Coreference and Language Acquisition
Sergio Baauw & Denis Delfitto
(Utrecht University)
1. Introduction
Young children between 3 and 7 years old are reported to often allow pronouns to corefer with
local c-commanding antecedents (1a), although they do not seem to have much trouble with
rejecting non-local antecedents for reflexive pronouns (1b) (Chien & Wexler 1990, among
others).
(1)
a.
b.
The boy touched him.
The boy touched himself.
(around 50% adult-like)
(almost 100% adult-like)
Although this phenomenon, which has come to be known as the Delay of Principle B Effect
(DPBE), has been attested for a variety of languages, including Dutch (Philip & Coopmans 1996),
English (Chien & Wexler 1990), Icelandic (Sigurjónsdóttir 1992) and Russian (Avrutin 1994), it
is not universal. The DPBE has not been found in the acquisition of Romance languages such as
Italian (McKee 1992), French (Hamann & Philip 1997) and Spanish (Padilla 1990; Baauw,
Escobar & Philip 1997), using constructions similar to (1a), as illustrated for Spanish by (2)1,2
(2)
La niña la señala.
(90% adult-like: Baauw, Escobar & Philip 1997)
The absence of a DPBE in Romance has been related to the fact that Romance weak pronouns
like Spanish la in (2) are clitics, unlike English pronouns such as him in (1a). Therefore we will
call this absence of a DPBE the Clitic Exemption Effect (CEE).
Interestingly, if the local subject is a quantifier, even English and Dutch speaking children
perform highly adult-like (Chien & Wexler 1990: Philip & Coopmans 1996):
(3)
Every boy touched him.
(85% adult-like)
Since quantifiers only allow bound-variable representations, children’s highly adult-like
performance on constructions like (3) suggests that they do not have problems with binding, but
with coreference (Grodzinsky & Reinhart 1993). Children know and obey the binding principles,
but allow coreference between co-arguments much more often than adults. In the adult language
coreference between co-arguments is limited to specific discourse contexts, like the ones in (5):
(5)
1
2
a.
Do you know what Mary and John have in common? Mary admires him and John
The DPBE is also absent in a Germanic language like Norwegian (Hestvik & Philip 1997).
Spanish children (like French and Norwegian children) do show a DPBE in Verbal Small Clauses (see Baauw,
Escobar & Philip 1997 for an account).
(i)
La niña la ve bailar.
(63% adult-like)
‘The girl sees her dance’
b.
c.
d.
admires him too.
Zelda’s husband is him (pointing context).
A. Is this speaker Zelda? B. How can you doubt it? She praises her to the sky. No
competing candidate would do that.
Everybody hates Lucifer. Only HE HIMSELF pities him / Even HE HIMSELF
hates him
In this paper we will try to account for the following two questions: (i) why does the
DPBE show up in the acquisition of some languages (i.e. why do children have problems with
coreference)?, and (ii) why do Romance children not show a DPBE (i.e. why do they not allow
unadult-like coreference between co-arguments)? In order to answer these two questions we will
first have to be more specific about what we understand under binding and coreference.
2. Binding and Coreference
2.1. Binding and the creation of reflexive predicates
With respect to binding we will adopt the following assumptions:
(6)
a.
b.
A NP α semantically binds pronoun β iff β and the trace of α are bound by the
same λ operator (Heim & Kratzer 1998).
When binding holds between two arguments of a predicate, a reflexive predicate
will be created, as a matter of definition.
If configuration (6a) is created by QR of the subject she in (7a), the result is an object of the form
(7b).
(7)
a.
b.
She praises her to the sky.
[λx: x praises x to the sky] (she)
However, the interpretation of (7a) as a reflexive predicate (7b) leads to ungrammaticality. This is
due to a constraint on the creation of reflexive predicates, namely Principle B of Reinhart &
Reuland’s (1993) alternative binding theory:
(8)
Reinhart & Reuland’s Binding Theory (R&R 1993)
Principle B: A reflexively interpreted semantic predicate must be reflexive-marked.
We could reinterpret Principle B as an interface filter that prohibits arity reduction in the syntax
(and the pragmatics, as we will see in section 2.2.) (inspired by Reuland 1998). This boils down
to saying that a relation cannot be reduced to a property in the course of the interpretation process,
unless it is already marked as a property in the lexicon: * λxλy (xRy) → λx (xRx).3 Therefore
3
Verbs like English wash and Dutch wassen in (i) are examples of verbs that have been reflexive marked in the
lexicon (= marked as a property). In Dutch, the object position of such verbs is occupied by the SE anaphor zich
(Everaert 1986). In English the object is left unexpressed.
(9) is excluded under the interpretation “[λx (xLx)] (John)”, since the relation “love” (λxλy
(xLy)) would be converted into a property.
(9)
John loves him.
However, relations can get a reflexive-like interpretation with the help of SELF-anaphors and
body-part reflexives. SELF and body-part reflexives are used to prevent a full reflexive predicate
from being created in syntax (cf. Reuland 1998).
(10)
John loves himself.
(10) is grammatical, because the predicate has the form “x R f(x)”, with the referent of f(x) quite
close to, but not identified with the referent of x (Postma 1997).
2.2. Coreference and the creation of reflexive predicates
In the case of coreference between an object pronoun and a local subject, the object pronoun is
interpreted as a free variable. Free variables pick up their antecedent from the discourse. We
assume, following Heim (1998), that pronouns that are interpreted as free variables establish a
relation with a referent by means of a hidden definite description, which Heim calls “guise”. If the
referent identified by the guise of the object pronoun and the referent identified by the guise of
the subject turn out to be the same, we say that the subject and the object corefer.
Coreference between co-arguments, however, is excluded, except in the contexts
mentioned in (5). In Grodzinky & Reinhart (1993) it was argued that the ungrammaticallity of
coreference in cases like (9) is not due to any principle of grammar, like Principle B, but to Rule
I, a constraint at the syntax/pragmatics interface which rules out coreference as a less economical
LF representation than a bound-variable representation if coreference and binding do not yield a
different interpretation.4 We will not follow this approach, since it is problematic for several
reasons, which we will not discuss in this paper (see Baauw & Delfitto, in progress). Instead, we
will argue that the ban on coreference in cases other than (5) can be derived from Principle B of
Reinhart & Reuland’s revised binding theory.
Concretely, we want to propose that coreference between co-arguments also leads to the
creation of a reflexive predicate, just like binding, to the effect that Principle B applies to these
cases of coreference as well (cross-modular Principle B). However, the creation of a reflexive
predicate by means of coreference can be achieved in two ways: (i) the object pronoun and the
subject refer to the same individual by using one single guise, or (ii) they do so by using two
(i)
a.
Jan wast zich.
b.
John is washing. (= John is washing himself)
This process of reflexive marking is lexically restricted. A verb like zien ‘see’, for instance, does not allow reflexive
marking in the lexicon:
(ii)
*Jan ziet zich.
John sees SE
4
Coreference does yield a different interpretation than binding in the cases under (5): the second sentence of (5a), for
instance, is not about self-admiration (bound variable reading of him), but about admiring John (coreference of John
and him).
different guises. The latter possibility follows from the fact that different guises may correspond
to a single referent (just like morning star and evening star refer to the same celestial body). We
propose the following:
(11)
a.
b.
If two guises are involved, the predicate is extensionally reflexive (12a).
If one guise is involved, the predicate is intensionally reflexive (12b):
(12)
a.
She praises her to the sky
|
|
F
G
|
|
z
z
(extensionally reflexive)
b.
She praises her to the sky
|
|
F
F
|
|
z
z
(intensionally reflexive)
Suppose that F represents the guise “the woman with the blue eyes” and G represents the
guise”the w oman with blond hair”. Since in (12b) coreference takes place by means of one single,
this kind of coreference holds in every possible world. This is not the case in (12a), where
coreference is realized by two different guises. In some possible world “the woman with blue
eyes” (F) and “the woman with blond hair” (G) refer to the same individual (z), but this need not
be the case in some other world, where F and G may refer to two different female individuals. We
propose that the cases where coreference is possible in the adult language (5), are all cases where
two guises are involved.
However, if coreference between co-arguments results in the creation of a reflexive
predicate, this will lead to a violation of Principle B, although this violation will be different for
cases of coreference involving one guise, than when coreference is realized with two guises. The
two kinds of Principle B violations are given in (13):
(13)
a.
b.
If a predicate is intensionally reflexive (12b), this will lead to a strong principle B
violation.
If a predicate is extensionally reflexive (12a), this will lead to a milder Principle B
violation.
The fact that (12a) still represents a Principle B violation (although a ‘weaker’ kind of violation)
explains why many of the cases of coreference listed under (5) sound rather ‘marked’. In fact,
some speakers may even find them unacceptable.5
5
Heim (1998) notes that some of the examples of (5), like (5a,d) are more marked than others, like (5b,c). We
suggest that the extend to which a case of coreference is judged as ‘marked’, depends on the ease by which the two
different guises involved can be kept distinct. For a more elaborated discussion, see Baauw & Delfitto, in progress.
3. Language Acquisition.
As briefly discussed in chapter 1, Dutch and English speaking children exhibit a DPBE in
constructions such as (14), while Spanish children almost always reject coreference between the
pronoun and the local subject.
(14)
a.
b.
John touched him. (roughly 50% adult like)
El niño le dibujó.
(90% adult-like)
‘The boy drew him.’
In this section we will show how the DPBE can be accounted for and why the Delay does not
show up in the acquisition of the Romance languages.
3.1. The DPBE
English children’s highly adult -like performance on sentences with quantified antecedents (see
(3)) clearly shows that children do not have problems with binding. Instead, they appear to allow
coreference of co-arguments in contexts where adults disallow this.
Recall from section 2. that coreference in the adult language involves the creation of a
mildly reflexive predicate, that is, a predicate whose arguments instantiate different guises that
have the same referent (12a). However, as we have seen, coreference is pragmatically constrained
in the adult language. It is only possible in cases where the discourse provides an extra guise for
the pronoun, as is the case in the contexts listed under (5). The question is now: why do children
allow coreference in cases where the context does not provide an extra guise for the pronoun? In
this section we want to put forward an account of the DPBE that is inspired by Avrutin (1994).
According to this approach, the DPBE is the result of problems children have with the application
of the pragmatic constraints on the introduction of new guises. More concretely, we will argue,
following Avrutin (1994), that children allow pronouns to introduce new guises because they fail
to make correct inferences about the application of pragmatic constraints on “inferential bridging”
(Clark 1977).
Avrutin argues that normally definite expressions, including pronouns, do not introduce
new guises. Instead, they refer to guises that have already been introduced in the discourse (“old
information”). This explains why (15b) is felicitous if it has been preceded by (15a), a
construction with an indefinite NP, but not when it is uttered out of the blue.
(15)
a.
b.
I have a new colleague.
He loves to play soccer.
However, it is not completely impossible for pronouns to introduce new guises. A pronoun can be
used deictically. When a pronoun is used deictically the guise that is introduced is connected
(“bridged”) to a visual situation, as an instance of “inferential bridging”. 6 However, this kind of
6
A more standard case of inferential bridging takes place when we relate bride in the second sentence of (i) to
wedding in first sentence of (i):
(i)
I attended a wedding recently. The bride was wearing a white dress.
bridging is only possible if the speaker is pointing at the object that the pronoun refers to. If the
speaker fails to point, the hearer will not be able to build a corresponding bridge. This means that
the constraints on bridging crucially involve the ability to make inferences about other speakers
discourse representations. We will argue, following Avrutin, that this is what children often fail to
do correctly: they fail to make correct inferences, due to a lack of computational resources, and
may conclude that the constraints on bridging are satisfied.7 Stated differently: children often
allow deictic bridging without pointing, due to difficulties they have with making inferences
about other peoples discourse representations.8 Suppose now that the “deictic guise” introduced
by the pronoun and the guise associated with the local subject have the same referent. In that case
a ‘mildly’ reflexive predicate is created:
(16)
John touched him.
|
|
A
B
|
|
z
z
(guise A = the male individual named John;
guise B = the male individual I am looking at (= the “de ictic guise”))
However, if children are able to create mildly reflexive predicates by having the pronoun
introduce a new guise, why do they not rule in a coreferent reading of (14a) 100% of the time?
Recall from the previous section that the creation of a mildly reflexive predicate leads to a
‘mild’ principle B violation. This explains why speakers of English consider many of examples
listed under (5) not to be fully grammatical. When a sentence is not fully grammatical, but not
radically ungrammatical either, some speakers will accept it, others will reject it. To children
basically the same applies: when they create a mildly reflexive predicate, they will have to decide
whether a mild Principle B violation is enough to make a sentence like (14a) false under the
coreferential interpretation. We suggest that they may decide either way: hence the 50% adult-like
performance.
In this case it is cultural knowledge that allows speakers to build the ‘bridge’ between two lexical items (or the guises
instantiated by the two lexical items): speakers share the knowledge that weddings involve the presence of a bride
(Heim 1982).
7
Note that not only children show a DPBE. This effect has also been observed in Broca aphasics, who are also
known to have limited computational resources.
8
There is some independent evidence in favor of this claim. Karmiloff-Smith (1979; 1980; 1986) shows that 4 and 5
year old children acquiring French or English often allow the deictic use of definite NPs and pronouns in cases where
adults disallow this (see also some of the experiments of Maratsos 1976 for English speaking children, and Bamberg
1987 for German speaking children). A typical example of this use is from an experiment conducted by KarmiloffSmith (1980) in which children had to tell what happened in a series of pictures. The pictures showed a story of a
balloon man who gives a green balloon to a boy. The boy loses the balloon and starts crying (B = the boy; M = the
balloon man; GB = green balloon):
“He ‘s (B) walking along…and he (B) sees a balloon man…and he (M) gives him (B) a green one…and he
(B) walks off home…and it (GB) flies away into the sky. So he (B) cries.”
3.2. The CEE.
If the source of the DPBE is basically non-linguistic, namely difficulties children have with
making inferences about other speaker’s discourse representations, why do Romance children not
exhibit this effect? As has been suggested in section 1., this developmental difference is related to
the different nature of weak object pronouns in Germanic and Romance. Romance weak
pronouns have been argued to be clitics, unlike English and Dutch object pronouns (McKee 1992;
Cardinaletti & Starke 1995; Cardinaletti & Starke 1996; Baauw, to appear).
However, in order to be able to account for the CEE, we will first have to be more precise
about what we understand under the term ‘clitic’. In Baauw (to appear) it is shown that not all
‘clitic-like’ pronouns cause a CEE in young children: Dutch 4 and 5 year olds show a DPBE in
constructions containing the weak object pronoun ‘m (‘him’). Th is suggests that a division
between ‘syntactic clitics’ and what we could call ‘PF clitics’ or regular ‘weak pronouns’ is
useful (Zwicky 1977; Rizzi 1986). In fact, in Baauw (to appear) it is argued, on the basis of crosslinguistic evidence, that whether the CEE shows up, depends on whether weak pronouns undergo
movement to the functional domain9:
(17)
The CEE is the result of clitic movement.
The question is now: why should movement be relevant to the appearance of the CEE? In the
account that we will present in this section, we will show that ‘moved clitics’ give rise to a CEE
because they are interpreted in a different way than VP-internal regular weak pronouns or strong
pronouns: unlike strong pronouns and regular weak pronouns, clitics can never be interpreted as
free variables. Since coreference crucially depends on the possibility of pronouns to be interpreted
as free variables, and the DPBE derives from problems children have with the application of
constraints on coreference, it follows that Romance children will not exhibit a DPBE in
constructions with clitic pronouns.10
9
Dutch weak object pronouns have been argued to undergo movement to the functional domain (Zwart 1992;
Cardinaletti 1992; 1994; Delfitto & Corver 1993). However, if we take scrambling in Dutch (which affects both NPs
and pronouns) and ‘generalized object shift’ (as in Icelandic) not to be the result of movement, but of base generation
of objects in different positions inside the VP (Neeleman 1994; Bayer & Kornfilt 1994; Neeleman & Weerman, to
appear), the Dutch child will have very limited exposure to data that could lead him/her to conclude that Dutch weak
pronouns involve (optional) movement to the functional domain. This is so because only in very limited cases do
weak pronouns end up in positions that are inaccessible to DPs, and even when they do, they only do so optionally.
Therefore Dutch children will analyze weak pronouns as VP internal regular weak pronouns, which explains why
they exhibit a DPBE (see Baauw, to appear).
10
This predicts that Romance children will exhibit a DPBE in constructions containing strong pronouns. This
prediction is corroborated by a pilot study carried out by Berger (1997) among Italian speaking children. Italian
speaking children appeared to allow coreference much more often in constructions containing strong pronouns, like
lui (ia), than in constructions containing clitics, like lo in (iib)
(i)
a.
Il ragazzo sta indicando lui.
b.
Il ragazzo lo sta indicando.
‘The boy is pointing at him.’
Spanish children, on the other hand, did not show a DPBE in constructions containing strong DO pronouns (Baauw
1998). However, unlike Italian strong DO pronouns, Spanish strong DO pronouns must be doubled by a clitic
pronoun.
Our account of the CEE is based on the following two assumptions on ‘clitic movement’:
First, we will assume that clitic movement is NP movement, in the sense of Sportiche (1992).
This means that the object clitic heads a functional projection, which we will call FP (following
Uriagereka 1995). A null element is generated in object position and is moved to [Spec, FP].
Second, we will assume that NP movement is empty variable movement, which is interpreted in
terms of λ- abstraction (Neeleman & Weerman, in press). Given these two assumptions the CEE
can be derived in the following way:
When the empty object variable moves from its argument position to [Spec, FP], a
variable chain is created. The tail of a variable chain must be interpreted as a predicative trace, i.e.
a variable bound by a λ-operator. In other words, NP movement corresponds to functional
abstraction in the semantics at the interface ((18) abstracts away from V-to-I movement):
(18)
a.
FP = λxλy (x señala y)
/ \
y
F’
/
\
la
VP = λx (x señala y)
/ \
V t
señala
__________
variable chain
b.
La niña la señala.
the girl her points-at
If the external argument (la niña ) is applied to this structure, this will inevitably result in a
Principle B violation. The reason is that under agreement of la and la niña the two variables will
be identified (x = y), to the effect that (18) will give rise to (19):
(19)
[λx (x señala x)] (la niña)
Since there is strong evidence that children know and obey Principle B (see the highly adult-like
performance of English and Dutch speaking children on constructions containing quantifiers),
Romance children will reject a reading of (18b) in which the object (la) and the local subject (la
niña) refer to the same individual, which is in line with our experimental results.
However, if clitics encode functional abstraction, one may wonder how they can refer to
objects outside the sentence, as in (20) where la in the second sentence refers to María in the first
sentence:
(20)
(ii)
Es imposible que María se haya ido al extranjero. Ayer la vi en casa.
`It is impossible that Maria went abroad. Yesterday I saw her at home.’
El niño le dibujó a él.
(84% adult like)
the boy himcl drew [acc] him
‘The boy drew him.’
Due to space limitation we will not discuss clitic doubling.
We suggest that in contexts which force disjoint reference (20) (and in ‘Oscar sentences’ (5); see
below), an empty topic is generated as an argument of the λ-abstract created by the variable
chain:
(21)
a.
[topic ec] Juan la vio en casa.
ec
John hercl saw at home
`John saw her at home’
b.
[λx (Juan vio x)] (ec)
This structure is very similar to (22), a case of clitic left dislocation. However, unlike an overt
(left dislocated) topic, the empty topic in (21) normally cannot introduce a new discourse topic,
since it lacks the necessary descriptive content. This explains why clitics generally refer to ‘old
information’ (familiarity conditi on).
(22)
a.
[topicA María] Juan
ACC. Maria John
la vio en casa. b.
hercl saw at home
[λx (Juan vio x)] (María)
However, (21) shows that although clitics cannot be interpreted as free variables, the
empty topic that functions as the argument of the λ-abstract can be interpreted as a free variable.11
This gives rise to a potential problem for our approach of the CEE: if the empty topic can be
interpreted as a free variable, why do Romance children not make use of an empty topic structure
in order to get a coreferential reading of (18b)? The fact that children apparently do not do this is
even more surprising given the fact that many Romance speaking adults do accept so called
‘Oscar sentences’. Oscar sentences provide evidence that empty topic structures can be used in
the Romance languages in order to get coreference between co-arguments. We have two possible
answers to this question:
One possibility is that coreference is disfavored with respect to binding if coreference is
arrived at through the interpretation of the pronoun as a bound variable. Notice that in order to get
coreference in (18b) λ-abstraction and λ-conversion would be required: the empty topic that ends
up covalued with the subject is the argument of a λ-abstract. This means that coreference in
Romance in fact involves binding + coreference, which is less economical than binding alone.
We might call this economy condition ‘Residual Rule I’, after the coreference rule introduced by
Grodzinsky & Reinhart (1993). We suggest that when the discourse provides evidence that
codetermination involves two different guises, as is the case in ‘Oscar sentences’, this economy
condition is overruled.
A second possibility (which we consider the preferred one) is that empty topics are
licensed by specific discourse requirements. When (18b) is uttered ‘out of the blue’, there is no
extra guise in discourse and no trigger for extra empty topic positions, to the effect that
coreference would be blocked. But when is an empty topic triggered? We propose that there are
basically two contexts that require the generation of an empty topic. Both have in common that
binding is either irrelevant or infelicitous. The first is the context where the two arguments of a
predicate refer to different individuals and hence to different guises. This is the case of (20),
11
Another possibility is that the empty topic it is bound by an element in a higher clause, like in (i):
(i)
María creía que Pedro la quería. (María = la)
‘Mary thought that Peter loved her.’
where the clitic refers to an object outside the sentence. Binding is irrelevant here. The second
case involve the so-called ‘Oscar sentences’ (or other sentences listed under (5)), which are
acceptable to many Romance speakers. Oscar sentences are embedded in contexts that provide
evidence that codetermination involves distinct guises. In these contexts binding cannot be
felicitous, since binding necessarily involves the use of a single guise. Romance speakers will
build up the extra empty-topic structure in order to avoid a single guise construal.
Notice that neither of these two contexts applies to the experimental task that the children
had to perform. Children just had to evaluate whether a sentence like (18b) corresponded to a
picture in which the subject (la niña) points at herself (while some other female individual was
watching her) (Truth Value Judgement Task; see Crain & Thornton 1998).12,13 This is not an
action that involves two individuals, nor does the discourse context force codetermination with
distinct guises, which means that a bound reading would not be infelicitous here, but it violates
Principle B.
4. Conclusion
In this paper we have tried to give an account of the DPBE in languages such as Dutch and
English and it’s absence in the Romance languages (CEE). We have a rgued that the DPBE is due
to the fact that children’s limited inferential capacity leads them to create ‘mildly reflexive’
predicates in contexts that are disallowed by adults. Romance children too have a more limited
inferential capacity, but they do not display a DPBE in constructions containing clitic pronouns
because (i) coreference in Romance is dependent on constructions involving empty topics, and
(ii) empty topics are only generated in contexts where binding is either irrelevant (reference to
different individuals) or infelicitous (‘Oscar sentences’).
More in general we conclude that coreference can be as ungrammatical as binding, in the
sense that coreference too can lead to a Principle B violation (cross-modular Principle B).
References.
Avrutin, S. (1994) Psycholinguistic Investigations in the Theory of Reference, Doctoral
Dissertation, MIT Working Papers in Linguistics.
Baauw, S. (1997) ‘Children’s Interpretation of Strong Pronouns in Spanish: Further evidence for
a Syntactic Source of the Delay of Principle B Effect,’ Ms., Utrecht University.
Baauw, S. (to appear) ‘The Role of the Clitic-Full Pronoun Distinction in the Acquisition of
Pronominal Coreference,’ BUCLD Proceedings 1998
Baauw, S. , Escobar & Philip (1997) ‘A Delay of Principle B Effect in Spanish Speaking
Children: The Role of Lexical Feature Acquisition,’ In: A. Sorace, C. Heycock & R.
12
The task is often presented to the child in the form of a ‘guessing game’, in which a puppet or one of the two
experimenters, has to ‘guess’ what happens on a picture that is shown to the child, but which is invisible to the
‘guesser’. The child has to judge whether the ‘guesser’ is correct or not.
13
All the studies cited in this paper made use of the Truth Value Judgement Task, with the exception of Padilla
(1990), who used an ‘Acting Out Task’.
Shillcock (Eds.), Proceedings of GALA `97, Human Communication Centre, Edinburgh.
Baauw & Delfitto (in progress)
Bamberg, M. (1987) The Acquisition of Narratives, Mouton de Gruyter, Berlin.
Bayer, J. & J. Kornfilt (1994) ‘Against Scrambling as an Instance of Move -Alpha,’ in: N. Corver
& H. van Riemsdijk (eds.), Studies on Scrambling; Movement and Non-Movement
Approaches to Free Word Order Phenomena, Mouton de Gruyter, Berlin.
Berger, C. (1997) Research Paper ‘Acquisition of Syntax,’ Utrecht University.
Cardinaletti, A. (1992) ‘On clitization in the Germanic Languages,’ ms. University of Venice and
Genova.
Cardinaletti, A. (1994) `The Internal Structure of Pronominal DPs, Linguistic Revieuw, 11, 195219
Cardinaletti, A. & M. Starke (1995) ‘The Tripartition of Pronouns and its Acquisition: Principle
B Problems are Ambiguity Problems, ‘ NELS 25, 1-12
Cardinaletti, A. & M. Starke (1996) ‘Deficient Pronouns: A View from Germanic,’ In: H.
Thráinsson, S.D. Epstein & S. Peters (eds.), Studies in Germanic Syntax II, 21-65,
Kluwer.
Chomsky, N. (1981) Lectures on Government and Binding, Foris, Dordrecht.
Chien Y.-C. & K. Wexler (1990) ‘Children’s Knowledge of Locality Conditions in Binding as
Evidence for the Modularity of Syntax and Pragmatics,’ Language Acquisition, 1, 225295.
Clark, H.H. (1977) ‘Bridging,’ In: P.N. Johnson -Laird & P.C. Wason (eds.), Thinking: Readings
in Cognitive Science, Cambridge University Press, Cambridge.
Crain, S. & R. Thorton (1998) Investigations in Universal Grammar. A Guide to Experiments on
the Acquisition of Syntax and Semantics, MIT Press, Cambridge, Mass.
Delfitto, D. & N. Corver (1993) ‘Feature Asymmetry and the Nature of Pronoun Movement,’
OTS Working Papers, Utrecht University.
Everaert, M. (1986) The Syntax of Reflexivisation, Foris, Dordrecht.
Grodzinsky, J. & T. Reinhart (1993) ‘The Innateness of Binding and Coreference,’ Linguistic
Inquiry, 24, 69-102.
Hamann, C.& W. Philip (1996) ‘The French Delay of Principle B Effect,’ ms. Utrecht University.
Heim, I. (1982) The Semantics of Definite and Indefinite Noun Phrases, Doctoral Dissertation,
University of Massachusetts, Amherst, MA.
Heim, I. (1998) ‘Anaphora and Semantic Interpretation: a Reinterpretation of Reinhart’s
Approach,’ In: U. Sauerland & O. Percus (eds.) The Interpretive Tract, MIT Working
Papers in Luinguistics, Vol. 25.
Heim, I. & A. Kratzer (1998) Semantics in Generative Grammar, Blackwell Textbooks in
Linguistics, Blackwell Publishers, Oxford.
Karmiloff-Smith, A. (1979) A Functional Approach to Child Language, Cambridge University
Press, Cambridge.
Karmiloff-Smith, A. (1980) ‘Psychological Processes Underlying Pronominalization and NonPronominalization in Children’s Connected Discourse,’ In: J. Kreiman & A.G Ojeda
(eds.), Papers from the Parasession on Pronouns and Anaphora, Chicago Linguistic
Society, Chicago, Illinois.
Karmiloff-Smith, A. (1986) ‘Some fundamental Aspects of Language Development after age 5,’
In: P. Flecher & M. Garman, Cambridge University Press, Cambridge.
McKee, C. (1992) ‘A Comparison of Pronouns and Anaphors in Italian and English Acquisition,’
Language Acquisition, 2, 21-54.
Maratsos, M. (1976) The Use in Definite and Indefinite Reference in Young Children, Cambridge
University Press, Cambridge.
Neeleman, A. (1994) Complex Predicates, Doctoral Dissertation, Utrecht University.
Neeleman, A. & F. Weerman (in press) Flexible Syntax: A Theory of Case and Arguments.
Padilla, J. (1990) On the Definition of Binding Domains in Spanish, Kluwer Academic
Publishers, Dordrecht/Boston/London.
Postma, G-J. (1997) ‘Logical Entailment and the Possessive Nature of Reflexive Pronouns,’ In:
H. Bennis, P. Pica & J. Rooryck (eds.), Atomism and Binding, Foris Publications,
Dordrecht.
Philip, W. & P. Coopmans (1996) ‘The role of lexical feature acquisition in the development of
pronominal anaphora,’ in: W. Philip & F. Wijnen (Eds.), Amsterdam Series on Child
Language Development, Vol. 5.
Reinhart, T. (1983) Anaphora and Semantic Interpretation, The University of Chicago Press,
Chicago.
Reinhart, T. & E. Reuland (1993) ‘Reflexivity,’ Linguistic Inquiry, 24, 657-720.
Reuland, E. (1998) ‘Primitives of Binding,’ Ms., Utrecht University
Rizzi, L. (1986) ‘On the Status of Subject Clitics in Romance,’ In: O. Jaeggli & C. Silva Corvalan (eds.), Studies in Romance Languages, Foris, Dordrecht.
Sigurjónsdóttir, S. (1992) Binding in Icelandic: Evidence from Language Acquisition, Doctoral
Dissertation, University of California.
Sportiche, D. (1992) ‘Clitic Constructions,’ Ms. UCLA.
Uriagereka, J. (1995) ‘Aspects of the Syntax of Clitic Placement in Western Romance,’
Linguistic Inquiry, 26, 79-123.
Zwart, J.W. (1992) ‘Notes on Clitics in Dutch,’ paper presented at the ESF Workshop on Clitics,
Lund May 22, Groningen University.
Zwicky, A. (1977) ‘On Clitics,’ Indiana University Club, Bloomington.