Académique Documents
Professionnel Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/263806929
CITATIONS READS
0 31
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Becky Halloran on 21 November 2016.
1. Introduction*
*
Becky Halloran, University of Florida, beckyhalloran@ufl.edu; Jason Rothman,
University of Reading. Thank you to Mike Iverson for his help and contributions as well
as the audience at BUCLD 37. Any and all errors are inadvertent and are completely our
own.
1
Here we assume that in Spanish (and other Romances languages) clitics head their own
functional projections (see footnote 2). We also assume, following traditional notions
within Minimalism, that the difference between Spanish and English here entails a formal
syntactic (uninterpretable) feature difference between the two languages. Whether this
feature is understood as a bona fide feature in the sense of Chomsky, or, in the view of
cartographic proponents, as a position in the syntactic structure (i.e. a functional head that
projects), is not important for our argumentation, and so we leave this issue aside.
infrequent in comparison to clitic solidarity as the ‘elsewhere’ condition.
Interestingly, grammatical exceptions to clitic solidarity, as just described, fall
outside general pedagogical rules on Spanish clitic placement, which
inaccurately maintain that multiple clitics always pattern together.
In this study, we test L2 Spanish knowledge of restrictions on clitic
solidarity; specifically, of argument structure constraints on the distribution of
multiple clitics. As we will see from the pattern of cross-sectional data that we
present, only the advanced learners as a group and some intermediate learners
show such knowledge despite the fact that the argument structure of ECM verbs
does not differ across English and Spanish. Our data show that the beginning
learners and some intermediate learners allow for multiple clitics in disaccord
with ECM argument structure, suggesting a more complex or nuanced account is
required if we do not want to abandon the possibility of L1 transfer of argument
structure.
We will argue that the L2 learners for which ECM argument structure is not
a factor for judgments on multiple clitic placement are simply following the
overgeneralized pedagogical rules; not necessarily because argument structure
does not transfer initially, but rather because the syntax of clitics is not yet part
of their interlanguage grammar. In other words, even if, as we suspect, the
argument structure for perception verbs has been transferred from English, these
learners cannot apply consistently the syntactic reflexes of such transfer to
something they have not yet truly acquired; namely, clitics and their placement.
Instead, they default to the overgeneralized pedagogical rules. Conversely, we
will argue that the learners for whom the ECM argument structure is
deterministic for their L2 judgments on the (im)possibility of clitic solidarity do
so precisely because the syntax of clitics is now in place in their interlanguage
grammar, leaving them no other recourse but to project beyond and even in
conflict with what is taught about clitic solidarity.
In sum, we will argue that the whole of the pattern revealed across the three
L2 proficiency groups examined suggests that the underlying syntax of clitics is
acquirable in adulthood by speakers whose L1s lack them. This is consistent
only with Full Accessibility accounts to UG-continuity in adulthood. Assuming
that verbal argument structure is transferred from the L1 and considering that
English argument structure of ECM verbs should also require the breaking of
clitic solidarity, we will discuss how these data highlight that not all transferred
properties can be applied at the beginning stages of L2 acquisition, specifically
when what is at stake is not the same across the two languages. Based on the
data presented here, we maintain that the ECM breaking of clitic solidarity will
only obtain with the consistency of native-conformity at a proficiency level
where the L2 grammar has a mental representation for clitics. This seems to be
possible at an advanced level of L2 proficiency.
2. Clitics
2.1 Clitics in Spanish
Clitics are weak pronouns that are independent morphemes, but are
phonologically dependent on a strong phrasal element; typically, a verb.
Essentially, this means they cannot exist independently or replace a lexical DP.
While various syntactic analyses for the syntax of clitics in Romance languages
have been presented, for the purposes of this study, we adopt Masullo’s (2004)
Minimalist analysis 2 , assuming Merge of the clitic and its argument with
movement caused by feature percolation. This analysis, in particular, provides
explanatory coverage on the restrictions on Merge and/or Movement that give
rise to the specific properties we examine herein.
Previously, we mentioned that the learning task for the
L1English/L2Spanish learner involves the acquisition of a new syntactic feature
as well as the distribution of the Spanish clitic system. This distribution, in what
we will refer to as ‘traditional’ constructions, is dependent on tense and
predicate complexity (see table 1). The distributional pattern (finite= proclisis,
non-finite= enclisis, finite+non-finite= 3seemingly optional proclisis or enclisis)
is language-specific and therefore must be ‘learned’ in the true sense of the
word.
2
Other viable syntactic analyses exist, notably those of Sportiche (1996) and Uriagereka
(1995). Both of these accounts are compatible with our analysis of the restrictions on
clitic solidarity; however, we adopt Masullo’s (2004) approach, as it is more current and
generally more explanatory.
3
We acknowledge lexical biases for proclisis or enclisis preference, e.g. Ir+a+infinitive
favors proclisis while preferir+ infinitive prefers enclisis, yet both are grammatically
available.
of Spanish as a second language, and is used without any reference to (possible)
exceptions in teaching clitic placement in the classroom. However, clitic
solidarity does not hold with particular periphrastic constructions; for example,
those with matrix ECM verbs, in which the two clitics are both accusative
arguments, each projected by one of the two verbs (see figure 1).
Due to the unique argument structure of ECM verbs, in these constructions
each clitic must first merge with the verb that selects it as an argument, forcing
the clitics to separate, and blocking feature percolation. In aux/modal+inf
constructions, the lexical verb projects both arguments and feature percolation is
possible. As a result, both proclisis and enclisis are available as grammatical
options as long as, in accord with clitic solidarity, the clitics stay together. In
traditional periphrastic constructions, (aux/modal+inf), the subject of the matrix
verb also serves as the subject for the embedded verb whereas this is not the
case in ECM constructions. ECM verbs require different subjects for the matrix
and embedded verbs, wherein the subject of the embedded verb must share
identity with the accusative object of the matrix verb (i.e., PRO is co-referential
with the object of the perception verb) (see (1-2)).
3. Prior Studies
The current study builds on these earlier works, examining the phenomenon
of clitic solidarity and syntactic restrictions to its application with a particular
subgroup of ECM verbs; namely, verbs of perception. The purpose of testing
the acquisition of these verbs is to determine whether L2 learners of Spanish
show sensitivity to the syntactic restrictions placed on clitic solidarity in these
constructions, despite their relative infrequency and overall inconsistency with
general pedagogical rules. We have two principal research questions:
Here we emphasize that several variables are at stake when considering the
learning task for the L1 English speaker. First, in order to have a target
representation of clitics, the learner must acquire a new functional category as
well as the features associated with this category4. At the same time, the L2er
must learn both the morphophonology of the clitic forms as well as the
distributional pattern of clitics in Spanish, which, being language-specific, do
not pertain directly to the underlying syntax of clitics. These pieces of the
learning task, therefore, involve learning in the true sense of the word and
knowledge of these aspects alone could be attributed to frequency in the input
and/or explicit instruction, as we know that clitic forms and the traditional
pattern of distribution are taught explicitly (even overgeneralized) in all second
language classrooms.
The restrictions placed on clitic solidarity in ECM constructions, however,
are universal rather than language-specific, and should theoretically fall out
from knowledge of both the argument structure of the lexical verbs at stake and
the underlying syntax of clitics. To be clear, in this case learners simply cannot
rely on instruction (as general pedagogical rules do not include these exceptional
cases) or frequency in the input (as these constructions, both in type and in
number, are much less frequent than ‘traditional’ periphrastic constructions).
4
Following Minimalism, differences across the world’s languages of this type, or
parametric differences, derive from variation in feature specifications or instantiation
across languages. As a result, English and Spanish must have some featural difference
for pronominal objects within the syntax proper, resulting in the former having strong
pronouns and the latter having clitics. Whatever the specifics of these features turn out to
be is precisely what the English natives need to acquire at the level of new L2 features.
Following approaches that claim that clitics are heads of their own functional projection,
this means that some uninterpretable featural difference between English and Spanish
obtains, whereby English lacks the feature that Spanish has that generates object clitics.
We could stipulate that there is an uninterpretable +clitic feature, and could end the story
there. Whether or not there is a clitic feature, or, the difference between English and
Spanish is better explained by another feature to be discussed in future syntactic theory,
is of little consequence for the present study. This is true since whatever the specific
feature turns out to be, it will be the case that Spanish has a feature that English natives
do not have in their L1 and must be acquired in the course of L2 Spanish interlanguage
development to attain a mental representation like that of native speakers.
Assuming full transfer models, the argument structure of the verbs would
transfer from English, as it is the same in both languages; however, clitics
cannot be transferred. Thus, despite the fact that the argument structure of the
lexical verbs may transfer from English, we hypothesize that learners must have
truly acquired the underlying representation of the clitics themselves in order to
consistently apply argument structure restrictions accordingly. In fact, since the
same logic holds for restructuring and causative contexts as well, what we
maintain here as an explanation would also account for the problems noted in
these specific contexts in previous research as early as Liceras (1985) and
Duffield and White (1999). In accord with this line of reasoning, demonstration
of accurate L2 clitic placement in ECM structures (and restructuring and
causative contexts) could serve as a better diagnostic for true acquisition of
clitics, precisely because knowledge of (production, comprehension , judgment
or all three) clitic climbing in finite and non-finite contexts could obtain from
both general frequency and/or explicit instruction.
Based on evidence in the literature, (Liceras, 1985; Bruhn-Garavito &
Montrul, 1996; Duffield & White, 1999), we predict that intermediate and
advanced L2 learners will display target-like knowledge of clitic placement in
‘traditional’ constructions, although we cannot deny that such knowledge might
be attributable to learning as opposed to acquisition in the way described above.
Based on previous evidence of L2 problems with clitic placement in causatives
and restructuring constructions, combined with the logic that knowledge of the
universal restrictions they (and ECM) place on clitic solidarity are a better
diagnostic of underlying representation, we can argue that clitics are late
acquired. In line with full UG-accessibility, we anticipate that sensitivity to the
restrictions placed on clitic solidarity in ECM constructions, despite instruction
to the contrary and limited input, will eventual be demonstrable in later stages of
L2 development.
5. Methodology
6. Results
4
4
3.5
3.5 3
3 2.5
Beginner
2.5 2 Beginner
2 Intermediate 1.5 Intermediate
1.5 1
1 Advanced Advanced
Native Control
Native Control
Figure 3 shows the results of all groups with all clitic placement
possibilities in ECM constructions 5 . For the beginner group, there are no
significant differences across the ECM constructions based on clitic position,
suggesting that although learners in this group appear to display target-like
knowledge of clitic placement in both traditional and ECM constructions when
considering their judgments of the grammatical placement tokens alone, they do
not make a clear distinction between grammatical and ungrammatical placement
in the case of the ECM constructions. The intermediate group makes a
distinction between all three placement options included in the analysis -
*enclisis/*proclisis/pro-endoclisis - (p<0.001). The advanced group makes a
distinction between *enclisis and the remaining clitic placement options
(p<0.001), and the native control group distinguishes between all three (p <
0.031).
4
3.5
3
Beginner
2.5
2 Intermediate
1.5
1 Advanced
Native Control
5
Constructions with mesoclitic placement were not included in any of the statistical
analyses discussed here, although they appear in the figures.
in modal+inf constructions. Following the argumentation we laid out above,
these are the crucial comparisons. Pairwise comparisons returned no significant
differences across groups in ECM constructions with pro-endoclisis, suggesting
that all groups show target-like knowledge of clitic placement in ECM
constructions, where clitic solidarity is broken. However, in modal+inf
constructions with *pro-endoclisis, the native and advanced groups perform
significantly differently than the beginners and intermediates (p<0.007). These
two groups (advanced and natives) reject the ungrammatical placement, whereas
the beginner and intermediate groups do not appear to do so.
So, while all groups seemingly accept pro-endoclisis in ECM constructions,
only the advanced experimental group is actually sensitive to the restrictions
placed on clitic solidarity exclusively in ECM constructions as this group
displays target-like knowledge in both modal+inf and ECM constructions,
making a clear distinction between the two based on argument structure
interactions with clitic distribution. As we are assuming full transfer of the
verbal argument structure in all experimental groups, we argue that the
distinction seen here between the performance of the advanced experimental
group and that of the beginner and intermediate groups is related to the
acquisition of the underlying syntactic representation of the clitics themselves.
4
3.5
3 Beginner
2.5 Intermediate
2
1.5 Advanced
1 Native
Finite *pro- Modal + nf ECM pro-
endoclisis *pro-endoclisis endoclisis
7. Conclusion
The data presented here provide clear evidence to answer our two principal
research questions. Our first question regards the learners’ ability to acquire
knowledge of clitic placement in traditional constructions. Our results show that
all experimental groups display target-like knowledge of clitic placement in
grammatical constructions across construction type. These results confirm those
of previous studies (Liceras, 1985; Bruhn-Garavito & Montrul, 1996; Duffield
& White, 1999) which purported that intermediate and advanced L2 learners are
able to acquire clitics and clitic placement knowledge in traditional
constructions. Further, our results show that even beginner L2 learners display
some knowledge of clitic placement in finite and modal+inf constructions.
While these data are consistent with prior results and suggest that L2 learners
are able to perform with native-like accuracy, this accuracy may be due, at least
in part, to explicit classroom instruction, true ‘learning’ of the language-specific
distribution rules, and high frequency in the input.
Our second question regards the learners’ ability to place clitics in ECM
constructions when universal restrictions apply, due to the argument structure of
the lexical verbs. Our data show that advanced L2 learners make a clear
distinction between ECM constructions and traditional modal+inf constructions
when faced with the pro-endoclisis clitic position, accepting pro-endoclisis
placement in ECM constructions where syntactic restrictions do not allow for
clitic solidarity, and, crucially, rejecting pro-endoclisis placement in modal+inf
constructions where clitic solidarity holds based on placement distribution
specific to Spanish. It appears that the beginner and intermediate groups,
despite displaying target-like knowledge of clitic placement in the grammatical
categories, do not make this distinction. This is consistent with previous
findings for causative and restructuring contexts by Liceras (1985) and Duffield
and White (1999). Since this distinction represents the knowledge that clitic
solidarity is highly constrained to very specific environments and is not
generalizable, we have argued throughout that this distinction should be the
measure through which the L2 representational status of clitics is assessed.
In summary, the data show that advanced learners are the only group truly
sensitive to the inherent restrictions placed on clitic solidarity (despite ambiguity
in the input, low frequency, and instruction to the contrary). Their judgments on
the placement of clitics in the crucial contexts are strong evidence to suggest
that they have successfully acquired the underlying syntax of clitics in Spanish.
This is to say, in the case of the advanced learners, they are truly treating clitics
as clitics. Without being able to rely on ‘rules’ of placement (which is possible
in the traditional constructions), these learners must have acquired an underlying
representation of clitics in order to demonstrate, beyond what they are taught,
how clitics interact in a native-like manner with the particular verbs at play. As
such, these results strongly support Full Access approaches to UG in adulthood.
References
Araya, Guillermo., Henk Haverkate, & Kitty van Leuven. (Eds.). (1980). Los clíticos en
el español actual. Amsterdam: Editorial Rodopi.
Castillo, Concha. (2001). The Configuration of ECM structures. Studia Linguistica, 55
(2), 113-139.
Duffield, Nigel & White, Lydia. (1999). Assessing L2 knowledge of Spanish clitic
placement: converging methodologies. Second Language Research, 15, 2, 133- 160.
Gass, Susan M. & Selinker, Larry. (2008). Second language acquisition: An introductory
course (Third Edition). Hillsdale, NJ: Lawrence Erlbaum Associates.
Hawkins, Roger, & Hattori, Hajime. (2006). Interpretation of English multiple wh-
questions by Japanese speakers: A missing uninterpretable feature account.
Second Language Research, 22(3), 269-301.
Liceras, Juana. M. (1985). The value of clitics in non-native Spanish, Second Language
Research, 1(2), 151-168.
Martins, Ana Maria. (2006). Aspects of Infinitival Construction in the History of
Portuguese. In Randall S. Gess & Deborah Arteaga (Eds.), Historical Romance
Linguistics (pp. 327- 355). John Benjamins.
Masullo, Pascual J. (2004). Clitics aren‘t Climbers! (Unpublished manuscript).
University of Pittsburg, Pittsburg.
Montrul, Silvina. (2004). The Acquisition of Spanish: Morphosyntactic development in
monolingual and bilingual L1 acquisition and adult L2 acquisition. Philadelphia,
PA: John Benjamins B.V.
Park, Insim. (2004). A study on ECM Infinitive Structure. Studies in Generative
Grammar, 14, 4, 497-517.
Rothman, Jason, Tiffany Judy, Pedro Guijarro-Fuentes, & Acrisio Pires. (2010). On the
(Un)-Ambiguity of Adjectival Modification in Spanish DPs. Studies in Second
Language Acquisition, 32, 47-77.
Sportiche, Dominique. (1996). Clitic constructions. In J. Rooryck and L. A. Zaring
(Eds.), Phrase structure and the lexicon (pp. 213-276). Dordrecht: Kluwer.
Schwartz, Bonnie D. & Sprouse, Rex A. (1996). L2 cognitive states and the Full
Transfer/Full Access model. Second Language Research 12: 40-72.
Thomas, Danielle. (2012). Grammatical Optionality and Variability in Bilingualism: How
Spanish-English bilinguals limit clitic-climbing. Dissertation: The University of
Toronto.
Tsimpli, Iianthi M., & Dimitrakopoulou, Maria. (2007). The interpretability hypothesis:
Evidence from wh-interrogatives in second language acquisition. Second Language
Research, 23(2), 215-242.
Uriagereka, Juan. (1995). Aspects of the syntax of clitic placement in western Romance.
Linguistic Inquiry, 26(1), 79-123.
White, Lydia. (2003). Second Language Acquisition and Universal Grammar.
Cambridge: Cambridge University Press.