Content uploaded by Roumyana Slabakova
Author content
All content in this area was uploaded by Roumyana Slabakova
Content may be subject to copyright.
Second Language Research
28(3) 319 –343
© The Author(s) 2012
Reprints and permission: sagepub.
co.uk/journalsPermissions.nav
DOI: 10.1177/0267658312447612
slr.sagepub.com
second
language
research
Clitic-doubled left dislocation
and focus fronting in L2
Spanish: A case of successful
acquisition at the syntax–
discourse interface
Roumyana Slabakova
and Paula Kempchinsky
University of Iowa, USA
Jason Rothman
University of Florida, USA and University of Ottawa, Canada
Abstract
This experimental study tests the Interface Hypothesis by looking into processes at the syntax–
discourse interface, teasing apart acquisition of syntactic, semantic and discourse knowledge.
Adopting López’s (2009) pragmatic features [±a(naphor)] and [±c(ontrast)], which in combination
account for the constructions of dislocation and fronting, we tested clitic left dislocation and
fronted focus in the comprehension of English native speakers learning Spanish. Furthermore,
we tested knowledge of an additional semantic property: the relationship between the discourse
anaphor and the antecedent in clitic left dislocation (CLLD). This relationship is free: it can
be subset, superset, part/whole. Syntactic knowledge of clitics was a condition for inclusion in
the main test. Our findings indicate that all learners are sensitive to the semantic constraints.
While the near-native speakers display native-like discourse knowledge, the advanced speakers
demonstrated some discourse knowledge, and intermediate learners did not display any discourse
knowledge. The findings support as well as challenge the Interface Hypothesis.
Keywords
clitic left dislocation, focus fronting, topicalization, left dislocation, information structure,
Spanish, clitics, syntax–discourse interface, syntax–semantics interface
Corresponding author:
Roumyana Slabakova, University of Iowa, 557 EPB, Iowa City, IA 52242, USA
Email: roumyana-slabakova@uiowa.edu
447612SLR28310.1177/0267658312447612Slabakova et al.Second Language Research
2012
320 Second Language Research 28(3)
I Introduction
Since the start of the 21st century, the Interface Hypothesis (IH) has been influential
within the generative second language acquisition paradigm (for a review of its tenets,
genesis and development, see Sorace, 2011). Although technically agnostic on whether
or not Universal Grammar is fully accessible in adulthood (see Sorace, 2011), it pro-
poses that syntax proper is not the bottleneck of adult second language (L2) acquisition.
Like other generative approaches, it derives from formal linguistic theory, specifically
the contemporary focus on linguistic interfaces (for a review, see Rothman and
Slabakova, 2011). Interfaces can be understood as points of interaction between mod-
ules or systems, where representations that are the output of one module or system
must be interpreted by another. The internal interfaces, as this term is widely used, are
those between modules of language, such as the syntax–semantics interface. External
interfaces, in contrast, involve language and non-linguistic cognitive systems, such as
the syntax–discourse interface. Figure 1 presents one possible configuration of lan-
guage architecture modules, the discourse–pragmatics box having its separate interface
with morpho-syntax.
Somewhat uniquely, the IH combines formal linguistic analysis with research and
concepts from psychology (and cognitive science more generally), building on the
notion of interfaces to propose a theory of possible second language acquisition (SLA)
attainment. In general terms, the IH predicts that second language learners will have
more difficulty in acquiring properties related to external interfaces than those related
to internal interfaces. In its current form (Sorace, 2011), the IH appeals to processing-
related explanations for the performance differences noted in near-native speakers of an
L2. Essentially, the IH maintains that there is a cognitive cost to having more than one
grammar represented in the mind, drawing on the concept of inhibitory control (e.g.
Green, 1986, 1998). Inhibitory control has long been discussed in psycholinguistics as
a deterministic factor in bilingual performance; it refers to the cognitive deactivation or
inhibition of the first grammar in the mind when the other is needed for a given linguis-
tic task. Linguistic inhibition is necessary for bilinguals under the generally accepted
idea that all grammars in the mind of bilinguals are simultaneously activated (Kroll et al.,
2008; Marian and Spivey, 2003; for a review, see Bialystok et al., 2009). Taking adult
• Discourse
• Pragmatics
• Morpho-syntax
• Grammatical
meanings and
features
• Concepts
• Lexical meanings • Semantics
• Phonetic system
• Prosody
Figure 1 Modular design of the language faculty, following Reinhart (2006)
Slabakova et al. 321
L2 acquirers as one particular case of bilingualism, these L2ers must use some cognitive
resources to suppress their first language (L1) grammar when engaged in the L2,
thereby taxing memory systems, executive function and attentional resource allocation.
It thus stands to reason that even in the case that underlying L2 linguistic representa-
tions are native-like, there will still be some target-deviant performance variability. The
IH builds on this bilingual cognitive effect, predicting that even the most advanced
learners (i.e. near-natives) will have some residual optionality in performance guided
by formal linguistic and cognitive/processing considerations.
Tsimpli and Sorace (2006) propose that purely syntactic operations and computa-
tions that derive from representational requirements of internal linguistic interfaces are
much less likely to show lingering optionality because, they claim, such operations
require less cognitive-processing capacity. Representations at the internal interfaces
involve linguistic features (formal, semantic and phonological), which are grammar
internal. In contrast, external interfaces – such as the syntax–discourse interface – map
linguistic representations to broader discourse representations that involve external
pragmatic conditions on contextual appropriateness. Integrating linguistic and non-
linguistic information reasonably imposes higher demands on processing resources. As
a result, L2 speakers are predicted to demonstrate some level of performance variability
with respect to properties at the syntax–discourse interface indefinitely, because of the
linguistic and cognitive complexities involved.
According to Sorace and Filiaci (2006), the IH does not contend that L2ers are a
priori unable to acquire target representations for syntax–discourse properties; rather,
residual optionality – as evidenced in linguistic performance – arises as a by-product of
the extra burden these properties entail for finite cognitive resources. Because the IH is
based on cognitive consequences of bilingualism, it makes predictions in addition to
and beyond L1 transfer. That is, some level of residual optionality is expected for syntax–
discourse properties irrespective of L1–L2 pairings, although L1 influence as a co-
occurring variable is not precluded. To date, Sorace and colleagues have produced an
impressive body of work to support the IH, mainly examining referential subject pro-
nominal distribution in contexts of topic shift and anaphora resolution (e.g. for L1
English / L2 Italian, see Belletti et al., 2007; Sorace and Filiaci, 2006; for L1 Russian /
L2 Greek, see Tsimpli and Sorace, 2006).
In recent years, more properties at the syntax–discourse interface across a wider array
of L1–L2 pairings have been examined, and the results have been interpreted as chal-
lenges to the IH. White (2011a) examines an impressive sampling of studies that test L2
acquisition across internal and external interfaces, and argues that when the data are
viewed together, the syntax–discourse interface, claimed by the IH to be the primary
locus of residual L2 optionality, does not seem to be more problematic than other inter-
faces. White shows that the studies surveyed attest to apparent successes and failures in
the L2 ultimate attainment of properties at both internal and external interfaces, as well
as of properties of narrow syntax. Moreover, a significant number of very recent studies
have provided evidence of native-like L2 performance for external interface properties.
Rothman (2009) examines contrastive focus and pronominal referential subject use in
highly advanced L2 Spanish. Rothman shows that some learners perform identically to
the native controls across several tasks. Slabakova and Ivanov (2011) examine the syntax
322 Second Language Research 28(3)
of clitics and the syntax–discourse interface knowledge of clitic left dislocation (CLLD)
in L2 Bulgarian and L2 Spanish. They show that L2 performance at the near-native
level does not vary from native speakers; that is, there is no evidence of residual option-
ality at the highest levels of L2 proficiency. Bohnacker (2010) examines the acquisition
of discourse-sensitive clause-initial properties in L2 Swedish. Her data demonstrate
substantial increases in clause-initial expletive subjects, clefts and lightweight given
elements as L2 proficiency increases, which she interprets as being indicative of devel-
opment towards the target. Donaldson (2011a, 2011b, in press) examines among other
properties clitic right dislocation (CLRD) and CLLD in near-native L2 French, and also
shows that residual optionality at external interfaces is not inevitable. Finally, Iverson
et al. (2008) examine discourse-sensitive mood alternations (indicative vs. subjunctive)
with epistemic predicates in advanced L2 Spanish, showing that some learners perform
like native speakers.
It must be noted that Sorace (2011) rejects the idea that the IH is a theory of lin-
guistic development, asserting instead that it is a hypothesis about SLA at the level of
ultimate near-native attainment. However, Lardiere (2011) and White (2011b) chal-
lenge the claim that the IH does not make developmental predictions, convincingly
arguing and demonstrating that it does. Many researchers since 2000 have derived
and applied developmental predictions of the IH, in particular the prediction that nar-
row syntactic properties will be acquired before interface-conditioned properties in
general, and that properties of the internal interfaces will be acquired before those of
the external interfaces. In fact, all the studies referenced in the preceding paragraph
show, cross-sectionally, protracted development of syntax–discourse properties when
measured against the acquisition of the purely syntactic properties at lower levels of
proficiency. For example, Slabakova and Ivanov (2011) and Rothman (2009) show
that the syntactic acquisition of clitic pronouns and null subjects, respectively, pre-
cedes the acquisition of their corresponding discourse-sensitive properties (but see
Lozano and Mendikoetxea, 2008, 2010).
In the present study we assume, with Lardiere (2011) and White (2011b), that the IH
can and does make developmental predictions. In particular, this is the prediction we are
testing:
[T]he hypothesis that narrow syntactic properties are acquirable completely in a second
language, even though they may exhibit significant developmental delays, whereas interface
properties involving syntax and another cognitive domain may not be fully acquirable. Let us
refer to this as the ‘interface hypothesis’. (Sorace and Filiaci, 2006: 340)
The study focuses on the acquisition of the discourse-sensitive properties of clitic left
dislocation (CLLD) and focus fronting (FF) in English-native learners of L2 Spanish at
various stages of proficiency, including a group of near-natives. As a cross-sectional
study, it allows us to examine discrete moments in the development of these properties,
while the results from the near-native group also allow us to test the ultimate attainment
predictions of the IH.
The grammatical structures under investigation are, prima facie, syntax–discourse
interface properties (although, as we will see, a semantic property is also implicated). As
Slabakova et al. 323
a theoretical point of departure, we adopt López’s (2009) conceptualization of the
syntax–discourse interface in general, as well as his specific analysis for the particular
properties tested.1 To be included in the experiment, all L2 participants – from intermedi-
ate to near natives – had to demonstrate knowledge of the syntax of clitics; thus, by virtue
of the study design, the learners displayed acquisition of the relevant syntactic proper-
ties, irrespective of proficiency level. Although intermediate learners did not show target
knowledge of all the syntax–discourse interface properties (appropriateness of the clitic-
doubled dislocations) or the semantics (constraints on the antecedent–dislocate relation-
ship that differentiate CLLD and CLRD), there is a clear trend towards full convergence
through the advanced level of proficiency, culminating in targetlike performance by
the near-natives. We will show that our findings present challenges to the IH, in that we
observe both a clear developmental trend and successful ultimate attainment. The article
is structured in the following way. Section II presents the formal theoretical analysis and
Section III then highlights the L2 learning task based on the formal description of Section
II. Sections IV and V describe the study’s methodology, present the study results and
their analysis, as well as provide a critical discussion and conclusions.
II Syntax and information structure
1 The interpretive properties and syntax of dislocations in Spanish
As briefly mentioned above, the syntactic structures investigated in this study are two
left-dislocation structures (clitic left dislocation and focus). We also investigated clitic
right dislocation and clausal alternations in new information or Rheme contexts (for
those results, see Slabakova et al., 2011).
Left dislocations are often termed ‘topicalization’ or ‘focalization’, according to the
information status of the dislocated constituent. However, the discourse definitions of
terms like ‘topic’ and ‘focus’ vary from one analysis to the next, and sometimes enter
into direct contradictions (see Casielles-Suárez, 2004; López, 2009). In our study we
have adopted the analysis of López (2009), which rejects an approach to left-peripheral
syntax in which notions such as ‘topic’ and ‘focus’ are formal features driving the syn-
tactic derivation.2 For López, these are at most descriptive terms. The crucial discourse/
information structure notions are discourse anaphor and contrast, represented in terms of
the primitive features [±a(naphor)] and [±c(ontrast)].
López proposes an approach to the syntax–discourse interface that assumes a prag-
matic module of computation; this module yields an intermediate level of representa-
tion between the syntactic structure and the discourse structure where the features
[±a] and [±c] are assigned. Individual lexical items enter into the syntactic derivation,
which proceeds in phases (Chomsky, 2001) and yields at the end of each phase – vP
and CP – a syntactic object Σ. This syntactic object is then inspected by the pragmat-
ics module, which assigns the relevant discourse features to constituents located in
syntactic positions at the phase edge, yielding the information structure Σ[p]. These
pragmatic structures are subsequently assembled into longer units of text in the discourse
module, in the form of discourse representation structures. The overall model is illus-
trated in (1):
324 Second Language Research 28(3)
(1) The syntax–discourse relationship (López, 2009: 23)
Table 1 Features bundles and constructions in López’s (2009) model
+contrast –contrast
+anaphor clitic left dislocation (CLLD) clitic right dislocation (CLRD)
–anaphor fronted focus (FF) rheme
More specifically, elements that move to the edge of the vP and agree with a doubl-
ing clitic are assigned the feature [+a]; elements that move to the left periphery
are assigned the feature [+c]. Non-dislocated elements receive no pragmatic fea-
tures; this includes constituents that represent new information (sometimes called
‘regular focus’ or ‘new information focus’; we use the descriptive term Rheme
here, which is taken from the theme–rheme dichotomy). This system thus defines
the combinations of syntactic constructions and discourse feature bundles that are
shown in Table 1.
We shall first illustrate these features with Spanish examples3 – where all capitals
stand for focus intonation – and then give a brief description of the actual syntactic
derivation. The feature [+anaphoric] reflects the fact that the phrase moved from its
underlying word order position, known as ‘the dislocate’, is obligatorily related to an
antecedent in the discourse. Clitic left dislocation (CLLD) is a [+anaphoric] construction.
However, the conditions on the anaphoric relationship may differ (Villalba, 2000). For
both CLRD and CLLD, the dislocate may be semantically identical to the discourse ante-
cedent; however, CLLD allows a broader anaphoric relation in that the dislocate can also
be in a subset, superset, or part/whole relationship with the antecedent, as in (2) and (3):
(2) CLLD with set–subset relationship between dislocate and antecedent:
[context: What did the movers do with the furniture?]
Las sillas las dejaron en el pasillo, pero no
the chairs CL left.3pl in the hallway, but NEG
sé dónde están las mesas.
know.1sg where are the tables
‘The chairs they left them in the hallway, but I don’t know what they did with the
tables.’
Slabakova et al. 325
(3) CLLD with whole–part relationship between dislocate and antecedent:
[context: What can we do with this table? It doesn’t fit through the doorway]
Mira, las patas las podemos doblar.
look the legs CL can.1pl fold
‘Look, the legs we can fold them.’
With respect to [±contrast], fronted focus (FF) and CLLD are positively marked. The
[+c] feature encodes the idea that the dislocated element identifies a missing element (in
semantic terms, a variable) within a presupposition.4 In (4) below, the fronted [+c] element
is la alfombra ‘the rug’, which has undergone focus fronting, while in (5) the fronted [+c]
element is the CLLD dislocate las sillas ‘the chairs’. In both cases, the preceding discourse
establishes the presupposition ‘John bought x.’ In (4), the identity of x does not require a
discourse antecedent; focus fronting yields the combination [–a,+c]. In (5) the identity of x
must have a discourse antecedent (CLLD yields [+a,+c]), where the relationship between
the antecedent and the CLLD dislocate can be one of those enumerated above. It is in this
sense that Arregi (2003) refers to CLLD as ‘contrastive topicalization’:
(4) Fronted focus
[context: John bought the furniture]
LA ALFOMBRA compró (no los muebles)
‘THE RUG he bought (not the furniture).’
(5) CLLD
[context: John bought the furniture]
Las sillas (sí que) las compró, pero las mesas …
‘The chairs he did (indeed) buy them, but the tables …’
With respect to the actual syntactic derivation of dislocation structures, López assumes
that movement of the dislocated constituent involves a first step of movement to the edge
of the verb phrase. To illustrate, we will enter into some detail on the derivation of a
CLLD example such as (2). In (6), the dislocated constituent has moved to the edge of the
verb phrase (specifically, vP), with the copy in the original position indicated with < >.
(6)
326 Second Language Research 28(3)
As (6) indicates, there is an Agree relationship between the clitic and the dislocated
DP las sillas. The clitic itself is adjoined to the v head of vP, so that in effect the dislo-
cated DP is also in an Agree relationship with that head. The vP syntactic object is then
inspected by the pragmatics module, assigns the discourse feature [+a] to the clitic in the
phasal head, and hence also to the dislocated DP, which is in an Agree relationship with
it; in this particular example, the DP is las sillas.
As the derivation proceeds to the next phase, the DP at the edge of vP – bearing its
[+a] feature – will subsequently move to the left periphery of the clause. López specifi-
cally argues that movement to the left periphery is movement to the edge of FiniteP
(FinP), yielding a structure such as (7).
(7)
Since FinP is assumed to be a phasal category, the syntactic structure in (7) is then
inspected by the pragmatics module, which assigns the feature [+c] to the constituent in
the specifier position of FinP. In this way the CLLD dislocate ends up with the pragmatic
features [+a,+c].
Note that the assignment of the [+c] feature is not identical to the assignment of the [+a]
feature, which is assigned to the phase head, and hence only indirectly to a constituent at
the edge of the phase that agrees with that head. This is how López’s analysis differentiates
focus fronting from CLLD. A focus fronted constituent, such as LA ALFOMBRA ‘the rug’
in (4), must also first move to the edge of the vP, by general locality conditions on syntactic
movement. However, there is no clitic in focus fronting, and hence no Agree relation
between the fronted constituent and the phasal head. Hence, this constituent does not get
marked [+a]. Subsequent movement of the fronted constituent to Spec,Fin – identical to
the movement of the CLLD dislocate in (7) – will result in the fronted constituent receiving
the feature [+c].
2 Dislocations in English
Prima facie, since it lacks clitics, English also lacks the CLLD construction. English
does, of course, have topicalization, and Romance CLLD is often seen as discourse-
equivalent to topicalization. However, English topicalization disallows a (resumptive)
pronoun referring to the dislocate:
Slabakova et al. 327
(8) Monica: Did you eat the salad?
Alfred: The lettuce I ate (*it). I didn’t like the olives.
Additionally, English appears to have a construction that looks similar to focus
fronting:
(9) FIDO they named their dog. (Prince, 1981: 259)
Nevertheless, it is difficult to make direct comparisons between the Spanish con-
structions and the English constructions. First of all, it is well known that English,
unlike Spanish (and Romance in general), allows only one preposed element per clause
(leaving aside hanging topics, see Emonds, 2004). However, the discourse status of this
preposed element is not clear. Gundel (1988) refers to ‘topic topicalization’ and ‘focus
topicalization’, whereby the discourse function of the latter is to link some gap in the
proposition with a contextually salient set of alternatives, while in the former, the
fronted element stands in an anaphoric relationship with some entity already estab-
lished in the discourse. Other work on preposing in English generally uses the term
‘focus preposing’ for Gundel’s ‘focus topicalization’, and Ward (1988 and subsequent
work) argues that preposing in English can be defined as either focus preposing or topi-
calization based on the discourse context and the intonational patterns of the sentence;
see de Swart and de Hoop (2000), who characterize English focus preposing as having
an element in topic position that receives focal stress. A careful perusal of the examples
labeled ‘focus preposing’ in Ward (1988) shows that they are more comparable to
Arregi’s (2003) terminology of ‘contrastive topicalization’, to which we alluded in the
previous section; i.e. picking out a token from a contextually salient set (a definition not
very different from Gundel’s focus topicalization), rather than correcting a constituent
in a previous utterance.5 Minimally, there do not appear to be unambiguous syntactic
diagnostics in English for ‘topicalization’ vs. ‘focus fronting’, unlike the case in Spanish
where the presence vs. lack of the clitic clearly distinguishes the two.6 English prepos-
ing and Spanish focus fronting are alike in their syntactic distribution: they are limited
to root clauses and to ‘root-like indirect discourse embedding’ contexts (Emonds,
2004), and in this respect both differ from CLLD, which is possible in a variety of sub-
ordinate clauses (see Haegeman, 2006).
Finally, English, like Spanish, also has a version of the Nuclear Stress Rule, placing
nuclear stress by default on the lowest (and therefore linearly last) constituent. For both
languages, then, if the direct object represents new information and there is no other mate-
rial within the VP, no further syntactic or phonological operations are necessary. However,
the typical strategy employed by English when the new information is not syntactically
the final element in the clause is stress shift (see, among others, Reinhart, 1995, 2006),
since English lacks any equivalent to Spanish p-movement (the scrambling of constituents
to allow the nuclear stress to fall on the discourse-appropriate constituent).
3 Learning tasks for native speakers of English
In this study, we endeavor to test the IH, and we have identified some Spanish construc-
tions whose felicity depends on the discourse context. We must establish whether
328 Second Language Research 28(3)
knowledge of these constructions is possible to transfer from the native language. Based
on the discussion in the previous section, English seems to have a direct equivalent of
one of the two Spanish constructions under investigation here: preposing with prosodic
emphasis on the dislocate, as in (9) above. However, if learners map English topicaliza-
tion as in (8) to Spanish CLLD, based on superficial similarity of the discourse con-
texts, the mapping will mislead them into rejecting clitics in CLLD.
In order to acquire the CLLD construction in its entirety, English learners must acquire
the syntactic properties of clitics, as well as the syntactic knowledge of the movement
operations involved in CLLD. In addition, the learners need to acquire the discourse
appropriateness of the clitic-doubled dislocations, which is a property at the syntax–
discourse interface. Such acquisition includes knowledge that a dislocate in an anaphoric
relationship with another previously mentioned constituent should be clitic-doubled,
while a dislocate that is not in such an anaphoric relationship cannot be doubled. Third,
the semantic freedom on the antecedent–dislocate relationship has to be learned; see
examples (2) and (3). Note that the latter is a subtle semantic property whose acquisition
requires sufficient experience with CLLD. At the same time, CLLD is a very common,
colloquial construction (López, 2009) and such linguistic experience with CLLD will be
available to learners, especially those in Spanish-speaking environments.
Keeping in mind the various considerations of what is available from the L1 grammar,
what has to be acquired in the target grammar, and the Interface Hypothesis, we predict
that:
1. Learners’ syntactic knowledge of clitics will not be extended to knowledge of
their information structure-marking properties at the intermediate levels of
proficiency.
2. Learners at more advanced proficiency levels will demonstrate knowledge of
both the syntax and discourse-appropriateness of the clitic-doubled dislocations.
3. Learners at more advanced proficiency levels will also demonstrate knowledge
of the semantic constraints on the clitic-doubled dislocations.
In addition, we can make order of acquisition predictions extrapolating from the
Interface Hypothesis, as below:
Narrow syntactic properties > syntax–semantics interface properties > syntax–discourse
interface properties
IV Experimental study
1 Participants and test materials
Eighty-eight individuals took an online battery of tests, posted on WebSurveyor, an
independent online survey service. Twenty-one were native speakers of Spanish
from different Spanish-speaking countries (1 from Argentina, 4 from Chile, 4 from
Colombia, 1 from Costa Rica, 2 from Cuba, 3 from Spain, 5 from Mexico and 1 from
Venezuela7), and 67 were adult learners of Spanish. An anonymous questionnaire with
personal questions ascertained that the learners had started acquiring Spanish after
Slabakova et al. 329
puberty (11 years of age). They reported Spanish as their second language, and none
reported speaking a third language better than Spanish. The large majority were
classroom learners. The near-natives reported more than 5 hours a week of Spanish
communication, with some of them significantly more, while most intermediate learn-
ers did not spend time communicating in Spanish at all. Most participants reported
education levels higher than high school (at least some college). The learners’ names
were not known to the investigators, to protect their identity;8 see Table 2.
The test materials included a proficiency test based on the DELE (official exam of
proficiency administered by the Spanish Ministry of Education), which is standardly
used by generative L2 Spanish research (e.g. White et al., 2004). The test contained 50
items (distributed between a cloze test format and multiple choice). The latter was used
to divide the 67 learners into proficiency levels: 21 individuals fell in the near-native
range (score range 47–50), 22 were classified as advanced learners (score range 40–47),
and 23 as intermediate learners of Spanish (score range 30–39).
The battery of tests also included a test on the syntax of clitics. This test was essential
in order to ascertain that learners had knowledge of the correct placement and obligatory
presence of clitics in some Spanish sentences. There is no point in testing learners’ knowl-
edge of the discourse and semantic constraints of clitics if they lack the prerequisite syn-
tactic knowledge. The test included 10 forced multiple-choice items. Each sentence had
five options to choose from, for a total of 50 points. Points were given for correctly
choosing the grammatical options and for correctly rejecting the ungrammatical choices.
The cut-off point for inclusion in the rest of the study was set at 35, or 70% accuracy. All
advanced and near-native participants scored over 45, or were 90% accurate, with a single
exception, an advanced-proficiency individual who scored 40, or 80%. All intermediate
learners scored above 35 points. Below are two examples from the clitic test.
(10) Dulce: ¿Te gustó algo de la tienda?
‘Did you like something from the store?’
Julia: Me encanta esa chaqueta, _______ pronto.
CL charm this jacket, _____ soon
‘I love this jacket, __________ (I-want to buy it) soon’
me quiero comprarla
me la quiero comprar
quiero comprármela
quiero comprar
quiero me la comprar
Table 2 Biographical data of participants
Average age Women/men Average age at start of Spanish
Native speakers (n = 21) 35.8 13/8 0.0
Near-natives (n = 22) 31.5 11/11 12.3
Advanced (n = 22) 27.9 10/12 11.4
Intermediate (n = 23) 22.3 17/6 13.1
330 Second Language Research 28(3)
(11) Lucía: ¿José le trajo los libros al profesor?
‘Did José bring the books to the professor?’
Pedro: Sí, _____. (Yes, he-brought them to-him)
se los trajo
los se trajo
trajo
setrajo los
trájoselos
The main experimental task was the felicity judgment task. It included 40 context-test
sentences combinations: 10 CLLD and 10 CLRD constructions, 5 fronted focus con-
structions and 5 Rheme constructions. Ten fillers were also included.9 The task was
untimed. A screen shot appears in Figure 2. Recall that we only report on CLLD and FF
in this article (for the other results, see Slabakova et al., 2011).
Each test item had the following structure: after a brief context, a short dialogue was
presented which contained two possible answers, our test sentences. Each one of the
answers had to be evaluated in felicity, on a scale of 4 (perfect) to 1 (very strange), but
participants could also choose ‘I don’t know.’ The 10 CLLD contexts were divided into
5 with identity between antecedent and dislocate, and 5 with a set–subset relationship
between antecedent and dislocate. Within each screen (context–dialog combination),
there were two answers to judge: one with a clitic and one without a clitic (see screen-
shot). In the end, we had 20 evaluation scores for the CLLD condition. In the 5 FF con-
texts, participants had to evaluate one sentence with clitic (unacceptable) and one without
clitic (acceptable).
The context stories and test sentences were presented both aurally and visually to all
participants. The aural presentation of test items was crucial for distinguishing CLLD
constructions, which lack emphatic stress on the fronted constituent, and FF construc-
tions, where the fronted constituent must be stressed and a doubling clitic is ungrammati-
cal. It was also important for the testing of CRLD items, which require an intonational
Figure 2 Screen shot of a felicity judgment task presentation, from the fronted focus condition
Slabakova et al. 331
break between the end of the clause and the right dislocated element. A negligible
number of answers displayed the ‘I don’t know’ option.
Below we give examples from the CLLD construction. The context and the question men-
tion furniture, which is the discourse antecedent. Juan’s answer contains reference to chairs,
which act as a subset of all possible furniture, the set. Thus answer A with a clitic and a left-
dislocated constituent was expected to be appropriate. The relationship between the anteced-
ent and the dislocate is not one of equivalence, but the CLLD allows for that. On the other
hand, answer B without a clitic is simply ungrammatical, and marked with an asterisk (*).
(12) Juan y Mónica están recién casados y se acaban de mudar a su nueva casa.
Mónica trabajó todo el día mientras Juan se quedó en casa y acomodó los
muebles. Mónica lo llamó para ver cómo estaban las cosas.
Juan and Monica are newlyweds and have just moved into their new home.
Monica worked all day long while Juan stayed at home and arranged the fur-
niture. Monica called him to see how things were going.
Mónica: Entonces ¿qué hiciste con los muebles?
Monica: So, what did you do with the furniture?
A) Juan: Bueno, las sillas las puse en la cocina, y los sillones los puse en la sala.
Juan: Well, the chairs them I put in the kitchen, and the armchairs them I put
in the living room.
B) Juan: * Bueno, las sillas puse en la cocina y los sillones puse en la sala.
Juan: Well, the chairs I put in the kitchen, and the armchairs I put in the living
room
In example (13), where the antecedent and the dislocate are identical (both are ‘the
vinegar’), answer A is still acceptable and answer B is ungrammatical due to the lack of
clitic-doubling.
(13) Mónica decidió quedarse en casa preparando la cena mientras su esposo Juan
fue al supermercado. Cuando regresaba a casa, Juan recibió una llamada
teléfonica de Mónica, quien quería asegurarse de que Juan había comprado lo
que necesitaba para preparar la ensalada.
Monica decided to stay at home making dinner while her husband Juan went to
the supermarket. When he was on the way home, Juan received a call from
Monica, who wanted to make sure that Juan had bought everything she needed.
Mónica: Entonces ¿compraste el aceite de oliva y el vinagre?
Monica: So, did you buy the olive oil and the vinegar?
A) Juan: El vinagre lo compré. Tenemos aceite en la casa.
Juan: The vinegar it I bought. I know we have oil at home.
B) Juan: * El vinagre compré. Tenemos aceite en la casa.
Juan: The vinegar I bought. I know we have oil at home.
In presenting the results, we will remind the reader of each construction tested.
332 Second Language Research 28(3)
2 Results
A general linear model ANOVA was conducted to compare the ratings for each construc-
tion included in the study described above. We were looking for significant contrasts
between ratings to acceptable and unacceptable test sentences within the same group of
participants, since we were interested in whether each group demonstrates a significant
contrast in their grammar on this particular construction. We further inspected pairwise
comparisons between specific types of sentences within the ratings of each participant
group.
a CLLD judgments: The mean group responses by the four participant groups in the
study for the CLLD structures are given in the two parts of Figure 3. Recall that we tested
sentences in four conditions: the semantic difference of equivalence or subset between
discourse antecedent and dislocate were crossed with the presence or absence of a clitic.
The ratings are of sentences such as in (14) below, this one from the set–subset condition.
The participants had to evaluate the answers with clitics, and the answers without clitics,
separately.
(14) Q: ¿Qué hiciste con los muebles?
what did-you with the furniture (antecedent and dislocate)
‘What did you do with the furniture?’
A: Las sillas * (las) puse en la cocina, y los sillones * (los)
the chairs them I-put in the kitchen, and the armchairs them
puse en la sala.
I-put in the living-room.
‘I put the chairs in the kitchen, and the armchairs in the living-room.’
A two-way repeated-measures ANOVA was performed with Type of sentence (with
clitic, appropriate; without clitic, inappropriate) as a within-participants variable and
group as a between-participants variable. There was an effect of type of sentence (F(3,
252) = 46.436, p < 0.0001), no effect of participant group (native, near-native, advanced,
intermediate) (F(3, 84) = 2.354, p = 0.078), and a significant interaction between type
and group (F(9, 252) = 9.208, p < 0.0001). We looked into the significant interaction
further.
In Table 3, we present the significant contrasts between the types of test sentences in
the grammar of the native speakers. The blank cells represent irrelevant comparisons.
Cells marked with note ‘a’ mark the contrasts between acceptable and unacceptable
presence of clitic; cells marked with note ‘b’ mark conditions that differ in the semantic
dimension. As expected, the native speakers distinguish between the acceptable and the
unacceptable test sentences, while they do not make a significant difference between the
two acceptable types (the two semantic types of antecedent–dislocate relationships with
clitics) and the two unacceptable types (the sentences without clitics). Thus the natives
demonstrate an important contrast in their grammar, observing the discourse appropri-
ateness of the clitic-doubled constructions but also showing that the semantic relation-
ship between the discourse antecedent and the dislocate in CLLD is not constrained to
Slabakova et al. 333
Figure 3 Mean ratings of two semantic types of CLLD acceptability in context
equivalence only. Table 4 shows that the exact same contrasts exist in the grammar of
the near-native speakers, making their grammar identical to that of the native speakers
with respect to these contrasts.
334 Second Language Research 28(3)
The advanced learners distinguish between sentences with clitics and ones without
clitics, but only in the set–subset condition; see Table 5. In the case of the equivalence
condition, the one which is supposed to be easier to detect, the advanced learners do
not reject the clitic-less sentences sufficiently resolutely, although they accept the
appropriate sentences with a high rating (M = 3.4 out of 4). Table 6 demonstrates that
these contrasts are not yet available in the grammars of learners of intermediate
proficiency.
In summary, the native and the near-native groups made all the relevant distinctions
in the CLLD constructions with respect to the obligatoriness of the clitic. The advanced
group also made this distinction, but only in the more difficult set–subset condition.
Moreover, all groups observe the semantic freedom the CLLD construction allows for
Table 4 Sidak’s post-hoc comparison: CLLD condition (near native speaker group)
* Equivalence +
No Clitic
Subset + Clitic * Subset + No Clitic
Equivalence + Clitic p < .0001an.s.b
* Equivalence + No Clitic n.s.b
Subset + Clitic p < .0001a
Note: see notes at Table 3.
Table 5 Sidak’s post hoc comparison: CLLD condition (Advanced group)
* Equivalence +
No Clitic
Subset + Clitic * Subset + No Clitic
Equivalence + Clitic n.s.an.s.b
* Equivalence + No Clitic n.s.b
Subset + Clitic p = .037a
Note: see notes at Table 3.
Table 3 Sidak’s post-hoc comparison: CLLD condition (native speaker group)
* Equivalence +
No Clitic
Subset + Clitic * Subset + No Clitic
Equivalence + Clitic p < .0001an.s.b
* Equivalence + No Clitic n.s.b
Subset + Clitic p < .0001a
Note: Equivalence stands for Antecedent and Dislocate being identical; Subset stands for the Dislocate
being a subset of the Antecedent. a Contrasts between acceptable and unacceptable presence of clitic;
b Conditions that differ in the semantic dimension.
Slabakova et al. 335
the relationship between discourse antecedent and dislocate, in the sense that they accept
the clitic-doubled dislocations as highly in the equivalence condition as they do in the
set–subset condition.10
b Fronted focus judgments: Fronted focus is a construction that is more or less equiva-
lent in Spanish and English.11 Example (15) illustrates the way FF was tested:
(15) Juan y Mónica invitaron a María a comer. La cena se sirvió en la terraza y todo
estaba muy rico. María felicitó a Juan por la sopa que había hecho. Cuando
Mónica escucha esto, responde:
Juan and Mónica invited Maria for dinner. Dinner was served on the terrace
and everything was very tasty. Maria complimented Juan on the soup he
cooked. When Monica hears this, she responds:
Mónica: LA CARNE preparó Juan, no la sopa.
THE MEAT prepared Juan, not the soup.
* LA CARNE la preparó Juan, no la sopa.
THE MEAT it prepared Juan, not the soup.
Figure 4 Mean ratings of Fronted Focus acceptability in context
Table 6 Sidak’s post hoc comparison: CLLD condition (intermediate group)
* Equivalence +
No Clitic
Subset + Clitic * Subset + No Clitic
Equivalence + Clitic n.s.an.s.b
* Equivalence + No Clitic n.s.b
Subset + Clitic n.s.a
Note: see notes at Table 3.
336 Second Language Research 28(3)
Mean judgments are displayed in Figure 4. In the fronted focus condition, the
ANOVA shows an effect of type of sentence (F(1,84) = 56.988, p < .0001) and an
effect of group (F(3,84) = 5.744, p < .001), but no significant interaction. The Sidak
post-hoc tests show that all participant groups demonstrate a significant contrast
between FF sentences with clitic (ungrammatical) and without clitic (grammatical),
with the exception of the advanced group, for which the contrast approaches signifi-
cance, p = 057. That is to say, all participants tended to reject this structure when the
dislocate was pronounced with contrastive focus intonation, but it was doubled by a
clitic; see Table 7.
c Individual results: In order to further examine the developmental trends, we analysed
the data looking at individual performances in the following way: if a participant had
scores differing by a full point in the anticipated direction, we considered them aware of
the expected distinction.
Clear developmental patterns in the CLLD conditions are evident in Table 8 below.
The large majority of the native speakers made the expected distinctions (20 out of 21,
or 95%), with this percentage decreasing as proficiency levels go down. It is interesting
to note that even among the intermediate learners there are individuals who are sensitive
to the clitic-doubling phenomenon. At the same time, the data in Table 8 underscore the
truism that group means often hide wide individual variation, and such individual con-
trasts should always be presented since we discuss the grammars of people, not of
groups.
V Discussion and conclusions
This experimental study looked at the acceptability of two constructions at the
syntax–discourse interface. In the dislocation with anaphoric relations to the preced-
ing context, CLLD, clitic-doubling is obligatory. In the FF dislocation marked by intona-
tion, clitic-doubling is ungrammatical. This is the construction where transfer from
English is possible since English uses the same device for contrastive focus marking.
In addition, we tested semantic constraints on the antecedent–dislocate relationship
within the dislocation construction.
The data uncovered highly differentiated behavior on the part of the native speakers.
We shall focus first on the CLLD construction, which was accepted with high reliability
by natives and learners alike. This construction is very frequent in vernacular, everyday
Table 7 Intra-group comparisons between mean ratings of grammatical and ungrammatical
sentences in the FF construction (significance set at the 0.05 level)
Fronted Focus ( No clitic vs. * Clitic)
Natives (n = 21) p < .0001
Near-natives (n = 22) p < .0001
Advanced (n = 22) p = .057
Intermediate (n = 23) p < .0001
Slabakova et al. 337
Spanish. It has been used before to test the Interface Hypothesis (Ivanov, 2009;
Valenzuela, 2005, 2006; see also Slabakova and Ivanov, 2011), since it is dependent on
knowledge of core syntactic properties (clitics) but also on sensitivity to topics marked
in the discourse. The predictions of the IH are that especially at near-native proficiency
levels, this construction may present residual performance difficulties to speakers, and
their acceptance of this construction may ultimately be measurably different from that of
native speakers. However, our findings show we did not find such difficulties.
In discussing what constitutes knowledge of a construction, Bley-Vroman’s (1983)
comparative fallacy emphasizes that research must establish that learners make a signifi-
cant distinction between ungrammatical and grammatical sentences within their own
grammars, and not necessarily compare their grammars to those of native speakers. We
already established in Section IV.2.a that the near-natives exhibited a significant contrast
in their grammars between the acceptable dislocations with clitics and the unacceptable
ones without clitics. Furthermore, they did so when the antecedent and the dislocate
were semantically equivalent as well as when they were in a set–subset relationship.
Now, we will further look at a comparison with native speakers, simply because the IH
proponents argue for, and use, such comparisons in their inspection of ‘native-like’ gram-
mars. Additional Sidak pairwise comparisons reveal that the near-native speakers as a
group did not differ in these choices from the native speakers (p = 1 for CLLD equiva-
lence; p = .132 for left dislocated object without clitic, with equivalence; p = 1 for CLLD
subset; p = .507 for left dislocated object without clitic, with subset dislocate). Therefore,
even by this ‘stricter’ measure our near-native participants satisfy the requirement for
being truly native-like.
What about the developmental trajectory of the learners? Recall that Sorace (2011)
argues that the predictions of the IH are most particularly relevant to near-native speak-
ers (as well as simultaneous bilinguals and first language attriters). However, many
researchers have questioned this limitation of the scope of the IH (Lardiere, 2011; White,
2011b), arguing that what is a problem for near-native speakers should be no less of a
problem for less advanced speakers.12 Moreover, we contend that if Sorace’s processing-
based rationale for why external interfaces are more likely to result in residual optional-
ity is on the right track, then we should observe a pattern of differential accuracy
cross-sectionally, whereby groups at lower levels of proficiency will perform less well
than the more advanced learners. With this in mind, we inspected the pairwise compari-
sons between the native speakers, the advanced learners and the intermediate learners.
There is indeed a significant difference in ratings between the native and the intermediate
Table 8 Number of individual subjects who demonstrate a contrast between acceptable and
unacceptable constructions within each condition (percentages in parentheses)
Subject groups Equivalence Set–subset
Natives (n = 21) 20 (95) 18 (85)
Near-natives (n = 22) 13 (59) 9 (43)
Advanced (n = 22) 7 (32) 8 (36)
Intermediate (n = 23) 3 (13) 6 (26)
338 Second Language Research 28(3)
groups (p = .001 for CLLD-equivalence; p = .0001 for left dislocation no clitic with
equivalence; p = .005 for CLLD-subset; p = .001 for left dislocation no clitic with subset
dislocate). However, there is no such difference between the ratings of the advanced and
the native group in accepting the constructions with clitics, although advanced learners
behave differently from the natives in rejecting the clitic-less test sentences (p = .734 for
CLLD-equivalence; p = .001 for left dislocation no clitic with equivalence; p = .987 for
CLLD-subset; p = .01 for left dislocation no clitic with subset dislocate). It is a well-
known effect in second language acquisition that learners first acquire a construction not
available from their own language and only then learn to reject the construction which is
equivalent to their native language with a similar degree of certainty (Trahey and White,
1993). In a sense, there are two ‘stages’ of development in the acquisition of any con-
struction: acceptance of a non L1-like construction, and rejection of a L1-like construc-
tion. Our advanced group seems to be at the lower stage, while our near-native group is
at the higher stage of knowledge. While we cannot confidently assert that the grammars
of the advanced speakers as a group are identical to the native grammars, a look at our
individual results reveals that there is a significant percentage of individual learners,
even among intermediate learners, who have successfully acquired the distinction under
investigation.
Our findings of successful acquisition are in agreement with previous studies on the
acquisition of CLLD (Ivanov, 2009; Valenzuela, 2005, 2006). They are also in agreement
with recent findings on the acquisition of the syntax–discourse interface in near-native
French (Donaldson, 2011a, 2011b, in press). In this study, however, we went beyond the
syntax–discourse properties to examine a third type of property: an additional semantic
constraint on the allowed relationship between antecedent and dislocate. Our near-native
and advanced learners were able to successfully calculate the discourse appropriateness
of CLLD even when the antecedent and the dislocate were loosely related. While the
latter is a more difficult semantic calculation than a similar calculation under complete
identity between discourse antecedent and dislocate, it does not appear to have had an
effect on the accuracy of our learners.
Let us now examine knowledge of the semantic freedom between antecedent and
dislocate in the CLLD in relation to the other types of knowledge. For this purpose, we
should compare ratings of acceptable test sentences with clitics in the equivalence and
the set–subset conditions, as well as ratings of the unacceptable sentences without clit-
ics in the two conditions. Please refer to Tables 3 to 6 for those comparisons in all
learner groups. All of the Sidak comparisons are non-significant. This fact suggests that
all groups judge acceptable and unacceptable sentences in the two semantic conditions
in the same way. In other words, no group is affected in their rating of test sentences by
whether or not the dislocate and antecedent are in a loose semantic relationship or not.
One could say that ‘semantics’ does not interfere with ‘discourse’ in an adverse way. In
this fashion, all learner groups, including the intermediate proficiency one, demonstrate
semantic knowledge comparable to that of the native speakers (a finding consistent
with the bottleneck hypothesis; Slabakova, 2008). Our research design does not allow
us to tease apart when this knowledge appeared: we are only in a position to establish
that at the time learners have the narrow-syntactic knowledge of clitics (as ascertained
by our clitic test), they also observe the semantic freedom of CLLD. Our predictions,
Slabakova et al. 339
based on the Interface Hypothesis, that there will be an order of acquisition among
narrow-syntactic, semantic and discourse-related properties, was partially supported.
Our findings suggest the following order:13
Narrow syntactic properties ~ syntax–semantics interface properties > syntax–discourse
interface properties
The FF construction expectedly did not present many difficulties to the learners. This
is because it is similar to the English construction with the same discourse effect and
transfer from the native language can aid even the low proficiency learners into complete
acquisition. One unexpected finding was the lack of significant contrast between accept-
able and unacceptable focus sentences in the advanced group (p = .057). However, the
statistical contrast approaches significance and the behavior of the intermediate group
suggests that learners do not have real problems with this construction.
In conclusion, we set out to test the IH in the Spanish L2 grammar of English native
speakers. The participants we included in the study were already accurate on the syntac-
tic behavior of clitics. Since there is a clear trend towards acquisition of the related
interface-conditioned structures, we can conclude that the developmental predictions we
have attributed to the IH are borne out to an extent. In our experiment, both the near-
native and the advanced groups demonstrated consistent sensitivity to the discourse
information structure by accepting left- and right-dislocated objects marked by a clitic.
The intermediate group did not reveal any sensitivity to clitics marking information
structure. The fact that the syntax–discourse properties were acquired after the syntax–
semantic properties across the different groups of learners supports the assertion that
internal interfaces may be less problematic than external ones. Considering group as well
as individual results, we conclude that knowledge at the syntax–discourse interface can
be attested as early as intermediate levels of proficiency, but is more prevalent at
advanced levels. No difference between native and near-native speakers was uncovered
with respect to the CLLD construction. Together with the other studies we cited, our
findings are a technical confirmation of the current version of the IH. To repeat the cita-
tion from Sorace and Filiaci (2006: 340), ‘interface properties involving syntax and
another cognitive domain may not be fully acquirable.’ Something that ‘may not’ happen
is equally likely with the same thing happening. We have established that the opposite of
‘may not be acquired’ is true for the constructions we studied, and we argue that proper-
ties at the syntax–discourse interface may or may not be successfully acquired. However,
the usefulness of a model that predicts that certain properties may or may not be acquired
is questionable. Next, we need to look at why some constructions may and others may
not be acquired.
We should be and are mindful of the limitations some will find in our dataset,
given that our experiments use behavioral measures in an offline format. As dis-
cussed in the introduction, the latest instantiations of the IH relate the so-called prob-
lems of residual optionality at the level of near-native speakers to cognitive,
processing limitations of bilingualism that are more directly testable using online
measures which more faithfully tap processing. Future research that tests the same
structures using processing methodologies is certainly warranted. However, we
340 Second Language Research 28(3)
should also keep in mind that prior to Sorace and Filiaci (2006), all evidence used to
support the IH, which by Sorace’s (2011) own account has not changed in its tenets since
its genesis, was exclusively derived from offline measures very close to what we have
used in the present study. Indeed, it will be interesting to see to what extent learners of
these same proficiency levels, especially in the cases where they apparently perform like
the natives, actually process these properties in a native-like manner. It seems reasonable
to expect, if processing these complex constructions were subject to an insurmountable
bilingual effect, that we would see some evidence of this in offline methodologies like
ours, much like Sorace and colleagues did in their studies before 2006 and in several
studies published since this shift in methodological preference.
Acknowledgements
We are grateful to the participants in this experiment and to all our colleagues who helped us
recruit them: Joyce Bruhn de Garavito, John Grinstead, Juana Liceras, Silvina Montrul, Elena
Valenzuela, and Bill VanPatten. Thanks are also due to five anonymous Second Language Research
reviewers, who helped us improve the clarity of the presentation.
Notes
1. One reviewer objects to our adoption of López’s analysis, rather than a cartographic-type
analysis (e.g. Rizzi, 1997) by which ‘topic’ and ‘focus’ are functional heads that enter into
feature-checking relations, maintaining that doing so forces the conclusion that the properties
in question are syntax–discourse interface properties, rather than properties within the domain
of core syntax. We must disagree: an analysis that identifies certain functional heads with
discourse labels does not by fiat make these phenomena part of core syntax. Simply acquiring
the correct syntactic representation of, say, a CLLD sentence in Spanish is not sufficient. The
sentence still has to be used in a discourse-appropriate environment, and ‘topic’ and ‘focus’
can only be defined with respect to some discourse. This poses a conceptual problem for
minimalist-based analyses; the relevant features can only enter the derivation as part of a lexi-
cal item, not a constituent. Various researchers, besides López (2009), have pointed out the
problems with this (see, for example, the collection of articles in van Craenenbroeck, 2009).
2. This approach is characteristic of the seminal work of Rizzi (1997) and subsequent research-
ers, according to which ‘topic’ and ‘focus’ are both designated functional heads and features
on the corresponding syntactic constituents, which must then move to these functional heads
to satisfy criterial requirements on matching such features. See also footnote 1.
3. Examples (2) to (5) are Spanish translations/adaptations of corresponding Catalan examples
in López (2009).
4. More precisely, the feature [+c] captures the idea that the fronting operation itself (CLLD or
focus fronting) simultaneously opens up a variable position and resolves it, unlike the vari-
able position in a wh-question, which is resolved only by the following sentence, not by the
wh-question itself, as in (8) below.
5. And in fact Choi (1997), in discussing example (8), refers to this as ‘contrastive focus’, which
again illustrates the lack of agreement on the meanings of the terms ‘topic’ and ‘focus’ dis-
cussed by López (2009).
6. The two constructions are also differentiated by obligatory postverbal subjects, when these
are overt, in focus fronting vs. the possibility of pre- or post-verbal subjects in CLLD. We did
not directly test this factor in our experiment.
Slabakova et al. 341
7. We aimed at testing native speakers from a variety of countries to get away from the possible
dialectal differences in judging clitic constructions. Since we did not control for country of
origin and the groups are too small, it is impossible to ascertain now whether there are signifi-
cant dialectal differences between the native speakers from the different countries. We chose
to test speakers of different dialects because they represent whom L2 learners in a classroom
setting have as teachers, over the course of the several years they are learning Spanish.
8. The validation of the test was provided by the software of WebSurveyor. By this we mean
that we ascertained, through the software, that none of the test participants took the test twice.
The software provides the internet address of the computer used to take the test, and roughly
the location. By internal review board (IRB) regulations, we are not allowed to retain any
personal data from test participants, so we deleted the IP addresses of the participants after
we made certain they did not coincide.
9. An anonymous reviewer questions the low number of tokens per condition and the low num-
ber of fillers. We reasoned that the tested constructions were different enough from each other
so that they could serve as each other’s fillers, to an extent. Furthermore, with audio-recorded
context and two sentences to judge per context, each test item took considerable time to eval-
uate. We did not restrict the time that participants needed to evaluate a sentence. Altogether,
the main test, clitic test, and proficiency test took between 40 to 60 minutes, which we feel is
the upper limit of taxing participant attention.
10. For the intermediate proficiency group, however, the non-significant difference between the
equivalence and subset conditions is offset by the fact that they do not make any significant
distinction in the data at all.
11. Assuming, as per the discussion in Section II.3, that English preposing with focused intona-
tion is comparable to Spanish focus fronting.
12. ‘I take it that part of Sorace’s point is that the IH endeavors to account for effects that are
both long-term and residual. But one can only fully establish this by considering devel-
opment as well as ultimate attainment, indeed by making comparisons between the two’
(White, 2011b: 109).
13. We stress the fact that developmental predictions are best examined with longitudinal
data. The order of acquisition here is extrapolated from cross-sectional data, hence merely
suggested.
References
Arregi K (2003) Clitic left dislocation is contrastive topicalization. In: Kaiser E and Arunachalan S
(eds) Proceedings of the 26th Annual Penn Linguistics Colloquium. University of Pennsylvania
Working Papers in Linguistics, volume 9: 31–44.
Belletti A, Bennati E, and Sorace A (2007) Theoretical and developmental issues in the syntax
of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory
25: 657–89 .
Bialystok E, Craik FIM, Green DW, and Gollan TH (2009) Bilingual minds. Psychological Science
in the Public Interest 10: 89–129.
Bley-Vroman R (1983) The comparative fallacy in interlanguage studies: The case of systematicity.
Language Learning 33: 1–17.
Bohnacker U (2010) The clause-initial position in L2 Swedish declaratives: Word order variation
and discourse pragmatics. Nordic Journal of Linguistics 33: 105–43.
Casielles-Suárez E (2004) The syntax–information structure interface: Evidence from Spanish and
English. Oxford: Routledge.
342 Second Language Research 28(3)
Choi H-W (1997) Information structure, phrase structure and their interface. In: Butt M and King
TH (eds) Proceedings of the LFG 97 Conference. Stanford, CA: CSLI Publications, 1–16.
Chomsky N (2001) Derivation by phase. In: Kenstowicz M (ed.) Ken Hale: A life in language.
Cambridge, MA: MIT Press, 1–51.
de Swart H and H de Hoop (2000) Topic and focus. In: Cheng L and Sybesma E (eds) The first
GLOT international state-of-the-article book. Berlin/New York: Mouton de Gruyter, 105–30.
Donaldson B (2011a) Left-dislocation in near-native French. Studies in Second Language
Acquisition 33: 399–432.
Donaldson B (2011b) Nativelike right-dislocations in near-native French. Second Language
Research 27: 361–90.
Donaldson B (in press) Syntax and discourse in near-native French: Clefts and focus. Language
Learning.
Emonds J (2004) Unspecified categories as the key to root constructions. In: Adger D, de Cat C,
and Tsoulas G (eds) Peripheries: Syntactic edges and their effects. Dordrecht: Kluwer, 75–120.
Green DW (1986) Control, activation and resource. Brain and Language 27: 210–23.
Green DW (1998) Mental control of the bilingual lexico-semantic system. Bilingualism: Language
and Cognition 1: 67–81.
Gundel JK (1988) The role of topic and comment in linguistic theory. New York: Garland.
Haegeman L (2006) Argument fronting in English, Romance CLLD, and the Left Periphery. In:
Zanuttini R, et al. (eds) Crosslinguistic research in syntax and semantics. Washington, DC:
Georgetown University Press, 27–52.
Ivanov I (2009) Second language acquisition of Bulgarian object clitics: A test case for the inter-
face hypothesis. Unpublished PhD thesis, University of Iowa, Iowa City, IA, USA.
Iverson M, Kempchinsky P, and Rothman J (2008) Interface vulnerability and knowledge of the
subjunctive/indicative distinction with negated epistemic predicates in L2 Spanish. EUROSLA
Yearbook 8: 135–63.
Kroll JF, Bobb SC, Misra MM, and Guo T (2008) Language selection in bilingual speech: Evidence
for inhibitory processes. Acta Psychologica 128: 416–30.
Lardiere D (2011) Who is the interface hypothesis about? Linguistic Approaches to Bilingualism
1: 48–53.
López L (2009) A derivational syntax for information structure. Oxford: Oxford University Press.
Lozano C and Mendikoetxea A (2008) Postverbal subjects at the interfaces in Spanish and Italian
learners of L2 English: A corpus analysis. In: Gilquin G, Papp S, and Díez-Bedmar MB (eds)
Linking up contrastive and learner corpus research. Amsterdam: Rodopi, 85–125.
Lozano C and Mendikoetxea A (2010) Interface conditions on postverbal subjects: A corpus study
of L2 English. Bilingualism: Language and Cognition 13: 475–97.
Marian V and Spivey MJ (2003) Competing activation in bilingual language processing: Within-
and between-language competition. Bilingualism: Language and Cognition 6: 97–115.
Prince EF (1981) Topicalization, focus-movement, and Yiddish-movement. Proceedings of the
Berkeley Linguistics Society 7: 249–64.
Reinhart T (1995) Interface strategies. OTS working papers TS-95-002. Utrecht: Utrecht
University.
Reinhart T (2006) Interface strategies. Cambridge, MA: MIT Press.
Rizzi L (1997) The fine structure of the left periphery. In: Haegeman L (ed.) Elements of grammar.
Dordrecht: Kluwer, 281–337.
Rothman J (2009) Pragmatic deficits with syntactic consequences: L2 pronominal subjects and the
syntax–pragmatics interface. Journal of Pragmatics 41: 951–73.
Rothman J and Slabakova R (2011) The mind–context divide: On linguistic interfaces and
language acquisition. Lingua 121: 568–76.
Slabakova et al. 343
Slabakova R (2008) Meaning in the second language. Berlin: Mouton DeGruyter.
Slabakova R and Ivanov I (2011) A more careful look at the syntax–discourse interface. Lingua
121: 637–51.
Slabakova R, Rothman J, and Kempchinsky P (2011) Gradient competence at the syntax–
discourse interface. EUROSLA Yearbook 11: 218–43.
Sorace A (2011) Pinning down the concept of ‘interface’ in bilingualism. Linguistic Approaches
to Bilingualism 1: 1–33.
Sorace A and Filiaci F (2006) Anaphora resolution in near-native speakers of Italian. Second
Language Research 22: 339–68.
Trahey M and White L (1993) Positive evidence and preemption in the second language
classroom. Studies in Second Language Acquisition 15: 181–204.
Tsimpli IM and Sorace A (2006) Differentiating interfaces: L2 performance in syntax
semantics and syntax discourse phenomena. In: Bamman D, Magnitskaia T, and Zaller C (eds)
Proceedings of the 30th annual Boston University Conference on Language Development
(BUCLD 30). Somerville, MA: Cascadilla Press, 653–64.
Valenzuela E (2005) L2 ultimate attainment and the syntax–discourse interface: The acquisition of
topic constructions in non-native Spanish and English. Unpublished PhD dissertation, McGill
University, Montreal, Quebec, Canada.
Valenzuela E (2006) L2 end state grammars and incomplete acquisition of the Spanish CLLD
constructions. In: Slabakova R, Montrul S and Prévost P (eds) Inquiries in linguistic
development: In honor of Lydia White. Amsterdam: John Benjamins, 283–304.
van Craenenbroeck J (ed.) (2009) Alternatives to cartography. Berlin/New York: Mouton de
Gruyter.
Villalba X (2000) The syntax of sentence periphery. Unpublished PhD dissertation, Universitat
Autònoma de Barcelona, Spain.
Ward GL (1988) The semantics and pragmatics of preposing. Unpublished PhD dissertation,
University of Pennsylvania, Philadelphia, PA, USA.
White L (2011a) Second language acquisition at the interfaces. Lingua 121: 577–90.
White L (2011b) The interface hypothesis: How far does it extend? Linguistic Approaches to
Bilingualism 1: 108–10.
White L, Valenzuela E, Kozlowska-MacGregor M, and Leung YK (2004) Gender and number
agreement in nonnative Spanish. Applied Psycholinguistics 25: 105–33.
