired.
Finally, Danish has +zero morphology, which implies that agrammatics will
omit inflection affixes and produce bare stems. However, there are no bare stems in TJ’s
speech production. This may be a positive effect of the eight months of rehabilitation TJ
has undergone.

K. R. C. 2001 93
7.3.3.4 Conclusions
TJ does not confirm the hypotheses of Trace Deletion and Tree-pruning. First of all, his
comprehension does not pattern the way that the Trace deletion Hypothesis would
predict. He only has some problems with the interpretation of subject-object relatives.
The hypothesis as such cannot explain this, but I fail to see any other and (importantly)
better explanation. For example, he does not perform according to a canonical word
order pattern, which would predict good performance on SVO and poor performance on
all others. This is not borne out. Neither does he perform according to canonical θ-role
order, which would predict that only AGENT-THEME order would be result in correct
performance. He performs correctly on for example passives, which have the non-
canonical THEME-AGENT order and he performs correctly on the sentences with and
EXPERIENCER subject, which were predicted to result in performance below chance.
Furthermore, subject-object relatives are not problematic because the relative pronoun
(the wh-element) and its antecedent do not have the same grammatical roles (subject
and object), because he has no problems with understanding for instance object-subject
relatives. The subject-object construction is special because the object moves across the
subject in the embedded clause and the relative pronoun and its antecedent do not have
the same grammatical roles. It therefore seems likely that it is a case of trace deletion
after all, as the movement of the object is crucial. Importantly, there is nothing in his
comprehension performance that falsifies the hypotheses.
His production shows a clear tendency to reduce the number of VPs in the
sentences, which is not predicted by the Tree-pruning Hypothesis. However, this
phenomenon does not falsify the hypothesis. He shows some signs of preposition drop
with a marked difference between the structural position of the preposition in question.
As predicted, he omits governed prepositions more frequently than ungoverned
prepositions. He shows no signs of impairment to the functional nodes CP, TnsP, AgrP,
and NegP and his test results can therefore neither verify nor falsify the Tree-pruning
Hypothesis.
I suspect that the reason why TJ’s test results can neither falsify nor verify any
of the hypotheses is that he has been in recovery for so long. At the time of testing TJ
had ceased to be a clear and representative case of Broca’s aphasia as he had almost
fully recovered. I see nothing in his performance that is seriously problematic for the
hypotheses. I discuss these matters in more detail in the following chapter.
K. R. C. 2001 94
8 Discussion: Cerebral Area and Function Revisited
TJ suffered massive hematomas in the areas of the perisylvian region, which means that
his entire grammatical neural ‘machinery’ was implicated, but not necessarily
destroyed. Once neurons are destroyed they are gone forever. They can not regenerate.
However, the connective system between neurons can regenerate or reorganize.
Disrupted connections can be ‘rebuilt’ (cf. Elman et al. 1996), and therefore brain
damage does not necessarily lead to permanent impairment if the lesion site is not too
great or encompassing with regards to the functional area in question. The function of
the affected area can sometimes be restored over time. Figure 11 on page 83 above is a
gross illustration of the region affected in TJ’s case. The medical report is not specific
or detailed enough to pinpoint the sites any more accurately than indicated with the light
shaded areas. Hence, the figure is not intended to be interpreted too literally. The point
is that his language areas, notably Broca’s area (and Wernicke’s area), are in the region
affected by the massive hematomas, which should reveal certain predictable behaviors
in the linguistic performance of the patient. As I showed in the previous section, TJ did
in fact show traces of such behavior, but due to eight months of intense training with
speech therapists he has come a long way. He has only fragments of aphasia left, but
importantly his performance did not falsify the theoretical apparatus posited in chapter 6
and sections 7.1 and 7.2.
His comprehension is impaired, but only with regards to subject-object relatives,
on which he performs 70% correct. His comprehension of all other constructions is not
impaired, or rather has been restored. Admittedly, these results deviate from the pattern
predicted by the Trace Deletion Hypothesis, which should have shown problems with
all constructions involving syntactic movement. However, it should again be kept in
mind that TJ has been in rehabilitation for eight months. Furthermore, it is important to
notice that the lowest performance is on a construction with movement of the object
across the subject. At least this is in accordance with the hypothesis. He had no
problems with actives (except for lexical passives, but see section 7.3.3.2 for
comments). Thus, his comprehension is near normal, with only chance performance on
subject-object relatives – a deficit, for which subtle theoretical mechanisms are required
to detect it. A distinction between center- and right-embedding is insufficient as a
measure of his performance: He correctly interprets object-subject and object-object
relatives, which are both right-embeddings (or right-branching structures); he also
K. R. C. 2001 95
correctly interprets subject-subject relatives, which are center-embeddings. Crucially, he
performs poorly on only subject-object relatives, which are also center-embeddings.
Hence, correlation between right-embedding and above chance performance and
between center-embedding and chance performance is not found.
On the production side
13
he also seems to have problems with object-relatives
(cf. TJ’s performance on subject-object relatives, object-object relatives, and object
clefts in Table 12). In 75% (9/12) of the wrongly produced, he changes the object
relative clause to a subject relative. In total, 42,9% (9/21) of the object relatives are
produced as subject relatives, which seems to be the preferred type of embedding. The
most notable problematic type is the same as the one causing problems in
comprehension: the subject-object relative. This reflects neither a problem with (and
hence avoidance of) center-embedding, as he performs near-normal on subject-subject
relative clauses, nor a strong preference for ‘basic order’ of AGENT and THEME, as for
example he performs 100% correct on the “blive” passives. It does, however, reflect a
preference for a basic word order (Subject-Verb-Object) in production, as is often
noted as a characteristic of Broca’s aphasia. For example:

Indeed, some patients (particular Broca’s) appear to overuse basic SVO, as
though this word order type provided a kind of “safe harbor” for sentence
planning. Such overuse is only evident in languages that permit pragmatic word
order variation [such as Danish, K. R. C.]; it could not be detected in a rigid
word order language like English. (Bates et al. 1991: 131. Emphasis added.)

This finding is compatible with the postulation that the front end of the perisylvian
region, which is one of the centers of TJ’s hematomas (see Figure 11), is crucially
involved in linguistics syntagmatic relations. Recall from chapter 5 above, that Bates et
al. (1991), Damasio (1992), Deacon (1997), and Pinker (1994) all shared a common
important feature in their respective theories on area and function: Broca’s area and its
vicinity are responsible for syntagmatic / relational / sequential computation of
grammatical structure. Importantly, this is also compatible with Grodzinsky’s

13
It should be noted that this production is repetition and spontaneous production. However, the fact that
he changes the sentences on repetition in a systematic way, e.g. the auxiliary verb reduction, shows that
he is not merely ‘parroting’, which would not necessarily involve his grammatical competence. Parrot-
like repetition would result in random errors (e.g. due to memory limitations) or no errors, if he were just
good at repetition. The systematic production errors suggest that TJ’s grammar is involved.
K. R. C. 2001 96
hypothesis that Broca’s area (and vicinity) is responsible for trace-antecedent (co-
indexing) chains in comprehension and for the functional projections in the syntactic
tree and their involvement in movement and hence chains.
Perhaps related to this is the fact that TJ has a tendency to use constructions
without auxiliary verbs. Leaving out the auxiliaries reduces the structure (minimizes the
size) of the syntactic tree, as the missing verbs do not project VPs. This means that less
syntagmatic / relational / sequential computation is required. However, the omission of
auxiliary verbs may be related to a lexical deficit. I shall return to this shortly.
As the CP node and all the nodes beneath it are restored in TJ no severe tense
inflection errors are found. None of the verbs expected to be finite and inflected for
tense were uninflected. Neither does TJ show any problems with correctly producing
adverbs, negations, and subjects. As mentioned above, this shows that the entire
syntactic representation is intact.
TJ does however make a few errors. He has a tendency to drop the verbal
passive morphology “–s”, he has a selective impairment to the past tense form of the
“slå” (hit), and he omits few prepositions; Governed prepositions are more frequently
omitted than ungoverned prepositions, cf. Table 16 on page 91. The omission of
governed prepositions may also be an expression of the disruption of relational
computation linked to a lesion in Broca’s area. The verb that heads the verb phrase in
which the omitted preposition was supposed to be subcategorizes for a PP. For example
the transitive verb “smile” (smile) subcategorizes for a PP headed by the preposition
“til” (at) with the THEME as complement:

(77) [
CP
Drengen
1
har … [
VP
t
1
smilet [
PP
til [
NP
pigen …]]]]
The boy has smiled at the girl

The point is that the verb determines the structure of the complement phrase that
follows it. As this is matter of grammatical structure, it demands some syntagmatic
computation. On the other hand, subcategorization is also very much a lexical matter, in
the sense that the number and syntactic categories of a verb’s arguments is determined
by its lexical entry, cf. section 2.1.2 above. So, some paradigmatic computation is
required as well. The lexical entry determines the presence of certain phrases as
opposed to others, cf. the Projection Principle (section 2.1.3). The X-bar component (cf.
K. R. C. 2001 97
section 2.1.1) of the grammar determines the structure and in interaction with Move-α
(and the rest of the grammar, such as the Case Filter) determines the structural position
of the phrases. TJ’s omission of governed prepositions is most likely caused by trauma
to Broca’s area, but may also be influenced by the damage to Wernicke’s area.
Let us for a moment consider an alternative model. Myers-Scotton & Jake
(2000) present a model of word selection or rather lemma activation. Lemma is another
word for lexical entry. The model, called the 4-M model, is based on a four-way
classification of morphemes, as opposed to the classical two-way model: function
words/morphemes (closed class) and content words/morphemes (open class). I shall
give a brief overview of their hypothesis. First consider the model for the production of
sentences:


Conceptual Level:
Speaker’s intentions → semantic/pragmatic feature bundles.

Lemma Level:
Content morphemes (directly selected) → Early system morphemes (indirectly selected)

Functional Level:
Formulator → Late system morphemes

Positional Level:
Phonetic/surface forms

Figure 12: Production process diagram. Based on Myers-Scotton & Jake 2000:
1056, figure 1.

On the conceptual level the speaker has some kind of intention to say something, which
leads to the activation of certain semantic/pragmatic feature bundles. In turn this
activation selects the appropriate lemmas, prototypically predicates (verbs) and their
arguments (typically NPs). These lemmas then potentially select certain system
morphemes. For example, if the speaker intends to say something like “She looks at
him” the verb “look” and its arguments “she” and “him” are directly selected by feature
bundle activation. They are called content morphemes and distinguish themselves from
the other types of morphemes by having the feature [+THEMATIC ROLE], i.e. they are
either theta-assignors or theta-assignees. Information on the categories of predicate and
arguments are also given by the feature bundles, such that “look” is the head of a VP
K. R. C. 2001 98
and the arguments are NPs. Furthermore, the verb (the head) selects the preposition
“at”, an early system morpheme. This is indirectly selected as it is the verb that ‘calls’ it
– it is not directly selected by feature bundle activation. Together the content
morphemes and the early system morphemes provide information to the formulator at
the functional level, which assemble larger hierarchical constituents. In order to do this,
some additional system morphemes are required (they are called by the formulator) such
as the agreement suffix (3
rd
person singular) on the verb. These late system morphemes
are divided into two categories: outsider morphemes and bridge morphemes. An
example of an outsider morpheme is the before mentioned English 3
rd
person singular
agreement suffix. They have to “look outside” their maximal projections (AgrP):

They depend on grammatical information outside of the immediate maximal
projection in which they occur. This information is only available when the
formulator sends directions to the positional/surface level for how maximal
projections are unified in a larger construction. (Myers-Scotton & Jake 2000:
1064)

In other words, the verbal suffix is dependent on movement in order to connect with the
verb.
Bridge system morphemes “connect content morphemes with each other without
reference to the specific semantic/pragmatic properties of a content head.” Examples of
this are the English preposition “of” in “friend of Tom” and the possessive suffix “-s” in
“Tom’s friend”, which mark the orders of head-complement and complement-head,
respectively. In summary, the distinction between the four types of morphemes can be
stated as follows:
K. R. C. 2001 99


[+/- Conceptually activated]

[+ Conceptual] [- Conceptual]
Content morphemes & Late system morphemes
Early system morphemes

[+/- Thematic role [+/- Looks outside
assignor/assignee] maximal projection]

[+ Thematic] [- Thematic] [- Outside] [+ Outside]
Content Early system Bridge Outsider
morphemes morphemes morphemes morphemes

Figure 13: Feature distribution and classification of morphemes.
Adapted from Myers-Scotton & Jake 2000: 1062, figure 2.


According to Myers-Scotton & Jake, patients with Broca's aphasia will produce content
morphemes more accurately than any of the system morphemes, and late system
morphemes will be missing or used less accurate. The late system morphemes are part
of the "structure-building apparatus" or syntagmatic computation: “The 4-M model
implicates late system morphemes, both outsiders and bridges, as expressions of the
relational aspects of language” (Myers-Scotton 2000: 1076). This fits very well with the
conclusions above (cf. the discussion in chapter 5): anterior and posterior lesions in the
language area leads to damage to syntagmatic and paradigmatic linguistic computation,
respectively.
Considering the data on TJ, the 4-M model provides the following classification
of impaired morphemes:




K. R. C. 2001 100
(78) Classification of impaired morphemes
Content morphemes: none
Early system morphemes: auxiliary verbs (aspect), "blive",
VP-internal prepositions
Bridge morphemes: passive "-s", adjoined prepositions
Outsider morphemes: tense

Content morphemes are predicates and arguments, the morphemes with the feature
[+THEMATIC ROLE]. Early system morphemes are those selected by the content
morphemes and part of the conceptually activated morphemes; the Danish auxiliary
verbs “have” and “blive” are aspect markers and must therefore be conceptually
activated. The VP-internal prepositions are subcategorized for by verbs (content
morphemes) and are therefore early system morphemes. I have categorized the passive
“-s” and the adjoined prepositions as bridge morphemes, because they merely signal the
‘direction’ of the predicate (i.e. the order of the arguments) in the same way as the
English genitive “-s” and possessive “of” signal complement-head and head-
complement order respectively. Tense inflections are outsider morphemes as they
depend on information outside their own maximal projection TnsP. Tense inflection is
dependent on the verb moving out of VP and into TnsP in main clauses; in embedded
clauses the tense inflection itself has to move into VP to connect with the verb. If
Danish had agreement morphology it would have been outsider morphemes as well.
Provided my classification is correct, the 4-M model correctly predicts that
content words are not impaired. The predicate and arguments are produced. On the
other possible extreme, the model predicts impairment affecting tense morphology,
which is also predicted in the Tree-pruning Hypothesis.
From hereon, I think the 4-M model loses compatibility: First, it is predicted that
early system morphemes are less impaired than late system morphemes. With regards to
prepositions this is not identical to what was predicted in the model outlined in the
previous chapters. Actually, it is quite the opposite: in the 4M model governed (VP-
internal) prepositions are early system morphemes and should therefore be less impaired
than ungoverned (adjoined) prepositions, which are late system morphemes. According
to Grodzinsky (1988) in agrammatism VP-internal prepositions are deleted/omitted,
while adjoined prepositions are not. This pattern is also reflected in the speech of TJ, cf.
Table 16 on page 91. As argued above, VP-internal prepositions depend on both
K. R. C. 2001 101
syntagmatic and paradigmatic computation: they are called by the verb, such that the
presence of a PP is specified, and they depend on the grammar to specify position in the
structure. This is not accounted for in the 4-M model, which appears to clusters
practically all of grammar into one: the formulator.
With regards to auxiliary verbs, TJ has a strong tendency to omit the verbs
related to aspect (+/- perfect), while the passive marker "blive" is not omitted. If the
auxiliary verbs are considered to be late system morphemes, i.e. part of the "structure
building apparatus", TJ's auxiliary verb omission, which I considered structure
reduction above, would be predicted correctly by the 4-M model. On the other hand,
auxiliaries must be considered early system morphemes, as they are "called by" the
main verb in order to specify semantic aspect. This is incompatible with the 4-M model
as well.
The prediction by the 4-M model that the passive morphology is impaired is
borne out in the speech of TJ.
On the whole, I think that the 4-M model is a clear and good proposal of the
process of morpheme selection. However, the predictions provided by the model are
incompatible with the data from TJ. This may, on the other hand, be an artifact of the
lack of elaboration an examples in the article by Myers-Scotton & Jake. It does,
however, once again support the distinction of syntagmatic (bridge and outsider
morphemes) and paradigmatic (content and early system morphemes) computation,
even though this distinction alone is insufficient. This has already been done by the
other approaches discussed so far (Bates et al. 1991, Damasio 1992, Deacon 1997,
Grodzinsky 2000, Pinker 1994). Furthermore, the model has nothing to say about on
issues of movement related to the impairment of the top-most nodes. In the 4-M model
agreement and tense, for example, are in the same category, and as such they must be
equally impaired. Evidence provided by e.g. Grodzinsky (2000) and Friedmann &
Grodzinsky (1997, 2000) point to a distinction between tense and agreement in
agrammatism (see also the discussion on the Split-inflection Hypothesis in section 6.2
above). For this reason, the 4-M model lacks breakdown compatibility. Furthermore,
using the 4-M model I fail to see how to account for the distinction in comprehension
between constructions with and without syntactic movement (of the underlying object),
as reported by Grodzinsky.
As stressed several times throughout the text, a mere distinction between
syntagmatic and paradigmatic is insufficient. For example, Broca's aphasics do not lose
K. R. C. 2001 102
all of their syntax (syntagmatic linguistic computation), only parts of it. The distinction
in comprehension between movement and non-movement clearly supports this. In
production, agrammatic speech is not a random flow of words. According to Bates et al.
(1991) agrammatics tend to overuse SVO order. Clearly not all relational aspects are
lost.
In Wernicke's aphasia, the sentences, however incomprehensible, are not formed
at random. The word substitutions are frequently within the proper semantic field, for
example "chair" for "table" and "knee" for "elbow". The phonemic substitutions, e.g.
"tubber" for "butter" also point to at least some sparing of paradigmatic relations, as the
words appear not to be completely random. For these reasons the functional distribution
within the language area has to be finer than assumed by a gross dichotomies such as
syntagmatic vs. paradigmatic, syntax vs. lexicon, anterior vs. posterior language area.
Even though this is hinted at in the 4-M model, it is still not elaborate enough to account
for all the phenomena associated with agrammatism (or the other types of aphasia for
that matter).
Grodzinsky proposed a syntactic approach to (Broca's) aphasia (cf. chapter 6
above), and I agree that the evidence seems to demand it. However little evidence of
aphasia left in TJ, his performance is not distributed randomly over syntagmatic
computation. He shows remnants of a movement-related deficit, only reflected in
subject-object relatives in comprehension, but in production all object relatives are
affected. Grodzinsky’s hypotheses, however, do not capture this late stage in recovery
in any precise manner, but still his deficit seems to be syntactic.
The lesion to his posterior region, I believe, is reflected in the few examples of
lexical deficits, such as his problems with the irregular inflection of the verb "slå" (hit)
in production, and perhaps his problems with the lexical passive verb "væmmes" (be
disgusted) in comprehension. He also makes some word substitutions in production, so
some aspects of word finding / paradigmatic relations appear to be affected.
K. R. C. 2001 103
9 Conclusions
Paradis (1998: 418) states that “[i]t is now known that the symptoms of agrammatism
will vary in accordance with the structure of each language.” Grodzinsky (2000: 15)
relates this variation to the zero-morphology parameter, such that agrammatism is
manifested differently depending on the morphology of the language: if the language
has +zero-morphology agrammatics will tend to show omission of inflection, whereas if
the language has –zero-morphology agrammatics will substitute inflection.
Furthermore, the deficit will vary according to the word order of the language. The fact
that this variation can be captured in the same grammatical framework that accounts for
normal competence makes the theory breakdown-compatible. The lesion site thus has
some sort of language-specific and grammar-specific function. As agrammatism is
caused by trauma to Broca's area, the aspects of grammar affected must be located in
that same area. In comprehension agrammatism is manifested as a disruption of trace-
antecedent chains, and in production it is manifested as pruning of the topmost
functional nodes of the syntactic representation. So, the apparatus responsible for these
aspects of grammar must be located in Broca's area and its vicinity. As trace-deletion
and tree-pruning are clearly two different representational aspects of grammar they can
hardly be computed by the same neural mechanism. For this reason they may be located
in close vicinity of each other instead of in the very same place. As Grodzinsky (2000:
18) puts it, it suggests "anatomical proximity, but functional separation." As other
aspects of grammar are not disrupted by damage to Broca's area, only the mechanisms
underlying the specific syntactic abilities affected in the deficit are located there,
nothing else.
A look at Figure 6 on page 41 above, in which cerebral areas within the
language zone are correlated with different types of aphasia, confirms the internal
modular structure of language. Different aspects of grammar are affected depending on
lesion site, and therefore all of grammar cannot be located in on spot. In fact:

[M]ost human linguistic abilities, including most syntax, are not located in the
anterior language areas – Broca’s area and deeper white matter, operculum,
and anterior insula. (Grodzinsky 2000: 17)

K. R. C. 2001 104
It is up to further investigation to discover the connections between different specific
aspects of grammar and different types of aphasia in correlation with lesion site.
The human grammatical competence is distributed over the perisylvian region of
the left hemisphere, but localized within this region, as argued in sections 3.2 and 3.3.
This is the case in normal subjects without history of early childhood lesions in the left
hemisphere or any other cerebral pathology. If normal development is not disrupted,
language will be located in the language zone of the left hemisphere. This is what I
called the default brain plan in section 3.4. In other words, language is innate and
reflected in the architecture of the brain (as well as our speech apparatus). The
robustness of language in the face of severe retardation (cf. section 4.2) also support a
neural basis of language, i.e. innateness, as general learning in such subjects is severely
reduced. Furthermore, this points to external modularity of language, as general
intelligence and language are doubly dissociated. One may be present without the other
(cf. Table 4, page 42).
As such the structure of the language zone constrains the range of possible
human languages - it constitutes the universal grammar. Language acquisition is, in
loose terms, a matter of adjusting the grammar to the one language spoken in the
ambient society. Furthermore, as this language capacity can (supposedly) be described
by a comprehensive linguistic theory, the very same theory should be able to describe
all the reported types of aphasia (and recovery). Furthermore, such a theory should be
able to make predictions regarding the performance of aphasics in any language, in
which such research has not yet been made.
In order to test the hypotheses of Grodzinsky’s syntactic approach to Broca’s
aphasia, I applied the framework to the Danish language and tested the predictions thus
made on TJ, a Danish agrammatic aphasia patient. However, TJ was an unclear case for
three reasons. First, his impairment was caused by hematomas in multiple places.
Second, his medical report was rather unclear on the exact locations of these
hematomas. Third, and most important, TJ had been in recovery for too long to show
any clear signs of agrammatism (which, according to his speech therapist, had been
apparent at earlier stages). Though his test results can neither falsify nor completely
verify the hypotheses, they point to a syntactic deficit – specifically related to movement
and trace-antecedent chains. TJ’s deficit was at the time of testing reduced to a problem
with object relatives, which involves movement of the object across the subject. As I
have argued, his entire clause structure is intact, and therefore his production deficit
K. R. C. 2001 105
cannot be account for by reference to tree pruning. His comprehension deficit can not be
accounted for by mere reference to trace deletion, which however, I think is still
involved. Therefore, we have to consider what the comprehension and production
deficits have in common: object relatives. What do the other constructions have that the
object relatives do not have? A) All the constructions predicted to yield above chance
performance in comprehension (cf. Table 10 on page 71) are not movement derived. B)
The three types of passives (verbal, “blive”, and psychological) have a preposition to
assign AGENT/EXPERIENCER grammatically. In object relatives the object is not assigned
case directly by the verb but through θ-transmission from the trace (both comprehension
and production contains movement derived constructions; the former due to someone
else’s intact grammar, the latter due to the intact CP in TJ). If the trace is deleted or the
chain otherwise broken this transmission is no longer possible and the cognitive Default
Strategy assigns AGENT to the leftmost NP in a θ-position. This is predicted to lead to
competition (two AGENTS) in e.g. verbal passives. However, it seems that in TJ the
presence of the adjoined AGENT overrides this potential competition. I propose that the
direct θ-assignor-assignee connection has been strengthened, such that the grammatical
assignment has priority over the cognitive assignment. This would account for the fact
that TJ only performs poorly on constructions where all θ-roles has to be assigned
through θ-transmission or cognitive assignment. If I am correct, this underlying deficit
of θ-transmission should be incorporated in the framework such that different stages of
recovery are captured as well as the most severe stage, on which Grodzinsky’s theory is
based. Recovery would thus go through at least the following stages (some may recover
fully, while others may not and thus stagnate at a prior stage):

a) Stages of severe agrammatism: Trace deletion, tree pruning.
b) Rebuilding of the syntactic tree and recovery of comprehension and production –
not necessarily in that order.
c) Strengthened grammatical θ-role assignment, which is manifested as reduced error
rates.
d) Reconstruction of trace-antecedent chains.
e) Recovery.

K. R. C. 2001 106
The strengthening of the direct grammatical θ-assignment will result in decreasing error
rates as there is less competition between the two identical θ-roles in e.g. reversible
passives; furthermore, it may lead to the reconstruction of grammatical θ-chains, such
that the correct θ-roles will be assigned to the appropriate NPs. In turn, the Default
Strategy will be down prioritized and eventually abandoned. Due to the reconstructed θ-
chains object-relatives will reappear in the speech production.
Granted that my proposal of the strengthening of direct θ-assignment is correct
(keeping in mind that TJ may not at all be a good example of agrammatism) it implies
that Grodzinsky’s theory has not yet fully gained Breakdown Compatibility. His theory
currently captures only some of the earlier stages of recovery. Though my proposals
may remedy some of the problems it still has nothing to say about the systematic
omission of auxiliary (aspect) verbs. There is still much work to be done.

K. R. C. 2001 107
10 References

Archangeli, Diana & D. Terrence Langendoen (Eds.) (1997) Optimality Theory. An
Overview, Oxford: Blackwell Publishers
Balogh, Jennifer E. & Yosef Grodzinsky (2000) Levels of linguistic representation
in Broca’s aphasia: Implicitness and referentiality of arguments. In Bastiaanse &
Grodzinsky (Eds.) (2000) [Pdf version available at http://freud.tau.ac.il/~yosef1/]
Baron-Cohen, S., A. M. Leslie, U. Frith (1985) Does the autistic child have a
‘theory of mind’? Cognition 21: 37-46
Bastiaanse, R. & Y. Grodzinsky (Eds.) (2000) Grammatical Disorders in Aphasia: a
Neurolinguistic Perspective, London: Whurr Publishers
Bates, Elizabeth, Beverly Wulfeck, and Brian MacWhinney (1991) Cross-linguistic
research in aphasia: an overview. Brain and Language 41: 123-148
Belletti, Adriana (1990) Generalized Verb Movement. Aspects of Verb Syntax,
Turin: Rosenberg & Sellier
Bellugi, U., S. Marks, A. Bihrle, and H. Sabo (1988) Dissociation between language
and cognitive functions in Williams syndrome. In Bishop & Mogford (Eds.) (1988:
177-189)
Bickerton, Derek (1988) Creole Languages and the Bioprogram. In F. Newmeyer
(ed.) Linguistics: The Cambridge Survey. Vol.2, Cambridge: Cambridge University
Press: 268-284.
Bickerton, Derek (1990) Language and Species, Chicago: The University of
Chicago Press
Bishop, D.V.M. (1989) Autism, Asperger’s syndrome and semantic-pragmatic
disorder: Where are the boundaries? British Journal of Disorders of Communication
24: 107-121
Bishop, Dorothy & Kay Mogford (1988) Five questions about language acquisition.
In Bishop & Mogford (Eds.) (1988: 239-260)
Bishop, Dorothy & Kay Mogford (Eds.) (1988/1993) Language development in
exceptional circumstances, Hove: Psychology Press Ltd.
Bishop, Dorothy (1988) Language development after focal brain damage. In Bishop
& Mogford (Eds.) (1988: 203-219)
K. R. C. 2001 108
Burton, Strang & Jane Grimshaw (1992) Coordination of VP-internal subjects.
Linguistic Inquiry 23 (2): 305-313
Calvin, William H. & George A. Ojemann (1994) Conversations with Neil’s Brain,
Reading MA: Addison-Wesley
Chomsky, Noam (1981) Lectures in Government and Binding, Dordrecht: Foris
Chomsky, Noam (1993) A minimalist program for linguistic theory. In Hale, K. & J.
Keyser (Eds.) The View from Building 20, Cambridge, Mass.: MIT Press, 1-52
Chomsky, Noam (1995) The Minimalist Program, Cambridge Mass: MIT Press
Christensen, Ken R. (2000) The co-evolution of language and the brain: A review of
two contrastive views (Pinker & Deacon). To appear in Grazer Linguistische
Studien [http://www-gewi.kfunigraz.ac.at/ling/gls/]
Comrie, Bernard (1989) Language Universals and Linguistic Typology, 2
nd
Edition,
Oxford: Blackwell
Corballis, Michael C. (1991) The Lopsided Ape. Evolution of the Generative Mind,
Oxford: Oxford University Press
Corina, David (1998) The processing of sign language. Evidence from aphasia. In
Stemmer & Whitaker (1998: 313: 329)
Crowley, Terry (1997) Introduction to Historical Linguistics, 3
rd
Edition, Oxford:
Oxford University Press
Damasio, Antonio R. & Hanna Damasio (1992) Brain and language. Scientific
American, September 1992: 63-71
Damasio, Antonio R. (1992) Aphasia. The New England Journal of Medicine 326,
8, 531-539
Damasio, Antonio R. (1994) Descartes’ Error, New York: Grosset/Putnam
Darwin, Charles (1859/1964) On the Origins of Species. Cambridge, Mass.: Harvard
University Press.
Deacon, Terrence (1997) The Symbolic Species. The Co-evolution of Language and
the Human Brain, London: Penguin
Dehaene, S., E. Dupoux, J. Mehler, L. Cohen, E. Paulesu, D. Perani, P.-F. van de
Moortele, S. Lehéricy & D. Le Bihan (1997) Anatomical variability in the cortical
representation of first and second language. NeuroReport 8: 3809-3815
Donald, Merlin (1991) Origins of the Modern Mind. Three Stages in the Evolution
of Culture and Cognition, Cambridge, Mass.: Harvard University Press
K. R. C. 2001 109
Drai, Dan & Yosef Grodzinsky (1999) Comprehension regularity in Broca’s
aphasia? There’s more to it than you ever imagined. Brain and Language 70: 139-
143
Dunbar, R.I.M. (1993) Co-evolution of neocortex size, group size and language in
humans. Behavioral and Brain Sciences 16 (4): 681-735
Edelman, Gerald M. (1998) Building a picture of the brain. Dædalus 127: 37-69
Elman, Jeffrey L., Elizabeth A. Bates, Mark H. Johnson, Annette Karmiloff-Smith,
Domenico Parisi & Kim Plunkett (1996) Rethinking Innateness. A Connectionist
Perspective on Development, Cambridge Mass.: MIT Press
Fay, W.H. (1988) Infantile autism. In Bishop & Mogford (Eds.) (1988: 190-202)
Fodor, Jerry A. (1983) The Modularity of Mind, Cambridge Mass: MIT Press
Fodor, Jerry A. (1985) Précis of the modularity of mind. Behavioral and Brain
Sciences 8: 1-42
Foley, William A. (1997) Anthropological Linguistics. An Introduction, Oxford:
Blackwell
Friedmann, Na’ama & Yosef Grodzinsky (1997) Tense and agreement in
agrammatic production: pruning the syntactic tree. Brain and Language 56: 397-425
Friedmann, Na’ama & Yosef Grodzinsky (2000) Split inflection in neurolinguistics.
In M.-A. Friedemann & L. Rizzi (Eds.) (2000) The Acquisition of Syntax: Studies in
Comparative Developmental Linguistics, New York: Longman. [Pdf-version
available at http://www-psy.ucsd.edu/~naama/]
Frith, Uta (1989) Autism. Explaining the Enigma, Oxford: Basil Blackwell
Gazzaniga, Michael S. (1989) Organization of the human brain. Science 245, 947-
952
Gleitman, Lila R. & Mark Liberman (Eds.) (1995) An Invitation to Cognitive
Science, 2
nd
Edition. Volume 1: Language, Cambridge Mass.: MIT Press
Gopnik, M. & M. Crago (1991) Familial aggregation of a developmental language
disorder. Cognition 39: 1-50
Grimshaw, Jane (1991) Extended Projections, MS., Brandeis University
Grimshaw, Jane (1997) Projection, heads, and optimality. Linguistic Inquiry 28,3:
373-422
Grodzinsky, Yosef (1984) The syntactic characterization of agrammatism.
Cognition 16: 99-120
K. R. C. 2001 110
Grodzinsky, Yosef (1986) Language deficits and the theory of syntax. Brain and
Language 27: 135-159
Grodzinsky, Yosef (1988) Syntactic representations in agrammatic aphasia: the case
of prepositions. Language and Speech 31 (2): 115-134
Grodzinsky, Yosef (1989) Agrammatic comprehension of relative clauses. Brain
and Language 37: 480-499
Grodzinsky, Yosef (1990) Theoretical Perspectives on Language Deficits,
Cambridge Mass.: MIT Press
Grodzinsky, Yosef (1991) There is an entity called agrammatic aphasia. Brain and
Language 41: 555-546
Grodzinsky, Yosef (1995a) A restrictive theory of agrammatic comprehension.
Brain and Language 50: 27-51
Grodzinsky, Yosef (1995b) Trace deletion, θ-roles, and cognitive strategies. Brain
and Language 51: 469-497
Grodzinsky, Yosef (2000) The neurology of syntax: language use without Broca’s
area. Behavioral and Brain Sciences 23 (1): 1-71
Grodzinsky, Yosef, Kenneth Wexler, Yu-Chin Chien, Susan Marakovitz & Julie
Solomon (1993) The breakdown of binding relations. Brain and Language 45: 396-
422
Haegeman, Liliane (1994) Introduction to Government and Binding Theory, 2
nd

Edition., Oxford: Blackwell
Harley, Trevor A. (1995) The Psychology of Language, Hove: Psychology Press,
Ltd.
Jackendoff, Ray S. (1972) Semantic Interpretation in Generative Grammar,
Cambridge Mass.: MIT Press
Jakobson, Roman (1956) Two aspects of language and two types of aphasic
disturbances. In R. Jakobson & M. Halle, (Eds.) Fundamentals of Language, The
Hague: Mouton
Kager, René (1999) Optimality Theory, Cambridge: Cambridge University Press
Kertesz, Andrew & Patricia McCabe (1975) Intelligence and aphasia: performance
of aphasics on Raven’s Coloured Progressive Matrices (RCPM). Brain and
Language 2: 387-395
K. R. C. 2001 111
Larson, R. K. (1988) On the double object construction. Linguistic Inquiry 19(3):
335-391
Lenneberg, E. H. (1967) The Biological Foundations of Language, New York:
Wiley
Lieberman, P. (1984) The Biology and Evolution of Language. Cambridge Mass.:
Harvard University Press
Menn, Lise, M. O'Connor, Loraine K. Obler & Audrey Holland (1995) Non-fluent
Aphasia in a Multilingual World, Amsterdam: John Benjamins Publishing Company
Müller, Ralph-Axel (1996) Innateness, autonomy, universality? Neurobiological
approaches to language. Behavioral and Brain Sciences 19: 611-675
Myers-Scotton, Carol & Janice L. Jake (2000) Four types of morpheme: evidence
from aphasia, code switching, and second-language acquisition. Linguistics 38(6):
1053-1100
Paradis, Michel (1998) Language and Communication in Multilinguals, in Stemmer
& Whitaker (Eds.) (1998: 417-430)
Pinker, Steven & Paul Bloom (1990) Natural language and natural selection.
Behavioral and Brain Sciences 13: 707-784
Pinker, Steven (1994) The Language Instinct, London: Penguin
Pinker, Steven (1995) Language acquisition. In Gleitman & Liberman (Eds.) (1995)
Pinker, Steven (1997) How the Mind Works, London: Penguin
Pollock, Jean-Yves (1989) Verb movement, universal grammar, and the structure of
IP. Linguistic Inquiry 20(3): 365-424
Rondal, J.A. (1988) Down’s syndrome. In Bishop & Mogford (Eds.) (1988: 165-
176)
Smith, Neil & Ianthi-Maria Tsimpli (1995) The Mind of a Savant. Language
Learning and Modularity, Oxford: Blackwell
Sobotta, J. & H. Becher (1975) Atlas of Human Anatomy, 9
th
English Edition,
München: Urban & Schwarzenberg
Stemmer, B. & H.A. Whitaker (Eds.) (1998) Handbook of Neurolinguistics, New
York: Academic Press
Stromswold, Karin, David Caplan, Nathaniel Alpert, and Scott Rauch (1996)
Localization of syntactic comprehension by positron emission tomography. Brain
and Language 52, 452-473
K. R. C. 2001 112
Van der Lely, H.K.J. & L. Stollwerck (1996) A grammatical specific language
impairment in children: an autosomal dominant inheritance? Brain & Language 52:
484-504
Van der Lely, H.K.J., S. Rosen & A. McClelland (1998) Evidence for a grammar
specific deficit in children. Current Biology 8, 23: 1253-1258
Vikner, Sten (1989) Object shift and double objects in Danish. Working Papers in
Scandinavian Syntax 44: 141-155
Vikner, Sten (1995) Verb Movement and Expletive Subjects in the Germanic
Languages, Oxford: Oxford University Press
Vikner, Sten (1999) Ledstillingen i dansk og government & binding. In Jensen, Per
Anker & Peter Skadhauge (Eds.) Sætningsskemaet i Generativ Grammatik, Institut
for Erhverssproglig Informatik og Kommunikation, Syddansk Universitet, Kolding
Vikner, Sten (2000) V°-to-I° Movement and Do-Insertion in Optimality Theory. To
appear in Legendre, Grimshaw, & Vikner (Eds.): Optimality-theoretic Syntax,
Cambridge, Mass.: MIT Press
Wexler, Ken (1994) Optional infinitives, head movement and the economy of
derivations. In Lightfoot, David & Norbert Hornstein (eds.) Verb Movement,
Cambridge: Cambridge University Press, 305-350
K. R. C. 2001 113
11 Appendix A: Sentence-to-Picture Matching Test

1. Pigen skubber drengen 1. Pigen skubber drengen
2. Drengen skubber pigen
3. Drengen skubbes af pigen
4. Pigen skubbes af drengen
5. Vis mig pigen som skubber drengen
6. Vis mig drengen som skubber pigen
7. Det er pigen som skubber drengen
8. Det er drengen som skubber pigen
9. Drengen bliver skubbet af pigen
10. Pigen bliver skubbet af drengen
11. Vis mig drengen som pigen skubber
12. Vis mig pigen som drengen skubber
13. Det er drengen som pigen skubber
14. Det er pigen som drengen skubber
15. Pigen som skubber drengen er sur
16. Drengen som skubber pigen er sur
17. Drengen som pigen skubber er sur
18. Pigen som drengen skubber er sur
2. Pigen kysser drengen 1. Pigen kysser drengen
2. Drengen kysser pigen
3. Drengen kysses af pigen
4. Pigen kysses af drengen
5. Vis mig pigen som kysser drengen
6. Vis mig drengen som kysser pigen
7. Det er pigen som kysser drengen
8. Det er drengen som kysser pigen
9. Drengen bliver kysset af pigen
10. Pigen bliver kysset af drengen
11. Vis mig drengen som pigen kysser
12. Vis mig pigen som drengen kysser
13. Det er drengen som pigen kysser
14. Det er pigen som drengen kysser
15. Pigen som kysser drengen er glad
16. Drengen som kysser pigen er glad
17. Drengen som pigen kysser er glad
18. Pigen som drengen kysser er glad
3. Hunden bider katten 1. Hunden bider katten
2. Katten bider hunden
3. Katten bides af hunden
4. Hunden bides af katten
5. Vis mig hunden som bider katten
6. Vis mig katten som bider hunden
7. Det er hunden som bider katten
8. Det er katten som bider hunden
9. Katten bliver bidt af hunden
10. Hunden bliver bidt af katten
11. Vis mig katten som hunden bider
12. Vis mig hunden som katten bider
13. Det er katten som hunden bider
14. Det er hunden som katten bider
15. Hunden som bider katten er sort
16. Katten som bidder hunden er sort
17. Katten som hunden bider er sort
18. Hunden som katten bider er sort
K. R. C. 2001 114

4. Pigen løfter drengen 1. Pigen løfter drengen
2. Drengen løfter pigen
3. Drengen løftes af pigen
4. Pigen løftes af drengen
5. Vis mig pigen som løfter drengen
6. Vis mig drengen som løfter pigen
7. Det er pigen som løfter drengen
8. Det er drengen som løfter pigen
9. Drengen bliver løftet af pigen
10. Pigen bliver løftet af drengen
11. Vis mig drengen som pigen løfter
12. Vis mig pigen som drengen løfter
13. Det er drengen som pigen løfter
14. Det er pigen som drengen løfter
15. Pigen som løfter drengen er glad
16. Drengen som løfter pigen er glad
17. Drengen som pigen løfter er glad
18. Pigen som drengen løfter er glad
5. Pigen slår drengen 1. Pigen slår drengen
2. Drengen slår pigen
3. Drengen slås af pigen
4. Pigen slås af drengen
5. Vis mig pigen som slår drengen
6. Vis mig drengen som slår pigen
7. Det er pigen som slår drengen
8. Det er drengen som slår pigen
9. Drengen bliver slået af pigen
10. Pigen bliver slået af drengen
11. Vis mig drengen som pigen slår
12. Vis mig pigen som drengen slår
13. Det er drengen som pigen slår
14. Det er pigen som drengen slår
15. Pigen som slår drengen er sur
16. Drengen som slår pigen er sur
17. Drengen som pigen slår er sur
18. Pigen som drengen slår er sur
6. Hunden snuser til katten 1. Hunden snuser til katten
2. Katten snuser til hunden
3. Katten bliver snuset til af katten
4. Hunden bliver snuset til af katten

K. R. C. 2001 115

7. Pigen peger på drengen 1. Pigen peger på drengen
2. Drengen peger på pigen
3. Drengen bliver peget på af pigen
4. Pigen bliver peget på af drengen
5. Pigen griner ad drengen
6. Drengen griner ad pigen
7. Drengen bliver grinet ad af pigen
8. Pigen bliver grinet ad af drengen
8. Pigen smiler til drengen 1. Pigen smiler til drengen
2. Drengen smiler til pigen
3. Drengen blivet smilet til af pigen
4. Pigen bliver smilet til af drengen
9. Pigen vinker til drengen 1. Pigen vinker til drengen
2. Drengen vinker til pigen
3. Drengen bliver vinket til af pigen
4. Pigen bliver vinket til af drengen

K. R. C. 2001 116

10. Drengen er inspireret af pigen 1. Drengen er inspireret af pigen
2. Pigen er inspireret af drengen
11. Hunden er irriteret på katten 1. Hunden er irriteret på katten
2. Katten er irriteret på hunden
3. Hunden frygter katten
4. Katten frygter hunden
5. Katten frygtes af hunden
6. Hunden frygtes af katten
12. Hunden er forbavset over katten 1. Hunden er forbavset over katten
2. Katten er forbavset over hunden
K. R. C. 2001 117

13. Drengen elsker pigen 1. Drengen elsker pigen
2. Pigen elsker drengen
3. Pigen elskes af drengen
4. Drengen elskes af pigen
5. Drengen er begejstret for pigen
6. Pigen er begejstret for drengen
7. Drengen synes om pigen
8. Pigen synes om drengen
9. Drengen er interesseret i pigen
10. Pigen er interesseret i drengen
11. Drengen beundrer pigen
12. Pigen beundrer drengen
13. Pigen beundres af drengen
14. Drengen beundres af pigen
14. Drengen hader pigen 1. Drengen hader pigen
2. Pigen hader drengen
3. Pigen hades af drengen
4. Drengen hades af pigen

15. Drengen væmmes ved pigen 1. Drengen væmmes ved pigen
2. Pigen væmmes ved drengen
3. Drengen afskyr pigen
4. Pigen afskyr drengen
5. Pigen afskys af drengen
6. Drengen afskys af pigen

K. R. C. 2001 118
12 Appendix B: Sentence Types and Tokens

1. Pigen skubber drengen
2. Drengen skubber pigen
3. Pigen kysser drengen
4. Drengen kysser pigen
5. Hunden bider katten
6. Katten bider hunden
7. Pigen løfter drengen
8. Drengen løfter pigen
9. Pigen slår drengen
10. Drengen slår pigen
1. Simple Active:

sub verb obj
Agent Theme






sub verb prep obj
Agent Theme


11. Hunden snuser til katten
12. Katten snuser til hunden
13. Pigen peger på drengen
14. Drengen peger på pigen
15. Pigen griner ad drengen
16. Drengen griner ad pigen
17. Pigen smiler til drengen
18. Drengen smiler til pigen
19. Pigen vinker til drengen
Drengen vinker til pigen
2. Sub-Sub Relative:

sub1 [sub1 verb obj]
Agent Theme
1. Pigen som skubber drengen er sur
2. Drengen som skubber pigen er sur
3. Pigen som kysser drengen er glad
4. Drengen som kysser pigen er glad
5. Hunden som bider katten er sort
6. Katten som bidder hunden er sort
7. Pigen som løfter drengen er glad
8. Drengen som løfter pigen er glad
9. Pigen som slår drengen er sur
10. Drengen som slår pigen er sur
3. Obj-Sub Relative:

obj1 [sub1 verb obj]
Agent Theme
1. Vis mig pigen som skubber drengen
2. Vis mig drengen som skubber pigen
3. Vis mig pigen som kysser drengen
4. Vis mig drengen som kysser pigen
5. Vis mig hunden som bider katten
6. Vis mig katten som bider hunden
7. Vis mig pigen som løfter drengen
8. Vis mig drengen som løfter pigen
9. Vis mig pigen som slår drengen
10. Vis mig drengen som slår pigen
4. Subject Cleft:

N° °° °1 [sub1 verb obj]
Agent Theme
1. Det er pigen som skubber drengen
2. Det er drengen som skubber pigen
3. Det er pigen som kysser drengen
4. Det er drengen som kysser pigen
5. Det er hunden som bider katten
6. Det er katten som bider hunden
7. Det er pigen som løfter drengen
8. Det er drengen som løfter pigen
9. Det er pigen som slår drengen
10. Det er drengen som slår pigen
K. R. C. 2001 119

Lexical Passive:

sub verb prep obj
Experiencer Theme

Drengen synes om pigen
Pigen synes om drengen
Drengen væmmes ved pigen
Pigen væmmes ved katten
Adjectival Passive:

sub copula adj oblique
Experiencer Theme
1. Drengen er interesseret i pigen
2. Pigen er interesseret i drengen
3. Drengen er inspireret af pigen
4. Pigen er inspireret af drengen
5. Drengen er begejstret for pigen
6. Pigen er begejstret for drengen
7. Hunden er irriteret på katten
8. Katten er irriteret på hunden
9. Hunden er forbavset over katten
10. Katten er forbavset over hunden
Psychological predicates:

sub verb obj
Experiencer Theme
1. Drengen beundrer pigen
2. Pigen beundrer drengen
3. Drengen elsker pigen
4. Pigen elsker drengen
5. Drengen hader pigen
6. Pigen hader drengen
7. Drengen afskyr pigen
8. Pigen afskyr drengen
9. Hunden frygter katten
10. Katten frygter hunden
1. Drengen bliver skubbet af pigen
2. Pigen bliver skubbet af drengen
3. Drengen bliver kysset af pigen
4. Pigen bliver kysset af drengen
5. Katten bliver bidt af hunden
6. Hunden bliver bidt af katten
7. Drengen bliver løftet af pigen
8. Pigen bliver løftet af drengen
9. Drengen bliver slået af pigen
10. Pigen bliver slået af drengen
"Blive" Passive:

sub aux verb prep oblique
Theme Agent






sub aux verb prep prep oblique
Theme Agent
11. Katten bliver snuset til af katten
12. Hunden bliver snuset til af katten
13. Drengen bliver peget på af pigen
14. Pigen bliver peget på af drengen
15. Drengen bliver grinet ad af pigen
16. Pigen bliver grinet ad af drengen
17. Drengen blivet smilet til af pigen
18. Pigen bliver smilet til af drengen
19. Drengen bliver vinket til af pigen
Pigen bliver vinket til af drengen
K. R. C. 2001 120

Sub-Obj Relative:

sub1 [obj1 sub verb]
Theme Agent
1. Drengen som pigen skubber er høj
2. Pigen som drengen skubber er høj
3. Drengen som pigen kysser er glad
4. Pigen som drengen kysser er glad
5. Katten som hunden bider er sort
6. Hunden som katten bider er sort
7. Drengen som pigen løfter er glad
8. Pigen som drengen løfter er glad
9. Drengen som pigen slår er sur
10. Pigen som drengen slår er sur
Obj-Obj Relative:

obj1 [obj1 sub verb]
Theme Agent
1. Vis mig drengen som pigen skubber
2. Vis mig pigen som drengen skubber
3. Vis mig drengen som pigen kysser
4. Vis mig pigen som drengen kysser
5. Vis mig katten som hunden bider
6. Vis mig hunden som katten bider
7. Vis mig drengen som pigen løfter
8. Vis mig pigen som drengen løfter
9. Vis mig drengen som pigen slår
10. Vis mig pigen som drengen slår
Object cleft:

N° °° °1 [obj1 sub verb]
Theme Agent
1. Det er drengen som pigen skubber
2. Det er pigen som drengen skubber
3. Det er drengen som pigen kysser
4. Det er pigen som drengen kysser
5. Det er katten som hunden bider
6. Det er hunden som katten bider
7. Det er drengen som pigen løfter
8. Det er pigen som drengen løfter
9. Det er drengen som pigen slår
10. Det er pigen som drengen slår
Verbal Passive:

sub verb prep oblique
Theme Agent
1. Drengen skubbes af pigen
2. Pigen skubbes af drengen
3. Drengen kysses af pigen
4. Pigen kysses af drengen
5. Katten bides af hunden
6. Hunden bides af katten
7. Drengen løftes af pigen
8. Pigen løftes af drengen
9. Drengen slås af pigen
10. Pigen slås af drengen
Psychological Passive:

sub verb prep oblique
Theme Experiencer
1. Pigen beundres af drengen
2. Drengen beundres af pigen
3. Pigen elskes af drengen
4. Drengen elskes af pigen
5. Pigen hades af drengen
6. Drengen hades af pigen
7. Pigen afskys af drengen
8. Drengen afskys af pigen
9. Katten frygtes af hunden
10. Hunden frygtes af katten

K. R. C. 2001 121
13 Appendix C: Repetition Test: Structure Distribution


Legend:

adj.pas: Adjectival Passive PAST_PF+: Past Double Perfect
lex.pas: Lexical Passive PAST_PF: Past Perfect
obj.clf: Object Cleft PAST: Past
o_o.rel: Object-Object Relative PRES_PF+: Present Double Perfect
o_s.rel: Object-Subject Relative PRES_PF: Present Perfect
psy.pas: Psychological Passive PRES: Present
psy.ver: Psychological Verb FUTURE: Future
sim.pas: Simple Active
sub.clf: Subject Cleft
s_o.rel: Subject-Object Relative
sim.pas: Simple ("blive") Passive
s_s.rel: Subject-Subject Relative
vrb.pas: Verbal Passive


adj.pas FUTURE Hunden bliver temmelig irriteret på katten
adj.pas PAST Hunden var ret forbavset over katten
adj.pas PAST_PF Pigen havde nok været inspireret af drengen
adj.pas PRES_PF Pigen har ofte været begejstret for drengen
adj.pas PRESENT Drengen er meget begejstret for pigen

lex.pas FUTURE Drengen vil nok synes om pigen
lex.pas PAST Pigen væmmedes nok ved drengen
lex.pas PAST_PF Pigen havde nok syntes om drengen
lex.pas PAST_PF+ Drengen havde nok haft syntes om pigen
lex.pas PRES_PF Pigen har nok væmmedes ved drengen
lex.pas PRES_PF+ Pigen har nok haft syntes om drengen
lex.pas PRESENT Drengen synes vist om pigen

o_o.rel FUTURE Vis mig katten som hunden sikkert vil bide
o_o.rel PAST Vis mig drengen som pigen vredt skubbede
o_o.rel PAST_PF Vis mig pigen som drengen ofte havde slået
o_o.rel PAST_PF+ Vis mig pigen som drengen vredt havde haft slået
o_o.rel PRES_PF Vis mig pigen som drengen sikkert har kysset
o_o.rel PRES_PF+ Vis mig pigen som drengen sikkert har haft kysset
o_o.rel PRESENT Vis mig drengen som pigen tit kysser

o_s.rel FUTURE Vis mig hunden som først vil bide katten
o_s.rel PAST Vis mig drengen som vredt skubbede pigen
o_s.rel PAST_PF Vis mig pigen som vredt havde slået drengen
o_s.rel PAST_PF+ Vis mig pigen som vist havde haft slået drengen
o_s.rel PRES_PF Vis mig pigen som nemt har løftet drengen
o_s.rel PRES_PF+ Vis mig pigen som jo har haft løftet drengen
o_s.rel PRESENT Vis mig drengen som blidt kysser pigen
K. R. C. 2001 122
obj.clf FUTURE Det er katten som hunden nok vil bide
obj.clf PAST Det er drengen som pigen vist slog
obj.clf PAST_PF Det var pigen som drengen vist havde skubbet
obj.clf PAST_PF+ Det var pigen som drengen vist havde haft skubbet
obj.clf PRES_PF Det er pigen som drengen nok har løftet
obj.clf PRES_PF+ Det er pigen som drengen nok har haft løftet
obj.clf PRESENT Det er drengen som pigen tit kysser

psy.pas FUTURE Hunden vil sikkert blive frygtet af katten
psy.pas PAST Drengen elskedes sikkert af pigen
psy.pas PAST_PF Drengen var vist blevet beundret af pigen
psy.pas PRES_PF Pigen er nok blevet beundret af drengen
psy.pas PRESENT Pigen elskes jo af drengen

psy.ver FUTURE Katten vil sikkert frygte hunden
psy.ver PAST Drengen elskede nok pigen
psy.ver PAST_PF Pigen havde nok elsket drengen
psy.ver PAST_PF+ Pigen havde nok haft elsket drengen
psy.ver PRES_PF Pigen har vist afskyet drengen
psy.ver PRES_PF+ Pigen har vist haft afskyet drengen
psy.ver PRESENT Drengen afskyr vist pigen

s_o.rel FUTURE Katten som hunden sikkert vil bide er sort
s_o.rel PAST Drengen som pigen vredt slog er sur
s_o.rel PAST_PF Pigen som drengen ofte havde skubbet var sur
s_o.rel PAST_PF+ Pigen som drengen ofte havde haft skubbet var sur
s_o.rel PRES_PF Pigen som drengen vist har kysset er glad
s_o.rel PRES_PF+ Pigen som drengen vist har haft kysset er glad
s_o.rel PRESENT Drengen som pigen gerne kysser er glad

s_s.rel FUTURE Hunden som jo vil bide katten er sort
s_s.rel PAST Drengen som jo skubbede pigen er sur
s_s.rel PAST_PF Pigen som nok havde slået drengen var sur
s_s.rel PAST_PF+ Pigen som nok havde haft slået drengen var sur
s_s.rel PRES_PF Pigen som tit har kysset drengen er glad
s_s.rel PRES_PF+ Pigen som nok har haft kysset drengen er glad
s_s.rel PRESENT Drengen som tit kysser pigen er glad

sim.act FUTURE Drengen vil sikkert slå pigen
sim.act FUTURE Pigen vil ikke løfte drengen
sim.act PAST Drengen pegede vist på pigen
sim.act PAST Pigen kyssede ikke drengen
sim.act PAST_PF Hunden havde nok snuset til katten
sim.act PAST_PF Pigen havde ikke vinket til drengen
sim.act PAST_PF+ Hunden havde nok haft snuset til katten
sim.act PAST_PF+ Pigen havde ikke haft vinket til drengen
sim.act PRES_PF Drengen har tit smilet til pigen
sim.act PRES_PF Pigen har ikke skubbet drengen
sim.act PRES_PF+ Drengen har tit haft smilet til pigen
sim.act PRES_PF+ Pigen har ikke haft skubbet drengen
sim.act PRESENT Drengen kysser gerne pigen
sim.act PRESENT Katten snuser ikke til hunden
K. R. C. 2001 123
sim.pas FUTURE Drengen vil sikkert blive slået af pigen
sim.pas FUTURE Pigen vil ikke blive peget på af drengen
sim.pas PAST Drengen blev ofte peget på af pigen
sim.pas PAST Pigen blev ikke kysset af drengen
sim.pas PAST_PF Hunden var nok blevet bidt af katten
sim.pas PAST_PF Pigen var ikke blevet smilet til af drengen
sim.pas PRES_PF Drengen er sikkert blevet smilet til af pigen
sim.pas PRES_PF Pigen er ikke blevet slået af drengen
sim.pas PRESENT Drengen bliver tit grinet ad af pigen
sim.pas PRESENT Katten bliver ikke bidt af hunden

sub.clf FUTURE Det er hunden som sikkert vil bide katten
sub.clf PAST Det er drengen som vredt skubbede pigen
sub.clf PAST_PF Det var pigen som vist havde skubbet drengen
sub.clf PAST_PF+ Det var pigen som vist havde haft skubbet drengen
sub.clf PRES_PF Det er pigen som nemt har løftet drengen
sub.clf PRES_PF+ Det er pigen som jo har haft løftet drengen
sub.clf PRESENT Det er drengen som gerne kysser pigen

vrb.pas PAST Drengen kyssedes gerne af pigen
vrb.pas PAST Drengen skubbedes jo af pigen
vrb.pas PRESENT Drengen løftes tit af pigen
vrb.pas PRESENT Hunden bides vist af katten

Excluded due to ungrammaticality:

adj.pas PAST_PF+ *Pigen havde nok haft været inspireret af drengen
adj.pas PRES_PF+ *Pigen har vist haft været begejstret for drengen
psy.pas PAST_PF+ *Drengen havde vist været blevet beundret af pigen
psy.pas PRES_PF+ *Pigen har nok været blevet beundret af drengen
sim.pas PAST_PF+ *Hunden havde nok været blevet bidt af katten
sim.pas PRES_PF+ *Drengen har sikkert været blevet smilet til af pigen
vrb.pas FUTURE *Drengen vil senere kysses af pigen
14

vrb.pas PAST_PF *Drengen havde haft skubbedes jo af pigen
vrb.pas PAST_PF+ *Hunden havde vist haft bides af katten
vrb.pas PRES_PF *Drengen har løftes tit af pigen
vrb.pas PRES_PF+ *Hunden har vist bides af katten

Extra added to reach 100 sentences:

sim.act PAST Drengene jagtede tit pigerne
sim.act PAST Pigerne drillede ofte drengene
sim.act PAST Drengene kyssede gerne pigerne
sim.act PAST Pigerne slog ofte drengene
sim.act PAST Hundene knurrede vredt af kattene
sim.act PAST Kattene væsede vredt af hundene


14
The sentence is grammatical if the meaning of “vil” (will) is volition instead of future.

K. R. C. 2001 124
14 Appendix D: TJ's Comprehension Test Data


Type Order Token Picture Performace
+1000 >=correct *=fail
adj.passive 1015. Hunden er irriteret på katten [A] >B] [ ] [*]
adj.passive 1018. Drengen er interesseret i pigen >A] [B] [ ] [ ]
adj.passive 1041. Hunden er forbavset over katten >A] [B] [ ] [ ]
adj.passive 1043. Pigen er begejstret for drengen [A] >B] [ ] [ ]
adj.passive 1060. Pigen er inspireret af drengen [A] >B] [ ] [*]
adj.passive 1085. Drengen er begejstret for pigen >A] [B] [ ] [ ]
adj.passive 1097. Katten er forbavset over hunden [A] >B] [ ] [ ]
adj.passive 1102. Katten er irriteret på hunden >A] [B] [ ] [ ]
adj.passive 1128. Pigen er interesseret i drengen [A] >B] [ ] [ ]
adj.passive 1144. Drengen er inspireret af pigen >A] [B] [ ] [ ]
lex.passive 1033. Pigen væmmes ved drengen >A] [B] [ ] [ ]
lex.passive 1036. Drengen synes om pigen >A] [B] [ ] [ ]
lex.passive 1131. Drengen væmmes ved pigen [A] >B] [*] [*]
lex.passive 1137. Pigen synes om drengen [A] >B] [ ] [ ]
obj.cleft 1031. Det er drengen som pigen skubber >A] [B] [ ] [*]
obj.cleft 1042. Det er pigen som drengen slår >A] [B] [ ] [*]
obj.cleft 1052. Det er drengen som pigen løfter [A] >B] [ ] [ ]
obj.cleft 1055. Det er pigen som drengen løfter >A] [B] [ ] [ ]
obj.cleft 1075. Det er hunden som katten bider [A] >B] [ ] [ ]
obj.cleft 1083. Det er pigen som drengen kysser >A] [B] [ ] [ ]
obj.cleft 1084. Det er drengen som pigen slår [A] >B] [ ] [ ]
obj.cleft 1092. Det er drengen som pigen kysser [A] >B] [ ] [ ]
obj.cleft 1107. Det er pigen som drengen skubber [A] >B] [ ] [ ]
obj.cleft 1118. Det er katten som hunden bider >A] [B] [ ] [ ]
obj.obj.relative 1008. Vis mig katten som hunden bider >A] [B] [ ] [ ]
obj.obj.relative 1011. Vis mig pigen som drengen slår >A] [B] [ ] [ ]
obj.obj.relative 1019. Vis mig pigen som drengen løfter >A] [B] [ ] [ ]
obj.obj.relative 1035. Vis mig drengen som pigen løfter [A] >B] [ ] [ ]
obj.obj.relative 1088. Vis mig drengen som pigen slår [A] >B] [ ] [ ]
obj.obj.relative 1105. Vis mig pigen som drengen kysser >A] [B] [ ] [ ]
obj.obj.relative 1113. Vis mig drengen som pigen kysser [A] >B] [ ] [ ]
obj.obj.relative 1132. Vis mig pigen som drengen skubber [A] >B] [ ] [ ]
obj.obj.relative 1135. Vis mig drengen som pigen skubber >A] [B] [ ] [ ]
obj.obj.relative 1139. Vis mig hunden som katten bider [A] >B] [ ] [ ]
obj.sub.relative 1005. Vis mig drengen som slår pigen >A] [B] [ ] [ ]
obj.sub.relative 1022. Vis mig pigen som kysser drengen [A] >B] [ ] [ ]
obj.sub.relative 1040. Vis mig hunden som bider katten >A] [B] [ ] [ ]
obj.sub.relative 1048. Vis mig drengen som skubber pigen [A] >B] [ ] [ ]
obj.sub.relative 1063. Vis mig drengen som kysser pigen >A] [B] [ ] [ ]
obj.sub.relative 1069. Vis mig pigen som skubber drengen >A] [B] [ ] [ ]
obj.sub.relative 1079. Vis mig pigen som løfter drengen [A] >B] [ ] [ ]
obj.sub.relative 1100. Vis mig katten som bider hunden [A] >B] [ ] [ ]
obj.sub.relative 1120. Vis mig pigen som slår drengen [A] >B] [ ] [ ]
obj.sub.relative 1126. Vis mig drengen som løfter pigen >A] [B] [ ] [ ]
psych.passive 1001. Hunden frygtes af katten [A] >B] [*] [ ]
psych.passive 1020. Pigen afskys af drengen [A] >B] [ ] [ ]
psych.passive 1027. Drengen elskes af pigen [A] >B] [ ] [ ]
psych.passive 1056. Katten frygtes af hunden >A] [B] [ ] [*]
psych.passive 1059. Pigen hades af drengen [A] >B] [ ] [ ]
psych.passive 1064. Drengen beundres af pigen [A] >B] [ ] [ ]
psych.passive 1101. Pigen elskes af drengen >A] [B] [ ] [ ]
psych.passive 1133. Drengen hades af pigen >A] [B] [ ] [ ]
psych.passive 1140. Drengen afskys af pigen >A] [B] [ ] [ ]
psych.passive 1141. Pigen beundres af drengen >A] [B] [ ] [ ]
psych.predicate 1004. Drengen hader pigen [A] >B] [ ] [ ]
psych.predicate 1017. Pigen elsker drengen [A] >B] [→] [ ]
psych.predicate 1026. Hunden frygter katten >A] [B] [ ] [ ]
psych.predicate 1032. Katten frygter hunden [A] >B] [ ] [ ]
psych.predicate 1045. Pigen beundrer drengen [A] >B] [→] [ ]
psych.predicate 1047. Drengen elsker pigen >A] [B] [ ] [ ]
psych.predicate 1077. Drengen beundrer pigen >A] [B] [ ] [ ]
psych.predicate 1086. Pigen afskyr drengen >A] [B] [ ] [ ]
psych.predicate 1094. Pigen hader drengen >A] [B] [ ] [ ]
psych.predicate 1103. Drengen afskyr pigen [A] >B] [ ] [ ]
sim.active 1007. Pigen løfter drengen [A] >B] [ ] [ ]
sim.active 1013. Pigen griner ad drengen >A] [B] [ ] [ ]
sim.active 1016. Drengen griner ad pigen [A] >B] [ ] [ ]
sim.active 1024. Drengen vinker til pigen >A] [B] [ ] [ ]
K. R. C. 2001 125
sim.active 1044. Katten bider hunden [A] >B] [ ] [ ]
sim.active 1046. Drengen smiler til pigen >A] [B] [ ] [ ]
sim.active 1051. Katten snuser til hunden [A] >B] [ ] [ ]
sim.active 1068. Pigen slår drengen [A] >B] [ ] [ ]
sim.active 1076. Hunden snuser til katten >A] [B] [ ] [ ]
sim.active 1078. Pigen peger på drengen >A] [B] [ ] [ ]
sim.active 1080. Pigen smiler til drengen [A] >B] [ ] [ ]
sim.active 1095. Pigen kysser drengen [A] >B] [ ] [ ]
sim.active 1096. Hunden bider katten >A] [B] [ ] [ ]
sim.active 1098. Drengen løfter pigen >A] [B] [ ] [ ]
sim.active 1106. Drengen peger på pigen [A] >B] [ ] [ ]
sim.active 1108. Pigen skubber drengen >A] [B] [ ] [ ]
sim.active 1109. Pigen vinker til drengen [A] >B] [ ] [ ]
sim.active 1111. Drengen slår pigen >A] [B] [ ] [ ]
sim.active 1117. Drengen skubber pigen [A] >B] [ ] [ ]
sim.active 1138. Drengen kysser pigen >A] [B] [ ] [ ]
sim.passive 1014. Hunden bliver bidt af katten [A] >B] [ ] [ ]
sim.passive 1025. Drengen bliver løftet af pigen [A] >B] [ ] [ ]
sim.passive 1053. Drengen bliver skubbet af pigen >A] [B] [ ] [ ]
sim.passive 1057. Drengen blivet smilet til af pigen [A] >B] [ ] [ ]
sim.passive 1058. Pigen bliver løftet af drengen >A] [B] [ ] [ ]
sim.passive 1065. Pigen bliver kysset af drengen >A] [B] [ ] [ ]
sim.passive 1066. Pigen bliver smilet til af drengen >A] [B] [ ] [ ]
sim.passive 1071. Pigen bliver vinket til af drengen >A] [B] [ ] [ ]
sim.passive 1087. Drengen bliver grinet ad af pigen >A] [B] [ ] [ ]
sim.passive 1090. Katten bliver snuset til af hunden >A] [B] [ ] [ ]
sim.passive 1104. Pigen bliver skubbet af drengen [A] >B] [ ] [ ]
sim.passive 1110. Pigen bliver slået af drengen >A] [B] [ ] [ ]
sim.passive 1114. Katten bliver bidt af hunden >A] [B] [ ] [ ]
sim.passive 1115. Hunden bliver snuset til af katten [A] >B] [ ] [ ]
sim.passive 1116. Pigen bliver grinet ad af drengen [A] >B] [ ] [ ]
sim.passive 1125. Pigen bliver peget på af drengen [A] >B] [ ] [ ]
sim.passive 1127. Drengen bliver vinket til af pigen [A] >B] [ ] [ ]
sim.passive 1129. Drengen bliver peget på af pigen >A] [B] [ ] [ ]
sim.passive 1134. Drengen bliver slået af pigen [A] >B] [ ] [ ]
sim.passive 1142. Drengen bliver kysset af pigen [A] >B] [ ] [ ]
sub.cleft 1002. Det er katten som bider hunden [A] >B] [ ] [ ]
sub.cleft 1006. Det er drengen som løfter pigen >A] [B] [ ] [ ]
sub.cleft 1030. Det er drengen som skubber pigen [A] >B] [ ] [*]
sub.cleft 1054. Det er pigen som kysser drengen [A] >B] [ ] [ ]
sub.cleft 1067. Det er drengen som slår pigen >A] [B] [ ] [ ]
sub.cleft 1081. Det er hunden som bider katten >A] [B] [ ] [ ]
sub.cleft 1082. Det er pigen som skubber drengen >A] [B] [ ] [ ]
sub.cleft 1099. Det er pigen som løfter drengen [A] >B] [ ] [ ]
sub.cleft 1124. Det er drengen som kysser pigen >A] [B] [ ] [ ]
sub.cleft 1136. Det er pigen som slår drengen [A] >B] [ ] [ ]
sub.obj.relative 1009. Pigen som drengen løfter er glad >A] [B] [ ] [ ]
sub.obj.relative 1028. Pigen som drengen skubber er sur [A] >B] [*] [*]
sub.obj.relative 1034. Drengen som pigen løfter er glad [A] >B] [*] [ ]
sub.obj.relative 1050. Drengen som pigen kysser er glad [A] >B] [ ] [*]
sub.obj.relative 1061. Pigen som drengen kysser er glad >A] [B] [ ] [ ]
sub.obj.relative 1072. Drengen som pigen slår er sur [A] >B] [ ] [ ]
sub.obj.relative 1089. Hunden som katten bider er sort [A] >B] [*] [ ]
sub.obj.relative 1091. Pigen som drengen slår er sur >A] [B] [ ] [ ]
sub.obj.relative 1093. Katten som hunden bider er sort >A] [B] [ ] [*]
sub.obj.relative 1112. Drengen som pigen skubber er sur >A] [B] [ ] [ ]
sub.sub.relative 1003. Hunden som bider katten er sort >A] [B] [ ] [ ]
sub.sub.relative 1010. Pigen som løfter drengen er glad [A] >B] [ ] [ ]
sub.sub.relative 1021. Drengen som løfter pigen er glad >A] [B] [ ] [ ]
sub.sub.relative 1023. Drengen som slår pigen er sur >A] [B] [ ] [ ]
sub.sub.relative 1029. Drengen som skubber pigen er sur [A] >B] [ ] [ ]
sub.sub.relative 1038. Pigen som slår drengen er sur [A] >B] [ ] [ ]
sub.sub.relative 1039. Pigen som kysser drengen er glad [A] >B] [ ] [ ]
sub.sub.relative 1049. Pigen som skubber drengen er sur >A] [B] [ ] [ ]
sub.sub.relative 1062. Drengen som kysser pigen er glad >A] [B] [ ] [ ]
sub.sub.relative 1122. Katten som bider hunden er sort [A] >B] [ ] [ ]
verbal.passive 1012. Drengen kysses af pigen [A] >B] [ ] [ ]
verbal.passive 1037. Katten bides af hunden >A] [B] [ ] [ ]
verbal.passive 1070. Pigen slås af drengen >A] [B] [ ] [ ]
verbal.passive 1073. Drengen skubbes af pigen >A] [B] [ ] [ ]
verbal.passive 1074. Drengen slås af pigen [A] >B] [ ] [ ]
verbal.passive 1119. Hunden bides af katten [A] >B] [ ] [ ]
verbal.passive 1121. Pigen løftes af drengen >A] [B] [ ] [ ]
verbal.passive 1123. Pigen kysses af drengen >A] [B] [ ] [ ]
verbal.passive 1130. Pigen skubbes af drengen [A] >B] [ ] [ ]
verbal.passive 1143. Drengen løftes af pigen [A] >B] [ ] [ ]
K. R. C. 2001 126
15 Appendix E: TJ's Production Test Data

Legend:

AP: Adjectival Passive 1: Past Double Perfect
LP: Lexical Passive PAST infl. 2: Past Perfect
OC: Object Cleft 3: Past
OO: Object-Object Relative 4: Present Double Perfect
OS: Object-Subject Relative PRES. infl. 5: Present Perfect
PP: Psychological Passive 6: Present
PV: Psychological Verb 7: (Present) Future
SA: Simple Active
SC: Subject Cleft
SO: Subject-Object Relative
SP: Simple ("blive") Passive
SS: Subject-Subject Relative
VP: Verbal Passive


O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

00 SA 3 SA 3
Drengene jagtede tit pigerne
"Drengene jagtede tit pigerne"

01 P SA 3 SA 3
Kattene hvæsede vredt af hundene
"Kattene hvæsede vredt af hundene"

02 P LP 7 LP 7
Drengen vil nok synes om pigen
"Drengen vil nok synes om pigen"
OK
03 VP 6 VP 6
Hunden bides vist af katten
"Hunden bides vist af katten"
OK
04 SA 3 SA 3
Pigerne slog ofte drengene
"Pigerne slog ofte drengene"
OK
05 P AP 6 AP 6
Drengen er meget begejstret for pigen
" Drengene er meget begejstret for pigerne"

06 P SA 3 SA 3
Hundene knurrede vredt af kattene
"Hundene knurrede vredt af kattene"

07 PV 7 PV 7
Katten vil sikkert frygte hunden
"Katten vil øøh … øh katten vil øh måske
frygte hunden"

08 PV 6 PV 6
Drengen afskyr vist pigen
"Drengen afskyr vist pigen"

09 OC 6 OC 6
Det er drengen som pigen tit kysser
"Det er drengen som pigen tit kysser"

10 OO 5 OO 5
Vis mig pigen som drengen sikkert har kysset
" Vis mig pigen som drengen har kysset"

K. R. C. 2001 127
O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

11 P AP 5 AP 6
Pigen har ofte været begejstret for drengen
"Pigen er ofte begejstret for drengene"
-1
12 OC 5 OC 3
Det er pigen som drengen nok har løftet
"Det er pigen øh øøh drengen øh øh sikkert
løftede"
-1
13 P SA 1 SA 5
Hunden havde nok haft snuset til katten
"Hunden har ofte øh snuset til hunden"
-1
14 OS 6 ? ?
Vis mig drengen som blidt kysser pigen
"Vis mig drengen som øøh … … øøh øh ja
det er jeg kan ikke huske det drengen som
ofte … … ja det kan jeg ikke huske"
(Vis mig drengen som ofte)
-1
15 SO 4 ? ?
Pigen som drengen vist har haft kysset er glad
"Pigen som ofte har … øh døhdøhdøh
drengen"
(Pigen som ofte har _ drengen)
-1 -1
16 P SP 6 SP 6
Drengen bliver tit grinet ad af pigen
"Drengen bliver tit grinet ad pigen"
-U
17 N SP 5 SP 5
Pigen er ikke blevet slået af drengen
"Pigen er ikke blevet slået af drengen"

18 SS 1 SO 5
Pigen som nok havde haft slået drengen var
sur
"Pigen som øøh måske har øh blevet slået af
drengen øh og det kan jeg ikke huske det
sidste"
(Pigen som måske er blevet slået af drengen
[var sur])
-1


19 PP 5 PP 5
Pigen er nok blevet beundret af drengen
"Pigen er øh er øøh ofte blevet beundret af
drengen"

20 N SA 5 SP 5
Pigen har ikke skubbet drengen
"Pigen har ikke blevet skubbet af drengen"
(Pigen er ikke blevet skubbet af drengen)

21 P SA 4 SA 5
Drengen har tit haft smilet til pigen
"Drengen øh har tit smilet til pigen"
-1
22 OS 3 OS 3
Vis mig drengen som vredt skubbede pigen
"Vis mig drengen som vredt skubbede med
pigen"
+G
23 OS 2 OS 3
Vis mig pigen som vredt havde slået drengen
"Vis mig pigen som vredt slåede drengen"
-1
24 SC 6 SC 6
Det er drengen som gerne kysser pigen
"Det er drengen som gerne kysser pigen"

K. R. C. 2001 128

O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

25 P SA 5 SA 5
Drengen har tit smilet til pigen
"Drengen har tit smilet til pigen"

26 PV 1 PV 4
Pigen havde nok haft elsket drengen
"Pigen har nok øh haft elsket drengen"

27 SS 7 SS 3
Hunden som jo vil bide katten er sort
"Hunden som ville bide katten var sort"

28 SA 7 SA 7
Drengen vil sikkert slå pigen
"Drengen vil sikkert slå pigen"

29 OC 3 SC 6
Det er drengen som pigen vist slog
"Det er drengen som vist slår pigen"

30 PV 5 PV 5
Pigen har vist afskyet drengen
"Pigen øh har vist afskyet drengen"

31 P AP 2 SP 5
Pigen havde nok været inspireret af drengen
"Pigen øøh har ofte øøh … har ofte øøh …
kigget på pigen"
+G
32 SC 3 SC 3
Det er drengen som vredt skubbede pigen
"Drengen som vredt skubbede pigen"
((Det er) drengen som vredt skubbede pigen)

33 SO 2 AP 3
Pigen som drengen ofte havde skubbet var sur
"Pigen var døøh … ofte … skrækket af …
drengen"
(Pigen var ofte forskrækket over drengen)
+G
34 OO 7 OO 3
Vis mig katten som hunden sikkert vil bide
"Vis mig katten som øh … bed øh hunden"
-1
35 P LP 1 ? ?
Drengen havde nok haft syntes om pigen
"Drengen øh … har ofte øh … døhdøh
kvinden"
(Drengen har ofte bla-bla kvinden)
-1 -1 -1
36 P SP 3 SP 3
Drengen blev ofte peget på af pigen
"Drengen blev ofte peget på af pigen"

37 SP 7 SP 7
Drengen vil sikkert blive slået af pigen
"Drengen vil øøh … øh øøh måske blive slået
af pigen"
(Drengen vil måske blive slået af pigen)

38 OO 1 OS 3
Vis mig pigen som drengen vredt havde haft
slået
"Vis mig pigen som øøh vredt slog slåede
drengen"
-2
39 P AP 3 AP 3
Hunden var ret forbavset over katten
"Hunden var ret forbavset over katten"

K. R. C. 2001 129

O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

40 PV 2 SA 3
Pigen havde nok elsket drengen
"Pigen var ofte øh nej pigen øøh elskede ofte
drengen"
(Pigen elskede ofte drengen)
-1
41 N SA 4 SA 5
Pigen har ikke haft skubbet drengen
"Pigen har ikke skubbet drengen"
-1
42 VP 6 VP 6
Drengen løftes tit af pigen
"Drengen løftes tit af pigen"
OK
43 P LP 4 ? ?
Pigen har nok haft syntes om drengen
"Pigen har ofte øøh … øøh … ikke tænke
men pigen er ofte … … … … ofte øh ja det
ka du ka j' ikke huske det sidste"
(Pigen har ofte)
-1 -1 -1
44 P LP 3 ? 3
Pigen væmmedes nok ved drengen
"Pigen vænnede sig ofte øh øh af drengen"
(pigen vænnede sig ofte [ACTIVE
REFLEXIVE] af drengen [PASSIVE])
-1

45 SS 4 SS 3
Pigen som nok har haft kysset drengen er
glad
"Pigen som øh ofte øh kyssede drengen er
døhdøh"
(Pigen som ofte kyssede drengen er bla-bla)
-2
46 N SA 7 SA 7
Pigen vil ikke løfte drengen
"Pigen vil ikke løfte drengen"

47 P N SA 6 SA 6
Katten snuser ikke til hunden
"Katten snuser ikke til hunden"

48 SC 2 SC 3
Det var pigen som vist havde skubbet drengen
"Det var pigen som øøh … skubbede til
drengen"
-1 +G
49 OO 3 OO 3
Vis mig drengen som pigen vredt skubbede
"Vis mig pigen som drengen skubbede …
vredt skubbede"
(Vis mig pigen som drengen vredt skubbede)

50 PP 6 PP 3
Pigen elskes jo af drengen
"Pigen elskede jo af drengen"
-1
51 N SP 3 SP 3
Pigen blev ikke kysset af drengen
"Pigen blev ikke kysset af drengen"

52 P LP 2 ? ?
Pigen havde nok syntes om drengen
"Pigen har ofte øh … af drengen . Pigen … øh
nej det kan jeg ikke"
(Pigen har ofte _ af drengen)
-1 -1
K. R. C. 2001 130
O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

53 SC 1 SC 3
Det var pigen som vist havde haft skubbet
drengen
"Det var pigen som ofte skubbede drengen"
-2
54 SO 7 OS 3
Katten som hunden sikkert vil bide er sort
"Katten ville bide hunden som var sort"

55 P LP 5 ? ?
Pigen har nok væmmedes ved drengen
"Pigen var of' nej … væmmede drengen.
Pigen øøh var ofte nej. Pigen ofte øøh
væmmede drengen"
(Pigen ofte væmmede drengen)
-1 -G -1
56 SA 6 SA 6
Drengen kysser gerne pigen
"Drengen kysser gerne pigen"

57 SA 3 SA 6
Pigerne drillede ofte drengene
"Pigerne drillede ofte drengene"

58 SS 5 SS 3
Pigen som tit har kysset drengen er glad
"Pigen som tit kyssede drengen var glad"
-1
59 PV 4 PV 5
Pigen har vist haft afskyet drengen
"Pigen har vist øøh har vist øøh afskyet nej
var vist afskyet drengen"
(Pigen har vist afskyet drengen)
-1
60 P AP 7 AP 7
Hunden bliver temmelig irriteret på katten
"Hunden bliver temmelig irriteret på katten"

61 PV 3 PV 3
Drengen elskede nok pigen
"Drengen elskede nok pigen"

62 P N SP 2 SP 2
Pigen var ikke blevet smilet til af drengen
"Pigen var ikke blevet smilet af drengen"
-G
63 PP 3 PP 3
Drengen elskedes sikkert af pigen
"Drengen elskedes vist af pigen"
OK
64 P SA 3 SA 3
Drengen pegede vist på pigen
"Drengen pegede vist på pigen"

65 OO 2 OS 3
Vis mig pigen som drengen ofte havde slået
"Vis mig pigen som ofte slåede drengene
drengen"
-1
66 SC 7 SC 7
Det er hunden som sikkert vil bide katten
"Det er hunden som sikkert vil bide katten"

67 OO 4 OS 3
Vis mig pigen som drengen sikkert har haft
kysset
"Vis mig pigen som vist øøh kiggede efter
drengen"
-1 +G
K. R. C. 2001 131

O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

68 SO 3 S-
rel
3
Drengen som pigen vredt slog er sur
"Drengen … øh som vist … øh øh kiggede
efter pigen"
(Drengen som vist kiggede efter pigen) [NP
[Sub-rel]
+G
69 SS 2 SS 2
Pigen som nok havde slået drengen var sur
"Pigen som øh ofte havde slået drengen var
sur"

70 P SP 5 SP 3
Drengen er sikkert blevet smilet til af pigen
"Drengen øh blev smilet af pigen nej nej det
kan jeg ikke huske"
(Drengen blev smilet af pigen)
-1 -G
71 SO 5 ? ?
Pigen som drengen vist har kysset er glad
"Pigen som …mem mem … ti pige drengen
som ret glad. Jeg kan ikke huske det
midterste"
-1 -1
72 OC 1 SC 2
Det var pigen som drengen vist havde haft
skubbet
"Det var pigen som vist havde skuffet
skubbet"
(Det var pigen som vist havde skubbet)
[INTRANS]
-1
73 OC 4 OC 5
Det er pigen som drengen nok har haft løftet
"Det er ret pigen som drengen of' øøøh … har
skubbet"
(Det er pigen som drengen ofte har skubbet)
-1
74 P LP 6 LP 6
Drengen synes vist om pigen
"Drengen synes vist om pigen"
OK
75 OS 4 OS 5
Vis mig pigen som jo har haft løftet drengen
"Øh vis mig pigen som har løftet drengen"
-1
76 P N SP 7 SP 7
Pigen vil ikke blive peget på af drengen
"Pigen vil ikke blive peget af drengen"
-G
77 N SP 6 SP 6
Katten bliver ikke bidt af hunden
"Katten bliver ikke bidt af hunden"

78 P N SA 2 SA 2
Pigen havde ikke vinket til drengen
" Pigen havde ikke vinket a' drengen"
(Pigen havde ikke vinket ad drengen)

79 SA 3 SA 3
Drengene kyssede gerne pigerne
"Drengene kyssede gerne pigerne"

80 SS 6 SS 3
Drengen som tit kysser pigen er glad
"Drengen som tit kyssede pigen er glad"

K. R. C. 2001 132

O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

81 VP 3 SA 3
Drengen kyssedes gerne af pigen
"Drengen kyssede gerne pigen"
-1
82 SP 2 SP 2
Hunden var nok blevet bidt af katten
"Hunden var nok bidt af katten"
(Hunden var nok [blevet] bidt af katten)
-1
83 P SA 2 SA 2
Hunden havde nok snuset til katten
"Hunden havde nok snuset at katten"
(Hunden havde nok snuset til katten)

84 N SA 3 SA 3
Pigen kyssede ikke drengen
"Pigen kyssede ikke drengen"

85 VP 3 VP 3
Drengen skubbedes jo af pigen
"Drengen skubbede jo af pigen"
-1
86 SS 3 SS 3
Drengen som jo skubbede pigen er sur
"Drengen som skubbede pigen er sur"

87 SC 5 SC 3
Det er pigen som nemt har løftet drengen
"Det er pigen som løftede drengen"
-1
88 SO 6 SO 3
Drengen som pigen gerne kysser er glad
"Drengen som pigen kyssede er gerne glad"

89 SO 1 SS 3
Pigen som drengen ofte havde haft skubbet
var sur
"Pigen som ofte øøh skubbede øh drengen var
sur"
(Pigen som ofte skubbede drengen var sur)
-2
90 OS 7 OS 3
Vis mig hunden som først vil bide katten
"Vis mig hunden som øøh ofte øøh bed
hunden"
-1
91 PP 2 PP 2
Drengen var vist blevet beundret af pigen
"Drengen var vist blevet beundret af af
hunden nej jo … eller pigen. Det kan jeg ikke
huske"
(Drengen var vist blevet beundret af pigen)

92 OC 7 OC 7
Det er katten som hunden nok vil bide
"Det er katten som hunden nok vil bide"

93 OO 6 OO ?
Vis mig drengen som pigen tit kysser
"Vis mig drengen som pigen ofte øøh ja det
kan jeg ikke huske det sidste ord"
"Vis mig drengen som pigen ofte [VERB])
-1
94 PP 7 PV 3
Hunden vil sikkert blive frygtet af katten
"Hunden vil sikkert blive øh … frygtet katten.
Øh hunden ville sikkert øøh frygte katten"
(Hunden ville sikkert frygte katten)

K. R. C. 2001 133
O
r
d
e
r

G
o
v
e
r
n
e
d

P

N
e
g
a
t
i
o
n

T
a
r
g
e
t

T
y
p
e

T
a
r
g
e
t

T
e
n
s
e

P
r
o
d
u
c
e
d

T
y
p
e

P
r
o
d
u
c
e
d

T
e
n
s
e





Target sentences
"Produced sentences"
(Interpretation by analyzer)


A
u
x

V
e
r
b

D
r
o
p

M
a
i
n

V
e
r
b

D
r
o
p

P
r
e
p
.

D
r
o
p

P
a
s
s
i
v
e

m
o
r
p
h
.

95 OS 1 OS 5
Vis mig pigen som vist havde haft slået
drengen
"Vis mig pigen som vist slåede drengen nej
vist har slået drengen"
(Vis mig pigen som vist har slået drengen)
-1
96 P N SA 1 SA 5
Pigen havde ikke haft vinket til drengen
"Pigen har ikke vinket til drengen"
-1
97 OC 2 OC 3
Det var pigen som drengen vist havde skubbet
"Det var pigen som drengen vist skubbede"
-1
98 SC 4 SC 5
Det er pigen som jo har haft løftet drengen
"Det var det er pigen som har løftet drengen
vist har løftet drengen"
(Det er pigen som vist har løftet drengen)
-1
99 OS 5 OS 3
Vis mig pigen som nemt har løftet drengen
"Vis mig pigen som øh nemt løftede drengen"
-1

K. R. C. 2001

2

Contents:
1 Preliminaries ........................................................................................................... 4 1.1 Introduction........................................................................................................... 4 1.2 Acknowledgements............................................................................................... 4 1.3 Abbreviations........................................................................................................ 5 2 Linguistic Theory .................................................................................................... 6 2.1 Modular Organization of the Grammar ................................................................ 7 2.1.1 X-bar Theory................................................................................................. 8 2.1.2 The Lexicon ................................................................................................ 10 2.1.3 The Projection Principle ............................................................................. 11 2.1.4 Extended Projection Principle..................................................................... 11 2.1.5 Move-α & Trace ......................................................................................... 12 2.1.6 Case Theory ................................................................................................ 12 2.1.7 Theta Theory............................................................................................... 13 2.1.8 The VP-Internal Subject Hypothesis .......................................................... 14 2.1.9 Parameters................................................................................................... 15 3 Normal Brains: The Language Area................................................................... 17 3.1 Evolution of Language........................................................................................ 17 3.2 Localization, Distribution and Lateralization ..................................................... 21 3.3 The Language Zone ............................................................................................ 22 3.4 Language Acquisition ......................................................................................... 24 4 Language Breakdown........................................................................................... 27 4.1 Mental Retardation and Impaired Language: Autism......................................... 27 4.2 Mental Retardation and Spared Language.......................................................... 28 4.2.1 Down’s Syndrome ...................................................................................... 28 4.2.2 Williams Syndrome .................................................................................... 30 4.2.3 Christopher – a Linguistic Idiot Savant ...................................................... 31 4.3 Impaired Language and Spared Cognition ......................................................... 32 4.3.1 Aphasia ....................................................................................................... 32 4.3.1.1 Broca’s Aphasia ...................................................................................... 33 4.3.1.2 Wernicke’s Aphasia ................................................................................ 34 4.3.1.3 Global Aphasia ....................................................................................... 35 4.3.1.4 Conduction Aphasia................................................................................ 35 4.3.1.5 Transcortical Aphasia ............................................................................. 36 4.3.1.6 Anomia.................................................................................................... 36 4.3.2 Specific Language Impairment ................................................................... 37 4.4 Discussion........................................................................................................... 39 5 6 Cerebral Area and Function ................................................................................ 43 A Syntactic Approach to Broca’s Aphasia ......................................................... 49

K. R. C. 2001

3

6.1 Comprehension: The Trace Deletion Hypothesis ............................................... 50 6.2 Production: The Tree-Pruning Hypothesis ......................................................... 56 7 Is Grodzinsky’s thesis applicable to Danish? ..................................................... 61 7.1 The Danish Language ......................................................................................... 61 7.2 Predictions About Agrammatism in Danish ....................................................... 67 7.2.1 Comprehension ........................................................................................... 68 7.2.2 Production................................................................................................... 71 7.2.3 Summary..................................................................................................... 76 7.3 Empirical Tests ................................................................................................... 77 7.3.1 Comprehension: the Sentence-to-Picture Matching Test ........................... 77 7.3.2 Production: Repetition Test ........................................................................ 79 7.3.3 Patient TJ .................................................................................................... 82 7.3.3.1 Medical history ....................................................................................... 82 7.3.3.2 Comprehension Test Results................................................................... 83 7.3.3.3 Production Test Results .......................................................................... 85 7.3.3.4 Conclusions............................................................................................. 93 8 9 Discussion: Cerebral Area and Function Revisited........................................... 94 Conclusions.......................................................................................................... 103

10 References............................................................................................................ 107 11 Appendix A: Sentence-to-Picture Matching Test ............................................ 113 12 14 15 Appendix B: Sentence Types and Tokens......................................................... 118 Appendix D: TJ's Comprehension Test Data................................................... 124 Appendix E: TJ's Production Test Data........................................................... 126 13 Appendix C: Repetition Test: Structure Distribution..................................... 121

I describe the modular structure of the grammar and introduce the modules (sub-theories) relevant for this paper.K. and to Gerald Fischer (speech therapist at Kommunehospitalet. In chapter 6 I present a syntactic approach to Broca's aphasia based on the general framework of generative grammar. The survey of language breakdown will also provide evidence for the internal modular structure of the human language competence. i. R. one of which is that language is reflected in the architecture of the brain and therefore linguistic theory can be used to predict specific symptoms of language breakdown in any language. which I use in chapter 7 to make some predictions about agrammatism in Danish. the localization of language in the language zone on the left hemisphere. It is about the connection between neurology and linguistic theory. Chapter 3 is about the biology of language . In chapter 9 I draw my conclusions. 1.1 Introduction This study is about the neural predisposition of the brain underlying human speciesspecific linguistic competence. Based on the previous chapter and the test results I return to the discussion on brain area and function in chapter 8. as the latter must be compatible with the patterns found in language deficits. Chapter 5 is a discussion on the correlation between lesion site and language function. 1. who volunteered to help me by taking the language tests.2 Acknowledgements Thanks to TJ. C. and language acquisition as viewed from the brain.the evolution of language. followed by my acknowledgements and a list of abbreviations. I discuss how findings from language impairments can play a crucial part in determining the proper linguistic theory. discussing the dissociation between language and intelligence as reflected in developmental deficits such as Down's syndrome and in acquired deficits. These predictions are tested empirically on a Danish subject with some aphasic symptoms due to lesions in the language area. Aarhus) for establishing contact with . I then turn to language impairment in chapter 4. I give a general introduction to the grammatical framework of generative grammar in chapter 2. The structure of the study is as follows: First.e. 2001 4 1 Preliminaries This chapter contains the preliminaries: first a general introduction to the structure of the paper. Finally. different types of aphasia.

Thanks to my supervisors: Peter Bakker (Department of Linguistics.3 Abbreviations Adject AdjP Adv AdvP AgrP AUX C Compl Conj CP DP e FEM I INFL INF IP LF MASC N NegP NP Adjectival Adjective Phrase Adverb Adverbial Phrase Agreement Phrase Auxiliary verb Complementizer Complement Conjunction Complementizer Phrase Determiner Phrase Empty X0 Feminine Inflection Inflection Infinitive Inflection Phrase Logical Form Masculine Noun Negation Phrase Noun Phrase Obj P PASS PERF PERF+ PERS PF PL. University of Aarhus) for his invaluable help on the specifics of grammatical structure. for example on the interpretation of TJ's medical file.K. University of Aarhus) for being supportive and helpful beyond duty. 2001 5 TJ. C. and for being very helpful. PP PRES Psych PT PTCP SING Spec Sub t TnsP V VP X0 Object Preposition Passive Perfect Double perfect Person Phonetic Form Plural Preposition Phrase Present Psychological Past Participle Singular Specifier Subject trace Tense Phrase Verb Verb Phrase Head of an XP . 1. R. and Sten Vikner (Department of English.

Generative linguistics takes as its goal to give an account of the linguistic competence of humans. In my description I only include the principles relevant for the task at hand and only the necessary elaboration. This grammar consists of two parts: First. . The following is only a brief and very general overview. Vikner 1999). and therefore my grammatical analyses will be very much influenced by Government and Binding Theory (Chomsky 1981. They are motivated by the simple fact that language is universal. From this follows that the goal of generative linguistics is not to describe the details of one specific language but rather to formulate what principles determine the grammar of any language.g. Language acquisition is viewed as the setting of these parameters on the universal grammar by exposure to the ambient language. This is motivated by the fact that languages are different in various but not arbitrary ways. there is a set of parameters whose settings determine the specifics of any language of the world. but I shall give more specific definitions and introduce further principles when required throughout the text. such as the basic word order. the basic underlying universal principles that are shared by all languages of the world. as these frameworks all share some of the same core principles. 2001 6 2 Linguistic Theory The linguistic framework in which I shall be working is generative grammar. for example Subject–Verb–Object in Danish. For general overviews I refer the reader to Grodzinsky 1990 (chapter 2 – an excellent introduction). 1995) and Optimality Theory (e. and as language is a universal human trait this competence is conceived as a universal grammar (UG). Grimshaw 1997. Haegeman 1994 and Vikner 1995 (chapter 2). it will also be compatible with more recent linguistic developments such as the Minimalist Program (Chomsky 1993. Archangeli & Langendoen 1997. At least in principle. Vikner 2000). R. C. Second. Haegeman 1994. I shall try to avoid being too loyal to any specific version of generative grammar while trying to retain the principles underlying most recent versions of the theory within the principles and parameters frame. Kager 1999.K.

or Phonological Form (abbreviated PF). In the figure below the principles are placed in boxes. there is the logico-semantic representation called Logical Form (LF). i. there is a strong connection between “the boy kisses the girl” and “the girl is kissed by the boy”. For example. and the arrows point to the levels at which they apply. Third. cf. R. but the relation is clearly felt. C. First. the principles do not necessarily apply at all levels. They are not quite synonymous. their domain (I leave out principles not presented in this paper): . A module is a mechanism. there is an underlying structure called the D-structure. and the thematic relations (the kisser and the kissed) between the participants (the boy and the girl) are the same. Anticipating the following descriptions of the principles involved. As will be explained below. some of the elements of the sentence are moved to other positions in order to reach the S-structure from D-structure. which is responsible for a designated domain.1 Modular Organization of the Grammar The grammar generates several levels of representation for every sentence. The levels are organized as follows: D-structure S-structure PF LF Figure 1: The organization of the grammar The principles of grammar are modularly organized. the difference in structure between the active and the passive. Therefore. This is posited to account for the structural (and semantic) connection between corresponding actives and passives. 2001 7 2.e. The scenario (the kissing).K. The active and passive sentences are two instantiations of the second level of representation: S-structure. which is the form used for phonetic interpretation. Fourth. I just list the principles here without going into details. as they may or may not share the same domain or even parts of the same domains. there is a phonological representation.

specifier. The lines indicate the movement of the verbal inflection “-ed” and the noun phrases “the boy” and “the girl”. R. in which I have left out some syntactic elaboration that will be introduced later (I exclude PF and LF throughout the text. The important thing to notice in this example is the common underlying structure. C. 2001 8 X-bar Theory Theta Theory Case Theory D-structure S-structure (Extended) Projection Principle Move-α PF LF Figure 2: The modular organization and the domains of application of the grammatical principles (adapted from Grodzinsky 1990: 25) Consider the following example. complement. To explain this and the following principles it may be useful to consider an example: (1) The boy kissed the girl from Sweden . 2. as they are not crucial for my account): “The boy kissed the girl” D-structure S-structure “The girl was kissed by the boy” [ – ed [ [the boy] kiss [the girl] ]] [[the boy] – [ – kiss-ed [the girl]] [[the girl] was [ – kissed – ]] by the boy Table 1: syntactic representations.1 X-bar Theory Important to the theory are the notions of head.1. while the “–“ indicate the empty space left by the movement.K. and projection. This is an oversimplification. but the concepts will become clear as the principles are presented.

In the sentence above “the girl” constitutes a noun phrase.e. which means that all phrases must abide by it. PP (preposition phrase). All the phrases. and compl is the term for complement. C. “the”).g. In addition to these lexical categories (they are lexical because the heads are words) there are functional heads. the highest level is called NP (for noun phrase). Finally. The head of the noun phrase is called N0.g. The determiner specifies the head. it is not just any girl. which is headed by the verbal inflection. . The complement of the head “girl” is the preposition phrase PP “from Sweden”. The elements are hierarchically ordered in the following way: (2) NP spec N’ N0 compl The girl from Sweden In the representation spec is the term used for specifier. AdvP (adverbial phrase). regardless of the type of the head. depending on the head. Hence. which is represented in the following way (a sort of ‘flattened tree’): 1 I leave out the notion of determiner phrases DPs where the head is the determiner (e. which in turn is headed by the preposition “from” taking the NP “Sweden” as complement (abbreviated as a triangle in (2) below). the zero level of the noun phrase. and hence the phrase it heads becomes a noun phrase. The X-bar schema is a constraint on all syntactic categories. This is also called the maximal projection of the head. and the N’ is the next higher level called N-bar (which is the level that has become the name of the theory: this level is present in all phrases so in an XP (e. for example IP (inflectional phrase). 2001 9 A head is an element that gives a larger unit its characteristics. The head of the phrase is the noun “girl”1.K. which takes a NP as complement ([DP the [NP girl]]) and use the classic term NP instead. share same structure: the X-bar structure. R. i. abbreviated NP. the phrases are: NP (noun phrase). a noun phrase or a preposition phrase this level is called X-bar). VP (verb phrase). which is therefore abbreviated X. but the one “pointed out” by the specifier. and AdjP (adjective phrase).

The lexical entry specifies that it is a verb (V). Consider the verb. 2001 10 (3) [XP spec [X’ X0 compl]] For example: (4) [NP the [N’ girl [COMPL from Sweden]]] (5) [NP boring [N’ books [COMPL about linguistics]]] I shall use a slightly abbreviated version of this notation leaving out the X’ branching orthographically as shown in (6) and exemplified in (7): (6) [XP spec X compl] (7) [NP boring books [PP about linguistics]] In this NP. The structure of the clause includes functional as well as lexical categories. which is the entity doing the kissing (AGENT). In other words. A word’s lexical entry specifies its syntactic category and its semantic meaning. This is illustrated in: (8) [CP that [IP [NP the boy] . For example an embedded clause like “(they said) that the boy kissed the girl” is introduced by the complementizer phrase CP headed be the complementizer “that”. This in turn takes a VP as its complement. the predicate “kiss” has . In addition it selects another NP.K. “boring” is the specifier of N0. or rather the predicate “kiss” in the example in (1) above. Henceforth I shall refer to the specifier position of an XP as [spec. “books”. and that takes a nominal complement (NP) – the verb subcategorizes for an NP. R.[VP kissed [NP the girl]]]] 2.2 The Lexicon Words are stored in a mental dictionary called the lexicon. which can be read as “specifier of XP”. XP]. and the preposition phrase (PP) is the complement. which takes as its complement an inflectional phrase IP headed by the verbal inflection. C.1. which is the entity receiving the kiss (THEME).

Grimshaw 1991). Leaving out. the latter to be further split up in section 6. defined below). .2 below) do not contribute to the thematic information as such. and it may be represented like this: (9) kiss: V <AGENT (NP). C. the clause is an Extended Projection of the verb. to obey the Projection Principle lexical information must be syntactically represented. for example one of the arguments would result in ungrammatical sentence. IP] position assigns NOMINATIVE case to the subject. THEME (NP)> 2. as indicated with an asterisk *: (11) *[VP kissed [NP the girl]] 2. The VP contains the core thematic information: the predicate and its arguments (under the VP-internal Subject Hypothesis.K. R. the functional categories (CP and IP. but tense inflection locates the event in time and the [spec.4 Extended Projection Principle According to the Extended Projection Principle clauses are projected by the verb.1. In short. lexical information is also reflected in the structure of the clause: (10) [NP The boy] [VP kissed [NP the girl]] So. First. 2001 11 two arguments. From a semantic point of view. So.1.3 The Projection Principle The semantics of the predicate (the verb) determines to a large extent the structure of the sentence. the specifics of the predicate (its lexical entry) determine the number of arguments and their categories and semantic roles. hence the Extended Projection Principle (cf. The lexical entry specifies their syntactic categories and their semantic roles.

K. R. C. 2001

12

2.1.5 Move-α & Trace α As shown in Table 1 above, the connection between D-structure and S-structure involves certain changes to the positions of some of the constituents. These changes are known as transformations or movements. The principle (or sub-component) of grammar responsible for these transformations is the rule of Move-Alpha or Move-α: (12) Move-α α “Move anything anywhere” These movements are constrained by the other modules of grammar in order to rule out ungrammatical forms, such as “*the the kissed boy girl” and “*was kissed the girl by the boy”. It is the interaction of Move-α with the other syntactic principles that results in grammatical strings. It is responsible for the mapping between active and passive, for example. Movement of constituents results in a phonologically silent but structurally represented position, which is filled with a construct, called a trace (t). The moved constituent is the antecedent of the trace and together they form a chain. The two are further linked together by a shared index. Consider again the example from (8) (without the CP): (13)

[IP [NP The boy]1 t2 [VP t1 kissed2 [NP the girl]]]

The verbal inflection that headed the IP has moved to the verb “kiss” in the VP, and has left a co-indexed trace (t2) behind in I0. The subject “the boy” has also undergone movement and left a trace, and this movement is due to the Case Filter, which I shall now explain.

2.1.6 Case Theory This is the component of grammar specifying the structural position in which lexical NPs may appear. According to this theory all NPs must be assigned abstract case. In some languages this case is overtly realized as case endings, e.g. German, while in others it is invisible, e.g. English. Certain elements are case assignors, such as tense

K. R. C. 2001

13

inflection, verbs, and prepositions. Thus, syntactic structures containing NPs unmarked for abstract case are labeled ungrammatical and filtered out by what is called the Case Filter at the level of S-structure, cf. Figure 1: (14) Case Filter: *NP [+lexical, -case] In other words: all lexical NPs must have case. In the example in (13) above, the subject was forced to move to [spec, IP] (specifier of the inflection phrase IP) in order abide the Case Filter. If it remains in its base position inside VP (I shall return to the position of the subject shortly; so far, suffice it to say that the subject is moved) it will not be case marked and the structure will be filtered out. The subject case NOMINATIVE is assigned to [spec, IP], and object case ACCUSATIVE is assigned by the verb to its complement “the girl”. Thus, the interaction of the Case Filter and Move-α determines the position of the subject.

2.1.7 Theta Theory Arguments are assigned semantic / thematic roles, known as theta-roles or θ-roles, such as
AGENT

(the ‘doer’ of the action specified by the predicate, e.g. “the boy” in (1)) or

THEME

(the entity affected by the action). A principle called the Theta Criterion

(Chomsky 1981) ensures that all arguments are assigned one and only one θ-role and that all θ-roles of a predicate are assigned to appropriate structures in a one-to-one relation. This can be illustrated as follows: (15) Argument1 – θ-roleα Argument2 – θ-roleβ θ-Roles constitute a universal set including the following: (16) AGENT: The one intentionally doing or initiating the action expressed by the predicate.

K. R. C. 2001

14

THEME:

The entity affected by the action or state expressed by the predicate.

EXPERIENCER: The entity experiencing the psychological state expressed by the predicate. There are other θ-roles, but the ones defined above are the only ones needed for my purposes in this study. For further information, see e.g. Haegeman (1994: 49ff) and Jackendoff (1972).

2.1.8 The VP-Internal Subject Hypothesis According to the VP-Internal Subject Hypothesis, the subject is base generated inside the verb phrase in [spec, VP] (specifier of VP), cf. Burton & Grimshaw (1992). As explained above, to abide the Case Filter the subject moves to [spec, IP] to be assigned NOMINATIVE case leaving behind a trace, with which it shares a common index (i): (17) IP NPi I0 ti V0 I’ VP V’ compl

The idea is that in the underlying structure (the D-structure) the subject (NPi in the diagram) is generated in [spec, VP], where V0 assigns a θ-role to the NP (in other words, the verb theta-marks the subject). In English NOMINATIVE is assigned to [spec, IP] by I0 (cf. Haegeman 1994):

NOMINATIVE (18) [CP spec C0 [IP spec I0 [VP spec V0 [IP he1 “he-NOM loves her-ACC” t2 [VP t1 compl]]]

loves2 her]]]

Danish parameters will be set to SVO and V2. Another parameter concerns word order: in e. but in Danish and Swedish the verb is always second. cf. R. such Diyrbal (a language of the northeastern Queensland in Australia. 2001 15 In Danish. The parameter involved ‘decides’ what assigns NOMINATIVE case: in Danish C0 is the NOMINATIVE-assignor and in English it is I0. Comrie 1989: 104) and Bandjalang (spoken in the northern New South Wales in Australia. IP] by C0 (e. C. Other languages have both types.g.g. NOMINATIVE is assigned to [spec. Vikner 1995: 54ff. while an English child will have the same parameters set to SVO and non-V2 (this is a simplification but it will suffice).g. where the verb must follow the subject: Topic-Subject-Verb.9 Parameters As mentioned above. Danish and Swedish the finite verb is always the second constituent in the sentence (preceded by the subject or a topicalized element). the grammar consists of principles and parameters.K. All four language shave Subject-Verb-Object (SVO) word order.g. This is not the case in English and French.): NOMINATIVE (19) [CP spec C0 [IP spec I0 [VP spec V0 compl]]] t2 [VP t1 t2 hende]]] [CP han1 elsker2 [IP t1 “he-NOM loves her-ACC” 2. the object is topicalized the subject follows the verb: TopicVerb-Subject. German. cf. English and French are not. while this is not the case in English and French. Crowley 1997: 137) has ABSOLUTIVE-ERGATIVE. For example English. An example of the latter is reflected in the difference between (18) and (19) above. cf. . Hebrew and Danish all have a NOMINATIVEACCUSATIVE case system while for instance Chukchi (a language of the north-east Siberia. Still another example of a parameter is the type of case system a language has. Comrie 1989: 113). This parameter is the V2 (verb second) parameter: Danish and Swedish are V2 languages. so if e. in the grammar of a child acquiring e.1. Briefly.

The English gloss of the ERGATIVE version of “she pushed him” would be “her pushed he”. i. In chapter 6 I shall argue that the modular structure of the grammar is relevant and even crucial in the definition and diagnostics of aphasia. In the ACCUSATIVE system the subject in both transitive and intransitive clauses is always in NOMINATIVE case.e. In the ERGATIVE system the subject of an intransitive clause has the same case marking as the object of a transitive clause. the principles and theories listed here are sub-components of a modularly structured grammar. . R. In interaction they generate the grammatical structures of a language. 2001 16 Grammatical Roles Accustive System Ergative System INTRANSITIVE “she pushed” Sub NOM ERG TRANSITIVE “she pushed him” Sub Obj NOM ACC ABS ERG Table 2: (NOMINATIVE-) ACCUSATIVE and (ABSOLUTIVE-) ERGATIVE cases systems. ERGATIVE case.K. C. In summary.

in one place or in many places. In time language has become represented in the architecture of the human brain: speciesspecific. It is clear though that the two kinds of wings are not homologous. is the language organ and I support this claim with a brief survey of language acquisition from a neural perspective. this chapter is about the biology of language. Deacon 1997. Pinker & Bloom 1990). The following is a brief discussion of how this competence could possibly as well as probably have come to be innate. the universal grammar described above is innate. and innate. I think. In short. This competence is both unique to mankind and universal among the peoples of the world. . This area. This leads to the next section in which I zoom in and look at an area in the left hemisphere. R. Together these two facts. which most likely started with a cognitive ‘arms race’ leading to selection for bigger brains. In other words. Some aspects of animal communication and the rudimentary symbolic communication (laboriously) taught to apes may have some analogies to human language in the same way that bat wings are analogous to the ‘wings’ of flying squirrels – both are used for flying. speciesspecificity and universality. C.1 Evolution of Language As mentioned in the preceding chapter. Donald 1991. has to be supported by a feasible account of the origins of such a competence. 1995). 2001 17 3 Normal Brains: The Language Area In this chapter I first discuss the evolution of language and the brain. Universality covers the observation of the fact that all peoples of the world have a language – no mute and non-signing tribe has ever been discovered. Pinker 1994. Then I move on to discuss where language functions are located in the brain: nowhere. universal. which means sharing the same basic structural design stemming from a common ancestor. point to some kind of specific human capacity or faculty represented in the brain as some sort of neural circuitry that is unique and universal to humans and dedicated to language processing.K. Bat wings are modifications of the hands of the common mammalian ancestor and the wings of the flying squirrel are modifications of the rib cage (cf. This species-specificity naturally entails that language is different from any other kind of communication in the animal kingdom (cf. I shall argue. 3. Such a claim. everywhere. the goal of generative linguistics is to provide an account of the human linguistic competence.

while not paying the costs of participation. Deacon 1997. who cooperated in hunting and nurturing. and furthermore ensure that all members of the group honor them. a larger brain does not inevitably produce support for language. so does the structural complexity: there are more social relations to keep track of. C. 1995) that the onset of human evolution and the break off from the chimps was the emergence of bipedalism in the Australopithecus Afarensis approximately 4 million years before present. in which the primary initiator was not monogamy. language is not the result of a larger brain – whales for example have quite larger brains than humans. What initiated the evolution of symbolic communication (besides the obvious benefit of sharing information) was the need to outwit potential social cheaters: those that reap the benefits of this cooperative hunting and nurturing. Actually our brain is three times larger than expected on the basis of our body size (cf. so to speak (unlike large flocks of cattle. Pinker & Bloom 1990. there is a strong connection between group size. Dunbar 1993. The human brain is large in relation to the body – a phenomenon known as encephalization. 2001 18 It is generally believed (cf. Donald (1991). However. Pinker (1984) proposes a different theory. In either case there was selection pressure for organisms with communicative skills. This lead to the evolution of symbolic communication that would enable them to establish and maintain symbolic social relations. R. Following his argument. as Dunbar (1993) points out. Donald 1991. According to one theory (Deacon 1997) what was special about the early hominid social groups was that they were monogamous couples living together in larger groups. Of course. The early humans on the other hand. This led to a cognitive ‘arms race’ which in turn lead to the enlargement of the human brain. This calls for greater . However.K. but they do not have a language. the evidence for a social structure capable of supporting language evolution is better with the later Homo Habilines. had much contact as they were living in groups. or rather marital contracts. monogamy is restricted to animals living in deserted and preferably vast areas where there is little contact with other members of the species. Pinker 1994. but in this respect we clearly differ from the other apes. See also Bickerton (1990). This trait is not unique to mankind (sparrows and mice for example have even bigger brains in relation to their bodies than humans). and Foley (1997) for discussions on the evolution of language. which has very little structure). brain size and language. which so to speak freed the hands for tool use. Deacon 1997). As primate groups grow in size. In other words.

i. the auditory areas in the superior temporal gyrus including Wernicke’s area (see Figure 3 below) is enlarged. This means natural selection has given the prefrontal cortex dominance over the rest of the brain. 2001 19 cognitive abilities. In the text the term posterior refers to areas (in the figure) to the right of the motor area.K. Due to the encephalization a relatively smaller portion of the cortex of the human brain is devoted to controlling bodily functions. R. this is ‘survival of the fittest’ played out on a developmental neuronal level. the body being relatively small in relation to the brain. sexual versus natural selection respectively (Darwin 1859). it is unresolved whether this results from the complexity of living in groups of monogamous couples as Deacon claims or from the cognitive demands of outsmarting uncooperative social parasites as Pinker claims. However. . C. In development the different parts of the brain connect to all other parts of the brain. The black line surrounds the so-called perisylvian region. while the black names merely indicate location. wins the control over it – in other words. Through a process named neural Darwinism (Edelman 1998) the largest part. which in turn gives rise to the enlargement of the brain. However. Figure 3: Idealized map of cerebral areas in the left hemisphere.e. The size of the motor areas is thus actually reduced in relation to the rest of the brain. I suspect the truth lies somewhere in between. The names in white indicate functional areas. Deacon 1997:217). The human brain is not just an ‘inflated’ chimp brain. Not all of it is larger than would be expected for a primate of human size. which thus has the largest number of connections to a given area. and most important the prefrontal cortex including Broca’s area (both areas are further discussed below) is twice its expected size (cf.

Deacon argues that the cortical areas in the left perisylvian region (see Figure 3) are not specialized for language but that they are computational centers also utilized for linguistic purposes. R.e. Language acquisition is thus based on a specialized learning capacity for symbolic systems (below I shall argue that this is not altogether plausible). Christensen 2000 for a discussion on the coevolution of language and the brain and a comparison of Pinker 1994 and Deacon 1997). represented in the neural architecture of the brain. In other words. Lieberman 1984). which together define the large array of possible human vowel sounds. which is also responsible for language. Furthermore. Pinker 1994). To Pinker language acquisition is not dependent on a general symbolic learning capacity. language and the brain have co-evolved: language has shaped the brain and vice versa (cf.K. i. This obvious benefit for our phoneme inventory comes with a severe disadvantage for breathing and swallowing. these areas are specifically dedicated to language. the most obvious adaptation of the human body to language is the (again) species-specific position of the larynx deep in the throat. Physiologically.e. C. It therefore seems natural to say that language has been internalized. humans are predisposed to language in the sense that they have a special learning algorithm for language. and therefore the brain is already prepared for linguistic input. This explains why language is universal and species-specific and independent on general learning capacities. consisting of setting the parameters of the innate universal grammar to match the surrounding language. i. which gives us two cavities (mouth and throat). Any way we slice it. as there is a risk of food falling into our lungs (cf. Both cross-linguistic and cross-time variation is constrained by the specialized structures of the human brain. According to Pinker (1994) on the other hand the left perisylvian region is the language organ. since neither typological language variation (Comrie 1989) nor diachronic language change (Crowley 1997) is completely arbitrary and unconstrained. it seems natural and plausible to claim that there is such a thing as a universal grammar (cf. 2001 20 To Deacon (1997) all this indicates that the prefrontal cortex houses a ‘symbolic processor’ (and (some) support comes from Donald 1991). Below I shall argue that the areas in the left perisylvian region indeed are specialized for language. . but rather a matter of instinctive and automatic acquisition of the ambient language.

K. right-. C. C. These deficits are nonlinguistic. as the correlation of aphasia and right-hemisphere damage is very rare.e. Deacon 1997). R. The second possibility. or in both hemispheres. Language is a cognitive module. is not plausible either.] has an important role in communication but makes no syntactic contribution to language use. left-. or 4: in several places. as damage to the brain can lead to selective impairment of language. 2001 21 3. and encountered only if the patient suffers from early left hemisphere damage (cf. Right hemidecortication (removal of the entire cortex in one hemisphere) and lesions in the right hemisphere cause some semantic and/or pragmatic disorders. a somewhat self-contained subsystem of the human mind (see also Pinker 1994. Bishop 1988). the evidence is that this side of the brain [i. 3: in one place. Donald 1991: 80 and Calvin & Ojemann 1994: chapter 4).2 Localization. I shall return to the matters of modularity and dissociation between language and intelligence in the following sections. It would be very strange to diagnose a person as language impaired if he or she has problems with getting the point of jokes and stories or just speaks in a flat and monotonous manner.e. as I shall discuss in more detail in chapter 4. such as problems with understanding narratives and jokes (cf. Grodzinsky (2000: 19) concludes: Thus. the right side. This leaves us with the last two possibilities: 3: language is localized in one area. The first answer is not an option. and impaired general intelligence need not have an impact on language. Language is independent and distinct from ‘general cognition’. 1985). as language per se is not damaged. This is of course closely related to the notion of modularity (Fodor 1983. . K. A logical place to start is to find out if language is represented in the right. a syndrome where the patient’s speech is flat and lacks prosody (cf. Distribution and Lateralization Where is language represented in the brain? There are four logically possible answers: 1: nowhere. and 4: language is distributed over several areas. as I shall show below. and aprosodia. The implication is that language acquisition is done by a general learning mechanism and hence dependent on general intelligence and vice versa. or not lateralized. the left. 1997). R. ‘everywhere’. 2: everywhere. Damasio 1992: 537. Studies of brain damage leading to language disorders have shown that language normally is localized in the left hemisphere. i.

Bishop 1988. in the region surrounding the Sylvian fissure (cf. which should account for the above mentioned problems with narratives and prosody for patients with right-hemisphere injuries. Damasio 1992. he states that “one can safely assume that the RH [right hemisphere. cf. the left hemisphere is speed optimized. R. based on evidence from bilingual aphasics Paradis (1998) places what he calls implicit linguistic competence in the left hemisphere. i. The left side is specialized in short time domains. Prosody is a feature of the inter-sentential domain. 2001 22 The discussion so far points to a localization of language in the left hemisphere.] is crucially involved in the processing of pragmatic aspects of language use” (Paradis 1998: 422). Calvin & Ojemann 1994: chapter 3 and Donald 1991:80). Deacon (1997) views the difference between the hemispheres not as linguistic versus non-linguistic computation. Gazzaniga 1989). but as a difference in speed of computation. and this is supported by results from tests where one hemisphere is anesthetized prior to brain surgery (this is called the Wada test. Regarding the right hemisphere. R. also known as the Sylvian fissure (see Figure 4 below). Broca’s area (the left inferior frontal gyrus) is especially active during tasks involving syntactic analysis (Stromswold et al. In other words. PET studies have shown that there is increased blood flow. which is sometimes though rarely the case.3 The Language Zone Within the left hemisphere language functions are localized in an area around sulcus lateralis.e. This is done in order to discover whether language functions are located in the left (as is almost always the case) or in the right. Much in the same line as Deacon. Bishop 1988: 206). K. C. which should account for the breakdown of syntax and morphology in left-hemisphere injuries (cf. 1996). which is the long ‘gorge’ that constitutes the upper ‘border’ of the temporal lobe. In the vast majority of cases where brain damage causes language deficits it is located the perisylvian region.K. hence cerebral activation. in this area during linguistic processing. C. Calvin & Ojemann 1984. The right hemisphere is specialized in large time domains (linguistic and non-linguistic). . Anesthetizing the ‘linguistic’ hemisphere causes language impairment (cf. Deacon 1997: 316). Damasio & Damasio 1992. as it spans the entire utterance regardless of the number of clauses. 3.

R. C. is often discussed as if it consisted of a collection of separate parts or areas. According to Paradis (1998: 422) (see also Menn et al. Dehaene et al.K. (1997) investigated a group of French-English bilinguals. Gazzaniga (1989) has shown what this region of the brain would look like if it had been ironed flat. the subjects' native language (L1) was located in the perisylvian region. The subjects were scanned in a fMRI (functional magnetic resonance imaging) while listening to stories in the two languages. much like if you take a large piece of cloth and crumble it up into a ball. English). French) showed a weaker but similar pattern. Therefore. . much of the surface is not ‘visible’ and the topmost parts seem adjacent. but sometimes they may appear to be separate. which he labels the language zone (Gazzaniga 1989: 950). whereas the second language (L2. This is apparent in Figure 4. For the same reason Pinker (1994) and Grodzinsky (2000) call this region the language organ. The brain does not have a flat surface. This area. Further evidence comes from studies of bilinguals. All the areas that are associated with language (and language breakdown. The results of the tests showed a consistent and significant increase in activity in the left perisylvian region during comprehension of the subjects' first language (L1. 2001 23 Figure 4: Language areas (from Pinker 1994:308). the left perisylvian region. where the subjects had learned their L2. see below) form a contiguous area. Others have found more consistent results showing that both L1 and L2 are located in the language zone. Dehaene et al. Whatever the reasons for the less consistent pattern of L2 in their investigation. suspect that the latter may be due to different teaching strategies in the schools. its surface is full of convolutions. where the different areas around the Sylvian fissure seem to be delimited in a nice clear-cut fashion. However.

In this section I briefly discuss the processes in the brain during language acquisition.1. Elman et al. At birth. The synapses.fine-tuning the phonetic inventory one word stage vocabulary spurt. 3.2. 2001 24 1995) both languages are represented in the language areas of the left hemisphere in bilingual as well as unilingual individuals. I have summed up the correlation between developmental neural and linguistic events in the following table: Age birth around 9 months around 12 months around 18 months around 36 months around 48 months peak in overall brain metabolism Neural development Completed cell formation and migration Adult distribution of metabolism. long-distance connections established Peak in number of connections within and between cortical regions Linguistic development left hemisphere specialization suppression of non-native sounds .rapid increase in number of syntactic constructions and complexity successful language acquisition Table 3: Developmental neural and linguistic events. which houses the long-distance connections between the major regions is not complete until around nine months of age. which are the connections between the neurons in the cortex. R. depending on the region of the brain.K. Huge numbers of neurons die from birth to around the age of seven.primitive syntax grammar explosion . the brain is already biased towards a left-hemisphere specialization for language (cf. Bishop 1988: 212). . Müller 1996: section 3.4 Language Acquisition The maturation of the human brain is the driving force in language acquisition. At approximately two years of age the brain has 50% more synaptic connections than the adult brain. C. 1995). The white matter (underlying the cortex or gray matter). are formed and migrate from the place where they are generated into their proper places. continue to develop and peak in number sometime between nine months and two years. 1996: chapter 5. Before birth all the nerve cells of the brain. Pinker 1994: chapter 9. peaking at around 4 years of age. The metabolic activity reaches an adult level at around nine months and rapidly exceeds it. two word stage . The synaptic connections decrease in number until adolescence. After birth the size of the brain and the thickness of the cerebral cortex increase in the first year. when the brain falls back to the adult level of metabolism (for elaboration see Deacon 1997: chapter 6. the neurons.

functional reorganization is possible in the very early stages. After this period successful first language acquisition is rare. Elman et al 1996). As Bishop states: Provided that the traditional language areas were spared. 1996:288). According to Bishop (1988) the majority of children suffering left hemidecortication or brain damage within the first couple of years of life do not develop aphasia. This period may be called the critical period (Lenneberg 1967) or the sensitive period (e.2. The neurons themselves are thus in place at around seven months after conception (cf. R.g. which is not the case if both L1 and L2 are acquired within the period. The brain does not consist of a vast number of identical neurons – it is not an equipotential mass (Müller 1996: section 3. The ability to recover rapidly decreases with age and chances are best before the age of ten. This period is the time window in which the brain is open to successful language acquisition and recovery from lesions. This reorganization is dependent on the neural plasticity. i. (Bishop 1988: 207) It appears then that there is some sort of 'default brain plan'. while adjusting / reorganizing itself to match the ambient language. lesions that encroached upon Broca's area or Wernicke's area were likely to bring about a functional reorganization of the brain. but the number of neurons and the number and strength of the connections between them need to be adjusted to suit the environment. with right-hemisphere specialization for language. even extensive early lesions of the left hemisphere did not result in right-hemisphere language representation. As mentioned above. the neurons migrate from the place where they are generated to their proper locations. The default brain develops following the bioprogram (Bickerton 1988).e. which is highest in infancy and rapidly decreases during the first years of life. Also. C.1) or a meatloaf (Pinker 1994). . there is a strong correlation between early damage to the language area and right-hemisphere language specialization. reaching the adult low at around puberty. This is due to the fact that the child brain is very plastic. Furthermore. However. Elman et al.K. The important point is that the brain seems to be 'prepared' for language but still dependent on stimuli from the environment. 2001 25 Evidence for the early left-hemisphere bias of language at birth comes from cases of brain damage. bilingual speakers’ first language will usually permanently influence the acquisition of their second language (for example through accent) after this sensitive period.

Within the left hemisphere. i. language functions are distributed over the language zone. In the next chapter. I have argued that evolution has lead to a neural predisposition to language learning / acquisition. C.e. which leads to neural malformation. […] How many times has this point been made in the last 30 years? How many times in the last three centuries? How many times is one going to have to make it again?” (Fodor 1985: 36) So far. . language is localized in the region surrounding the Sylvian fissure – the language zone. or by damage to previously normal adult brains. Thus. This predisposition is reflected in the architecture of the human brain in that linguistic functions are located in the left hemisphere. This fits very well with the facts of brain development. innateness does not mean that everything is pre-wired in the brain of the infant: No nativist has ever supposed that innate capacities are unaffected or unformed by environmental interactions. Furthermore. or genetics and environment must contribute in order to secure successful language acquisition. the critical period. I discuss what happens when the language zone is damaged – either by developmental deficits. language is left lateralized. This interaction between (genetic) neural predisposition and environment has to take place within a limited time frame. R. a process to which the environment provides crucial input.K. This neural specialization is the basis of language acquisition. to be successful. Both nature and nurture. 2001 26 Recall from chapter 2 that in generative linguistics a language is viewed as parameter setting on a universal grammar. The infant brain develops according to some kind of a default plan / bioprogram. Language acquisition is viewed as the process in which these parameter settings become fixed.

It is a brief survey of language deficits. and internally. Of the minority that does develop speech approximately 75% go through a prolonged period of echolalia. From this survey I shall conclude like others before me that there is a double dissociation between language and intelligence. below 60. R. that is. According to Fay (1988) mutism is a frequent characteristic symptom associated with autism. I link the different kinds of acquired language impairments to different areas of the brain – or rather to a certain region in the left hemisphere: the language organ. autistic people tend to lack communicative intent. 2001 27 4 Language Breakdown This chapter is about the kinds of impairment or trauma that lead to the breakdown of language. the repetition of words or fragments of sentences just uttered by others. hearing and intelligence. and Specific Language Impairment – and other are acquired deficits due to brain damage – aphasia and anomia. 4. which again supports the modularity hypothesis. Furthermore.g.1 Mental Retardation and Impaired Language: Autism Autism is a severe developmental deficit that typically appears some time during the first three years of life. but this deafness is restricted to linguistic sounds. . Finally. which is (sometimes) acquired or spared in spite of brain damage.K. while they may show signs of intolerance to other environmental sounds. Hence. language acquisition is not a general learning problem for some central processor. They tend to be socially withdrawn. Their speech is often monotonous and mechanical.Q. Interestingly. some of which are developmental deficits – autism. C. People suffering from an autistic disorder (Bishop 1989) tend to be (but are not always) mentally retarded with an I. This also points to modularity – externally. The dissociation between language and intelligence is used to establish a typology of language impairment and mental retardation. They lack what is known as a theory of mind. they lack empathy and the awareness of other people as conscious beings with feelings and emotion. in that there exist deficits specific to certain aspects of grammar. Williams syndrome. such as vacuum cleaners. Language is innate. which is also evident from the very robustness of language. in that it is independent of e. Down’s syndrome. some appear to be deaf as they do not respond when spoken to.

animals. 2001 28 As is evident from this short description of autism. . the latter sometimes quite clearly intact.1 Down’s Syndrome As in autism. They are not isolated cases. 4. Frith 1989: 68-81). which includes Broca’s area. which is used to memorize long lists of names. who is a so-called idiot savant. etc. and then discuss a patient called Christopher. According to Deacon (1997: 275) (based on Damasio 1994). The intellectual maturation tends to be complete at around 12 to 15 years of physical age. autistic people tend to have smaller cerebellums and brain stems. Other kinds include mathematical geniuses who can mentally multiply large numbers at a rapid speed. In short. 4. telephone numbers. Down’s syndrome and Williams syndrome. and insanity – all examples of non-normal intelligence/cognition and undamaged language. psychopathy. people suffering from Down’s syndrome (also known as mongolism) are mentally retarded with an upper limit of a mental age comparable to that of a normal 4 to 5 year-old child. All three are instances of mental retardation and spared language. R. actually. subjects suffering from Down’s syndrome do not lack communicative intent and / or ability.K. Unlike autistic patients.2. this dissociation between language and mental retardation or alternative intelligence is abundant: dementia. but the only direct evidence of cortical involvement in the deficit seems to focus on the reduced blood flow of the prefrontal lobes. C. Neurologically. and musical prodigies. who seem to have an ‘island’ of spared skill on a background of generally poor learning abilities. bus routes.2 Mental Retardation and Spared Language In the following three subsections I first discuss two developmental deficits. it is a mentally retarding deficit causing low intelligence and a blocking or mal-acquisition (abnormal acquisition leading to abnormal language) of language per se – not just of speech. autism is a case of mental retardation and impaired language. schizophrenia. Yet another strange characteristic of the syndrome is the so-called idiot savants (see Frith 1989: 84). Often this involves very good long-term memory. the basis of autism is still not completely resolved (cf. Most children with Down’s syndrome begin to speak between the age of two and four years in spite of many physical abnormalities.

The experimenter then asked the subjects where Sally would look for the marble.K. hearing loss due to malformations of the inner ear. Sally placed a marble in her basket and left the scene. The correct response was to point at the basket and the wrong to point at the box. a larynx often positioned too high in the throat. The test consisted of a marble and two dolls. while only 20% of the autistic subjects were correct. C. According to Rondal (1988: 171) their utterances tend to be short and telegraphic with somewhat limited morpho-syntactic elaboration. Then Anne took the marble out of the basket and hid it in her box. 2001 29 Among the numerous factors working against them are pathologies such as an undersized mouth cavity. auxiliaries. and they show limited use of ‘function words’. after which Sally returned. R. Baron-Cohen et al. However. 86% of the normal subjects and 85% of the Down’s syndrome children were correct. a protruding and edematous tongue. such as prepositions.Q. it is surprising that individuals with Down’s syndrome develop language at all. 64). Furthermore. they are very close to normal people regarding the theory of mind.Q. but it goes through the same stages as normal subjects. which however say nothing about morpho-syntactic elaboration. pronouns and conjunctions. due to the physiological deficiencies. These claims are supported by measures of MLU (mean length of utterance). 14 Down’s syndrome children (mean I. Should we need another proof of the robustness of language in the face of biological and psychological hazards. . Down’s syndrome subjects amply supply this proof. language acquisition is generally delayed. (1985) tested 20 autistic children (mean I. or about telegraphic speech for that matter. abnormal lips. copulas. use of function words. In fact. (Rondal 1988: 166) Even though there is inter-individual variability within the syndrome. They have a reduced lexicon but it is used and understood correctly. 82). hence the test is called the ‘Sally-Anne’ Test. etc. the only example in the text (Rondal 1988: 170) includes all of these ‘function words’. Down’s syndrome subjects tend to be very social and generally happy. As Rondal puts it: Given the number and severity of factors that militate against them. As opposed to autistic patients. speech production is impaired. and 27 clinically normal children. hypotonia (reduced sensitivity to stimulation) of the speech muscles. named Sally and Anne.

their range of vocabulary is above their mental age. which includes a star-like pattern in the iris. In spite of this visuospatial impairment their facial recognition is unimpaired. vulture. a narrow face. A striking dissociation exists between their inability to draw e. Another characteristic of Williams syndrome is the unusual vocabulary for the mental age of the patients. and notably very loquacious. They are severely mentally retarded with an I. And in sharp contrast to the autistic patients. passives and various types of embeddings. They have difficulties with dressing themselves. sabretooth. antelope. and generally grammatically correct.Q. As opposed to Down’s syndrome patients. and they are friendly and highly social people.K. R. and they are able to use these structures productively and appropriately. For this reason they are sometimes called “elfin-like” or “elfinpeople” (or as Pinker points out. Their visuospatial capacity is also severely impaired. sharp chins. Their expressive language is complex in terms of morphological and syntactic structures. at approximately 50 and are unable to live independent lives even as adults. and thick lips (cf. recognizable drawing resembling what was intended. horse. but apart from that their language (both comprehension and production) appears to be intact and normal. Williams syndrome subjects suffer from supravalvular aortic stenosis (a narrowing of the aorta) and from abnormalities in the metabolism of both calcium and calcitonin (Bellugi et al. a Williams patient would mention such (to Europeans unusual) animals as hippopotamus. they look like Mick Jagger). tense and aspect markers. finding their way. 2001 30 4. when they draw an elephant or a bicycle it is not a cohesive.2. cat. Language acquisition is delayed. remembering routines.g. an elephant and their ability to describe it – which they do eloquently. This rare and genetically based metabolic disorder is called hypercalcemia and has various effects on the organism: . Bellugi et al. 1988: 183). eyebrow hair growth towards the nose. a broad forehead. For example. 1988: 178. “Complex structures in the spontaneous speech of the Williams syndrome children are abundant” (Bellugi et al. Whereas a normal child would mention a dog. when asked to mention some animals. but a collection of parts (though correct parts). condor. such as inflection. etc. etc. chihuahua. their language use is clearly not echolalic or formulaic. tying their shoes. a flat nasal bridge. 1988: 178). Pinker 1994: 52). cow. C.2 Williams Syndrome Williams patients are characterized by a peculiar facial appearance. etc.

He is institutionalized because he cannot look after himself. Second. C.3 Christopher – a Linguistic Idiot Savant Smith & Tsimpli (1995) describe a very special and quite unique person called Christopher. Third. though no one knows why it has the effects it does. 1985) autistic subjects. He is mentally retarded and has an I. as mentioned above.K.e. write and communicate (though with some differences in degree ranging from fluency to bare elements) in fifteen to twenty different languages for which he has had no formal training. An EEG (electroencephalogram) scan showed slow waves in the frontal lobes. he failed on an equivalent to the ‘SallyAnne’ test (see section 4.2.1 above) showing an apparent lack of a theory of mind. and most interesting. cf. which is characteristic of (and perhaps specific to. p. section 3. reduced activity there. This is very surprising because autistic subjects tend to lack any form of communicative intent and ability. First.1 above). and it acts in complex ways on the brain. Christopher is a linguistic ‘idiot savant’ (cf. He has not been diagnosed as autistic. and if they have any it is . R. 4).Q. (Pinker 1994: 52) As Deacon (1997:269) explains. As Smith & Tsimpli put it. this is the direct opposite neural pattern of the autistic subjects. but several tests and scans have shown that he has suffered hydrocephalic brain damage and severe neural impairment of his motor co-ordination. while the prefrontal lobes are spared. but some evidence points in this direction. Baron-Cohen et al. and internal organs during development. Interestingly. Frith 1989): he cannot button his shirt or tie his shoes but he can read. 4. some cerebral atrophy with wide sulci over both hemispheres. 2001 31 The syndrome seems to be associated with a defective gene on chromosome 11 involved in the regulation of calcium. skull. around 60 (42-76 depending on the specific test. i. postmortem and MRI analyses of Williams patients’ brains have revealed reduction of the entire posterior lobes. the mentioned reduced activity in the frontal lobes is consistent with the reduced cortex characteristic of (at least some types of) autism. The brains of Williams syndrome patients are even more ‘front heavy’ than normal brains.2. which already have relatively large frontal lobes (cf. he has a single spared (or overly developed) mastery or skill in the face of otherwise retarded intelligence: language. his medical and neurological history is rather opaque. ibid.

Christopher is a clear example of a person whose dissociation between language and intelligence is very clear. which also fits his very social nature but not his lack of theory of mind. I mention Christopher in this separate section because he is not a prototypical case of autism but represents a ‘borderline’ case with elements characteristic of both Williams syndrome and autism. . Aphasia is “the loss of or impairment of language abilities following brain damage” (Pinker 1994: 473). I shall briefly discuss the different types in the same order in the following subsections. I give a survey of the different types of aphasia. First. Damasio 1992: 536. The former is characterized by spared linguistic abilities in the face of severe mental retardation. conduction -. six subtypes are distinguished (cf.K. C. which are all caused by brain damage and then I discuss the developmental deficit called Specific Language Impairment.3. global -.3 Impaired Language and Spared Cognition In the next subsections I discuss the kinds of deficits that selectively damage language while leaving general cognition (relatively) intact. Bishop 1988:204): Broca’s -.1 Aphasia The term aphasia actually covers a number of disorders that have as common ground of intact non-verbal cognition and impaired language. it is an acquired disorder of linguistic processing. The latter is however a disorder with much interpersonal variance. as these two traits are among the hallmarks of autism. Typically. and transcortical aphasia. and he clearly has the will to communicate (as do Williams patients). 2001 32 severely impaired. or SLI. and as he lacks the theory of mind and has a single ‘super skill’. This single linguistic skill is characteristic of Williams syndrome. In other words. which is not characteristic of autism. Wernicke’s -. R. he certainly has to be considered autistic to a certain degree. and anomia. 4. 4.

Furthermore. Brodmann’s areas 44 and 45 (also called the left inferior frontal gyrus). which is a drastic loss of speech fluency making speech effortful and telegraphic. i. (beneath the cortex towards the inner parts of the brain). their ability to assemble phonemes into words is also defective. which roughly coincide with functional areas. just to the right of Broca’s area). The left arrow points to area 44.e. prepositions. which is the lower part of the motor strip. The symbols indicate different cortical cell structures. The hallmark of Broca’s aphasia is agrammatism. Broca’s area. C. Fig 7) 2 The vicinity refers to three areas: the operculum.1 Broca’s Aphasia Damage to Broca’s area. auxiliary verbs and tense-inflection. . below). which is a group of convolutions at the base of the Sylvian fissure to the below to the right of Broca’s area. a feature shared with Wernicke’s aphasics (cf. 2001 33 4. R.K. an impairment of the subject’s grammar which manifests itself as an inability to organize words into grammatical sentences and an improper use or non-use of grammatical words and morphemes. The right arrow points to the posterior (rightmost) part of area 22 – this posterior part is Wernicke’s area. and its vicinity2 results in Broca’s aphasia. Finally.1. such as conjunctions. From Sobotta & Becher (1975: 6. The insula. see Figure 5 below. Figure 5: Brodmann’s areas. it refers to the subjacent white matter.3.

1. Broca’s aphasics have difficulties with the interpretation of semantically reversible passives. which cause them to . The mechanism underlying this selective impairment has been the focus of research done by Yosef Grodzinsky (e. 4. The woman was touched by the man. chapter 2) as well as “be kissed” (THEME). 1986. An example of a non-reversible is “John kicked the ball”. which is effortful and slow. (20) (21) (22) The car was driven by the woman. or “dog” for “queen”. This type of aphasia is characterized by the fluency of the speech of the subjects and is therefore sometimes called fluent aphasia. 2001 34 Comprehension is also impaired. For example they might say “chair” instead of “table”. however. In this sense it can be said to be the complement of Broca’s aphasia. 2000) and it is the topic of chapter 6 below so I will not go into further detail here. such as (20). because only “John” can be a “kicker”. see Figure 5 above) and its adjacent areas. Broca’s aphasics have little or no difficulties with nonreversible passives where only one noun phrase can be the AGENT. 1990. much to their own surprise and dismay.K. Another aspect of this disorder is the aphasic’s inability to repeat sentences they hear. C. The man touched the woman. The passive counterpart in (22). 1995a+b. but they are characterized by their lack of sense and frequent neologisms and word substitutions. R.3.g. It does not cause any problems for the aphasic. does. Wernicke’s aphasics suffer from verbal or semantic paraphasia – they have great difficulty selecting the appropriate words that accurately match the intended meaning. Example (21) is an active semantically reversible sentence.2 Wernicke’s Aphasia Wernicke’s aphasia is caused by damage to the posterior region of the left auditoryassociation cortex (Brodmann’s Area 22. They are even unable to repeat the types of sentences that they fully understand. “knee” for “elbow”. The utterances are fluent and more or less grammatical. Speech is laborious and produced with normal (or even above normal) speed and intonation. where both “John” and “Jill” can be “kissers” AGENTS (cf. Broca’s aphasics are well aware of their impairment and are often depressed (Damasio 1992). which are sentences like “John kissed Jill”. They also suffer from phonemic paraphasia.

C. Their non-deliberate (automatic) speech is preserved. and they seem to have no understanding of grammatical words or complex sentences. How can they not be aware of their impairment? Indeed. 4. they use many (unintelligible) neologisms such as “robbli” for “queen” (cf.3. verbs. As their utterances tend to be grammatical with ‘distorted’ words. It is caused by damage to one of two loci: (1) The supramarginal gyrus (Brodmann’s area 40.3 Global Aphasia The combination of Broca’s and Wernicke’s aphasia (i. and the lack of capacity for verbatim repetition.1.4 below) and anomics (see section 4.4 Conduction Aphasia This kind of aphasia shares three features with Broca’s and Wernicke’s aphasia: phonemic paraphasia (word substitution). and sometimes their inability to understand what others are saying makes them anxious and agitated and perhaps even paranoid (Damasio 1992: 534). Conduction aphasia is distinguished by the relative preservation of auditory comprehension (unlike Wernicke’s aphasia) and speech production (unlike Broca’s aphasia).6 below) are not intellectually impaired. Comprehension in Wernicke’s aphasics is also impaired. These structures may be used repeatedly and inappropriately in a vague attempt to communicate.3. However. which perhaps relates to general intelligence. 2001 35 make phonemic substitutions or distortions such as “tubber” for “butter” or “leasing” for “ceiling”. naming problems. On the production side global aphasics’ deliberate speech is limited to a few words and sentences.3. studies have shown that there are some differences in intelligence in aphasics: Broca’s and conduction aphasics (see section 4. They are able to appropriately use an inventory of expletives such as “god damn it” with correct inflection and articulation. whereas Wernicke’s and global aphasics (see below) have been shown to have below normal intelligence (cf. just . 4. Their auditory comprehension is limited to a small set of nouns.1. Furthermore. located in the posterior region of the temporal lobe. both comprehension and speech.1.e. Wernicke’s aphasics are less apt to become depressed and frustrated than Broca’s aphasics are.1.3. Kertesz & McCabe 1975). damage to both Broca’s area and Wernicke’s area) is called Global aphasia and is the almost complete loss of language. They seem to be less aware of their impairment. Harley 1995: 280. Pinker 1994: 310-311). and it has often been assumed that Wernicke’s aphasia is a semantic deficit. and idioms.K. R.

They are fluent patients with normal comprehension and no severe substitutions of words or inflections.e. 45. It is caused by damage to the left frontal cortex above and in front of (and sometimes involving) Broca’s area (i. see Figure 5. which can be found in isolation or accompanying Wernicke’s and Broca’s aphasia. cf. Damasio (1992: 534). 44. parts of Brodmann’s areas 9.3. also known as Brodmann’s area 39. R. see Figure 5 above.K. Literally. Bates et al. see Figure 4 in section 3. Neurologically. It is divided into two variants: a sensory and a motor variant. 2001 36 above Wernicke’s area. Harley 1995: 275). parts of Brodmann’s areas 21. cf. 37. as well as in their ability to recognize spoken or written names of things (cf. All types of aphasia have distinct symptoms and are all associated with distinct areas in the perisylvian region. Transcortical motor aphasia is somewhat the mirror image of the sensory variant. and 46) (cf. (2) The left primary auditory area (Brodmann’s areas 41 and 42 just left of Wernicke’s area. See also Figure 4).1.6 Anomia Anomia is a naming disorder. Speech is nonfluent and comprehension is largely intact. page 33). Pure anomics do not suffer from the symptoms of the other kinds of aphasia.3 (page 23). They have problems with naming and finding the right words (cf. anomia means “no-name-ia”. Figure 5). Damasio 1992: 535). the name of the deficit speaks for itself.e.1. 22. 1991: 144). 4. and 40). . 4. This is summarized in Figure 6 on page 41 below. Grodzinsky 2000: 20). C. anomia is caused by trauma to an area to the right of Wernicke’s area called the angular gyrus (cf. Transcortical sensory aphasia is fluent and involves impaired comprehension. 39. Anomia can be described as a chronical state of “it’s just on the tip of my tongue”. Transcortical aphasia is distinguished by the (relative) preservation of verbatim repetition (cf.3. People with anomia are impaired in their ability to name objects or pictures of objects. As Pinker (1994: 311) states.5 Transcortical Aphasia So far the types of aphasia discussed all involved an impairment of the capacity for repetition. It is caused by damage to temporal or parietal cortex in the vicinity of Wernicke’s area (i. Harley 1995: 272). the insula and the underlying white matter (see footnote 2.

K. R. C. 2001

37

4.3.2 Specific Language Impairment Specific language impairment (SLI) is characterized by severe impairment in the development of language comprehension and production without any mental retardation, motor, hearing, social, or emotional disorders that could account for the impairment. The child is otherwise normal, only language is impaired. A most interesting aspect of this disorder is that it is not caused by brain damage: Few children with specific language impairment have any history of brain disease or any hard neurological signs, and techniques such as CT scan or electroencephalography (EEG) reveal abnormalities only in a minority of children with particularly severe problems. Overall the evidence for an acquired brain lesion as the cause of specific language impairment is slim. (Bishop & Mogford 1988: 258) However, SLI tends to run in families, which points to a genetic explanation for the impairment. Of course, the mere fact that a form of behavior runs in families does not prove a genetic foundation. Consider for instance recipes, lullabies, stories, etc. which are transmitted in families, but friends, neighbors, colleagues and such may also “be contaminated” or “inherit” the use of them. SLI can reasonably be said to have a genetic cause as it runs in the family much in the same way as psoriasis, which affects only some (not all) descendants of a common ancestor and does not afflict close age-mates, friends or other family members. For example, in one single large family of 30 members, the KE family (Gopnik & Crago 1991), half of the members were language impaired, which is quite many as SLI affects only about 7% of children in general (van der Lely et al 1998: 1253). SLI appears to be a rather heterogeneous disorder. Subjects may be with or without articulatory, phonological, or comprehension impairments (cf. van der Lely & Stollwerck 1996), and according to Bishop & Mogford (1988: 259) it is largely defined by exclusion – i.e. if the impairment is not caused by any of the other syndromes it is labeled SLI. Future research may answer the question whether SLI is a cover term for several distinct disorders or a single deficit. In fact, several homogeneous subgroups have been identified, cf. van der Lely & Stollwerck (1996: 486). Other findings point to

K. R. C. 2001

38

a single disorder with a common underlying deficit with several manifestations (as I will argue below that Broca’s aphasia is). However, SLI subjects are mostly characterized by varying degrees of grammatical and morphological impairment in comprehension and/or expression, i.e. a grammar deficit. Van der Lely et al. 1998 have investigated a case of pure grammatical SLI. The subject named AZ has a non-verbal I.Q. of 119, i.e. with intelligence above average. His grammatical skills (morpho-syntax), on the other hand, are severely impaired. AZ has been tested on a battery of tests in three domains: grammatical abilities, non-grammatical language abilities, and non-verbal cognitive abilities. Analysis of AZ’s speech has revealed that he uses only short sentences with frequent (70-80%) omission of inflection (‘-s’ for person and tense) and omission of phrases (“the dog was poking [his head] in [-to the jar]”). He uses very few embedded sentences (2 of 26) and only simple phrases (nothing like “the small black dog”), and he has severe problems in producing wh-questions. In fact, 83% of the wh-sentences contain errors, for example omission or ungrammatical use of auxiliaries, and he has severe problems with morphology and inflection. Furthermore, he is impaired in his ability to assign reference to pronouns and reflexives, when syntactic grammatical knowledge is crucial for the judgement. On the other hand, when non-grammatical knowledge is sufficient for the assignment AZ’s performance is normal (96% correct). For example, in “Grandpa says Grandma is pinching him” the pronoun “him” can only refer to “grandpa” because it refers to a male and therefore only semantic knowledge is necessary for the assignment. On the other hand, in a sentence like “John says that Jack is pinching him/himself” the pronoun “him” can only refer to “John” and the reflexive “himself” only to “Jack”. Here grammatical knowledge is crucially required for the right interpretation as both “John” and “Jack” are males and hence match the semantics of both “him” and “himself”. Here, AZ’s performance did not differ significantly from chance, cf. Van der Lely et al. (1998: 1255). AZ has achieved an overall non-verbal I.Q. ranging from 119 to 131, clearly above average. He shows no deficits in any non-grammatical domain. His impairment is restricted to his grammar. This resembles the diagnosis for Broca’s aphasia, which is also associated with grammatical impairment and normal intelligence – even though Broca’s aphasia is most clearly characterized by reduced speech fluency. However, it is important to distinguish between acquired deficits like aphasia, and developmental deficits like SLI, autism and

K. R. C. 2001

39

Down’s syndrome. SLI is not caused by cerebral trauma. According to Bishop & Mogford (1988: 259), “it is possible that specific language impairment in children reflects an underlying immaturity of neurological development of those brain areas concerned with language development”.

4.4 Discussion
We have seen that autistic subjects suffer from a developmental deficit that, in most cases, renders them mentally retarded with an I.Q. below 60. They are socially withdrawn and have no theory of mind. They lack the ability and intent to communicate and mostly they are mute but those who do acquire language are echolalic and speak monotonously and mechanically. Autism is associated with reduced activity and/or size of the frontal lobes, which result in the blocking or mal-acquisition of language. Down’s syndrome is also a developmental deficit that causes mental retardation resulting in an I.Q. of about 60. Unlike the autistic subjects, Down’s syndrome subjects are very social and possess a theory of mind. Their lexicon is reduced compared to normal subjects, but they are able to use and understand it correctly. Furthermore, Down’s syndrome is clear evidence for the robustness of language, as the subjects acquire and use language in spite of massive physical handicaps in the speech production and perception systems. In comparing autism and Down’s syndrome it is clear that language is independent of intelligence. Both syndromes show severe mental retardation, but they differ on a very important point: one blocks out language while the other does not. Hence, language is not dependent on normal or high intelligence and it is not blocked by low intelligence. Therefore, language acquisition is not based on general learning, as these syndromes clearly limit learning abilities in almost every other field and the subjects are generally poor learners. This is an (indirect) argument in favor of the innateness hypothesis of language for two reasons: one, some sort of neural specialization must be a prerequisite for language learning since general learning cannot apply; and two, there is dissociation between intelligence and linguistic competence. Another conclusion that might (incorrectly) be drawn from this comparison is that language acquisition is dependent on the theory of mind. Perhaps, one might argue, if one does not acknowledge other people as agents and speakers, the required parameter setting would not take place, and hence leave the child without a language. However intuitive and reasonable this may seem the hypothesis does not hold.

Broca’s aphasia. It appears to be caused by an immaturity of the development of the language areas. 2001 40 Christopher is a special case of autism (though he is not a typical case and has not been clinically diagnosed as autistic. of around 50. as the deficit was confined to morpho-syntactic aspects of language.K. C.Q. above normal but a severe grammar deficit. speech is effortless and fluent and it is produced with normal or above normal speed. which has features from both associated types of aphasia: both production and comprehension are severely limited. This also points to a modular internal structure of language. Thus. Comprehension and repetition is also impaired. Williams syndrome is also a developmental deficit. of normal size) – quite the opposite pattern of autism. Their comprehension is also impaired. caused by damage to Wernicke’s area and its vicinity. The subjects have an I. The combination of damage to both Broca’s and Wernicke’s area results in global aphasia. caused by damage to either supramarginal gyrus or to the left primary auditory . which is characterized by normal or even high intelligence and a selectively impaired language. but is nonsensical and full of semantic and phonemic substitutions and neologisms. and they are very social and loquacious. because he has language as his special ‘super skill’. language impairments caused by brain damage. We turn now to the different kinds of aphasia. but speaks several languages. cf. Smith & Tsimpli (1995) and section 4. 1998 of pure grammatical SLI offers further support for the above mentioned dissociation between language and intelligence: The subject AZ had an I. R. caused by damage to Broca’s area and its vicinity. The last of the developmental deficits I discuss is SLI. But again. He has reduced activity in the frontal lobes. is characterized by agrammatism and a drastic loss of speech fluency.Q. Grammatical impairment seems to be a defining characteristic of SLI. while the frontal lobes are spared (i. his test results point to autism. They have an unusual vocabulary. but he is clearly not deprived of language. The study by Van der Lely et al.2. the case of the autistic Christopher provides counter evidence. This pattern leads Deacon (1997) to propose a symbolic processor in the frontal lobes that controls language. Their language is intact.e. In this way he is a counter example to the hypothesis. Their posterior lobes are reduced. Conduction aphasia. which result in effortful and telegraphic speech. both in terms of comprehension and production. SLI has no neuropathological basis. clearly above their mental age. though the deficit is rather nonhomogenous and has several manifestations. In Wernicke’s aphasia.3 above). as it is not caused by deformity of the brain. language is independent of the theory of mind. He lacks the theory of mind.

I have summarized the loci of the different types of aphasia in Figure 6 below: Figure 6: Cerebral regions and their associated types of aphasia. Finally.K. caused by damage to the angular gyrus. anomia. The two variants of transcortical aphasia are characterized by the ability to repeat spoken sentences in contrast to the other types of aphasia. is a naming disorder. It also offers some support for the modularity hypothesis. or rather mental retardation. Anomics understand nouns and names correctly but have problems finding the right ones for expression. R. This typology is summarized in Table 4 below. which clearly shows the double dissociation between the two. naming problems. C. an inability of verbatim repetition. and interestingly a sparing of production and comprehension. 2001 41 area. 3) It is now possible to establish a typology of language and general intelligence. is associated with phonemic paraphasia (substitutions). The categories ‘mentally normal’ and ‘mentally retarded’ are not . Fig. Based on Sobotta & Becher (1975: 4.

htm].I. Aphasia e.L.g. as for example Wernicke’s aphasia is also associated with reduced I. . because they refer to cases not covered in this study. as a consequence of the trauma.K. blindness due to brain damage S.dk/engelsk/engsv/medfoedt. C. 2001 42 intended to be interpreted too literally.g. R.3 3 For a similar classification see Vikner’s web-site [http://www. insanity.hum. psychopathy Down’s syndrome Williams syndrome ? Autism Table 4: A typology of language impairment and mental retardation.au. MENTALLY NORMAL NORMAL / GOOD LANGUAGE MENTALLY RETARDED IMPAIRED / POOR LANGUAGE nerologically normal nerologically abnormal nerologically normal nerologically abnormal Normal subjects e. Some areas are shaded.Q.

(Damasio 1992: 534) At a first sight. C. cf.3.g. For this reason this kind of clinical description and definition is insufficient in that it seems to presuppose that language is a unitary skill (underlying the processes of comprehension and production). the characteristic agrammatism of Broca’s aphasia). not all of it. Broca’s aphasia is not a loss of syntax and likewise Wernicke’s aphasia is not a loss of semantics. grammatical structure (cf. as for example aspects of syntax are impaired in e. A description of the function of a cerebral area in terms of language processes. the coupling of phonological representations with semantic concepts. page 33) is part of the network concerned with the relational aspects of language – that is. is misconceived. as mentioned in section 3.]. in the form of an internal auditory representation or of vocalization. Broca’s aphasia. […] Wernicke’s area is no longer seen as a center for word selection. According to Damasio (1992: 534). footnote 2. However. For similar reasons a description in terms of syntax versus semantics is also insufficient. R. Pinker labels the left perisylvian region the language organ. Rather. the classic clinical description of language is based on the processes of comprehension and production. On the other hand. such as production or comprehension. K. or in terms of traditional linguistic distinctions. R. Pinker (1994) seems to imply a classic localizationist approach: grammar in Broca’s area and lexicon in Wernicke’s area. Wernicke’s area is […] a processor of speech that allows sounds to be mapped as words and to be used subsequently to evoke conceptual meanings [i. morphemes and verbs (Broca’s aphasics have difficulties especially with verbs. it appears that once a word is selected for possible use in an utterance. none of the above mentioned language impairments involve only a complete and undifferentiated loss of only linguistic comprehension or only production.e. are the functions of Broca’s and Wernicke’s areas? Let us consider some proposals. then.K. Broca’s area (and its vicinity. In other words. such as syntax and semantics. cf. and hence he cannot at the same time lump all of grammar into one distinct small area like Broca’s area and the . But. C. What. Broca’s area is concerned with syntax and morphology. Wernicke’s area is part of the network required to implement its constituent speech sounds. 2001 43 5 Cerebral Area and Function As is clear from the survey of the different types of aphasia. Bates et al 1991).

then what would be the function of the rest of the language area (the perisylvian region)? He clearly does not make this mistake: So. which they both state is part of the network responsible for ‘relational aspects’ of language. When asked to distinguish grammatical from ungrammatical sentences. […] And. which then must be handled by something else.K. some kinds of grammatical abilities seem to survive damage to Broca’s area. Rather. based on work by Roman Jakobson (1956). C. Wernicke’s area he assigns a role in looking up words in the lexicon and sending them on to other areas of the grammar. If he did. p. For this reason it is not possible to determine exactly what Damasio has in mind. Wernicke’s and Broca’s areas are responsible for the computation of paradigmatic and syntagmatic relations respectively and are as such bottlenecks in larger computational chains. . Damage to Broca’s area alone usually does not produce long-lasting severe aphasia. is quite compatible with Damasio’s (1992) model described above. 2001 44 entire lexicon into Wernicke’s area. the lexicon itself can be distributed over a much larger area. This model. there is no language-specific cerebral region. (Pinker 1994: 309) Still. at least for Pinker. In Pinker’s view. To him. Both authors seemingly agree that this phonological form is implemented by the grammar in some sort of an internal representation of the utterance (which. R.310). Deacon (1997). some Broca’s aphasics can detect even subtle violations of the rules of syntax […]. would correspond to PF in Figure 1 in chapter 2). has a different approach. so the role of Broca’s area in language is maddeningly unclear (ibid. This is at least partly done in Broca’s area. most surprisingly of all. Thus. nor do all aphasics detect them. is Broca’s area the grammar organ? Not really. to put it crudely. sentence structure and morphology. a word’s lexical entry includes both semantic content and phonological form. such as sequential order. I think. aphasics do not detect all ungrammaticalities. Damasio (1992) is typical of clinical papers on these matters in that no particular linguistic theory is mentioned and no linguistic examples are included in the text. Damasio suspects that Wernicke’s area has something to do with assigning a chosen word its proper phonological form but does not mention lexical semantics. Wernicke’s area is just the “librarian”.

which Deacon correlates with the diminished frontal lobes in autism. (Deacon 1997: 305-306) To Deacon. a lack of drive to communicate (mutism. can be caused by damage to the internal cerebral surface of the left hemisphere in the area including the anterior cingulate gyrus and supplementary motor area. He explains the two different kinds of relationships as follows: In the most general sense. Cases of idiot savants (Fay 1988.g. that Williams patients have normal frontal lobes and (near) normal language. Damasio (1992:537). adverbs. that normal subjects with normal frontal lobes have normal language. and that autistic patients have abnormal frontal lobes and abnormal language do not place language in the frontal lobes. sections 4. nouns. R. Coupled with the fact that trauma to the perisylvian region causes language deficits.2. verbs. not autism). art. The facts just stated cannot account for the aphasic effects of brain damage to the language area. this leads him to postulate that the frontal lobes house the central symbolic processor and that Broca’s and Wernicke’s areas are computational centers for information from the frontal lobes.1). Furthermore. Frith 1989) clearly cast a shadow of doubt on this.1 and 4. music. the frontal lobes house a symbolic processor. cf. C. […] Syntagmatic operations are reflected in the complementary relationships between words from different parts of speech (e. But the mere facts. adjectives. I find this at least dubious. it should be kept in mind that there is a difference between acquired impairment and developmental deficits. 2001 45 linguistic as well as non-linguistic. or articles) and the way these different classes of words alternate in sequence in a sentence. and Deacon correlates this fact to the species-specificity of language – or symbolic communication and thinking. How can one explain autistic geniuses of e. Deacon’s proposal fits quite well with the finding that there is a severe reduction in blood flow in the frontal lobes of autistic subjects with severe communicative impairments and with the spared frontal lobes of William’s syndrome patients (cf. As Broca’s area is part of the region affected in autism one would predict that the capacity for syntagmatic computation per se should be severely impaired as well. section 3.. or .2 respectively).g. Again. all words of the same part of speech are paradigmatic of each other to some degree since they can substitute for one another.K. The frontal lobes are very large in proportion to the rest of the brain in humans (cf.

2001 46 mathematics? These abilities clearly involve syntagmatic computations.g. . issue 41. but a much less severe effect on Italian. cf. (Deacon 1997: 307) 4 Bates et al. C. The important point is that a lesion in Broca’s area has different effects on e. English (and German) uses fixed word and phrase position to signal such grammatical relationships as possession. The same relationships are signal by inflectional affixes in highly inflected languages like Italian (and Latin). question vs. the claim is.3 above). will have great effect on English. depending on the relative importance of positional or inflectional tricks for cuing [sic] grammatical decisions in different languages. then the parts of this module map in very different ways to different grammatical operations. Christopher (who I argued is autistic in section 4. Smith and Tsimpli (1995). which in turn would be more effected than English (regarding syntax) by a damage to Wernicke’s area. (1991)4 who show that there is cross-linguistic variation in the correlation between lesion site and aphasia type. As support for his hypothesis he cites findings by Bates et al. active vs. as they cannot be described as mere skills of substitution. since his abilities involve grammatical skills as well. and if so (as he actually has) he should only be able to compile lists of words in paradigmatic relation to each other (due to the sparing of Wernicke’s area). R. subordination. statement. which supposedly is responsible for syntagmatic relations. English and Italian. Deacon interprets this as evidence for his claim that there is no language-specific module: So if there is a grammar module.2. This sort of module is a will-o’-the-wisp. or in other words. and these can differ from language to language” (1997: 306). It is rather superfluous to state that music and mathematical equations are sequentially ordered and dependent on syntagmatic computation. Christopher’s linguistic “super skill” is clearly not limited to recitation of huge lists of remembered words. passive.K. damage to Broca’s area. and clearly contrary to the facts. This prediction is not borne out. should not be able to learn language. (1991) is actually a summary of findings by a number of researchers published as articles in a special issue of Brain and Language. as paradigmatic operations. Deacon states that “Syntactic operations and grammatical judgements can involve many different syntagmatic and paradigmatic processes. Hence. Furthermore.

4). R.K. The mere distinction between syntagmatic and paradigmatic relations within a semiotic system has little explanatory or descriptive value. due to the co-evolution of language and brain. lumping together all aspects of grammar into one is a misconception. he claims that language is acquired so easily and fast because humans have a special kind of learning ability to do so. Actually. I have already argued that this dependency is false (cf. such as syntax vs. Nevertheless. at least in linguistic circles (as far as I know). as language is clearly located in the language zone of the left hemisphere (the frontal lobes may. Deacon’s proposal. As already mentioned. Grammatical abilities are not lost in an either/or manner. C.1 above) as well as externally (cf. however. still be crucially involved in symbolic cognition as such). they share a common feature: they all place a responsibility for syntagmatic. most clearly reflected in the theory of mind. 2001 47 But this variation is only strange if one assumes that language competence is divided into only two modules: grammar and lexicon. Pinker. I don’t think that there is much in favor of Deacon’s hypothesis that Broca’s area and Wernicke’s area are but computational centers that also compute linguistic signals from the frontal lobes. one of the cornerstones of generative grammar is that language is modular. His claim that the frontal lobes house the central processor that is also responsible for language is also rather dubious. paradigmatic. Damasio is a clinician and it is unclear to which linguistic framework he would adhere. Furthermore. That is. is a prerequisite for language. relational. What is lacking in the three proposals is a grounding in a specific linguistic theory. and Deacon are all too unspecific in their proposals on the functions of the areas inside the language zone. believes in this kind of gross localizationist approach anymore. lexicon or syntagmatic vs. No one. In fact. as I have shown. a framework that provides the proper descriptive tools to give an account beyond the inadequate dichotomies. Though Pinker adheres to the Principles and Parameters framework (see chapter 2) he does not go sufficiently into detail in his 1994 account. Damasio. chapter 4).e. has little to offer but . Deacon claims that general symbolic cognition. i. section 2. In the following chapter I shall return to this aspect with a refined theory of grammar. or sequential computation of grammatical structure on Broca’s area (and vicinity). internally (cf. general intelligence. such that a lesion in Broca’s area would lead to a complete loss of all grammatical competence. section 4. which is capable of describing the linguistic competence of normal language users and which accounts for the observed breakdown patterns – a theory that is breakdown compatible (Grodzinsky 1990: 111).

I think. which also provides a description of normal speakers’ competence in terms of a specific grammatical framework.K. However. this phenomenon is accountable for by a generative framework. The fact that language deficits vary cross-linguistically is. 2001 48 the old distinction between syntagmatic and paradigmatic relations. C. . R. the only linguistic phenomenon he does account for and that is not covered by the others. as I shall show below.

There was a disruption. C.g. while the latter is the result of damage to a hitherto normal brain. (Emphasis added. and thus did not appear to have “asyntactic comprehension”.K. and the conception of the two being localized in two distinct areas of the left hemisphere. 2001 49 6 A Syntactic Approach to Broca’s Aphasia The distinction between syntax and lexicon.) (The meaning of the loose terms “many syntactic constructions” and “certain aspects of syntax” will become clear below. and that distinctions within syntax were needed to account for the comprehension deficit. It appears that the anterior part is responsible for the syntagmatic aspects of linguistic computation. does not lead to the loss of all of syntax. It was becoming clear.) Let me once again point out the important distinction between developmental and acquired impairment. As Grodzinsky (2000: 4) states: “On testing. which clearly includes syntax. that a distinction between different levels of linguistic analysis would not suffice. Bates et al 1991) as well as pathologically. then. Damage to this area. Therefore. The theory must meet the Criterion of Breakdown-Compatibility: . Language functions are distributed over the language zone. R. was argued in the preceding chapters to be insufficient and wrong on empirical grounds. it is obvious to consider Broca aphasics’ linguistic competence as a normal grammar with some damaged part or parts – not as a completely different grammar. Broca’s aphasics showed near-normal abilities in comprehension and grammaticality judgement on many syntactic constructions. As Broca’s aphasia is an acquired impairment. but it was restricted to certain aspects of syntax. the theory that accounts for the linguistic competence in normal people has to able to account for the competence in aphasics as well. both cross-linguistically (as shown in e. just as they were for speech production. The former is the result of a brain that is different from a normal brain. however.

of course.K. 6. The scientific field concerned with the linguistic abilities of the human species is. as already mentioned.2 I present the Tree-pruning Hypothesis. Dutch. Based on research both by the author himself as well as others in a diversity of languages including Japanese. their entire grammar cannot be lost. Then. The following domains of grammar are supposedly left unimpaired in processes of . Russian. First. it is natural to approach the domain of aphasia from a syntactic point of view. Granted that this is true. and I shall argue that the deficits to both of these domains are due to selective impairments to aspects of grammar. the production errors are caused be a disruption to the syntactic tree. see also Grodzinsky 1986: 137) The accounts in the preceding chapters have primarily been clinical accounts. Grodzinsky (2000. Italian and English.1 I present an account of the comprehension deficit. their understanding is not entirely impaired. in which one or more nodes (maximal projections. linguistics. in order to properly describe what impact trauma to different regions of the brain have on peoples linguistic competence. 2001 50 (23) The Criterion of Breakdown-Compatibility: Every pattern of impairment and sparing of linguistic ability must be accounted for in a natural. in section 6. see also 1990) argues that in fact most of the Broca aphasics’ grammar is intact. such as word order. in section 6. Therefore. (Grodzinsky 1990: 111. Chinese. which accounts for the production impairment in Broca’s aphasia. R. cf. the framework of linguistics should also be applied to language impairments such as aphasia. and their speech production also present intact features of grammar. Hebrew. In the next sections agrammatic comprehension and production will be viewed from a syntactic point of view. In fact. the hypothesis is that the comprehension problems associated with Broca’s aphasia are due to a loss or deletion of traces in the syntactic representation. Briefly. chapter 2) are missing from the representation. C. Hence. This deletion of traces is what causes the subsequent misinterpretation of thematic roles. non-ad hoc fashion. According to the hypothesis. and in turn to aid in establishing proper diagnostics. the hypothesis is called the Trace Deletion Hypothesis.1 Comprehension: The Trace Deletion Hypothesis As Broca’s aphasics have some linguistic competence.

iii. where a noun phrase is moved from object position to subject position leaving a trace: (24) [The boy]1 was pushed t1 by the girl. and argument structure. which are constraints on anaphoric relations between pronouns and reflexives and their antecedents (Chomsky 1981. Grodzinsky & Reinhart 1993). C. THEME etc. They are able to detect violations of subcategorization.g.).e. Grodzinsky 1995: 31. Balogh & Grodzinsky 2000. However. such as grammatical case (for example ACCUSATIVE or DATIVE) which is typically assigned to canonical positions. R. They are also able to detect violations of phrase structure rules (cf.K. Broca’s aphasics know the thetaroles of predicates and are able to assign them directly to positions. The module called the Theta-theory is also intact. They are able to compute (interpret) some intra-sentential dependencies. iv. 1993). In comprehension. Haegeman 1994). certain aspects of pronominal reference are impaired. Broca’s aphasics can construct basic syntactic trees for simple sentences not containing intra-sentential dependencies (such as trace-antecedent relations). Broca’s aphasics suffer severe difficulties with constructions involving syntactic movement. In contrast to these spared abilities. i. ii. Grodzinsky et al. for example active sentences. but they have to do with discourse (pragmatic skills) and not with intra-sentential binding relations (cf. They seem to have no impairment to the part of the lexicon that interfaces with sentence grammar. 2000: 4) (I shall deal with production below): i. They are also able to handle binding relations (cf. such as the object position. Grodzinsky 1995a). . such as passives. such as ii-iv. the verb "to eat" subcategorizes for a NP such as "a cake". e. 2001 51 comprehension in Broca’s aphasia (cf. how many and what kinds of arguments (AGENT. which is evident from comprehension tasks on simple structures such as active sentences.

. C. who pushed the boy it is the girl who pushed the boy the boy was interested in the girl the boy admires the girl [the boy]1 was pushed t1 by the girl [the boy]1 who the girl pushed t1 was tall show me [the boy]1 who the girl pushed t1 it is [the boy]1 who the girl pushed t1 [the girl]1 is admired t1 by the boy Theta-Structure AGENT-THEME AGENT-THEME AGENT-THEME AGENT-THEME EXP. the role of THEME is assigned by the verb to the position to its right – the typical object position of actives. Grodzinsky 1995b: 491 or Balogh & Grodzinsky 2000: 17). Haegeman 1994) the argument to the left (in this case "the girl") which is optionally adjoined in a PP. IP] to be assigned NOMINATIVE.THEME EXP. The experiment is supposed to probe the patient’s abilities on theta-assignment to structural positions. subject case (see chapter 2 for more information). Canonically.g. Performance above chance above chance above chance above chance above chance above chance Chance Chance Chance Chance below chance Table 5. The passive morphology absorbs (cf. and those where the patients perform below chance (for statistical details. see e. Note also that the θ-role EXP.K. R. at chance level.. is short for EXPERIENCER.THEME THEME-AGENT THEME-AGENT THEME-AGENT THEME-AGENT THEME-EXP. In order to abide the Case Filter. and will be described in due course below. For example. which depicts the semantic content of the sentence. Numerous experiments (see for example Grodzinsky 1990 and 1995a) on semantically reversible sentences have supported the claim that transformation (i.. chance refers to 30%70% correct performance. the patient hears the sentence "the boy pushed the girl" and has to choose between a picture where a boy is pushing a girl and a picture where a girl is pushing a boy. Roughly. The experiments typically involve the sentence-to-picture-matching test. movement) implies comprehension problems. I have summarized these findings in the following table: Construction Active Subject-subject relative Object-subject relative Subject cleft Adjectival passive Psychological verb Passive Subject-object relative Object-object relative Object cleft Psychological passive Example the girl pushed the boy the girl. 2001 52 In (24) the NP "the boy" is the THEME and "the girl" is the AGENT. who pushed the boy.e. is angry show me the girl. For expository reasons the total syntactic complexity of the examples is left out. the one doing the pushing. in which the patient hears a sentence and then he/she has to point out a picture. the NP the boy is moved to [spec. The data categorizes sentential structures into three sets: Those where the patients perform above chance.

passives. hate. 2001 53 The term psychological verb refers to non-agentive verbs. From the table above it is clear that the distinguishing factor dividing the syntactic constructions is movement. Third. C. R. Keep in mind that traces are constructs (cf. the hypothesis does not predict chance performance in itself in tests on e. and deletion does not refer to some sort of mental eraser removing ts from trees in the patient’s head. The Trace-Deletion Hypothesis alone. Second. for three reasons. 1990. The point is that the thetatransmission fails and this can be accounted for by saying the traces of the moved NPs are deleted. and admire. The agrammatic tree either lacks the traces of the moved NPs or the links between the trace and its antecedent is broken. love. the moved NP will not have a theta-role assigned to them because the trace that would normally transmit the role is deleted. In either case. Perhaps a note on the term ‘deleted’ is in order. I will not go into further discussion on the two possibilities (no traces or broken chains . chapter 2). so kick is an agentive verb. and hence . You can deliberately kick someone (you are the AGENT). This finding gives rise to the Trace-Deletion Hypothesis (I refer the reader to Grodzinsky 1986. the theta-transmission fails. under the VP-internal Subject Hypothesis (see chapter 2) all constructions involve movement of the subject. only that they are problematic and that comprehension is impaired. however. 1995a+b and 2000 for elaboration): (25) The Trace-Deletion Hypothesis: In agrammatism. You can not deliberately love someone. in the representations of a sentence in Broca’s aphasics. So.K. but you can experience (feel) love for someone (you are the EXPERIENCER). is not enough to account for all the constructions in the table. such as feel. In comprehension the hearer constructs a mental grammatical representation of the utterance. it does not predict that the performance on psychological passives would be below chance. which can be represented orthographically as a syntactic tree. First.note that chains are instances of the relational aspects of language discussed in the previous chapter). traces in θ-positions (structural positions to which θ-roles are assigned) are deleted from the syntactic representations.g. The other types all involve movement of the ‘underlying’ object to subject position and hence the representations have traces. In the types on which agrammatic aphasics perform above chance (near normal) there are no traces in the examples given in the table. which denote experience rather that action.

The NP John.K. This strategy applies after the grammar has built the syntactic representation with the missing theta-role(s). C. Let’s consider an example: (28) AGENT. In normal as well as in aphasic speakers. such as world knowledge – this is called the informational encapsulation of the grammar (cf. What remedies these problems is the application of a heuristic non-linguistic strategy that assigns NPs theta-roles according to their linear position. This strategy is based on knowledge of the world. which means that the initial (argument) NP will be assigned the role of AGENT: (26) The Default Strategy: Phrasal constituents with no theta-role are assigned one by default. 2001 54 the prediction is that all constructions would be problematic. which is not borne out. by linear considerations (NP1=AGENT). Grodzinsky 1990). Fodor 1983. is moved and its trace is deleted (illustrated with _) and therefore theta-transmission is impossible leaving the NP without a θ-role. Then the Default Strategy kicks in and assigns it the role of agrammatic subject is forced to guess which is correct. R. This leads to chance performance as the [The girl] is admired _ by the boy THEME AGENT EXPERIENCER EXPERIENCER (normal assignment) (agrammatic assignment) . in Broca’s aphasia the Default Strategy is applied blindly resulting in counterintuitive assigning: (27) John was killed _ by Bill THEME AGENT AGENT AGENT (normal assignment) (agrammatic assignment) by the In this example Bill is not moved and is therefore assigned the θ-role AGENT grammar. The representation now has two AGENTS. The second problem was with psychological passives. the information the grammar uses in the construction of representations is not available to higher cognition. such as frequency of occurrence / canonical order of roles. on the other hand. Therefore. which are in competition.

Jackendoff (1972) ranks the θ-roles in a hierarchy according to thematic salience: (29) The Thematic Hierarchy: AGENT > EXPERIENCER > LOCATION. Another such hierarchy is the Animacy Hierarchy. The third problem was of a theoretical nature. R. GOAL > THEME Thus. and EXPERIENCER. which is not predicted by such an account. hence only one interpretation is made in (28). he or she will guess. The two θ-roles in the agrammatic representation are not identical like they were in (27).K. According to such an account. when the aphasic encounters a structure that does not have the AGENT first. as was the case in (27) above. even though it may seem a bit odd to consider admiration as something one deliberately does (I deliberately admire her / I admire her on purpose). This supports the transformational account outlined here (see also Grodzinsky 1989). The agrammatic interprets such sentences directly complementary of normal language users: In (28) the girl is doing the admiring. not the NP the boy even though it is assigned the role of EXPERIENCER by the grammar as this role is lower ranked. and this would result in chance performance. This consistent reversal of the roles in sentences with psychological verbs is the reason why an account merely based on canonical order fails. According to the VP-internal subject hypothesis the subject moves out of the verb phrase (in which it is base . However. Comrie 1989: 128): (30) 1ST / 2ND PERS. In my account I restrict myself to NPs that are animate. see Grodzinsky 1995b. A NP is higher in animacy (more animate) the more to the left it is on the following hierarchy (cf. PRONOUNS > OTHER HUMAN NPS > ANIMAL NPS > INANIMATE NPS For an account of experiments involving non-agentive animate and inanimate NPs and therefore θ-roles other than AGENT. THEME. SOURCE. 2001 55 Here the result of the application of the strategy no longer leads to competition between two identical roles. in (28) the girl is doing the admiring because of the role as AGENT. To account for this we have to consider how the different theta-roles are related. not the boy. C. the NP assigned the highest ranking is ‘the doer’. the data shows performance below chance and a reversal of the roles.

somewhat reminiscent of the Jakobsonian syntagmatic vs.g. Without the Default Strategy the subject would not be assigned a θ-role. Mostly. The Default Strategy states that constituents without grammatically assigned θ-roles are assigned one by linear considerations. and thus compensates for the lost grammatical θ-transmission and leading to the right The interaction of the two parts lead to chance performance on movement derived clauses (except psychological passive.2 Production: The Tree-Pruning Hypothesis The production side of the deficit does not mirror the comprehension side. Through the application of the Default Strategy the subject is assigned AGENT in non-problematic actives: (31) [John [(t) killed Bill]] AGENT AGENT THEME THEME (normal assignment: θ-transmission from t) (agrammatic assignment: Default Strategy) Under normal circumstances the subject is assigned its θ-role by transmission from its trace. section 4. the account of agrammatic comprehension consists of two interacting parts: the Trace Deletion Hypothesis and the Default Strategy. which is associated with below chance performance). I have . as there is cross-linguistic variation within this deficit (e. However. traces in θ-positions are deleted from the syntactic representation.K.1 above). it is characterized as improper use or omission of grammatical words and morphemes (cf. In agrammatic comprehension. al 1991 and Grodzinsky 1990. this is not possible as the trace is deleted (or the trace-antecedent chain is broken). The same θ-role is assigned. In short. Bates et. however. but for a different reason. AGENT. Bates et al. which leads to separation of θ-assignor and θ-assignee.3. which in turn results in NPs without θ-roles. 6. 2000). After all. R. (1991) relate this variation to differences in competition between two emergent connectionist neural networks. paradigmatic computation of information that Deacon (1997) supports. a number of researchers have pointed out that this is not altogether true. The strategy assigns it interpretation. 2001 56 generated) to receive NOMINATIVE. such that the first NP (in a θposition) is assigned AGENT. it was long believed that Broca’s aphasia was an impairment of speech production with a sparing of comprehension. According to the Trace Deletion Hypothesis. C.

English and Japanese) +Zero-morphology -Zero-morphology omission substitution (e. There is thus no violation of constraints on lexical well-formedness. 2001 57 already explained why this approach.g. R. Grodzinsky’s studies (1984.e. substitution errors occur instead: (32) (e.K. Otherwise the inflectional morpheme is substituted. I think. if the language has – zero-morphology. as omission of inflection only occurs if the resulting word is a real word. Hebrew) Agrammatics do not produce non-words (cf. In both nominal and verbal inflected elements morphemic omission is observed if the bare stem is a real and well-formed word in the current language.e. I guess six month… my mother pass away (Omission of number and tense inflection) Japanese: inorimasu (correct: inorimasushita) I-prayed I-pray (Omission of tense inflection) . C. Therefore. Italian. It obeys the important Criterion of BreakdownCompatibility. The differences between languages are constrained by the typological space defined by the possible parametric variation of the universal grammar. and 2000) have shown a correlation between zero-morphology and omission of inflection in agrammatics (i.e.g. in my opinion. if the omission of inflection results in nonwords. Russian. Otherwise. Grodzinsky’s approach to this is. Here are some examples of omission and substitution respectively (taken from Grodzinsky 1990: 52): (33) Omission: English: Uh. it should not be surprising to find that at least some aspects of agrammatism reflect this parametric variation. is wrong. This supports the proposed connection between the typological parameter (+/. 1990. giving rise to grammatical aberration. more elegant. people suffering from agrammatism) in a variety of languages.zero-morphology) and observed agrammatic behavior. Grodzinsky 1990: 52). i. if the language has +zero-morphology. i. oh.

These findings offer support for the Split-Inflection Hypothesis (Pollock 1989). Impaired agreement implies impaired tense. Icelandic. however. On the other hand. is never found. while not having problems with agreement (based on both new tests as well as retrospective literature reviews). not vice versa.) Red (FEM. we my husband and I”) (Substitution of agreement inflection: tiyl-nu > tiyl-u) The deficit is.my and-I (“They took a walk. Dutch. The omission/substitution is determined by the elements’ structural position. there is dissociation between tense and agreement inflection. 2001 58 Substitution: Russian: grustnaja malchik sad (FEM. more restricted than to the correlation between omission/substitution and +/-zero-morphology. Arabic.PAST. according to which the INFL node IP (inflection phrase) in (34) is split up into an agreement phrase AgrP and a tense phrase TnsP in (35) (NegP is short for negation phrase): (34) (35) [CP spec C0 [IP spec I0 [NegP spec N0 [VP spec V0 compl ]]]] [CP spec C0 [TnsP spec Tns0 [NegP spec N0 [AgrP spec Agr0 [VP spec V0 compl ]]]]] .) went (Substitution of agreement inflection: ross-o > ross-a) Hebrew: tiylu anaxnu ba’ali ve’ani ‘take-a-walk’. According to Grodzinsky and Friedmann (1997. Japanese has shown that agrammatics may be impaired in tense.PL. 2000). inflection: grustn-iy > grustn-aja) Italian: Cappuceto rossa andava Little Ridinghood (MASC.3rd PERS. C. Friedmann & Grodzinsky (1997. cf. R. we husband. Italian.) boy (MASC. This runs contrary to the common belief that Broca’s aphasics have equal problems with all functional categories. impaired agreement and spared tense. the opposite. French. Finnish. German. Swedish.K.) (Substitution of agr. Evidence from Hebrew. 2000).

A gardener may trim a tree to control its growth by cutting off branches. . The hypothesis is further supported by the finding that the verb stays in situ (in its base position) instead of moving up the tree6: In Germanic languages such as Dutch. which means that the TnsP and CP ‘branches’ are missing.g. Dutch and German children place the finite verb in V2 position. a non-finite form always appears in a sentence-final position. 6 This phenomenon is also known from studies of language acquisition (cf. This is called pruning. Children also produce sentences with uninflected verbs in situ.K. C. Based on these findings Grodzinsky (2000:16) states the following hypothesis: (36) The Tree-Pruning Hypothesis: Agrammatic aphasic patients produce trees that are intact up to the Tense node and “pruned”5 from this node up Or in more formal terms (from Friedmann & Grodzinsky 2000:25 [pdf version]): a) T is impaired in agrammatic production b) An impaired node cannot project any higher. and thus prevent he tree from growing higher. 2001 59 The ordering or ranking of the functional categories is not uncontroversial. If the AgrP node is impaired. If the tense node is impaired. R. It also allows for varying degrees of severity. Crucially. German. and errors of agreement will occur accompanied by errors of tense. if the verb moves it will be finite (inflected for tense). something which I will come back to in the next chapter. and hence only tense errors will occur (no agreement errors). while the non-finite forms are sentence final. TnsP and CP are disrupted. indicating that the verb has not moved up the 5 The term “pruned” refers to the way trees are trimmed. such verbs are also called root infinitives. the syntactic tree is disrupted from AgrP and up. patients frequently use the infinitive instead of the inflected verb. In the same way the syntactic tree in agrammatic production is cut or rather destroyed from the tense node up. and Icelandic. This dissociation between tense and agreement suggests a possible deficit that involves the tense node but not the agreement node. Wexler (1994) argues that children go through an “optional infinitive” stage. The same results are predicted in agrammatism: if the verb cannot move due to impaired nodes under the Tree-pruning Hypothesis it stays in situ in the infinitive form in V0. Wexler 1994). The interesting point is the position of the verbs: e. in which they utter both sentences with finite verbs and sentences with non-finite verbs (infinitives). English and French speaking children put the finite verb before the negation and the non-finite (infinitive) after the negation.

K. R. C. 2001

60

tree to C, where tensed verbs in matrix clauses of V2 languages should move. (Friedmann & Grodzinsky 2000: 8 [pdf version]) Furthermore, Grodzinsky & Friedmann (1997, 2000) show that agrammatics make no Wh-questions and no embedded sentences involving a complementizer in CP. Both structures depend on the presence of CP, since wh-elements (such as English who and which) normally move to [spec, CP], and complementizers like that are generated in C0. Correlated with the disruption of tense and the sparing of agreement, the absence of embeddings supports the Tree-Pruning Hypothesis. As mentioned, their model also allows for varying degrees of impairment. In this (rather simplified) tree structure the double slashes (//) indicate possible points of impairment: (37)

[CP // [TnsP // [NegP [AgrP // [VP

Agrammatics may be impaired from agreement up (deficit to agreement, tense, and CP), from the tense node up (deficit to tense and CP), or just to the topmost node CP (see Grodzinsky & Friedmann 1997). In the next chapter I want to apply the Trace-Deletion Hypothesis, the Default Strategy and the Tree-Pruning Hypothesis to Danish, which is a verb-second (V2) language. Because of its V2 status certain predictions can be made about Danish agrammatic speech production. Furthermore, I present an alternative order of the functional categories TnsP and AgrP. I also show how differences in degrees of severity are predicted to be reflected in the speech production of Danish Broca aphasics.

K. R. C. 2001

61

7 Is Grodzinsky’s thesis applicable to Danish?
Linguistic competence is universal to humans and only humans (the hypotheses of Universality and Species-Specificity) and based on common cerebral predisposition (the Innateness Hypothesis). Therefore a theory assumed to be able to account for language breakdown has to be applicable to all languages of the world. For this reason it is important to test whether Grodzinsky’s hypotheses are applicable to Danish (as well as to every other language).

7.1 The Danish Language
Danish is the official language of Denmark. It belongs to the East Scandinavian language group of the Germanic family. It has a fairly strict word order, SVO (SubjectVerb-Object), in both main and embedded clauses. Moreover, Danish is a verb-second (V2) language, which means that the finite verb is always the second constituent of the main clause. Yes/no questions constructions constitute the only exception, where the verb is sentence initial (see (42) in Table 6 below). Furthermore, if another constituent is fronted (topicalized), it triggers inversion of the subject and the finite verb, which places the finite verb in second position (V2) and the subject in third ((40) in Table 6 below). In embedded clauses fronting and inversion are prohibited. In brief, the linear word order in Danish is as follows7:

7

I leave out structures that contain both a direct and an indirect object. Either the indirect object is realized as a preposition phrase or as part of a double object construction (cf. Larson 1988, Vikner 1989), in which the VP has the following structure:

For example: [CP han1 har2 [IP t1 t2 [VP t1 t2 [VP t1 købt3 [VP hende t3 [NP en gave]]]]]] he has bought her a present

K. R. C. 2001

62

Conj

Sub/Topic
Jeg I (Sub) jeg I (Sub) Pærer Pears (Obj+Top) Hvad What

Verb-finite
spiser eat-PRES har have-PRES spiser eat-PRES køber buy-PRES køber Buy-PRES

(Sub)

Adv
endda even tit often også also

Neg
ikke not

Verb-nonfinite
smagt taste-PT.PTCP

Obj
bananer bananas ferskener peaches

(38) (39) (40) (41) (42)
Og And Men But

jeg I du? you? du you

også also

æbler? apples?

Table 6: Topological analysis of Danish main clauses.

Conj

Sub
jeg I der who du you

Adv
endda even tit often aldrig never

Neg
ikke not

Verb-finite
spiser eat-PRES har have-PRES prøver try-PRES

Verb-nonfinite
smagt taste-PT.PTCP

Obj
bananer bananas ferskener peaches

(43) (44) (45)

du ved you know Piger Girls …

at that

hvis if

Table 7: Topological analysis of Danish embedded clauses.

The structural analysis of main clauses and embedded clauses are given below in (46) and (47) respectively, with the split-IP order proposed by Pollock’s (1989) (the symbol ‘–‘ indicates an unfilled specifier position. The word “ikke” is the Danish negation equivalent to the English “not”.) For expository reasons I have left out the structure of the complement (compl) of the verb. (46) Main Clause (preliminary version): [CP Conj [CP Sub1 Verb2 [TnsP t1 t2 [NegP [AdvP Adv] [NegP (ikke) t2 [AgrP - t2 [VP t1 t2 Compl ]]]]]]] For example: [CP og [CP jeg1 spiser2 [TnsP t1 t2 [NegP [AdvP faktisk] [NegP ikke t2 [AgrP - t2 [VP t1 t2 bananer]]]]]]] “And I actually do not eat bananas” (lit.: and I eat actually not bananas)

t2 [VP t1 t2 katten]]]]]] “The dog actually didn’t bite the cat.” (Literally: the dog bit actually not the cat. Perhaps a description of the derivation of a standard main clause is in order. one for each argument “hunden” og “katten” (their structure is left out for expository reasons). in the presence of a coordinating conjunction (it must be coordinating as it occurs in a main clause) another CP is projected. Consider the following example: (48) [CP hunden1 bed2 [TnsP t1 t2 [NegP [AdvP faktisk] [NegP ikke t2 [AgrP .K.: that I actually not eat bananas) Note the CP-recursion in the main clause (46): In V2 languages the verb always moves from V0 through Agr0. Vikner (1995). thus satisfying the . 2001 63 (47) Embedded Clause (preliminary version): [CP Conj [TnsP Sub1 t2 [NegP [AdvP Adv] [NegP (ikke) t2 [AgrP . Note also that in embedded clauses the tense and agreement inflection is lowered to the verb in V0 instead of being attached to the verb by movement out of V0.) First of all. The former is placed left of V0 receiving the AGENT role and the latter to the right receiving the role of THEME. SO. cf.” (lit.t2 [VP t1 Verb2 Compl ]]]]]] For example: [CP at [TnsP jeg1 t2 [NegP [AdvP faktisk] [NegP ikke t2 [AgrP . the verb “bide” projects a verb phrase VP and subcategorizes for two NPs. R. Neg0 and Tns0 to C0 in main clauses. C. because the first CP is ‘occupied’ by the verb.t2 [VP t1 spiser2 bananer]]]]]] “That I actually do not eat bananas. which is headed by the conjunction (the complementizer). following the Projection Principle.

If the verb moves to Tns0 before Agr0 then the tense inflection should be closer to the verb stem than the agreement inflection. The argument is based on the ordering of the inflection on the verb. negation and agreement. C. 2001 64 Theta Criterion. The subject moves from [spec. ii. The sentential adverbial phrase (AdvP) is adjoined to the NegP8. [AgrP [TnsP [NegP ]]] rather than [TnsP [Neg [AgrP ]]] as Pollock 1989 proposes. I only include sentence adverbials and leave out structures with VP-adverbials. it is widely accepted that Belletti’s order is correct. but open to parametric variation.K. Haegeman (1994). Due to the Extended Projection Principle the clause structure is projected by the verb.e. However. Pollock’s order somehow has to be possible. This is the case in for example German: “Du glaub-t-est” ‘you think-PAST-2nd PERS. which can only appears after the negation: jeg spiser ikke langsomt “I don’t eat slowly” (literally: I eat not slowly) *jeg spiser langsomt ikke . TnsP] to [spec. As mentioned in the section 6. VP] to [spec. CP]. which are adjoined to VP: … [NegP [AdvP Adv ] [NegP spec Neg0 [VP [AdvP Adv ] [VP spec V0… / \ Sentence adverbial VP-adverbial: An example of a VP-adverb is “langsomt” (slowly). TnsP] where it is assigned NOMINATIVE case. This re-ordering (Belletti 1990) has serious consequences: i. This is known as the Mirror Principle. as it is applicable to agrammatic production in the languages studied by Grodzinsky (e. and finally assuming V2 position in C0. 2000).2. which is vacant as no other topic is fronted. NegP. The verb is moved up to Agr0 and via Neg0 on to Tns0 to receive tense inflection. and AgrP is not uncontroversial. in order to maintain neurological support. This may imply that the ranking of the functional categories is not universal. This order should reflect the movement of the verb. 8 In order not to complicate matters unnecessarily. Haegeman 1994 for discussion). The object is assigned ACCUSATIVE by the verb in V0. i. R. and Vikner (1995) have a different order of tense.SING’ (cf. Under the Breakdown Hypothesis. the above ordering of the functional categories TnsP. and the Case Filter forces the subject to move from its base position in [spec. Belletti (1990).g.

Neg0] like English “not” and French “pas”. It is attached to (incorporated into) the verb. R. resulting in e.t2 [NegP [AdvP Adv] [NegP (ikke) t2 [VP t1 Verb2 Compl ]]]]]] For example: [CP at [AgrP hunden1 t2 [TnsP t1 t2 [NegP [AdvP faktisk] [NegP ikke t2 [VP t1 bed2 katten]]]]]] “That the dog actually didn’t bite the cat” (Literally: that the dog actually not bit the cat) .2 above this has never been found (Friedman & Grodzinsky 1997. 2001 65 iii. Perhaps there are agrammatic speakers of some language that is only poorly studied. but as mentioned in section 6. 2000). It follows from the possibility of parametric variation in the ordering of the functional categories that the Tree-Pruning Hypothesis will give rise to language specific variation in agrammatism. Movement of the verb V0 is cyclic and is done step-wise and cannot skip a position (this is known as the Head Movement Constraint). final version (replaces (46)): [CP Conj [CP Sub1 Verb2 [AgrP t1 t2 [TnsP . The English negation “n’t” (and the French “ne”) is a Neg0 element. Hence.e. impairment to the Agreement phrase alone with a sparing of Tense is possible. final version (replaces (47)): [CP Conj [AgrP Sub1 t2 [TnsP . a head. This may be an artifact of the languages studied. i.K. For this reason. Otherwise it would block verb movement. This gives us the following structure.g. C.t2 [NegP [AdvP Adv] [NegP (ikke) t2 [VP t1 t2 Compl ]]]]]]] For example: [CP hunden1 bed2 [AgrP t1 t2 [TnsP t1 t2 [NegP [AdvP faktisk] [NegP ikke t2 [VP t1 t2 katten]]]]]] “The dog actually didn’t bite the cat” (Literally: the dog bit actually not the cat) (50) Structure of Danish Embedded Clauses. On the other hand. in which the order of TnsP and AgrP is reversed in the line of Belleti: (49) Structure of Danish Main Clauses . main clause structure version 2. “didn’t” (thanks to Sten Vikner for pointing this out). if no such language can be found then we will be forced to reevaluate the internal structure of IP. it is necessary to assume that negation in Danish (“ikke”) is in [spec.

VP] to [spec. I have left out the structure of the complement (compl) and the adverbial phrase AdvP: Figure 7: Structure of Danish main clauses. 2001 66 The figures below show the structure of main clauses and embedded clauses respectively. AgrP]. . R. where the subject moves from [spec. Figure 8 shows the movement in embedded clauses.K. C. For expository reasons. which result in the obligatory V2 word order. In Figure 7 shows the movement of the subject and the verb in main clauses from their base positions inside VP to their surface positions in CP. which remains in its base position in V0. and the verbal inflection moves downwards to the verb.

Depending on the specifics of the Danish typological parameter-settings.K. Agr0] (in Danish by C0 and in English by Agr0 for reasons not discussed here. certain predictions can be made about the performance of agrammatic aphasics in Danish (as well as in any other language). see also section 2. predictions can be made with regards to omission or substitution of inflection based on the presence or absence of zero-morphology. while other types will turn out normal. the aphasic’s comprehension of certain types of sentences will be impaired. as the hypothesis states that certain parts of the grammatical representation (syntactic tree) are disrupted.2 Predictions About Agrammatism in Danish Based on the interaction between the Trace-Deletion Hypothesis and the Default Strategy and on the Tree-Pruning Hypothesis. the structure of the sentence will be the crucial factor in predicting impairment. R.8). cf. Furthermore. In production. Note that NOMINATIVE is assigned to [spec. Vikner 1995. 7. .1. C. 2001 67 Figure 8: Structure of Danish embedded clauses.

v2 [NegP . C.v2 [VP t1 skubber2 [NP drengen]]]]]] er høj vis mig pigen1 [CP som1 e [AgrP t1 v2 [TnsP .v2 [VP t1 v2 [NP drengen]]]]]]]] [CP pigen1 peger2 [AgrP t1 v2 [TnsP .v2 [VP t1 v2 [PP på [NP drengen]]]]] pigen peger på drengen the girl points AGENT Subject -subject relative pigen1 [ som1 skubber drengen ] er høj the girl Objectsubject relative vis mig pigen1 [ som1 skubber drengen ] show me Subject cleft det er pigen1 [ som1 skubber drengen ] it is the girl who pushes the boy THEME AGENT the girl who pushes the boy THEME AGENT who pushes the boy THEME is tall AGENT at the boy THEME pigen1 [CP som1 e [AgrP t1 v2 [TnsP .K. R.1 Comprehension First of all.2.v2 [NegP .v2 [NegP . let us take a look at the structure of the relevant clauses in Danish in the following two tables: Type Simple active pigen skubber drengen the girl pushes AGENT the boy THEME Structure [CP pigen1 skubber2 [AgrP t1 v2 [TnsP . 2001 68 7.v2 [NegP .v2 [VP t1 skubber2 [NP drengen]]]]]] (Continued on next page) .v2 [NegP .v2 [VP t1 skubber2 [NP drengen]]]]]] det er pigen1 [CP som1 e [AgrP t1 v2 [TnsP .

v2 [VP [VP t1 v2 t1 ] [PP af [NP pigen]]]]]] drengen1 skubbe-s the boy THEME t1 af pigen (“the boy is pushed by the girl”) AGENT push-PRES.PASS by the girl “Blive” passive [CP drengen1 bliver2 [AgrP t1 v2 [TnsP .v2 [NegP .v2 [NegP .v2 [VP [VP t1 v2 [VP t1 peget [PP på t1]]] [PP af [NP pigen]]]]]]] drengen1 bliver peget på t1 af pigen the boy THEME is pointed at by the girl AGENT (Continued on next page) .v2 [VP t1 v2 [AdvP t1 interesseret [PP i [NP pigen]]]]]]]] drengen er interesseret i pigen the boy is interested in the girl THEME EXPERIENCER Psych.v2 [NegP .v2 [VP [VP t1 v2 [VP t1 skubbet t1]] [PP af [NP pigen]]]]]]] drengen1 bliver skubbet t1 af pigen the boy THEME is pushed by the girl AGENT [CP drengen1 bliver2 [AgrP t1 v2 [TnsP .v2 [VP t1 v2 [PP om [NP pigen]]]]]]] drengen synes om pigen the boy thinks about the girl THEME (“the boy likes the girl”) EXPERIENCER Adject. 2001 69 Type Lexical passive Structure (continued from above) [CP drengen1 synes2 [AgrP t1 v2 [TnsP . R. EXPERIENCER Type Verbal passive Structure [CP drengen1 skubbes2 [AgrP t1 v2 [TnsP . C.K.v2 [NegP .verb [CP drengen1 beundrer2 [AgrP t1 v2 [TnsP .v2 [NegP .v2 [VP t1 v2 [NP pigen]]]]]] drengen beundrer pigen the boy admires the girl THEME Table 8: Danish clause structure 1. Passive [CP drengen1 er2 [AgrP t1 v2 [TnsP .v2 [NegP .

which is moved out of [spec. The double occurrence of the AGENT-role leads to chance performance. admire-PRES.v3 [NegP .K. and below chance is expected to be the same as in English.v2 [NegP .v3 [VP t2 skubber3 t1 ]]]]] er høj drengen1 [ som1 pigen skubber t1 ] er høj the boy who the girl pushes is tall THEME AGENT Objectobject relative vis mig drengen1 [CP som1 e [AgrP pigen2 v3 [TnsP . is assigned the by the Default Strategy and no comprehension problem arises.v3 [VP t2 skubber t1]]]]]]]] det er drengen1 [ som1 pigen skubber t1 ] it is the boy who the girl pushes THEME AGENT Psych. Table 9 contains the structures expected to be problematic. The exception to this is. the . R.v3 [VP t2 skubber3 t1]]]]]]]] vis mig drengen1 [ som1 pigen skubber t1 ] show me the boy who the girl pushes THEME AGENT Object cleft det er drengen1 [CP som1 e [AgrP pigen2 v3 [TnsP . 2001 70 Type Subject -object relative Structure (continued from above) drengen1 [CP som1 e [AgrP pigen2 v3 [TnsP . Table 8 contains the structures on which agrammatics are expected to perform above chance. The object has not undergone movement and is therefore assigned its role of THEME grammatically. The prediction is that due to the Trace-Deletion Hypothesis the θ-role can no longer be transferred through the link to the trace and the result is a wrong assignment by the Default Strategy.v3 [NegP . C. as all the clauses have THEME AGENT- theta-structure. The distribution of performance rates between above chance.v3 [NegP . VP] under the VPAGENT-role internal Subject Hypothesis. All the clauses involve movement of the object.PASS by the boy The structures in the two tables above are all semantically reversible transitives. as in English. at chance. Passive [CP pigen1 beundres2 [AgrP t1 v2 [TnsP . The subject.v2 [VP [VP t1 v2 t1 ] [PP af [NP drengen]]]]]]] pigen1 beundre-s the girl THEME t1 af drengen EXPERIENCER Table 9: Danish clause structure 2.

K.2. from the agreement (AgrP) node up. or to the CP alone depending on the degree of severity of the deficit.2 Production The Tree-Pruning Hypothesis predicts impairment to the top-most nodes of the syntactic tree – more precisely. R.subject relative Subject cleft Lexical passive Adject. These predictions about Danish aphasic comprehension (which in fact is identical to the pattern observed in English) are summarized in the following table: Structure Simple active Subject -subject relative Object. C. from the tense (TnsP) node up. 2001 71 psychological passive in which the result of θ-role assignment (in interaction with the Thematic Hierarchy) gives rise to the reversal of the roles (or rather their salience) resulting in below chance performance. In the illustration below the points of breakdown is indicated with double lines: . Passive Psychological verb Verbal passive “Blive” passive Subject-object relative Object-object relative Object cleft Psychological passive THEME-EXPERIENCER AGENT-EXPERIENCER THEME-AGENT AGENT-AGENT EXPERIENCER-THEME AGENT-THEME AGENT-THEME AGENT-THEME Normal Assignment Aphasic Assignment Predicted Performance above chance chance below chance Table 10: Predicted Danish aphasic performance on comprehension tests. 7.

I return to this problem shortly).t1 *[NegP [AdvP Adv] [NegP . (51) (52) (53) (54) *[AgrP – Verb1 [TnsP – t1 [NegP [AdvP Adv] [NegP . So in practice it will only be possible to place the site of breakdown in TnsP or CP (or actually perhaps only in CP. However. Figure 9) in the entire TnsP is missing (and so are AgrP and CP) and the verb cannot move farther than to Neg0 and thus appears uninflected for tense – i.e. where it is inflected for tense (perhaps the verb can move to Agr0 (51) but there would be no way of telling. Figure 9 above) the verb will only be able to move as far as to Tns0. the verb always (in normal language users) moves to C0 (see Table 8 and Table 9). as Danish is a V2 language. and as there is no agreement inflection in Danish. in the infinitive. R.K. C.t1 [VP {sub} t1… [VP {sub} t1… *[TnsP – Verb1 [NegP [AdvP Adv] [NegP . For this reason it is not possible to decide on the relative ordering of AgrP and TnsP.Verb1 [VP {sub} t1… *[VP {sub} Verb… . 2001 72 (51) (52) (53) Figure 9: Possible points of breakdown in the syntactic representation in Danish (the numbers refer to examples below). If the tense node is disrupted (53) (cf. This will reveal itself in the following way: If the impairment is restricted to CP (51) (cf. it is not possible to tell whether the agreement node is intact if the tense node is impaired.

This would be distinguished from (53) by the position of the verb in relation to the negation in negative clauses and to the medial adverb (adjoined to VP) in positive clauses. if NegP is impaired the verb follows it. listed here in order of degree of severity of impairment (from least to most severe): (55) Normal: Unimpaired CP [CP han1 skubber2 [AgrP t1 t2 [TnsP .K. i. for the simple reason that there would be no base position for the negation. An impaired NegP would result in non-normal use or non-use of negation. If the NegP is intact the verb precedes the adverb or negation. where the first two are phonetically indistinguishable (for the sake of brevity I have left out the adjoined AdvP). Haegeman 1994 and Vikner 1995). it also affects other aspects of syntax related to the node in question. R. Being an argument. The subject is dependent on C0 to assign NOMINATIVE case in Danish. as shown in (54). 2001 73 There is of course another logical possibility. which is movement from one argument position to another. in which case the verb remains in its base position inside VP. C. directly from .t2 [NegP ikke t2 [VP t1 t2 pigen ]]]]] he push-PRES not the girl (56) Impaired CP *[AgrP – skubber1 [TnsP – t1 [NegP ikke t1 [VP {han} t1 pigen ]]]] push-PRES not {he} the girl (57) Impaired AgrP *[TnsP – skubber1 [NegP ikke t1 [VP {han} t1 pigen ]]] push-PRES not {he } the girl (58) Impaired TnsP *[NegP ikke skubbe1 [VP {han} t1 pigen ]] not push-INF {he} the girl (59) Impaired NegP *[VP {han} skubbe hende … {he} push-INF her The consequences of disruptions of the syntactic tree not only affects verbmovement. The impairment may theoretically be from the negation phrase (NegP) up.e. There are thus four possibilities. the subject normally undergoes A-movement (cf. which is not shown in Figure 9.

in languages without zeromorphology. (32) above. Arguments depend on case in order to pass the Case Filter. even in this case the subject will not be overt (phonetically realized). [spec. The fact that it is overt (phonetically realized) may be due to an impaired Case Filter (the same would be the case if the subject was adjoined to the topmost node). Lexical subjects as well as pronouns should not be overt without case. Perhaps the subject is ‘named’ or stated first and then the rest of the sentence. cf. cf.K. 2001 74 [spec. AgrP]. and [spec. the subject is fronted and overt due to knowledge other than grammatical (such as pragmatics and meta-linguistic knowledge). However. If none of these positions are available the subject will have nowhere to go. . This means that theoretically the subject will be dropped. This movement is only possible in (51)/(56) where the [spec. As Danish has +zero-morphology. VP]. In other words. and assign them a default case […]”. so Friedmann & Grodzinsky (2000: 14-15 [pdf version]) proposes that “the agrammatics may use subjects as topicalized elements. AgrP] not via the intermediate specifier positions. and if CP is missing C0 will not be present to assign case. Here is an example of a clause pruned at the tense node: (60) *[NegP SUB1 VERB2 [VP t1 t2 compl… the boy push (inf.1). Therefore. R.e. An admittedly a bit ad hoc solution to this problem might be that the subject is placed in front of the sentence after the representation has left the grammar (which is informationally encapsulated. i.) the girl Another problem with this proposal is that only [spec. I have put the subjects in (51)-(54) and (56)-(59) above in “curly” brackets ({}). The subject and the rest of the sentence may be uttered in two distinct but related parts instead of in one construction. i. I see no apparent solution to this problem. C. because it has not been assigned NOMINATIVE case. Presently. VP] to [spec. section 6. AgrP] position is intact. without breaking a general rule of movement: arguments move to and from argument positions only. According to Friedmann and Grodzinsky (2000) subject pronouns are only dropped when verbal inflection is substituted. if the predictions hold.e. Danish speaking aphasics should not drop subject pronouns. so to speak. CP] are argument positions. which in Danish is NOMINATIVE. This means that the subject is moved to the topmost available specifier. in the same non-linguistic (non-grammatical) way as with the Default Strategy.

and therefore all main clauses in Danish will also be affected. Studying agrammatism may provide the evidence needed to verify one and falsify the other. which in fact is the strategy I have used so far in this section. where the complementizer or wh-element should be. R. Another effect of pruned trees is that “Wh-questions and embedded clauses are nonexistent or completely ill-formed in the speech of the patients” (Grodzinsky 2000: 16). a V2 language.e [TnsP . some embeddings are .K. in the Minimalist Program this is called movement as last resort).) It is up to empirical research to choose between the two different models of verb movement outlined here: either the verb always moves if possible or it only moves when necessary. If the CP is missing the verb will not be able to move out of VP unless one assumes that the verb always moves if possible.t2 [NegP ikke t2 [VP t1 t2 pigen ]]]]] he push-PRES not the girl (62) Agrammatic main clause: impaired tree *[han] ([AgrP . The problem with subject movement just discussed is related to a problem of verb movement. As the C0 position is not present to motivate movement. 2001 75 Above I briefly mentioned that perhaps it would only be possible to place the locus of breakdown in CP. which means that the verb always moves to C0 in main clauses. However. the verb will most like remain uninflected in its base position in V0. Danish is thus quite vulnerable to agrammatism.e) [NegP ikke e [VP {han} skubbe pigen ]] he not {he} push-INF the girl (I have placed the subject of (62) in front of the sentence in square brackets to indicate the indeterminacy whether it is cognitively or grammatically fronted. in generative grammar it is generally assumed that movement does not take place unless it is necessary (for example. This is due to (again) the missing CP. Any disruption of the syntactic tree will affect CP. as in theory any structural degree of severity will have the same effect on main clauses: (61) Normal main clause: intact tree [CP han1 skubber2 [AgrP t1 t2 [TnsP . Danish is. as mentioned. C. However. That means that relatives (with wh-elements or complementizers) and clefts (see Table 8 above for subject relatives and clefts and Table 9 for object relatives and clefts) will be either nonexistent or ill-formed. either it is a freestanding constituent or else it occupies the top-most available specifier position (or it is adjoined).

R. as it sometimes is in English: (63) han ved [CP (at) [AgrP hun1 t2 [TnsP . However.e [TnsP .2. 2001 76 constructed without complementizers or wh-elements.t2 [NegP ikke t2 [VP t1 kommer2 ]]]]] he know-PRES (that) she not come-PRES “He knows she’s not coming” Embedded clauses with optional complementizers should not be as vulnerable to impairment as other sentence types. because the CP is not involved in verb movement and therefore the verb should theoretically be inflected for tense and agreement depending on which node is impaired. for example the Danish complementizer “at” (that) is optional.3 Summary All of these predictions of the previous section can be summed up in a series of questions that need to be answered through empirical tests in order to either validate or falsify the hypotheses of trace deletion and tree-pruning underlying agrammatism in Broca’s aphasia. TnsP].e [VP {han} vide[AgrP hun1 t2 [TnsP . So if C0 is missing so may the subject – unless it may occupy a nonargument position. A Danish agrammatic person with (only) an impaired CP node would probably utter (63) as: (64) *[han][AgrP . i. such as [spec.K. C. In comprehension. the subject depends on C0 to assign NOMINATIVE.t2 [NegP ikke t2 [VP t1 kommer2 he know-INF she not come-PRES “He know she’s not coming” 7. as it cannot be fronted (the main clause is already there). do Danish agrammatics perform according to the distribution outlined in Table 10 above: No syntactic movement Syntactic movement Psychological passive No comprehension problem comprehension problem comprehension problem above chance performance chance performance below chance performance The following questions deal with production: .

Second. R.3. and that the value and weight of my empirical data would have been significantly increased had I had several test subjects. For various reasons. this test proceeds as follows. As described in section 4.K. 7.3 Empirical Tests Agrammatism manifests itself differently with regards to comprehension and production. v. iv. of the two helpful places one institution had no patients with Broca’s aphasia and the other currently treated only one agrammatic aphasic patient – the patient tested here. vii. is it inflected for tense? What is the position of the verb relative to the medial adverb or negation? What is the degree of severity? Do speakers drop the subject? If not. 2001 77 ii. Consider the following example. Do Danish aphasics produce constructions without movement of the finite verb? If so. is it fronted? Danish has zero-morphology. C. my efforts to find more than one patient for this study were in vain. First of all. I am very much aware that a single case is not much to base any statistics on. iii.1 Comprehension: the Sentence-to-Picture Matching Test The comprehension test is a sentence-to-picture matching test. the linguistic tests used to examine agrammatism are also different: one for comprehension and one for production. The patient is presented with the pictures below depicting a boy loving a girl (A) and a girl loving a boy (B): . the institutions and speech therapists responsible for the treatment and rehabilitation of aphasics were not open to outside research by linguists. The patient then hears a semantically reversible sentence and the task is then to point out the picture that depicts the meaning of the sentence. and then I discuss the data I have obtained from applying the test to a patient with agrammatism. vi. implying omission of inflection. First the patient is presented with two pictures that semantically mirror each other. Therefore. First I describe the tests I have used. are these instead produced as infinitives? If the verb has moved.1 above.3. True or false? 7.

rare in the sense of distribution. The few lexical passives are actually quite frequently used. in this case the correct answer is of course (B). I have devised 15 sets of pictures that correspond to the two interpretations of semantically reversible predications. as it is quite rare in Danish . 2001 78 Figure 10: Picture of “the boy loves the girl” and “the girl loves the boy”. the patient hears the sentence "The girl loves the boy" and has to point out the corresponding picture. Next. C. Lexical passive is an exception with only four tokens.PASS with the girl “The boy is disgusted at the girl” Note that the lexical passive does not involve movement of the object (see Table 8) unlike for example the verbal passive (see Table 9). I have constructed the set of sentences in such a way that that each type of construction is represented by ten token sentences. R. see appendix B. such that each of the two pictures of the same set is represented equally frequent. such as: (65) Drengen væmme-s The boy ved pigen disgust-PRES. left and right column respectively. The pictures and the sentences that have the semantic meaning corresponding to the pictures can be found in appendix A. Another important thing is that I have included ten additional tokens of simple actives and ten "blive" passives that have a verb that subcategorizes for a preposition .K. I have also constructed 144 semantically reversible sentences that can be matched with the pictures.

but not for example A-B-A-A-A-A-B-B. which may otherwise become biased towards either (A) or (B) if longer series were allowed.3. it is ensured that both intended and produced are known. which he or she then has to repeat. In the following example the first preposition "på" is subcategorized for by the verb and it is therefore inside the VP that is projected by the verb. also to avoid any bias or other unpredicted influence on the results. The sentences are presented to the patient in a random order. sentences that relate to the same pair of pictures (such as the pair in Figure 10) do not occur in series of more than two. The latter has a sequence of more than three consecutive identical correct answers: 4 As in a row.1. The sentences are distributed over the 13 structural types (see Table 10 above). For example. C. The patient hears a sentence. Section 2. I have constructed a set of 100 sentences listed in appendix C. By doing a repetition test. either (A) or (B). Furthermore.2). Moreover. The terms VP-internal and adjoined to VP will suffice. it can be difficult sometimes to determine what was intended when the utterance is severely malformed. 2001 79 phrase (cf. R. Each type is represented by seven tokens.2 Production: Repetition Test The production test is a repetition test. This is done to ensure that all types of structure are present in the corpus. For elaboration see Haegeman (1994)).K. one for each of the tense/aspects: . while the second preposition "af" heads the PP adjoined to the VP (structural information not central to this point is left out for expository reasons): (66) Pigen1 bliver [VP [VP t1 peget på t1 ] [PP af drengen]] the girl is pointed at by the boy The VP-internal preposition is governed by the verb (I shall not go into further formal detail of government relations. It will thus be possible to see whether the presence or non-presence of a governed proposition affects comprehension. such that the correct answer. 7. This is done to secure that the intact comprehension of certain structures that happens to be presented consecutively does not affect performance. of the presented sentences appear in series of no more than three. the correct answers may come in sequences such as A-A-B-AB-B-A-A-A-B.

such as “jeg har haft ringet [men du var ikke hjemme]” (‘I have phoned you [but you weren’t home]’).PTCP the girl “The boy had pushed the girl” Past perfect Drengen hav-de The boy skubb-et pigen have-PAST push-PAST. I include the construction knowing it may be controversial.PTCP the girl “The boy had pushed the girl” Past Drengen skubb-ede pigen The boy push-PAST the girl “The boy pushed the girl” Present double perfect Drengen har-Ø The boy haft skubb-et pigen have-PRES have-PAST. 2001 80 (67) Tense/Aspects: Past double perfect9: Drengen hav-de The boy haf-t skubb-et pigen have-PAST have-PAST. C.K. R.PTCP push-PAST.PTCP the girl “The boy has pushed the girl” Present perfect Drengen har-Ø The boy skubb-et pigen have-PRES push-PAST. most speakers would not doubt accept the double perfect in sentences about phone calls. .PTCP push-PAST. However.PTCP the girl “The boy has pushed the girl” 9 The double perfect constructions are not used by all Danish speakers and some even find them ungrammatical.

e. 2001 81 Present Drengen skubb-er pigen The boy push-PRES the girl “The boy pushes the girl” Future Drengen vil-Ø The boy skubb-e pigen will-PRES push-INF the girl “The boy will push the girl” In addition I have added a set of simple actives and "blive" passives with the negation "ikke" in order to be able to test whether the use of negation is intact. which is ungrammatical for two independent reasons: first. Most of the excluded sentences are in the passive voice and constructed with the double perfect. In order to detect movement of the verb all the sentences include sentence medial adverbs or negation. in the perfect aspect): (68) **Hunden hav-de vist haf-t bid-es af katten The dog have-PAST probably have-PAST.K. . The set includes sentences both with and without governed prepositions. I have included sentences both with and without VP-internal (governed) prepositions. in order to see whether one or the other is dropped. C. For example this verbal passive. second the passive cannot be constructed on a participle (i. the verbal passive is only possible in the present tense (and in past constructions involving the modality verbs “skulle” (should) or “ville” (would)).PTCT bite-PASS by the cat To reach a total of 100 sentences I have added six simple passives. However. eleven of the 105 sentences generated in the set had to be excluded due to ungrammaticality (see appendix C). According to Grodzinsky (1988) in agrammatics only the VP-internal prepositions are deleted (phonetically silent). R. while the adjoined prepositions that head phrases like “[the boy was hit] by the girl” are not. Finally.

I first provide his medical history and then the results of the test described in the previous sections. The other was a 26-year-old woman studying to become an architect at the Aarhus School of Architecture.K. stretching into the frontal-. 7. 7. R. Both subjects performed 100% correct on both the comprehension test and the repetition test.3 Patient TJ In order to test the hypotheses and predictions of section 7. Both tests were performed on two normal subjects with no prior history of brain damage or neural pathology.3. In addition he had a small hematoma in the right parietal lobe but. and temporal lobes. He suffered massive hematomas (bloodfilled swellings) to the areas in the perisylvian region. According to his medical report10.1 Medical history TJ is a 31-year-old (right-handed) male with a master's degree in art history.3. I have summarized the zone within which the hematomas in the left hemisphere have occurred in the following illustration: 10 The medical report is not listed as a reference due to the fact that it is confidential and not publicly accessible.2 above. who was hit by a truck while riding his bicycle. parietal-. the right hemisphere makes no grammatical contribution to language use. He had infarction (cell death) in most of his left temporal lobe. C. where the same structural type never occurs in sequences above two to avoid any bias. One was a 28-year-old male fellow student of linguistics at the University of Aarhus.2 I have tested a patient diagnosed as showing signs of Broca’s aphasia. 2001 82 The sentences are presented in random order. the accident caused massive trauma to his left hemisphere.3. as shown in section 3. .

3. The following table shows a summary of his performance on the comprehension test. 2001 83 Figure 11: Left-hemisphere damage in patient TJ. and at that stage he had improved significantly. C. Based on Sobotta & Becher (1975: 4. . 3) According to the medical report. each take with a duration of about half an hour.3. according to his speech therapist. Fig.2 Comprehension Test Results TJ was tested on the comprehension test described above.K. On the other hand. The shaded area is a gross summary of hematoma sites. The arrows point to major hematoma centers. while his impressive functions (comprehension) were relatively spared. 7. The medical report is rather vague with regards to the precise location of the hematomas. This had the disadvantage that he was no longer a clear example of Broca’s aphasia and therefore his test performance could not be expected to follow the predictions completely. it had the advantage that he was easily understandable. the neurologists diagnosed him as suffering from severe expressive aphasia. Due to fairly successful rehabilitation most of his linguistic abilities were restored to a near-normal level at the time of testing. which was applied twice with a week’s interval. R. I tested TJ eight months after the accident.

07/10 10/10. there are only four tokens of lexical passive. Furthermore. There is also a clear reduction in his performance on lexical passives. the important point is that the lowest performance (70%) is on a predicted type of construction. R. but then again this is based on one of only four tokens. However. However. 10/10 20/20. i. TJ performed above chance on all but one sentence type. I define ‘chance performance’ as between 30% and 70% knowing that this is a gross simplification of the mathematics of statistics. 10/10 10/10. 10/10 10/10. he failed on the same token both times: (69) Drengen væmme-s The boy ved pigen the girl disgust-PRES. 10/10 10/10. As can be seen in the table. . 03/04 10/10. Also evident from the table is that he does not perform according to a canonical/non-canonical word order or role order distinction. which is not much of a base for conclusions. 08/10 09/10. 20/20 07/10. 09/10 03/04. 2001 84 Type Simple active Subject–subject relative Object-subject relative Subject cleft Lexical passive Adjectival passive Psychological verb Verbal passive “Blive” passive Subject-object relative Object-object relative Object cleft Psychological passive Correct 20/20. 09/10 % 100 100 100 95 75 90 90 100 100 70 100 90 90 Performance Above chance Chance Above chance Table 11: TJ's performance on the Sentence-to-Picture Matching Test. 18/10 08/10.e. 10/10 10/10. 20/20 10/10. Neither does it matter whether the AGENT has the same or different grammatical 'roles' (subject / object) in the matrix clause and the embedded clause – only when the subject of the matrix clause is the object of the embedded clause.K.PASS at “The boy is disgusted at the girl” This seems to imply a problem with the lexical entry of the verb rather than the construction. subject-object relative. The fact that he performs normal on subject-subject relatives shows that he has no difficulties with center-embedding as such. C. if one only considers the percentage.

0% 100.1% 28.3% 14.7% 85.0% 71.7% (1/19) (1/7) (1/7) (0/7) (5/7) (0/5) (1/7) (1/4) (0/7) (6/7) 2 1 1 2 Object-object relative Object cleft Psychological passive 57.9% 1 1 Simple active Simple active 5 ? Changes 1 1 1 Simple passive Subject-object relative ? Correct 94.K.4% 80.0% 85.0% 85.7% 75. C.0% 42.0% (4/7) (2/7) (1/5) 4 1 2 1 Subject–subject relative Object-subject relative Adjectival passive ? Object-subject relative [NP N [object relative]] Object-subject relative Psychological active 71. In the table and henceforth the numbers in black circles refer to a set of sentences in the test data results in appendix E (they are also marked in the appendix): (70) #93 #14 #35 #52 #43 #15 #55 #71 #44 .7% 85. 2001 85 The presence or absence of governed (VP-internal) prepositions did not affect comprehension.3% 0. which took about twenty minutes.0% 28.3% 25.3% Simple passive’ should to read as ‘one sentence was produced as a simple Table 12: TJ's performance on the Repetition Test. The results are given in the table below (the ‘Changes’ column for example ‘1 active’): Type Simple active Subject–subject relative Object-subject relative Subject cleft Lexical passive Adjectival passive Psychological verb Verbal passive “Blive” passive Subject-object relative Error Rate 5. R.0% 0.7& 100.3. 7.3.0% 14.3% 14.0% 14.6% 100.6% 20.3 Production Test Results On the production side. TJ was tested once on the Repetition Test.4% 0.

TJ has a tendency to avoid the passive morphology “–s”. 12 11 The Danish “jo” is equivalent to the German “ja” and is hard to translate without any context. NOT to the passive “-et” (participle) inflected on the auxiliary. it can be translated as for example “after all. The omission of the passive ending is consistent with the prediction from above (page 57) that +zero-morphology leads to It should be noted that when talking about passive morphology. Depending on the context. .5%) are not produced correctly. which involves an auxiliary verb and no passive morphology. the passive morphology was omitted in five cases. whereas no errors were made on the "blive" passive. Consider for example this German example: Er ist ja ein Berliner. R. and 4 verbal passives: (71) Passive “-s” omission Type: Lexical passive: Psych. 2 psychological passives. C. The passive morphology was intended in three passive construction types: 7 lexical passives. eight (61. passive: Verbal passive: Error: 3 drop main V 2 drop passive morphology 1 drop passive morphology 2 drop passive morphology (#50 in appendix E) (#81 and #85) For example this verbal passive: (72) Intended (#85): drengen skubb-ede-s the boy Produced: *drengen skubb-ede jo the boy push-PAST af pigen after-all by the girl jo af pigen by the girl push-PAST-PASS after-all12 The entire verb was dropped in three instances.K. I am only referring to the Danish passive “-s” on the main verb. he is a Berliner” or “but he is a Berliner”. Of 13 intended instances of passive morphology11. 2001 86 This is done in order to keep the results and my treatment of them as transparent as possible and to be able to use the numbers as short hand references. such as “pigen blev skubb-et af drengen” (the girl was push-ed by the boy).

Furthermore. for example "kigget på" (looked at) instead of "inspireret af" (inspired by) (see #31 in appendix E). -main V -1 AUX -main V ? PAST PERF PAST PRES PERF ? PAST PRES (Continued on next page) . but there are examples of word substitutions (semantic paraphasia).3% correct.0% 0X 1X 3X 1X 4X 1X 2: PAST PERF [AUX V] 42. He performs near normal on object-subject relatives but almost consistently wrong on subject-object relatives.9% correct.9% 6X 6X 1X 1X 3: PAST [V] 90.K.2 and 7. The following table shows TJ's performance on the different tense/aspect constructions: Tense & AUX 1: PAST PERF+ [AUX AUX V] Correct Performance 0.9% 20X 2X PAST PERF+ PAST PERF PAST PRES PERF+ PRES PERF Produced Verb Reduction -1 AUX -2 AUX -1 AUX -1 AUX. which are also center-embedded. 2001 87 omission and not substitution in agrammatism. sections 6. Again. The performance on tense inflected forms is central to the Tree-pruning Hypothesis. C. Subject-subject relatives are center-embedded and he performs near normal on them. There are no examples of morphemic substitution in the test data. TJ has a strong tendency to use forms without movement of the (underlying) object as is evident from his performance on object relatives: Subject-object relative: 14. object-object relative: 42. both of which are right-embedded. he performs only 71.2. cf. a reference to centerversus right-embedding is insufficient.4% correct on object clefts.2 above. R. while his performance is only 14. which is clearly a reduction.3% correct on the subject-object relatives.

.6% V ? TOTAL 100 100 Table 14: Auxiliary verb reduction.9% 5X 6X 1X 1X 1X 6: PRES [V] 7: FUTURE (PRES) [AUX V] 64.3% 68.K.9% correct on the perfect constructions (PAST. -main V ? PAST PRES PERF PRES ? ? PAST PRES -main V -1 AUX Table 13: TJ's tense/aspect performance. All of these constructions involve an auxiliary verb.0% Correct Performance 2X 0X 6X 2X 5: PRES PERF [AUX V] 42.PERF and PRES. R. Only one is produced as a double perfect. C.PERF) and 64. the reduction of the number of VPs is still clear from the comparison of the remaining structures. he performs only 42. Whatever one’s opinion on the double perfect (see footnote 9 on page 80).0% 63. Clearly there is a tendency to reduce the number of auxiliary verbs. -main V -1 AUX -1 AUX -1 AUX -1 AUX.8% 3X 11X 2X 5X 9X ? PAST FUTURE PAST PRES PERF+ PRES PERF Produced Verb Reduction -2 AUX -1 AUX -1 AUX. This is summarized in the following table: Verbs AUX AUX V AUX V Intended 20 52 38 Produced 1 33 58 8 % 5. 2001 88 (Continued from above) Tense & AUX 4: PRES PERF+ [AUX AUX V] 0. TJ performs 0% correct on the double perfect constructions (PERF+) in both past and present tense. In addition. but in the wrong tense.53% 152.3% on the future constructions.

5% Table 15: The distribution of inflection for past and present in TJ's performance data. i. past or present. and TJ performed 100% correct.2% 81. There is no clear ’default’ tense. This is evident from Table 13 above. Importantly.7%). It is not the case that tense/aspect per se is omitted. 2001 89 TJ has a tendency to use constructions without auxiliary verbs. R.e. All the verbs are inflected for tense.9% correct on the constructions with only one verb and inflected for past tense. i. but a slight preference for past.8% correct on the same construction inflected for present. which was inflected for tense. to minimize the structure – recall from section 2 that each verb projects a VP. In three ) he omitted the entire embedded clause. In summary: Tense Inflection PAST PRESENT ? TOTAL 100 Intended 46 54 Produced 53 44 3 100 % 115.K. he produces no unmoved uninflected verbs (i. while he performs only 68. As mentioned all verbs are inflected. Next consider the distribution of inflection for past and present. where it is shown that TJ performs 90. In some (5/8) of the clauses where TJ omitted the main verb ( sentences ( ) he produced the auxiliary verb. The question mark indicates that the verb as well as the rest of the clause has been omitted. There are twelve clauses with the negation "ikke" (indicated with a P in the third column in appendix E). root infinitives). This shows that NegP is intact. C. The following changes in tense inflection are made: (73) 10x past 17x present 3x present present past ?( ) Note that present is changed more frequently than past: of 30 changes 20 are made to the present tense (66. which suggests that the tense node TnsP is intact.e.e. TJ's use of negation is normal. . only the auxiliary verbs. and therefore it is a case of structure reduction not tense/aspect omission.

TJ correctly produces embedded clauses. As mentioned. This structural difference between the two types of prepositions is also found in the speech of TJ: . In all but one sentence (99/100) the position of the adverb in relation the verb is correct. This shows that the CP node is intact. In section 4. According to Grodzinsky (1988) agrammatics frequently omit only one kind of preposition. as stated in the Tree-pruning Hypothesis (cf. The correct order of verb and adverb (VFINITE-Adv) shows that the verb has moved out of VP (as the AdvP is adjoined to a node (NegP) higher than VP the base-generated order is AdvVFINITE. Therefore TnsP must be intact (recall from section 7.e. of course.2 that it is not possible to tell whether AgrP is intact). This of course means that there must be a position to which the verb can move. That is. it is not possible to decide on the problem (from section 7.2. Figure 7 and Figure 8 in section 7. section 7. TnsP.e. Anyway.1 I mentioned that Broca’s aphasics have an improper use (or nonuse) of prepositions. i. This is also supported by the fact that the subject is always present and in its correct position – [spec. The only example of an incorrect adverb position is . i.1. On the basis of the evidence from TJ. R.3. 2001 90 All the 100 sentences in the test set has a sentence medial adverb. he has evidently undergone successful rehabilitation. cf.K. i. ungoverned by the VP (cf. which is adjoined to NegP. (36) in section 7. In his main clauses. In matrix clauses the adverbial follows the finite verb and in embedded clauses the adverb precedes the finite verb.e. and NegP are intact because “an impaired node cannot project any higher”. which should have had the matrix word order but instead has the word order of an embedded clause.3. while they produce those that are adjoined to the VP. C. because it contains the relative pronoun “som” (who) in [spec. also shows that both AgrP. as the entire tree structure is intact. see Figure 7 on page 66). the fact that he produces embeddings with a wh-element shows that the CP node is intact.2 above). the verb has undergone movement and is in V2 position. TJ has a very mild form of agrammatism. There is no example of a non-moved verb in the infinitive in the data. the prepositions that are inside the VP.2) of deciding between verb movement as ‘movement when possible’ or ‘movement when necessary’.2. CP] in the relative clauses.2.2 above). those that are governed by the VP. This. CP] – preceding the finite verb in C0.

however. The Danish verb “slå” (hit) is a very common word. There is also evidence of a single impaired lexical entry in TJ’s production. comes from the six cases where he substitutes a verb that does not subcategorize for a PP with one that does. The omission of a governed or ungoverned preposition is indicated with –G and –U respectively in the second column on the right. In appendix E the sentences with governed prepositions are marked with a P in the second column on the left. Still. if the object is present it is governed by the VP. which has irregular inflection. 2001 91 Preposition Governed Ungoverned Intended 27 19 Omitted 4 1 Error Rate 14. a marked difference between governed and ungoverned prepositions. only 14. Further evidence for TJ’s relatively restored ability to use prepositions correctly. i. Consider for example this simplified version of #31 from appendix E (slightly modified for expository reasons): (74) Target: pigen var [VP inspireret] the girl was (75) Produced: pigen har [VP kigget [PP på [NP drengen]]] the girl has looked at the boy inspired [PP af [NP drengen]] by the boy In (74) the verb “inspirere” (inspire) does not subcategorize for a preposition phrase. Compare [VP[PP]]) are marked with +G in the second column on the right.K. which is close to normal production. The governed prepositions are omitted almost three times as often as the ungoverned.8% of the governed prepositions are deleted. the optional PP is adjoined to VP. In appendix E such cases of substitution ([VP] [PP] ungoverned prepositions.3% Table 16: Preposition drop in TJ’s production. C. There is no ungrammatical use of governed or . Few prepositions are dropped but there is. In each of these cases he correctly produces the governed preposition. R.8% 5.e. In (75) the verb “kigge (på)” (look (at)) optionally subcategorizes for a PP.

i. but TJ only produced it once out of three intended and once before selfcorrecting it. “slå-ede” instead of the correct “slog”: Intended slået slået slået slået slå slog Slog Slog Produced slået slåede slog slåede slå slog slår kiggede efter slåede slået Correct Wrong Wrong. . Present Ptcp. Finally. 2001 92 the two common verbs “ligne” (resemble. R. look like) and “slå” (hit). #69) (#23. Correct Correct Correct Wrong Wrong Appendix E (#17. #65) (#38) (#95) (#28) (#04) (#29) (#68) Table 17: TJ’s use of the verb “slå” (hit).K. The correct form is “slog”. No other verb or other type of word was used in a similar manner. It seems that the lexical entry for this verb is selectively impaired. #37. Danish has +zero morphology. which have regular and irregular inflection respectively: (76) Tense: Infinitive Present Past Past Ptcp. Self-correction. In cases where he produced the wrong tense form and substituted it with the past form (4/8). Wrong Self-correction. C. #18. he produced the verb stem with the regular inflection. there are no bare stems in TJ’s speech production. However. which implies that agrammatics will omit inflection affixes and produce bare stems. Regular: lign-e lign-er lign-ede lign-et lign-ende Irregular: slå-Ø slå-r slog slå-et slå-ende What is interesting is TJ’s use of the past tense form of “slå”. This may be a positive effect of the eight months of rehabilitation TJ has undergone.e.

He shows no signs of impairment to the functional nodes CP. R. but I fail to see any other and (importantly) better explanation. he does not perform according to a canonical word order pattern.K. I suspect that the reason why TJ’s test results can neither falsify nor verify any of the hypotheses is that he has been in recovery for so long. The subject-object construction is special because the object moves across the subject in the embedded clause and the relative pronoun and its antecedent do not have the same grammatical roles. For example. Importantly. there is nothing in his comprehension performance that falsifies the hypotheses. which have the nonorder and he performs correctly on the sentences with and subject. as the movement of the object is crucial.3. First of all. because he has no problems with understanding for instance object-subject relatives.3. . He only has some problems with the interpretation of subject-object relatives. I discuss these matters in more detail in the following chapter. which is not predicted by the Tree-pruning Hypothesis. 2001 93 7. which would predict good performance on SVO and poor performance on all others. C. His production shows a clear tendency to reduce the number of VPs in the sentences. He performs correctly on for example passives. Neither does he perform according to canonical θ-role order. this phenomenon does not falsify the hypothesis. He shows some signs of preposition drop with a marked difference between the structural position of the preposition in question. I see nothing in his performance that is seriously problematic for the hypotheses. It therefore seems likely that it is a case of trace deletion after all. TnsP. and NegP and his test results can therefore neither verify nor falsify the Tree-pruning Hypothesis. subject-object relatives are not problematic because the relative pronoun (the wh-element) and its antecedent do not have the same grammatical roles (subject and object). which were predicted to result in performance below chance.4 Conclusions TJ does not confirm the hypotheses of Trace Deletion and Tree-pruning. which would predict that only canonical THEME-AGENT AGENT-THEME order would be result in correct performance. However. AgrP. he omits governed prepositions more frequently than ungoverned prepositions. As predicted. At the time of testing TJ had ceased to be a clear and representative case of Broca’s aphasia as he had almost fully recovered. his comprehension does not pattern the way that the Trace deletion Hypothesis would predict. This is not borne out. The hypothesis as such cannot explain this. EXPERIENCER Furthermore.

3. Figure 11 on page 83 above is a gross illustration of the region affected in TJ’s case. As I showed in the previous section. which should have shown problems with all constructions involving syntactic movement. which are both right-embeddings (or right-branching structures). The medical report is not specific or detailed enough to pinpoint the sites any more accurately than indicated with the light shaded areas. are in the region affected by the massive hematomas. but due to eight months of intense training with speech therapists he has come a long way. Admittedly. but see section 7. it should again be kept in mind that TJ has been in rehabilitation for eight months. At least this is in accordance with the hypothesis. and therefore brain damage does not necessarily lead to permanent impairment if the lesion site is not too great or encompassing with regards to the functional area in question. 2001 94 8 Discussion: Cerebral Area and Function Revisited TJ suffered massive hematomas in the areas of the perisylvian region. the figure is not intended to be interpreted too literally. His comprehension of all other constructions is not impaired. He had no problems with actives (except for lexical passives. notably Broca’s area (and Wernicke’s area). Furthermore. it is important to notice that the lowest performance is on a construction with movement of the object across the subject. However. which means that his entire grammatical neural ‘machinery’ was implicated. or rather has been restored.1 and 7.K. but not necessarily destroyed.2 for comments). He has only fragments of aphasia left. with only chance performance on subject-object relatives – a deficit. Once neurons are destroyed they are gone forever. Elman et al.2. his comprehension is near normal. but only with regards to subject-object relatives. but importantly his performance did not falsify the theoretical apparatus posited in chapter 6 and sections 7. A distinction between center. 1996). he also . C. Hence. R. on which he performs 70% correct. However. Disrupted connections can be ‘rebuilt’ (cf. these results deviate from the pattern predicted by the Trace Deletion Hypothesis. TJ did in fact show traces of such behavior.3. for which subtle theoretical mechanisms are required to detect it.and right-embedding is insufficient as a measure of his performance: He correctly interprets object-subject and object-object relatives. They can not regenerate. His comprehension is impaired. The point is that his language areas. the connective system between neurons can regenerate or reorganize. Thus. The function of the affected area can sometimes be restored over time. which should reveal certain predictable behaviors in the linguistic performance of the patient.

42. . due to memory limitations) or no errors.g. In 75% (9/12) of the wrongly produced. Such overuse is only evident in languages that permit pragmatic word order variation [such as Danish. C. The most notable problematic type is the same as the one causing problems in comprehension: the subject-object relative. Crucially. he performs poorly on only subject-object relatives. correlation between right-embedding and above chance performance and between center-embedding and chance performance is not found. 1991: 131. R. and object clefts in Table 12). and Pinker (1994) all shared a common important feature in their respective theories on area and function: Broca’s area and its vicinity are responsible for syntagmatic / relational / sequential computation of grammatical structure. It does. For example: Indeed. which are also center-embeddings. it could not be detected in a rigid word order language like English.K. which are center-embeddings.g. R. Importantly. which seems to be the preferred type of embedding. Emphasis added. if he were just good at repetition. C.9% (9/21) of the object relatives are produced as subject relatives. nor a strong preference for ‘basic order’ of AGENT and THEME. as he performs near-normal on subject-subject relative clauses. Hence. On the production side13 he also seems to have problems with object-relatives (cf. as for example he performs 100% correct on the “blive” passives. Recall from chapter 5 above. the auxiliary verb reduction. shows that he is not merely ‘parroting’. is crucially involved in linguistics syntagmatic relations. however. Damasio (1992). which would not necessarily involve his grammatical competence. this is also compatible with Grodzinsky’s 13 It should be noted that this production is repetition and spontaneous production. In total. reflect a preference for a basic word order (Subject-Verb-Object) in production.]. The systematic production errors suggest that TJ’s grammar is involved. Deacon (1997). (Bates et al. (1991). This reflects neither a problem with (and hence avoidance of) center-embedding. K. that Bates et al. some patients (particular Broca’s) appear to overuse basic SVO.) This finding is compatible with the postulation that the front end of the perisylvian region. e. as is often noted as a characteristic of Broca’s aphasia. TJ’s performance on subject-object relatives. 2001 95 correctly interprets subject-subject relatives. as though this word order type provided a kind of “safe harbor” for sentence planning. Parrotlike repetition would result in random errors (e. However. he changes the object relative clause to a subject relative. object-object relatives. the fact that he changes the sentences on repetition in a systematic way. which is one of the centers of TJ’s hematomas (see Figure 11).

cf. the Projection Principle (section 2. Governed prepositions are more frequently omitted than ungoverned prepositions. So.K. 2001 96 hypothesis that Broca’s area (and vicinity) is responsible for trace-antecedent (coindexing) chains in comprehension and for the functional projections in the syntactic tree and their involvement in movement and hence chains. the omission of auxiliary verbs may be related to a lexical deficit. This means that less syntagmatic / relational / sequential computation is required. He has a tendency to drop the verbal passive morphology “–s”. some paradigmatic computation is required as well. None of the verbs expected to be finite and inflected for tense were uninflected. it demands some syntagmatic computation. However. The lexical entry determines the presence of certain phrases as opposed to others. and subjects.1. as the missing verbs do not project VPs. negations. As mentioned above.2 above. The X-bar component (cf. The omission of governed prepositions may also be an expression of the disruption of relational computation linked to a lesion in Broca’s area. C. I shall return to this shortly. subcategorization is also very much a lexical matter. R. As this is matter of grammatical structure. Neither does TJ show any problems with correctly producing adverbs.1. For example the transitive verb “smile” (smile) subcategorizes for a PP headed by the preposition “til” (at) with the THEME as complement: (77) [CP Drengen1 har … [VP t1 smilet [PP til [NP pigen …]]]] The boy has smiled at the girl The point is that the verb determines the structure of the complement phrase that follows it. The verb that heads the verb phrase in which the omitted preposition was supposed to be subcategorizes for a PP. and he omits few prepositions.3). he has a selective impairment to the past tense form of the “slå” (hit). cf. On the other hand. . Table 16 on page 91. cf. Leaving out the auxiliaries reduces the structure (minimizes the size) of the syntactic tree. this shows that the entire syntactic representation is intact. Perhaps related to this is the fact that TJ has a tendency to use constructions without auxiliary verbs. TJ does however make a few errors. As the CP node and all the nodes beneath it are restored in TJ no severe tense inflection errors are found. section 2. in the sense that the number and syntactic categories of a verb’s arguments is determined by its lexical entry.

but may also be influenced by the damage to Wernicke’s area. The model.1) of the grammar determines the structure and in interaction with Move-α (and the rest of the grammar. In turn this activation selects the appropriate lemmas. Based on Myers-Scotton & Jake 2000: 1056. ⇓ Lemma Level: Content morphemes (directly selected) → Early system morphemes (indirectly selected) ⇓ Functional Level: Formulator → Late system morphemes ⇓ Positional Level: Phonetic/surface forms Figure 12: Production process diagram. is based on a four-way classification of morphemes. they are either theta-assignors or theta-assignees.K. For example. i. figure 1. Let us for a moment consider an alternative model.e. Information on the categories of predicate and arguments are also given by the feature bundles. C. called the 4-M model. These lemmas then potentially select certain system morphemes. Lemma is another word for lexical entry. 2001 97 section 2. R. as opposed to the classical two-way model: function words/morphemes (closed class) and content words/morphemes (open class). On the conceptual level the speaker has some kind of intention to say something. such as the Case Filter) determines the structural position of the phrases. such that “look” is the head of a VP . I shall give a brief overview of their hypothesis. TJ’s omission of governed prepositions is most likely caused by trauma to Broca’s area. They are called content morphemes and distinguish themselves from the other types of morphemes by having the feature [+THEMATIC ROLE].1. which leads to the activation of certain semantic/pragmatic feature bundles. Myers-Scotton & Jake (2000) present a model of word selection or rather lemma activation. prototypically predicates (verbs) and their arguments (typically NPs). First consider the model for the production of sentences: Conceptual Level: Speaker’s intentions → semantic/pragmatic feature bundles. if the speaker intends to say something like “She looks at him” the verb “look” and its arguments “she” and “him” are directly selected by feature bundle activation.

which mark the orders of head-complement and complement-head. R. Furthermore. They have to “look outside” their maximal projections (AgrP): They depend on grammatical information outside of the immediate maximal projection in which they occur. Together the content morphemes and the early system morphemes provide information to the formulator at the functional level. an early system morpheme. In summary. 2001 98 and the arguments are NPs. An example of an outsider morpheme is the before mentioned English 3rd person singular agreement suffix. This is indirectly selected as it is the verb that ‘calls’ it – it is not directly selected by feature bundle activation. (Myers-Scotton & Jake 2000: 1064) In other words.” Examples of this are the English preposition “of” in “friend of Tom” and the possessive suffix “-s” in “Tom’s friend”. the verb (the head) selects the preposition “at”. the verbal suffix is dependent on movement in order to connect with the verb. These late system morphemes are divided into two categories: outsider morphemes and bridge morphemes. some additional system morphemes are required (they are called by the formulator) such as the agreement suffix (3rd person singular) on the verb. In order to do this. Bridge system morphemes “connect content morphemes with each other without reference to the specific semantic/pragmatic properties of a content head. C. which assemble larger hierarchical constituents. respectively. This information is only available when the formulator sends directions to the positional/surface level for how maximal projections are unified in a larger construction.K. the distinction between the four types of morphemes can be stated as follows: .

respectively. Considering the data on TJ. 2001 99 [+/.Conceptually activated] [+ Conceptual] Content morphemes & Early system morphemes [.Looks outside maximal projection] [+ Thematic] Content morphemes [. patients with Broca's aphasia will produce content morphemes more accurately than any of the system morphemes. According to Myers-Scotton & Jake. This fits very well with the conclusions above (cf.Thematic role assignor/assignee] [+/. both outsiders and bridges.Thematic] Early system morphemes [. the 4-M model provides the following classification of impaired morphemes: . The late system morphemes are part of the "structure-building apparatus" or syntagmatic computation: “The 4-M model implicates late system morphemes. C. as expressions of the relational aspects of language” (Myers-Scotton 2000: 1076).K.Outside] [+ Outside] Bridge morphemes Outsider morphemes Figure 13: Feature distribution and classification of morphemes. R. and late system morphemes will be missing or used less accurate. figure 2. the discussion in chapter 5): anterior and posterior lesions in the language area leads to damage to syntagmatic and paradigmatic linguistic computation. Adapted from Myers-Scotton & Jake 2000: 1062.Conceptual] Late system morphemes [+/.

the Danish auxiliary verbs “have” and “blive” are aspect markers and must therefore be conceptually activated. the order of the arguments) in the same way as the English genitive “-s” and possessive “of” signal complement-head and headcomplement order respectively. the 4-M model correctly predicts that content words are not impaired. which are late system morphemes. This pattern is also reflected in the speech of TJ. From hereon. "blive". R. the model predicts impairment affecting tense morphology.K. the morphemes with the feature [+THEMATIC ROLE]. I have categorized the passive “-s” and the adjoined prepositions as bridge morphemes.e. it is quite the opposite: in the 4M model governed (VPinternal) prepositions are early system morphemes and should therefore be less impaired than ungoverned (adjoined) prepositions. because they merely signal the ‘direction’ of the predicate (i. As argued above. in embedded clauses the tense inflection itself has to move into VP to connect with the verb. Early system morphemes are those selected by the content morphemes and part of the conceptually activated morphemes. Tense inflection is dependent on the verb moving out of VP and into TnsP in main clauses. it is predicted that early system morphemes are less impaired than late system morphemes. while adjoined prepositions are not. Actually. 2001 100 (78) Classification of impaired morphemes Content morphemes: Early system morphemes: Bridge morphemes: Outsider morphemes: none auxiliary verbs (aspect). On the other possible extreme. Tense inflections are outsider morphemes as they depend on information outside their own maximal projection TnsP. If Danish had agreement morphology it would have been outsider morphemes as well. According to Grodzinsky (1988) in agrammatism VP-internal prepositions are deleted/omitted. cf. The VP-internal prepositions are subcategorized for by verbs (content morphemes) and are therefore early system morphemes. Provided my classification is correct. which is also predicted in the Tree-pruning Hypothesis. VP-internal prepositions passive "-s". The predicate and arguments are produced. With regards to prepositions this is not identical to what was predicted in the model outlined in the previous chapters. I think the 4-M model loses compatibility: First. Table 16 on page 91. C. VP-internal prepositions depend on both . adjoined prepositions tense Content morphemes are predicates and arguments.

part of the "structure building apparatus". TJ has a strong tendency to omit the verbs related to aspect (+/.2 above). Furthermore. 1991.g. i. C. With regards to auxiliary verbs. It does. would be predicted correctly by the 4-M model. R.e. However. Deacon 1997. For example. Damasio 1992. and they depend on the grammar to specify position in the structure. This is not accounted for in the 4-M model. while the passive marker "blive" is not omitted. using the 4-M model I fail to see how to account for the distinction in comprehension between constructions with and without syntactic movement (of the underlying object). In the 4-M model agreement and tense. 2001 101 syntagmatic and paradigmatic computation: they are called by the verb. For this reason. If the auxiliary verbs are considered to be late system morphemes. as they are "called by" the main verb in order to specify semantic aspect. for example.perfect). are in the same category. TJ's auxiliary verb omission.K. This has already been done by the other approaches discussed so far (Bates et al. 2000) point to a distinction between tense and agreement in agrammatism (see also the discussion on the Split-inflection Hypothesis in section 6. Furthermore. As stressed several times throughout the text. such that the presence of a PP is specified. the model has nothing to say about on issues of movement related to the impairment of the top-most nodes. be an artifact of the lack of elaboration an examples in the article by Myers-Scotton & Jake. however. I think that the 4-M model is a clear and good proposal of the process of morpheme selection. auxiliaries must be considered early system morphemes. This is incompatible with the 4-M model as well. On the whole. as reported by Grodzinsky. which I considered structure reduction above. Grodzinsky 2000. on the other hand. Broca's aphasics do not lose . The prediction by the 4-M model that the passive morphology is impaired is borne out in the speech of TJ. even though this distinction alone is insufficient. the 4-M model lacks breakdown compatibility. the predictions provided by the model are incompatible with the data from TJ. This may. Evidence provided by e. a mere distinction between syntagmatic and paradigmatic is insufficient. and as such they must be equally impaired. once again support the distinction of syntagmatic (bridge and outsider morphemes) and paradigmatic (content and early system morphemes) computation. Pinker 1994). Grodzinsky (2000) and Friedmann & Grodzinsky (1997. which appears to clusters practically all of grammar into one: the formulator. On the other hand.

posterior language area. Grodzinsky’s hypotheses. R. (1991) agrammatics tend to overuse SVO order. and perhaps his problems with the lexical passive verb "væmmes" (be disgusted) in comprehension. Clearly not all relational aspects are lost. agrammatic speech is not a random flow of words. such as his problems with the irregular inflection of the verb "slå" (hit) in production. however incomprehensible. The word substitutions are frequently within the proper semantic field.g. but still his deficit seems to be syntactic. For these reasons the functional distribution within the language area has to be finer than assumed by a gross dichotomies such as syntagmatic vs. However little evidence of aphasia left in TJ. In Wernicke's aphasia. only reflected in subject-object relatives in comprehension. syntax vs. The phonemic substitutions. as the words appear not to be completely random. his performance is not distributed randomly over syntagmatic computation. I believe. The distinction in comprehension between movement and non-movement clearly supports this. He also makes some word substitutions in production. but in production all object relatives are affected. He shows remnants of a movement-related deficit. anterior vs. the sentences. The lesion to his posterior region. is reflected in the few examples of lexical deficits. lexicon. for example "chair" for "table" and "knee" for "elbow". According to Bates et al. and I agree that the evidence seems to demand it. Even though this is hinted at in the 4-M model. "tubber" for "butter" also point to at least some sparing of paradigmatic relations.K. C. paradigmatic. chapter 6 above). Grodzinsky proposed a syntactic approach to (Broca's) aphasia (cf. . however. so some aspects of word finding / paradigmatic relations appear to be affected. e. In production. are not formed at random. only parts of it. do not capture this late stage in recovery in any precise manner. it is still not elaborate enough to account for all the phenomena associated with agrammatism (or the other types of aphasia for that matter). 2001 102 all of their syntax (syntagmatic linguistic computation).

confirms the internal modular structure of language. but functional separation. As Grodzinsky (2000: 18) puts it." As other aspects of grammar are not disrupted by damage to Broca's area. nothing else. As trace-deletion and tree-pruning are clearly two different representational aspects of grammar they can hardly be computed by the same neural mechanism.” Grodzinsky (2000: 15) relates this variation to the zero-morphology parameter. As agrammatism is caused by trauma to Broca's area. operculum. Furthermore. A look at Figure 6 on page 41 above. such that agrammatism is manifested differently depending on the morphology of the language: if the language has +zero-morphology agrammatics will tend to show omission of inflection. in which cerebral areas within the language zone are correlated with different types of aphasia. R. only the mechanisms underlying the specific syntactic abilities affected in the deficit are located there. So. whereas if the language has –zero-morphology agrammatics will substitute inflection. The fact that this variation can be captured in the same grammatical framework that accounts for normal competence makes the theory breakdown-compatible. the aspects of grammar affected must be located in that same area. are not located in the anterior language areas – Broca’s area and deeper white matter. The lesion site thus has some sort of language-specific and grammar-specific function. the apparatus responsible for these aspects of grammar must be located in Broca's area and its vicinity. and therefore all of grammar cannot be located in on spot. (Grodzinsky 2000: 17) . it suggests "anatomical proximity. For this reason they may be located in close vicinity of each other instead of in the very same place. Different aspects of grammar are affected depending on lesion site. In comprehension agrammatism is manifested as a disruption of traceantecedent chains. C. In fact: [M]ost human linguistic abilities. and anterior insula. including most syntax. and in production it is manifested as pruning of the topmost functional nodes of the syntactic representation. 2001 103 9 Conclusions Paradis (1998: 418) states that “[i]t is now known that the symptoms of agrammatism will vary in accordance with the structure of each language.K. the deficit will vary according to the word order of the language.

TJ’s deficit was at the time of testing reduced to a problem with object relatives. language will be located in the language zone of the left hemisphere.it constitutes the universal grammar. in which such research has not yet been made. R. Second. This is the case in normal subjects without history of early childhood lesions in the left hemisphere or any other cerebral pathology.2) also support a neural basis of language. this points to external modularity of language. C. Table 4. 2001 104 It is up to further investigation to discover the connections between different specific aspects of grammar and different types of aphasia in correlation with lesion site. in loose terms. and therefore his production deficit . had been apparent at earlier stages). Third. which involves movement of the object across the subject. his entire clause structure is intact. as general learning in such subjects is severely reduced. innateness. as general intelligence and language are doubly dissociated.K. I applied the framework to the Danish language and tested the predictions thus made on TJ. according to his speech therapist. Furthermore. page 42). they point to a syntactic deficit – specifically related to movement and trace-antecedent chains. Though his test results can neither falsify nor completely verify the hypotheses. The robustness of language in the face of severe retardation (cf. his impairment was caused by hematomas in multiple places. TJ had been in recovery for too long to show any clear signs of agrammatism (which. his medical report was rather unclear on the exact locations of these hematomas. The human grammatical competence is distributed over the perisylvian region of the left hemisphere. In order to test the hypotheses of Grodzinsky’s syntactic approach to Broca’s aphasia. the very same theory should be able to describe all the reported types of aphasia (and recovery). section 4.3.2 and 3.4. and most important. a Danish agrammatic aphasia patient. a matter of adjusting the grammar to the one language spoken in the ambient society. One may be present without the other (cf. Language acquisition is. language is innate and reflected in the architecture of the brain (as well as our speech apparatus). as this language capacity can (supposedly) be described by a comprehensive linguistic theory. but localized within this region. In other words. Furthermore. i. as argued in sections 3. As such the structure of the language zone constrains the range of possible human languages . As I have argued. First. If normal development is not disrupted. Furthermore. TJ was an unclear case for three reasons. This is what I called the default brain plan in section 3.e. However. such a theory should be able to make predictions regarding the performance of aphasics in any language.

R. b) Rebuilding of the syntactic tree and recovery of comprehension and production – not necessarily in that order. Therefore. Table 10 on page 71) are not movement derived. C. What do the other constructions have that the object relatives do not have? A) All the constructions predicted to yield above chance performance in comprehension (cf. . c) Strengthened grammatical θ-role assignment.g. such that the grammatical assignment has priority over the cognitive assignment. tree pruning.K. This is predicted to lead to in e. we have to consider what the comprehension and production deficits have in common: object relatives. This would account for the fact that TJ only performs poorly on constructions where all θ-roles has to be assigned through θ-transmission or cognitive assignment. 2001 105 cannot be account for by reference to tree pruning. it seems that in TJ the overrides this potential competition. verbal passives. I propose that the AGENT competition (two AGENTS) presence of the adjoined direct θ-assignor-assignee connection has been strengthened. while others may not and thus stagnate at a prior stage): a) Stages of severe agrammatism: Trace deletion. However. B) The three types of passives (verbal. which is manifested as reduced error rates. the former due to someone else’s intact grammar. “blive”. which however. If I am correct. and psychological) have a preposition to assign AGENT/EXPERIENCER grammatically. If the trace is deleted or the chain otherwise broken this transmission is no longer possible and the cognitive Default Strategy assigns AGENT to the leftmost NP in a θ-position. the latter due to the intact CP in TJ). this underlying deficit of θ-transmission should be incorporated in the framework such that different stages of recovery are captured as well as the most severe stage. d) Reconstruction of trace-antecedent chains. His comprehension deficit can not be accounted for by mere reference to trace deletion. In object relatives the object is not assigned case directly by the verb but through θ-transmission from the trace (both comprehension and production contains movement derived constructions. Recovery would thus go through at least the following stages (some may recover fully. I think is still involved. e) Recovery. on which Grodzinsky’s theory is based.

Though my proposals may remedy some of the problems it still has nothing to say about the systematic omission of auxiliary (aspect) verbs. R. the Default Strategy will be down prioritized and eventually abandoned. There is still much work to be done. In turn. 2001 106 The strengthening of the direct grammatical θ-assignment will result in decreasing error rates as there is less competition between the two identical θ-roles in e. . reversible passives. such that the correct θ-roles will be assigned to the appropriate NPs.g. His theory currently captures only some of the earlier stages of recovery. furthermore. C. it may lead to the reconstruction of grammatical θ-chains.K. Granted that my proposal of the strengthening of direct θ-assignment is correct (keeping in mind that TJ may not at all be a good example of agrammatism) it implies that Grodzinsky’s theory has not yet fully gained Breakdown Compatibility. Due to the reconstructed θchains object-relatives will reappear in the speech production.

D. S. Terrence Langendoen (Eds. In Bishop & Mogford (Eds. (1989) Autism. Leslie. and Brian MacWhinney (1991) Cross-linguistic research in aphasia: an overview.ac. A. Bihrle. U. Jennifer E. & Y. Dorothy (1988) Language development after focal brain damage. Derek (1988) Creole Languages and the Bioprogram. In Bastiaanse & Grodzinsky (Eds. In F. Brain and Language 41: 123-148 Belletti. Frith (1985) Does the autistic child have a ‘theory of mind’? Cognition 21: 37-46 Bastiaanse. C... Beverly Wulfeck. R. Grodzinsky (Eds. Hove: Psychology Press Ltd.) (1988: 177-189) Bickerton. Vol.) (2000) Grammatical Disorders in Aphasia: a Neurolinguistic Perspective.V. Bickerton.) (1988: 203-219) . Aspects of Verb Syntax. Dorothy & Kay Mogford (1988) Five questions about language acquisition. 2001 107 10 References Archangeli.) (2000) [Pdf version available at http://freud. Elizabeth.) (1988/1993) Language development in exceptional circumstances. Chicago: The University of Chicago Press Bishop.) (1988: 239-260) Bishop.) (1997) Optimality Theory. & Yosef Grodzinsky (2000) Levels of linguistic representation in Broca’s aphasia: Implicitness and referentiality of arguments. Marks.tau. Newmeyer (ed. Diana & D.M. and H. S. London: Whurr Publishers Bates. Derek (1990) Language and Species. U.2.K. Bishop. Oxford: Blackwell Publishers Balogh. In Bishop & Mogford (Eds. Cambridge: Cambridge University Press: 268-284. Dorothy & Kay Mogford (Eds. An Overview. A. Adriana (1990) Generalized Verb Movement.) Linguistics: The Cambridge Survey. M. Turin: Rosenberg & Sellier Bellugi. R.il/~yosef1/] Baron-Cohen. Asperger’s syndrome and semantic-pragmatic disorder: Where are the boundaries? British Journal of Disorders of Communication 24: 107-121 Bishop. In Bishop & Mogford (Eds. Sabo (1988) Dissociation between language and cognitive functions in Williams syndrome.

531-539 Damasio. Cambridge. 8. Le Bihan (1997) Anatomical variability in the cortical representation of first and second language. Evolution of the Generative Mind.: Harvard University Press. Charles (1859/1964) On the Origins of Species. C. 2nd Edition. E. & J. Terry (1997) Introduction to Historical Linguistics. Noam (1993) A minimalist program for linguistic theory. Merlin (1991) Origins of the Modern Mind.ac. J. Scientific American. Dupoux. In Stemmer & Whitaker (1998: 313: 329) Crowley. Cambridge. S. K. London: Penguin Dehaene. Antonio R. Deacon.. NeuroReport 8: 3809-3815 Donald. Antonio R.: Harvard University Press . The New England Journal of Medicine 326.) The View from Building 20. The Co-evolution of Language and the Human Brain. William H. Oxford: Oxford University Press Damasio. Paulesu. R. Cohen. Antonio R. (1994) Descartes’ Error. van de Moortele. Dordrecht: Foris Chomsky. D. Oxford: Blackwell Corballis. (2000) The co-evolution of language and the brain: A review of two contrastive views (Pinker & Deacon). Oxford: Oxford University Press Corina. Ojemann (1994) Conversations with Neil’s Brain. Mass. Ken R. Noam (1981) Lectures in Government and Binding. Mass. (1991) The Lopsided Ape. September 1992: 63-71 Damasio. Reading MA: Addison-Wesley Chomsky.: MIT Press.-F. E. 3rd Edition.kfunigraz. Terrence (1997) The Symbolic Species. P. (1992) Aphasia. Evidence from aphasia.K. In Hale. 1-52 Chomsky. Keyser (Eds. & Hanna Damasio (1992) Brain and language. Strang & Jane Grimshaw (1992) Coordination of VP-internal subjects. David (1998) The processing of sign language. Mass. Bernard (1989) Language Universals and Linguistic Typology. To appear in Grazer Linguistische Studien [http://www-gewi. Linguistic Inquiry 23 (2): 305-313 Calvin. Mehler. L. Cambridge. 2001 108 Burton. Noam (1995) The Minimalist Program. New York: Grosset/Putnam Darwin. & George A. Lehéricy & D. Michael C. S. Perani.at/ling/gls/] Comrie. Cambridge Mass: MIT Press Christensen. Three Stages in the Evolution of Culture and Cognition.

M. Science 245. An Introduction.. Dan & Yosef Grodzinsky (1999) Comprehension regularity in Broca’s aphasia? There’s more to it than you ever imagined. Domenico Parisi & Kim Plunkett (1996) Rethinking Innateness. Lila R. C. Brain and Language 56: 397-425 Friedmann. (1998) Building a picture of the brain. Cognition 39: 1-50 Grimshaw. group size and language in humans. R. R. Oxford: Basil Blackwell Gazzaniga. Jeffrey L. Na’ama & Yosef Grodzinsky (2000) Split inflection in neurolinguistics.: MIT Press Gopnik. Behavioral and Brain Sciences 8: 1-42 Foley.H. [Pdf-version available at http://www-psy. and optimality. Crago (1991) Familial aggregation of a developmental language disorder. A Connectionist Perspective on Development. 947952 Gleitman. & Mark Liberman (Eds. Friedemann & L.: MIT Press Fay. 2nd Edition. Elizabeth A. Jerry A. Brandeis University Grimshaw. Yosef (1984) The syntactic characterization of agrammatism. Cambridge Mass: MIT Press Fodor. (1983) The Modularity of Mind. Dædalus 127: 37-69 Elman. Cambridge Mass. Mark H. (1997) Anthropological Linguistics.M. & M. Volume 1: Language. Oxford: Blackwell Friedmann.edu/~naama/] Frith. heads. Jane (1997) Projection. W. Linguistic Inquiry 28.) (1995) An Invitation to Cognitive Science. (1988) Infantile autism.3: 373-422 Grodzinsky. MS.I. Uta (1989) Autism. Brain and Language 70: 139143 Dunbar. Jerry A. Rizzi (Eds.K.) (1988: 190-202) Fodor. Cambridge Mass.ucsd. In M. Explaining the Enigma. Na’ama & Yosef Grodzinsky (1997) Tense and agreement in agrammatic production: pruning the syntactic tree.-A. New York: Longman. 2001 109 Drai. Annette Karmiloff-Smith. Michael S. (1985) Précis of the modularity of mind. Jane (1991) Extended Projections.) (2000) The Acquisition of Syntax: Studies in Comparative Developmental Linguistics. Johnson. William A. Cognition 16: 99-120 .. Behavioral and Brain Sciences 16 (4): 681-735 Edelman. In Bishop & Mogford (Eds. Gerald M. (1989) Organization of the human brain. (1993) Co-evolution of neocortex size. Bates.

Behavioral and Brain Sciences 23 (1): 1-71 Grodzinsky.: MIT Press Grodzinsky. Susan Marakovitz & Julie Solomon (1993) The breakdown of binding relations. Brain and Language 41: 555-546 Grodzinsky. Brain and Language 51: 469-497 Grodzinsky. Brain and Language 27: 135-159 Grodzinsky. Language and Speech 31 (2): 115-134 Grodzinsky. Brain and Language 50: 27-51 Grodzinsky. 2nd Edition. R. Yosef (1995a) A restrictive theory of agrammatic comprehension. Liliane (1994) Introduction to Government and Binding Theory. (1972) Semantic Interpretation in Generative Grammar. (1995) The Psychology of Language. C. θ-roles. Hove: Psychology Press. Yosef (1991) There is an entity called agrammatic aphasia. Cambridge Mass. Brain and Language 37: 480-499 Grodzinsky. Yosef (2000) The neurology of syntax: language use without Broca’s area. Yosef (1988) Syntactic representations in agrammatic aphasia: the case of prepositions. Yosef (1995b) Trace deletion. 2001 110 Grodzinsky. (Eds. René (1999) Optimality Theory.K. Andrew & Patricia McCabe (1975) Intelligence and aphasia: performance of aphasics on Raven’s Coloured Progressive Matrices (RCPM). and cognitive strategies. Jackendoff. Cambridge: Cambridge University Press Kertesz.: MIT Press Jakobson. The Hague: Mouton Kager. In R. Trevor A.. Yu-Chin Chien. Yosef. Yosef (1986) Language deficits and the theory of syntax. Cambridge Mass. Halle.) Fundamentals of Language. Roman (1956) Two aspects of language and two types of aphasic disturbances. Kenneth Wexler. Jakobson & M. Ltd. Brain and Language 2: 387-395 . Yosef (1989) Agrammatic comprehension of relative clauses. Yosef (1990) Theoretical Perspectives on Language Deficits. Oxford: Blackwell Harley. Brain and Language 45: 396422 Haegeman. Ray S.

and Scott Rauch (1996) Localization of syntactic comprehension by positron emission tomography. Nathaniel Alpert.) (1995) Pinker. Steven & Paul Bloom (1990) Natural language and natural selection. Loraine K. Lise. Amsterdam: John Benjamins Publishing Company Müller. B. H.) (1988: 165176) Smith.) (1998: 417-430) Pinker. J. Linguistics 38(6): 1053-1100 Paradis. 452-473 . Jean-Yves (1989) Verb movement. E. O'Connor. 9th English Edition. C. & H. Brain and Language 52. universality? Neurobiological approaches to language. Steven (1994) The Language Instinct. Neil & Ianthi-Maria Tsimpli (1995) The Mind of a Savant.: Harvard University Press Menn. Steven (1995) Language acquisition. Behavioral and Brain Sciences 19: 611-675 Myers-Scotton. Jake (2000) Four types of morpheme: evidence from aphasia. universal grammar. (1984) The Biology and Evolution of Language. (1967) The Biological Foundations of Language.A.) (1998) Handbook of Neurolinguistics. Whitaker (Eds. code switching. Karin. M. New York: Wiley Lieberman. David Caplan. Oxford: Blackwell Sobotta. München: Urban & Schwarzenberg Stemmer. 2001 111 Larson. R. Becher (1975) Atlas of Human Anatomy. Michel (1998) Language and Communication in Multilinguals. In Bishop & Mogford (Eds. Behavioral and Brain Sciences 13: 707-784 Pinker.A. J. Linguistic Inquiry 20(3): 365-424 Rondal. In Gleitman & Liberman (Eds. London: Penguin Pollock. and the structure of IP. R. Ralph-Axel (1996) Innateness. autonomy. Cambridge Mass. Steven (1997) How the Mind Works. (1988) On the double object construction. (1988) Down’s syndrome. Linguistic Inquiry 19(3): 335-391 Lenneberg. Carol & Janice L. Obler & Audrey Holland (1995) Non-fluent Aphasia in a Multilingual World. Language Learning and Modularity. & H.K. P. K. New York: Academic Press Stromswold. and second-language acquisition. London: Penguin Pinker. in Stemmer & Whitaker (Eds.

David & Norbert Hornstein (eds. Sten (1999) Ledstillingen i dansk og government & binding..J. H. head movement and the economy of derivations. Current Biology 8. Ken (1994) Optional infinitives. R. Syddansk Universitet. Per Anker & Peter Skadhauge (Eds. In Lightfoot. Grimshaw.J.) Sætningsskemaet i Generativ Grammatik. 305-350 .K. Working Papers in Scandinavian Syntax 44: 141-155 Vikner. Oxford: Oxford University Press Vikner. Sten (1995) Verb Movement and Expletive Subjects in the Germanic Languages. 2001 112 Van der Lely.: MIT Press Wexler. McClelland (1998) Evidence for a grammar specific deficit in children.) Verb Movement. 23: 1253-1258 Vikner.K. Rosen & A. S. Cambridge: Cambridge University Press. To appear in Legendre. Kolding Vikner. Sten (2000) V°-to-I° Movement and Do-Insertion in Optimality Theory. C. In Jensen. Mass. & Vikner (Eds. H. Institut for Erhverssproglig Informatik og Kommunikation. & L. Stollwerck (1996) A grammatical specific language impairment in children: an autosomal dominant inheritance? Brain & Language 52: 484-504 Van der Lely. Cambridge.): Optimality-theoretic Syntax. Sten (1989) Object shift and double objects in Danish.K.

1. 8. 2. 2. 18.K. 3. 16. 3. 7. 3. 13. Pigen skubber drengen 1. 13. 15. 17. 6. 12. 11. 12. Hunden bider katten . 10. 6. 8. 11. 6. Pigen skubber drengen Drengen skubber pigen Drengen skubbes af pigen Pigen skubbes af drengen Vis mig pigen som skubber drengen Vis mig drengen som skubber pigen Det er pigen som skubber drengen Det er drengen som skubber pigen Drengen bliver skubbet af pigen Pigen bliver skubbet af drengen Vis mig drengen som pigen skubber Vis mig pigen som drengen skubber Det er drengen som pigen skubber Det er pigen som drengen skubber Pigen som skubber drengen er sur Drengen som skubber pigen er sur Drengen som pigen skubber er sur Pigen som drengen skubber er sur Pigen kysser drengen Drengen kysser pigen Drengen kysses af pigen Pigen kysses af drengen Vis mig pigen som kysser drengen Vis mig drengen som kysser pigen Det er pigen som kysser drengen Det er drengen som kysser pigen Drengen bliver kysset af pigen Pigen bliver kysset af drengen Vis mig drengen som pigen kysser Vis mig pigen som drengen kysser Det er drengen som pigen kysser Det er pigen som drengen kysser Pigen som kysser drengen er glad Drengen som kysser pigen er glad Drengen som pigen kysser er glad Pigen som drengen kysser er glad Hunden bider katten Katten bider hunden Katten bides af hunden Hunden bides af katten Vis mig hunden som bider katten Vis mig katten som bider hunden Det er hunden som bider katten Det er katten som bider hunden Katten bliver bidt af hunden Hunden bliver bidt af katten Vis mig katten som hunden bider Vis mig hunden som katten bider Det er katten som hunden bider Det er hunden som katten bider Hunden som bider katten er sort Katten som bidder hunden er sort Katten som hunden bider er sort Hunden som katten bider er sort 2. 14. 11. 18. 16. 12. 10. 13. 2. 8. 15. 4. 2001 113 11 Appendix A: Sentence-to-Picture Matching Test 1. 5. 9. 1. 14. C. 14. 17. 4. 5. 15. Pigen kysser drengen 3. 4. 7. 17. 7. 5. 16. 9. 10. 18. R. 9.

16. 16. R. 2. 9. 3. 6. 7. 2. 13. 12. 7. C. 8. 4. 2. 15. 17. Pigen løfter drengen Drengen løfter pigen Drengen løftes af pigen Pigen løftes af drengen Vis mig pigen som løfter drengen Vis mig drengen som løfter pigen Det er pigen som løfter drengen Det er drengen som løfter pigen Drengen bliver løftet af pigen Pigen bliver løftet af drengen Vis mig drengen som pigen løfter Vis mig pigen som drengen løfter Det er drengen som pigen løfter Det er pigen som drengen løfter Pigen som løfter drengen er glad Drengen som løfter pigen er glad Drengen som pigen løfter er glad Pigen som drengen løfter er glad Pigen slår drengen Drengen slår pigen Drengen slås af pigen Pigen slås af drengen Vis mig pigen som slår drengen Vis mig drengen som slår pigen Det er pigen som slår drengen Det er drengen som slår pigen Drengen bliver slået af pigen Pigen bliver slået af drengen Vis mig drengen som pigen slår Vis mig pigen som drengen slår Det er drengen som pigen slår Det er pigen som drengen slår Pigen som slår drengen er sur Drengen som slår pigen er sur Drengen som pigen slår er sur Pigen som drengen slår er sur Hunden snuser til katten Katten snuser til hunden Katten bliver snuset til af katten Hunden bliver snuset til af katten . 3. Hunden snuser til katten 1. 18. 14. 17. 10. Pigen slår drengen 6. 2001 114 4. 6. 14. 1. 3. 5. 15. 13. 9.K. 18. 12. 10. Pigen løfter drengen 5. 4. 11. 11. 8. 1. 4. 5.

Pigen peger på drengen Drengen peger på pigen Drengen bliver peget på af pigen Pigen bliver peget på af drengen Pigen griner ad drengen Drengen griner ad pigen Drengen bliver grinet ad af pigen Pigen bliver grinet ad af drengen 8. 3. 4. Pigen peger på drengen 1. 6. 3. 7. 8. C. 4. 2. Pigen smiler til drengen Drengen smiler til pigen Drengen blivet smilet til af pigen Pigen bliver smilet til af drengen 9. R. Pigen vinker til drengen Drengen vinker til pigen Drengen bliver vinket til af pigen Pigen bliver vinket til af drengen . 5. 4. Pigen vinker til drengen 1. Pigen smiler til drengen 1. 2. 2. 3.K. 2001 115 7.

Hunden er irriteret på katten Katten er irriteret på hunden Hunden frygter katten Katten frygter hunden Katten frygtes af hunden Hunden frygtes af katten 12. 2. Drengen er inspireret af pigen 1. 2. 2. C. 4. Hunden er forbavset over katten 1. 3. Hunden er irriteret på katten 1. Drengen er inspireret af pigen Pigen er inspireret af drengen 11. R. Hunden er forbavset over katten Katten er forbavset over hunden . 6. 2001 116 10. 5.K.

1. Drengen hader pigen 15. Drengen væmmes ved pigen 1. 6. 5. 4. 10. 6. 3. 3. 3. 11. C. 4. Drengen elsker pigen Pigen elsker drengen Pigen elskes af drengen Drengen elskes af pigen Drengen er begejstret for pigen Pigen er begejstret for drengen Drengen synes om pigen Pigen synes om drengen Drengen er interesseret i pigen Pigen er interesseret i drengen Drengen beundrer pigen Pigen beundrer drengen Pigen beundres af drengen Drengen beundres af pigen Drengen hader pigen Pigen hader drengen Pigen hades af drengen Drengen hades af pigen 14. Drengen elsker pigen 1. 5. R.K. 12. 8. Drengen væmmes ved pigen Pigen væmmes ved drengen Drengen afskyr pigen Pigen afskyr drengen Pigen afskys af drengen Drengen afskys af pigen . 2. 2. 4. 2. 7. 2001 117 13. 13. 9. 14.

Drengen som løfter pigen er glad 9. Sub-Sub Relative: sub1 [sub1 verb obj] Agent Theme 3.K. Subject Cleft: N°1 [sub1 verb obj] ° Agent Theme . Pigen som kysser drengen er glad 4. Pigen slår drengen 10. Katten som bidder hunden er sort 7. Pigen som slår drengen er sur 10. Pigen vinker til drengen Drengen vinker til pigen 1. C. Vis mig drengen som slår pigen 1. Pigen griner ad drengen 16. Det er pigen som løfter drengen 8. Det er drengen som kysser pigen 5. R. Drengen skubber pigen 3. Katten bider hunden 7. Pigen som skubber drengen er sur 2. Obj-Sub Relative: obj1 [sub1 verb obj] Agent Theme 4. Drengen peger på pigen 15. Drengen griner ad pigen 17. Pigen kysser drengen 4. Vis mig pigen som kysser drengen 4. Hunden snuser til katten 12. Vis mig hunden som bider katten 6. Drengen som skubber pigen er sur 3. Drengen som kysser pigen er glad 5. Katten snuser til hunden 13. Hunden bider katten 6. Det er drengen som skubber pigen 3. Vis mig drengen som skubber pigen 3. Vis mig pigen som skubber drengen 2. Vis mig pigen som løfter drengen 8. Pigen som løfter drengen er glad 8. Pigen skubber drengen 2. Pigen smiler til drengen 18. Hunden som bider katten er sort 6. Vis mig drengen som kysser pigen 5. 2001 118 12 Appendix B: Sentence Types and Tokens 1. Vis mig pigen som slår drengen 10. Det er drengen som slår pigen sub verb prep obj Agent Theme 2. Det er pigen som slår drengen 10. Vis mig drengen som løfter pigen 9. Drengen kysser pigen 5. Simple Active: sub verb obj Agent Theme 1. Det er katten som bider hunden 7. Drengen smiler til pigen 19. Det er pigen som skubber drengen 2. Vis mig katten som bider hunden 7. Drengen som slår pigen er sur 1. Det er hunden som bider katten 6. Det er pigen som kysser drengen 4. Det er drengen som løfter pigen 9. Pigen peger på drengen 14. Drengen slår pigen 11. Pigen løfter drengen 8. Drengen løfter pigen 9.

Katten bliver bidt af hunden 6. Hunden er irriteret på katten 8. Hunden bliver bidt af katten 7. Drengen hader pigen 6. Pigen er interesseret i drengen 3. Pigen elsker drengen 5. Pigen beundrer drengen 3. Pigen bliver smilet til af drengen 19. Drengen bliver slået af pigen 10. Drengen beundrer pigen 2. Pigen bliver kysset af drengen 5. Pigen er inspireret af drengen 5. 2001 119 Lexical Passive: sub verb prep obj Experiencer Theme Adjectival Passive: sub copula adj oblique Experiencer Theme Drengen synes om pigen Pigen synes om drengen Drengen væmmes ved pigen Pigen væmmes ved katten 1. Pigen bliver grinet ad af drengen 17. Hunden er forbavset over katten 10. Pigen hader drengen 7. Drengen afskyr pigen 8. Drengen bliver skubbet af pigen 2. Drengen elsker pigen 4. Pigen bliver peget på af drengen 15. Drengen blivet smilet til af pigen 18. Pigen bliver skubbet af drengen 3. Katten er forbavset over hunden 1. Hunden frygter katten 10. Drengen bliver peget på af pigen 14. R. Drengen er interesseret i pigen 2. Katten frygter hunden 1. Pigen bliver slået af drengen 11. Drengen bliver vinket til af pigen Pigen bliver vinket til af drengen Psychological predicates: sub verb obj Experiencer Theme "Blive" Passive: sub aux verb prep oblique Theme Agent sub aux verb prep prep oblique Theme Agent . Drengen bliver grinet ad af pigen 16. Drengen er begejstret for pigen 6. Katten er irriteret på hunden 9. Pigen afskyr drengen 9. Pigen er begejstret for drengen 7. Pigen bliver løftet af drengen 9. Drengen er inspireret af pigen 4. Drengen bliver kysset af pigen 4. Katten bliver snuset til af katten 12. Drengen bliver løftet af pigen 8.K. C. Hunden bliver snuset til af katten 13.

4. 9. 9. 10. 10. Drengen som pigen skubber er høj Pigen som drengen skubber er høj Drengen som pigen kysser er glad Pigen som drengen kysser er glad Katten som hunden bider er sort Hunden som katten bider er sort Drengen som pigen løfter er glad Pigen som drengen løfter er glad Drengen som pigen slår er sur Pigen som drengen slår er sur Vis mig drengen som pigen skubber Vis mig pigen som drengen skubber Vis mig drengen som pigen kysser Vis mig pigen som drengen kysser Vis mig katten som hunden bider Vis mig hunden som katten bider Vis mig drengen som pigen løfter Vis mig pigen som drengen løfter Vis mig drengen som pigen slår Vis mig pigen som drengen slår Det er drengen som pigen skubber Det er pigen som drengen skubber Det er drengen som pigen kysser Det er pigen som drengen kysser Det er katten som hunden bider Det er hunden som katten bider Det er drengen som pigen løfter Det er pigen som drengen løfter Det er drengen som pigen slår Det er pigen som drengen slår Drengen skubbes af pigen Pigen skubbes af drengen Drengen kysses af pigen Pigen kysses af drengen Katten bides af hunden Hunden bides af katten Drengen løftes af pigen Pigen løftes af drengen Drengen slås af pigen Pigen slås af drengen Pigen beundres af drengen Drengen beundres af pigen Pigen elskes af drengen Drengen elskes af pigen Pigen hades af drengen Drengen hades af pigen Pigen afskys af drengen Drengen afskys af pigen Katten frygtes af hunden Hunden frygtes af katten . 7. 7. 5. 9. 8. 8. 8. 4. 2. 5. 2. R. 8. 2. 6. 10. 6. 1. 6. 9. 8. 3. 3. 5. 7. 9. 6. 3. 1. 4. 2. 5. 3. C. 6. 4. 5. 4. 10. 1. 10. 2. 3. 7.K. 2001 120 Sub-Obj Relative: sub verb] sub1 [obj1 Theme Agent Obj-Obj Relative: obj1 [obj1 sub verb] Theme Agent Object cleft: sub verb] N°1 [obj1 ° Theme Agent Verbal Passive: sub verb prep oblique Theme Agent Psychological Passive: sub verb prep oblique Theme Experiencer 1. 7. 1.

pas PRES_PF lex.pas PRES_PF+ lex.pas: s_s.rel PAST_PF+ o_s.rel FUTURE o_o.rel FUTURE o_s.rel PRES_PF o_s.rel PAST_PF o_o.pas: obj.pas PAST_PF adj.pas PAST adj.rel PAST o_s.rel PAST_PF o_s.pas FUTURE adj.pas PRESENT lex.pas PAST_PF lex.K.pas: sub. C.rel: psy.clf: s_o. R.rel: sim.clf: o_o. 2001 121 13 Appendix C: Repetition Test: Structure Distribution Legend: adj.rel PRESENT o_s.ver: sim.pas FUTURE lex.rel: o_s.pas: psy.rel PRESENT Hunden bliver temmelig irriteret på katten Hunden var ret forbavset over katten Pigen havde nok været inspireret af drengen Pigen har ofte været begejstret for drengen Drengen er meget begejstret for pigen Drengen vil nok synes om pigen Pigen væmmedes nok ved drengen Pigen havde nok syntes om drengen Drengen havde nok haft syntes om pigen Pigen har nok væmmedes ved drengen Pigen har nok haft syntes om drengen Drengen synes vist om pigen Vis mig katten som hunden sikkert vil bide Vis mig drengen som pigen vredt skubbede Vis mig pigen som drengen ofte havde slået Vis mig pigen som drengen vredt havde haft slået Vis mig pigen som drengen sikkert har kysset Vis mig pigen som drengen sikkert har haft kysset Vis mig drengen som pigen tit kysser Vis mig hunden som først vil bide katten Vis mig drengen som vredt skubbede pigen Vis mig pigen som vredt havde slået drengen Vis mig pigen som vist havde haft slået drengen Vis mig pigen som nemt har løftet drengen Vis mig pigen som jo har haft løftet drengen Vis mig drengen som blidt kysser pigen .pas PRESENT o_o.pas PAST lex.rel PRES_PF+ o_o.rel PAST o_o.pas: lex.rel PRES_PF o_o.rel: vrb.pas PRES_PF adj.pas: Adjectival Passive Lexical Passive Object Cleft Object-Object Relative Object-Subject Relative Psychological Passive Psychological Verb Simple Active Subject Cleft Subject-Object Relative Simple ("blive") Passive Subject-Subject Relative Verbal Passive PAST_PF+: PAST_PF: PAST: PRES_PF+: PRES_PF: PRES: FUTURE: Past Double Perfect Past Perfect Past Present Double Perfect Present Perfect Present Future adj.pas PAST_PF+ lex.rel PRES_PF+ o_s.rel PAST_PF+ o_o.

R.rel PRES_PF+ s_s.ver PRES_PF psy.clf PRESENT psy.act PAST_PF sim.ver PRESENT s_o.rel PAST_PF s_o.clf PAST_PF obj.rel FUTURE s_o.clf PRES_PF obj.act PRES_PF sim.act PAST_PF sim.act PRES_PF sim.act FUTURE sim.ver PRES_PF+ psy.clf PAST obj.act PAST_PF+ sim.rel PRESENT s_s.clf FUTURE obj.rel PRESENT sim.act PRES_PF+ sim. 2001 122 obj.act PAST_PF+ sim.act PAST sim.ver PAST_PF psy.rel PAST_PF+ s_o.K.ver PAST psy.rel PAST s_s.clf PRES_PF+ obj.ver FUTURE psy.act PAST sim.rel PRES_PF+ s_o.rel PAST s_o.rel PRES_PF s_o.pas PAST psy.rel PAST_PF s_s.pas FUTURE psy.ver PAST_PF+ psy.clf PAST_PF+ obj.pas PRES_PF psy.act PRESENT Det er katten som hunden nok vil bide Det er drengen som pigen vist slog Det var pigen som drengen vist havde skubbet Det var pigen som drengen vist havde haft skubbet Det er pigen som drengen nok har løftet Det er pigen som drengen nok har haft løftet Det er drengen som pigen tit kysser Hunden vil sikkert blive frygtet af katten Drengen elskedes sikkert af pigen Drengen var vist blevet beundret af pigen Pigen er nok blevet beundret af drengen Pigen elskes jo af drengen Katten vil sikkert frygte hunden Drengen elskede nok pigen Pigen havde nok elsket drengen Pigen havde nok haft elsket drengen Pigen har vist afskyet drengen Pigen har vist haft afskyet drengen Drengen afskyr vist pigen Katten som hunden sikkert vil bide er sort Drengen som pigen vredt slog er sur Pigen som drengen ofte havde skubbet var sur Pigen som drengen ofte havde haft skubbet var sur Pigen som drengen vist har kysset er glad Pigen som drengen vist har haft kysset er glad Drengen som pigen gerne kysser er glad Hunden som jo vil bide katten er sort Drengen som jo skubbede pigen er sur Pigen som nok havde slået drengen var sur Pigen som nok havde haft slået drengen var sur Pigen som tit har kysset drengen er glad Pigen som nok har haft kysset drengen er glad Drengen som tit kysser pigen er glad Drengen vil sikkert slå pigen Pigen vil ikke løfte drengen Drengen pegede vist på pigen Pigen kyssede ikke drengen Hunden havde nok snuset til katten Pigen havde ikke vinket til drengen Hunden havde nok haft snuset til katten Pigen havde ikke haft vinket til drengen Drengen har tit smilet til pigen Pigen har ikke skubbet drengen Drengen har tit haft smilet til pigen Pigen har ikke haft skubbet drengen Drengen kysser gerne pigen Katten snuser ikke til hunden .rel FUTURE s_s.act FUTURE sim.rel PRES_PF s_s.act PRES_PF+ sim. C.act PRESENT sim.rel PAST_PF+ s_s.pas PAST_PF psy.pas PRESENT psy.

clf PRESENT vrb.pas PAST sim.pas PAST_PF+ psy.pas PRES_PF+ sim.pas FUTURE sim.pas PRES_PF+ *Pigen havde nok haft været inspireret af drengen *Pigen har vist haft været begejstret for drengen *Drengen havde vist været blevet beundret af pigen *Pigen har nok været blevet beundret af drengen *Hunden havde nok været blevet bidt af katten *Drengen har sikkert været blevet smilet til af pigen *Drengen vil senere kysses af pigen14 *Drengen havde haft skubbedes jo af pigen *Hunden havde vist haft bides af katten *Drengen har løftes tit af pigen *Hunden har vist bides af katten Extra added to reach 100 sentences: sim.act PAST sim.pas PAST vrb.pas PAST_PF+ vrb.act PAST Drengene jagtede tit pigerne Pigerne drillede ofte drengene Drengene kyssede gerne pigerne Pigerne slog ofte drengene Hundene knurrede vredt af kattene Kattene væsede vredt af hundene 14 The sentence is grammatical if the meaning of “vil” (will) is volition instead of future. C.pas PAST_PF+ sim. .clf PAST sub.pas PRESENT vrb.pas PRESENT Drengen vil sikkert blive slået af pigen Pigen vil ikke blive peget på af drengen Drengen blev ofte peget på af pigen Pigen blev ikke kysset af drengen Hunden var nok blevet bidt af katten Pigen var ikke blevet smilet til af drengen Drengen er sikkert blevet smilet til af pigen Pigen er ikke blevet slået af drengen Drengen bliver tit grinet ad af pigen Katten bliver ikke bidt af hunden Det er hunden som sikkert vil bide katten Det er drengen som vredt skubbede pigen Det var pigen som vist havde skubbet drengen Det var pigen som vist havde haft skubbet drengen Det er pigen som nemt har løftet drengen Det er pigen som jo har haft løftet drengen Det er drengen som gerne kysser pigen Drengen kyssedes gerne af pigen Drengen skubbedes jo af pigen Drengen løftes tit af pigen Hunden bides vist af katten Excluded due to ungrammaticality: adj.act PAST sim.pas FUTURE vrb. R.pas PRES_PF+ vrb.pas PRESENT sim.pas PRES_PF vrb.clf PRES_PF sub.pas PRES_PF sim.clf PRES_PF+ sub.clf FUTURE sub.pas PAST_PF sim.act PAST sim.pas FUTURE sim.K.pas PRESENT sub.pas PAST_PF vrb.pas PAST vrb.clf PAST_PF sub.clf PAST_PF+ sub.pas PAST sim.act PAST sim.pas PRES_PF sim.pas PAST_PF sim. 2001 123 sim.pas PAST_PF+ adj.pas PRES_PF+ psy.act PAST sim.

1100.passive psych. 1079. 1004.predicate psych.obj.active sim.cleft obj.cleft obj. 1105.obj.obj.relative obj.relative obj.relative obj.passive adj.relative obj. 1101. 1042.passive lex.cleft obj.obj.cleft obj. 1059.sub. 1036. 1020. 1022.passive psych. 1131. 1011.obj. 1060.predicate psych. 1137. 1083. 1128.sub.predicate psych. 1075. 1126.cleft obj.relative obj. 1077. 1033.relative obj. 1018.obj. 1041.passive lex. 1035.cleft obj. 1019. 1086. 1088.sub.active sim.passive adj. C.predicate psych. 1085.relative obj. 1107.passive psych. 1141.cleft obj.passive psych.passive psych. 1063.sub.relative obj.passive psych. Token Picture >=correct [A] >A] >A] [A] [A] >A] [A] >A] [A] >A] >A] >A] [A] [A] >A] >A] [A] >A] [A] >A] [A] [A] [A] >A] >A] >A] >A] [A] [A] >A] [A] [A] >A] [A] >A] [A] >A] [A] >A] >A] [A] [A] [A] >A] [A] [A] [A] >A] [A] [A] >A] >A] >A] >A] [A] [A] >A] [A] [A] >A] >A] >A] >A] [A] [A] >A] [A] >A] >B] [B] [B] >B] >B] [B] >B] [B] >B] [B] [B] [B] >B] >B] [B] [B] >B] [B] >B] [B] >B] >B] >B] [B] [B] [B] [B] >B] >B] [B] >B] >B] [B] >B] [B] >B] [B] >B] [B] [B] >B] >B] >B] [B] >B] >B] >B] [B] >B] >B] [B] [B] [B] [B] >B] >B] [B] >B] >B] [B] [B] [B] [B] >B] >B] [B] >B] [B] Performace *=fail [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [→] [ ] [ ] [→] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [*] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] adj. 1008. R.passive psych.relative obj.passive psych.relative psych.passive obj.obj.passive adj.obj. 1064. 1043.sub.passive adj.obj. 1048.relative obj.predicate psych. 1016. 1032. 1040. 1140.K. 1069.cleft obj. 1084.predicate psych. 1027. 1013.passive adj. 1144.sub. 1047. 1135.passive adj. 1120. 1118.predicate psych.sub. 1007. 1001.sub.predicate sim. 1113. 2001 124 14 Appendix D: TJ's Comprehension Test Data Type Order +1000 1015.cleft obj.relative obj.relative obj. 1024.predicate psych.sub. 1133.active sim.relative obj.passive adj. 1132. 1092.relative obj.relative obj.relative obj.passive adj.relative obj.relative obj. 1052. 1094. 1102. 1045.predicate psych.passive adj.obj.relative obj.passive psych. 1103.sub. 1055. 1097. 1017.active Hunden er irriteret på katten Drengen er interesseret i pigen Hunden er forbavset over katten Pigen er begejstret for drengen Pigen er inspireret af drengen Drengen er begejstret for pigen Katten er forbavset over hunden Katten er irriteret på hunden Pigen er interesseret i drengen Drengen er inspireret af pigen Pigen væmmes ved drengen Drengen synes om pigen Drengen væmmes ved pigen Pigen synes om drengen Det er drengen som pigen skubber Det er pigen som drengen slår Det er drengen som pigen løfter Det er pigen som drengen løfter Det er hunden som katten bider Det er pigen som drengen kysser Det er drengen som pigen slår Det er drengen som pigen kysser Det er pigen som drengen skubber Det er katten som hunden bider Vis mig katten som hunden bider Vis mig pigen som drengen slår Vis mig pigen som drengen løfter Vis mig drengen som pigen løfter Vis mig drengen som pigen slår Vis mig pigen som drengen kysser Vis mig drengen som pigen kysser Vis mig pigen som drengen skubber Vis mig drengen som pigen skubber Vis mig hunden som katten bider Vis mig drengen som slår pigen Vis mig pigen som kysser drengen Vis mig hunden som bider katten Vis mig drengen som skubber pigen Vis mig drengen som kysser pigen Vis mig pigen som skubber drengen Vis mig pigen som løfter drengen Vis mig katten som bider hunden Vis mig pigen som slår drengen Vis mig drengen som løfter pigen Hunden frygtes af katten Pigen afskys af drengen Drengen elskes af pigen Katten frygtes af hunden Pigen hades af drengen Drengen beundres af pigen Pigen elskes af drengen Drengen hades af pigen Drengen afskys af pigen Pigen beundres af drengen Drengen hader pigen Pigen elsker drengen Hunden frygter katten Katten frygter hunden Pigen beundrer drengen Drengen elsker pigen Drengen beundrer pigen Pigen afskyr drengen Pigen hader drengen Drengen afskyr pigen Pigen løfter drengen Pigen griner ad drengen Drengen griner ad pigen Drengen vinker til pigen . 1031.passive lex. 1056.cleft obj. 1005.passive lex.passive psych. 1139. 1026.

1110. 1050. 1136.active sim.passive sim.active sim.cleft sub. 1096.relative sub.obj. 1046.passive 1044.passive verbal. 1074. 1061.passive sim. C.sub. 1067.passive verbal. 1093.sub.cleft sub. 1049. 1082.sub. 1066.relative sub. 1054. 1076.passive sim. 1037.sub.relative sub. 1034. 1058. 1078. 1104. 1091.relative sub.passive sim.sub. 1065. R. 1138.active sim. 1123. 1057. 1090.obj.relative sub. 1038.active sim.passive verbal.cleft sub.passive sim.relative sub. 1073. 1129.K.passive sim.active sim.obj. 1099.passive sim.passive sim.relative verbal.passive sim.obj.active sim. 1023. 1108.passive verbal. 1071. 1116.relative sub. 1127. 1080.passive verbal. 1039.relative sub.passive sim.active sim. 1095. 1114. 1142.passive verbal.relative sub. Katten bider hunden Drengen smiler til pigen Katten snuser til hunden Pigen slår drengen Hunden snuser til katten Pigen peger på drengen Pigen smiler til drengen Pigen kysser drengen Hunden bider katten Drengen løfter pigen Drengen peger på pigen Pigen skubber drengen Pigen vinker til drengen Drengen slår pigen Drengen skubber pigen Drengen kysser pigen Hunden bliver bidt af katten Drengen bliver løftet af pigen Drengen bliver skubbet af pigen Drengen blivet smilet til af pigen Pigen bliver løftet af drengen Pigen bliver kysset af drengen Pigen bliver smilet til af drengen Pigen bliver vinket til af drengen Drengen bliver grinet ad af pigen Katten bliver snuset til af hunden Pigen bliver skubbet af drengen Pigen bliver slået af drengen Katten bliver bidt af hunden Hunden bliver snuset til af katten Pigen bliver grinet ad af drengen Pigen bliver peget på af drengen Drengen bliver vinket til af pigen Drengen bliver peget på af pigen Drengen bliver slået af pigen Drengen bliver kysset af pigen Det er katten som bider hunden Det er drengen som løfter pigen Det er drengen som skubber pigen Det er pigen som kysser drengen Det er drengen som slår pigen Det er hunden som bider katten Det er pigen som skubber drengen Det er pigen som løfter drengen Det er drengen som kysser pigen Det er pigen som slår drengen Pigen som drengen løfter er glad Pigen som drengen skubber er sur Drengen som pigen løfter er glad Drengen som pigen kysser er glad Pigen som drengen kysser er glad Drengen som pigen slår er sur Hunden som katten bider er sort Pigen som drengen slår er sur Katten som hunden bider er sort Drengen som pigen skubber er sur Hunden som bider katten er sort Pigen som løfter drengen er glad Drengen som løfter pigen er glad Drengen som slår pigen er sur Drengen som skubber pigen er sur Pigen som slår drengen er sur Pigen som kysser drengen er glad Pigen som skubber drengen er sur Drengen som kysser pigen er glad Katten som bider hunden er sort Drengen kysses af pigen Katten bides af hunden Pigen slås af drengen Drengen skubbes af pigen Drengen slås af pigen Hunden bides af katten Pigen løftes af drengen Pigen kysses af drengen Pigen skubbes af drengen Drengen løftes af pigen [A] >A] [A] [A] >A] >A] [A] [A] >A] >A] [A] >A] [A] >A] [A] >A] [A] [A] >A] [A] >A] >A] >A] >A] >A] >A] [A] >A] >A] [A] [A] [A] [A] >A] [A] [A] [A] >A] [A] [A] >A] >A] >A] [A] >A] [A] >A] [A] [A] [A] >A] [A] [A] >A] >A] >A] >A] [A] >A] >A] [A] [A] [A] >A] >A] [A] [A] >A] >A] >A] [A] [A] >A] >A] [A] [A] >B] [B] >B] >B] [B] [B] >B] >B] [B] [B] >B] [B] >B] [B] >B] [B] >B] >B] [B] >B] [B] [B] [B] [B] [B] [B] >B] [B] [B] >B] >B] >B] >B] [B] >B] >B] >B] [B] >B] >B] [B] [B] [B] >B] [B] >B] [B] >B] >B] >B] [B] >B] >B] [B] [B] [B] [B] >B] [B] [B] >B] >B] >B] [B] [B] >B] >B] [B] [B] [B] >B] >B] [B] [B] >B] >B] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [*] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [*] [ ] [*] [ ] [ ] [ ] [ ] [*] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] .active sim.passive sim. 1106.sub.cleft sub. 1089.passive sim. 1068. 1051. 1143.obj.active sim.cleft sub.passive verbal.passive sub.relative sub.active sim.active sim.relative sub. 1121.obj.passive sim.cleft sub.passive sim. 1125.passive sim.sub.active sim.obj.cleft sub.active sim. 1134. 1028.relative sub. 1119.passive sim.cleft sub.passive sim. 1081. 1025. 1111. 1070. 1124.cleft sub. 1130. 1062.active sim.relative sub.relative sub.relative sub.passive verbal. 1021. 1098.obj.sub. 1115. 1003.cleft sub. 1002. 1087.sub.relative sub.relative sub. 1053. 1012.obj.relative sub. 1072.sub. 1014. 1010. 1029. 1117.active sim.active sim. 1006. 1009.relative sub.passive sim. 1109. 1122. 1112. 2001 125 sim.passive sim.obj. 1030.passive verbal.

1: Past Double Perfect 2: Past Perfect 3: Past 4: Present Double Perfect 5: Present Perfect 6: Present 7: (Present) Future Produced Tense Governed P Target Type 00 01 02 03 04 05 06 07 P P P P SA SA LP VP SA AP SA PV 3 3 7 6 3 6 3 7 SA SA LP VP SA AP SA PV 3 3 7 6 3 6 3 7 08 09 10 PV OC OO 6 6 5 PV OC OO 6 6 5 Drengene jagtede tit pigerne "Drengene jagtede tit pigerne" Kattene hvæsede vredt af hundene "Kattene hvæsede vredt af hundene" Drengen vil nok synes om pigen "Drengen vil nok synes om pigen" Hunden bides vist af katten "Hunden bides vist af katten" Pigerne slog ofte drengene "Pigerne slog ofte drengene" Drengen er meget begejstret for pigen " Drengene er meget begejstret for pigerne" Hundene knurrede vredt af kattene "Hundene knurrede vredt af kattene" Katten vil sikkert frygte hunden "Katten vil øøh … øh katten vil øh måske frygte hunden" Drengen afskyr vist pigen "Drengen afskyr vist pigen" Det er drengen som pigen tit kysser "Det er drengen som pigen tit kysser" Vis mig pigen som drengen sikkert har kysset " Vis mig pigen som drengen har kysset" Prep. C. OK OK OK Main Verb Drop Produced Type Aux Verb Drop Target Tense . infl. R. 2001 126 15 Appendix E: TJ's Production Test Data Legend: AP: LP: OC: OO: OS: PP: PV: SA: SC: SO: SP: SS: VP: Adjectival Passive Lexical Passive Object Cleft Object-Object Relative Object-Subject Relative Psychological Passive Psychological Verb Simple Active Subject Cleft Subject-Object Relative Simple ("blive") Passive Subject-Subject Relative Verbal Passive PAST infl.K. PRES. Drop Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) Passive morph.

C. Main Verb Drop Produced Type Aux Verb Drop Target Tense . 2001 127 Produced Tense Governed P Target Type 11 12 P AP OC 5 5 AP OC 6 3 13 14 P SA OS 1 6 SA ? 5 ? 15 SO 4 ? ? 16 17 18 P N SP SP SS 6 5 1 SP SP SO 6 5 5 19 PP 5 PP 5 20 N SA 5 SP 5 21 22 P SA OS 4 3 SA OS 5 3 23 24 OS SC 2 6 OS SC 3 6 Pigen har ofte været begejstret for drengen "Pigen er ofte begejstret for drengene" Det er pigen som drengen nok har løftet "Det er pigen øh øøh drengen øh øh sikkert løftede" Hunden havde nok haft snuset til katten "Hunden har ofte øh snuset til hunden" Vis mig drengen som blidt kysser pigen "Vis mig drengen som øøh … … øøh øh ja det er jeg kan ikke huske det drengen som ofte … … ja det kan jeg ikke huske" (Vis mig drengen som ofte) Pigen som drengen vist har haft kysset er glad "Pigen som ofte har … øh døhdøhdøh drengen" (Pigen som ofte har _ drengen) Drengen bliver tit grinet ad af pigen "Drengen bliver tit grinet ad pigen" Pigen er ikke blevet slået af drengen "Pigen er ikke blevet slået af drengen" Pigen som nok havde haft slået drengen var sur "Pigen som øøh måske har øh blevet slået af drengen øh og det kan jeg ikke huske det sidste" (Pigen som måske er blevet slået af drengen [var sur]) Pigen er nok blevet beundret af drengen "Pigen er øh er øøh ofte blevet beundret af drengen" Pigen har ikke skubbet drengen "Pigen har ikke blevet skubbet af drengen" (Pigen er ikke blevet skubbet af drengen) Drengen har tit haft smilet til pigen "Drengen øh har tit smilet til pigen" Vis mig drengen som vredt skubbede pigen "Vis mig drengen som vredt skubbede med pigen" Vis mig pigen som vredt havde slået drengen "Vis mig pigen som vredt slåede drengen" Det er drengen som gerne kysser pigen "Det er drengen som gerne kysser pigen" -1 -1 -1 -1 -1 -1 -1 -1 +G -1 Prep.K. Drop -U Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) Passive morph. R.

K. 2001 128 Produced Tense Governed P Target Type 25 26 27 28 29 30 31 P SA PV SS SA OC PV 5 1 7 7 3 5 2 SA PV SS SA SC PV SP 5 4 3 7 6 5 5 P AP 32 SC 3 SC 3 33 SO 2 AP 3 34 35 P OO LP 7 1 OO ? 3 ? 36 37 P SP SP 3 7 SP SP 3 7 38 OO 1 OS 3 39 P AP 3 AP 3 Drengen har tit smilet til pigen "Drengen har tit smilet til pigen" Pigen havde nok haft elsket drengen "Pigen har nok øh haft elsket drengen" Hunden som jo vil bide katten er sort "Hunden som ville bide katten var sort" Drengen vil sikkert slå pigen "Drengen vil sikkert slå pigen" Det er drengen som pigen vist slog "Det er drengen som vist slår pigen" Pigen har vist afskyet drengen "Pigen øh har vist afskyet drengen" Pigen havde nok været inspireret af drengen "Pigen øøh har ofte øøh … har ofte øøh … kigget på pigen" Det er drengen som vredt skubbede pigen "Drengen som vredt skubbede pigen" ((Det er) drengen som vredt skubbede pigen) Pigen som drengen ofte havde skubbet var sur "Pigen var døøh … ofte … skrækket af … drengen" (Pigen var ofte forskrækket over drengen) Vis mig katten som hunden sikkert vil bide "Vis mig katten som øh … bed øh hunden" Drengen havde nok haft syntes om pigen "Drengen øh … har ofte øh … døhdøh kvinden" (Drengen har ofte bla-bla kvinden) Drengen blev ofte peget på af pigen "Drengen blev ofte peget på af pigen" Drengen vil sikkert blive slået af pigen "Drengen vil øøh … øh øøh måske blive slået af pigen" (Drengen vil måske blive slået af pigen) Vis mig pigen som drengen vredt havde haft slået "Vis mig pigen som øøh vredt slog slåede drengen" Hunden var ret forbavset over katten "Hunden var ret forbavset over katten" -1 -1 -1 -1 -2 Prep. Main Verb Drop Produced Type Aux Verb Drop Target Tense . C. R. Drop +G +G Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) Passive morph.

-1 -1 -1 Main Verb Drop Produced Type Aux Verb Drop Target Tense . C. 2001 129 Produced Tense Governed P Target Type 40 PV 2 SA 3 41 42 43 P N SA VP LP 4 6 4 SA VP ? 5 6 ? 44 P LP 3 ? 3 45 SS 4 SS 3 46 47 48 P N N SA SA SC 7 6 2 SA SA SC 7 6 3 49 OO 3 OO 3 50 51 52 P N PP SP LP 6 3 2 PP SP ? 3 3 ? Pigen havde nok elsket drengen "Pigen var ofte øh nej pigen øøh elskede ofte drengen" (Pigen elskede ofte drengen) Pigen har ikke haft skubbet drengen "Pigen har ikke skubbet drengen" Drengen løftes tit af pigen "Drengen løftes tit af pigen" Pigen har nok haft syntes om drengen "Pigen har ofte øøh … øøh … ikke tænke men pigen er ofte … … … … ofte øh ja det ka du ka j' ikke huske det sidste" (Pigen har ofte) Pigen væmmedes nok ved drengen "Pigen vænnede sig ofte øh øh af drengen" (pigen vænnede sig ofte [ACTIVE REFLEXIVE] af drengen [PASSIVE]) Pigen som nok har haft kysset drengen er glad "Pigen som øh ofte øh kyssede drengen er døhdøh" (Pigen som ofte kyssede drengen er bla-bla) Pigen vil ikke løfte drengen "Pigen vil ikke løfte drengen" Katten snuser ikke til hunden "Katten snuser ikke til hunden" Det var pigen som vist havde skubbet drengen "Det var pigen som øøh … skubbede til drengen" Vis mig drengen som pigen vredt skubbede "Vis mig pigen som drengen skubbede … vredt skubbede" (Vis mig pigen som drengen vredt skubbede) Pigen elskes jo af drengen "Pigen elskede jo af drengen" Pigen blev ikke kysset af drengen "Pigen blev ikke kysset af drengen" Pigen havde nok syntes om drengen "Pigen har ofte øh … af drengen . Pigen … øh nej det kan jeg ikke" (Pigen har ofte _ af drengen) -1 -1 OK -1 -1 -1 Prep.K. R. Drop +G Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) -2 -1 -1 Passive morph.

R. Drop -G -G +G Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) -1 -1 -1 -1 Passive morph. C. 2001 130 Produced Tense Governed P Target Type 53 SC 1 SC 3 54 55 P SO LP 7 5 OS ? 3 ? 56 57 58 59 SA SA SS PV 6 3 5 4 SA SA SS PV 6 6 3 5 60 61 62 63 64 65 P AP PV 7 3 2 3 3 2 AP PV SP PP SA OS 7 3 2 3 3 3 P N SP PP P SA OO 66 67 SC OO 7 4 SC OS 7 3 Det var pigen som vist havde haft skubbet drengen "Det var pigen som ofte skubbede drengen" Katten som hunden sikkert vil bide er sort "Katten ville bide hunden som var sort" Pigen har nok væmmedes ved drengen "Pigen var of' nej … væmmede drengen. Pigen øøh var ofte nej.K. Pigen ofte øøh væmmede drengen" (Pigen ofte væmmede drengen) Drengen kysser gerne pigen "Drengen kysser gerne pigen" Pigerne drillede ofte drengene "Pigerne drillede ofte drengene" Pigen som tit har kysset drengen er glad "Pigen som tit kyssede drengen var glad" Pigen har vist haft afskyet drengen "Pigen har vist øøh har vist øøh afskyet nej var vist afskyet drengen" (Pigen har vist afskyet drengen) Hunden bliver temmelig irriteret på katten "Hunden bliver temmelig irriteret på katten" Drengen elskede nok pigen "Drengen elskede nok pigen" Pigen var ikke blevet smilet til af drengen "Pigen var ikke blevet smilet af drengen" Drengen elskedes sikkert af pigen "Drengen elskedes vist af pigen" Drengen pegede vist på pigen "Drengen pegede vist på pigen" Vis mig pigen som drengen ofte havde slået "Vis mig pigen som ofte slåede drengene drengen" Det er hunden som sikkert vil bide katten "Det er hunden som sikkert vil bide katten" Vis mig pigen som drengen sikkert har haft kysset "Vis mig pigen som vist øøh kiggede efter drengen" -2 -1 Prep. -1 OK Main Verb Drop Produced Type Aux Verb Drop Target Tense .

2001 131 Produced Tense Governed P Target Type 68 SO 3 Srel 3 69 SS 2 SS 2 70 P SP 5 SP 3 71 SO 5 ? ? 72 OC 1 SC 2 73 OC 4 OC 5 74 75 76 77 78 P LP OS 6 4 7 6 2 LP OS SP SP SA 6 5 7 6 2 P N N SP SP SA P N 79 80 SA SS 3 6 SA SS 3 3 Drengen som pigen vredt slog er sur "Drengen … øh som vist … øh øh kiggede efter pigen" (Drengen som vist kiggede efter pigen) [NP [Sub-rel] Pigen som nok havde slået drengen var sur "Pigen som øh ofte havde slået drengen var sur" Drengen er sikkert blevet smilet til af pigen "Drengen øh blev smilet af pigen nej nej det kan jeg ikke huske" (Drengen blev smilet af pigen) Pigen som drengen vist har kysset er glad "Pigen som …mem mem … ti pige drengen som ret glad. R. Drop +G -G -G Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) -1 Passive morph. Jeg kan ikke huske det midterste" Det var pigen som drengen vist havde haft skubbet "Det var pigen som vist havde skuffet skubbet" (Det var pigen som vist havde skubbet) [INTRANS] Det er pigen som drengen nok har haft løftet "Det er ret pigen som drengen of' øøøh … har skubbet" (Det er pigen som drengen ofte har skubbet) Drengen synes vist om pigen "Drengen synes vist om pigen" Vis mig pigen som jo har haft løftet drengen "Øh vis mig pigen som har løftet drengen" Pigen vil ikke blive peget på af drengen "Pigen vil ikke blive peget af drengen" Katten bliver ikke bidt af hunden "Katten bliver ikke bidt af hunden" Pigen havde ikke vinket til drengen " Pigen havde ikke vinket a' drengen" (Pigen havde ikke vinket ad drengen) Drengene kyssede gerne pigerne "Drengene kyssede gerne pigerne" Drengen som tit kysser pigen er glad "Drengen som tit kyssede pigen er glad" -1 -1 -1 -1 -1 Prep. C.K. OK Main Verb Drop Produced Type Aux Verb Drop Target Tense .

-1 -1 Main Verb Drop Produced Type Aux Verb Drop Target Tense . Drop Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) 83 P SA 2 SA 2 84 85 86 87 88 89 N SA VP SS SC SO SO 3 3 3 5 6 1 SA VP SS SC SO SS 3 3 3 3 3 3 90 OS 7 OS 3 91 PP 2 PP 2 92 93 OC OO 7 6 OC OO 7 ? 94 PP 7 PV 3 Drengen kyssedes gerne af pigen "Drengen kyssede gerne pigen" Hunden var nok blevet bidt af katten "Hunden var nok bidt af katten" (Hunden var nok [blevet] bidt af katten) Hunden havde nok snuset til katten "Hunden havde nok snuset at katten" (Hunden havde nok snuset til katten) Pigen kyssede ikke drengen "Pigen kyssede ikke drengen" Drengen skubbedes jo af pigen "Drengen skubbede jo af pigen" Drengen som jo skubbede pigen er sur "Drengen som skubbede pigen er sur" Det er pigen som nemt har løftet drengen "Det er pigen som løftede drengen" Drengen som pigen gerne kysser er glad "Drengen som pigen kyssede er gerne glad" Pigen som drengen ofte havde haft skubbet var sur "Pigen som ofte øøh skubbede øh drengen var sur" (Pigen som ofte skubbede drengen var sur) Vis mig hunden som først vil bide katten "Vis mig hunden som øøh ofte øøh bed hunden" Drengen var vist blevet beundret af pigen "Drengen var vist blevet beundret af af hunden nej jo … eller pigen. C.K. Det kan jeg ikke huske" (Drengen var vist blevet beundret af pigen) Det er katten som hunden nok vil bide "Det er katten som hunden nok vil bide" Vis mig drengen som pigen tit kysser "Vis mig drengen som pigen ofte øøh ja det kan jeg ikke huske det sidste ord" "Vis mig drengen som pigen ofte [VERB]) Hunden vil sikkert blive frygtet af katten "Hunden vil sikkert blive øh … frygtet katten. 2001 132 Produced Tense Governed P Target Type 81 82 VP SP 3 2 SA SP 3 2 Prep. Øh hunden ville sikkert øøh frygte katten" (Hunden ville sikkert frygte katten) -1 -1 -2 -1 -1 Passive morph. R.

R. Drop Negation Order Target sentences "Produced sentences" (Interpretation by analyzer) Passive morph. Main Verb Drop Produced Type Aux Verb Drop Target Tense .K. C. 2001 133 Produced Tense Governed P Target Type 95 OS 1 OS 5 96 97 98 P N SA OC SC 1 2 4 SA OC SC 5 3 5 99 OS 5 OS 3 Vis mig pigen som vist havde haft slået drengen "Vis mig pigen som vist slåede drengen nej vist har slået drengen" (Vis mig pigen som vist har slået drengen) Pigen havde ikke haft vinket til drengen "Pigen har ikke vinket til drengen" Det var pigen som drengen vist havde skubbet "Det var pigen som drengen vist skubbede" Det er pigen som jo har haft løftet drengen "Det var det er pigen som har løftet drengen vist har løftet drengen" (Det er pigen som vist har løftet drengen) Vis mig pigen som nemt har løftet drengen "Vis mig pigen som øh nemt løftede drengen" -1 -1 -1 -1 -1 Prep.

Sign up to vote on this title
UsefulNot useful