Pronominal mental representations in advanced L2 and L3
learners of Spanish1
Cristóbal Lozano
University of Essex
[email protected]
Abstract
Evidence in recent second language research (SLR) leads to an apparent
contradiction. While some studies claim that advanced adult learners can indeed
achieve native-like competence, other studies suggest that they only achieve nearnative competence. The former studies focus on universal principles of universal
grammar (UG), whereas the latter investigate properties which UG allows to vary
(within limits) and attribute lack of native-like competence to L1 influence on the L2.
In this study we investigate whether this is the expected pattern, i.e., that advanced L2
speakers will always show native-like competence where principles are involved, yet
near-nativeness where the L1 differs parametrically from the L1 with respect to
functional features.
An experimental study consisting of three groups ((i) English and (ii) Greek natives
learners of Spanish as L2 and L3 respectively and (ii) Spanish natives) were tested on
some pronominal constructions where the presence of overt and null pronouns is
constrained by universal principles and language-specific/discursive constraints.
Results show that all learners behave similarly with respect to the universal
constructions, as expected. With discursive constructions, the Greek group behaves
similarly to the Spanish group as there is no parametric variation between Greek and
Spanish with respect to pronominal constraints. Crucially, although the English group
also behaves similarly to the Spanish group, their intuitions are statistically divergent
from those of Spanish natives. This is expected if L1-L2 parametric differences
obtain.
1. Introduction
Current approaches to post-childhood SLR are faced with three well-attested
problems (Hawkins, 2001a):
(i)
(ii)
1
Why is knowledge of language underdetermined by the input? That is, why do
learners seem to be sensitive to constructions they have never encountered in
the input?
Why is not some input properly represented? That is, why do learners seem
not to be sensitive to constructions they have abundantly been exposed to?
An earlier version of this paper was presented at the Second Language Acquisition Discussion Group
(University of Essex, November 2001) and at the Congreso internacional sobre nuevas tendencias de
la lingüística (Universidad de Granada, November 2001). I am grateful to colleagues there for their
comments. This research is funded by the ESRC, award no. R00429934281.
(iii)
Why does the L1 influence the L2 in some cases but not others? That is, what
linguistic areas are more vulnerable to L1 influence in second language
acquisition (SLA)?
Two intriguing (and somewhat contradictory) findings pervade recent SLR. Some
studies report that at advanced levels of proficiency post-childhood learners can
achieve native-like competence, showing similar mental representations to adult
native speakers (observation i above). However, other studies show that learners only
achieve near-native competence despite long immersion in the target language
(observation ii) due to L1-L2 differences (observation iii).
The first observation (native-like representations) concerns constructions that are
claimed to be part of Universal Grammar (UG) and that are typical ‘poverty-of-thestimulus’ cases (i.e., they are not present in the input the learners are exposed to).
Such constructions are universal invariants and learners show sensitivity to them,
even though they may not necessarily be instantiated in their L1. This is the case of
pronominal constructions in L1 English – L2 Japanese (Kano, 1997; 1998; Marsden,
1998) and L1 English – L2 Spanish (Pérez-Leroux & Glass, 1997; 1999).
The second observation (near-native representations) comes from a series studies
claiming that adult language learners will show persistent difficulty with L2
functional features that are not instantiated in their L1. This is the case even after long
immersion/exposure to the L2 (Franceschina, 2001; Hawkins, 2000; Hawkins &
Chan, 1997). The L1 is claimed to be responsible for such L2 representational
deficits, causing learners to show divergent representations from native speakers
(Sorace, 1993) and leading to persistent selective fossilisation (Hawkins, 2000).
This study deals with how the presence of overt and null pronominal subjects in
Spanish is determined by (i) universal constraints and (ii) language-specific/discursive
constraints governened by functional features. An experiment was conducted to test
advanced adult learners sensitivity to the two constraints. Results support the findings
of a number of previous (and somewhat unrelated) studies –learners show native-like
mental representations when universal constraints are involved, whereas they show
native-like mental representations with discursive constraints.
2. Distribution of overt and null pronouns in native Spanish
It is well known that in null-subject languages like Spanish and Greek, overt
pronominal subjects (e.g., él ‘he’, ella ‘she’; aftos ‘he’, afti ‘she’) can be phonetically
either overt, (1a) and (2a), or null, (1b) and (2b) in finite clauses. However, non nullsubject languages permit only overt pronouns (3a).
(1)
(2)
(3)
a. Él/ella tiene poco dinero
b. pro tiene poco dinero
a. Aftos/afti ehei liga lefta
b. pro ehei liga lefta
a. He/she has little money
b.* pro has little money
(Spanish)
(Greek)
(English)
Rizzi (1997:273) argues that the licensing of pro occur under agreement or
government1. Consider (4), where pro occupies [Spec, TP], which is governed by the
head T. The [+D] and [+AGR]2 features of the functional head T licence the null
pronoun3.
(4)
TP
Spec
pro
T’
T
VP
tiene
poco dinero
Whether a head can license pro is a language-specific property (i.e., amenable to
parametrisation). In Spanish and Greek, T is a proper licensor, whereas in English is
not.
Overt/null pronominal alternations, however, are not in free alternation in native
Spanish. There are at least two constraints regulating their co-occurrence, namely, the
overt pronoun constraint (OPC) and the null pronoun constraint (NPC).
2.1. The overt pronoun constraint (OPC)
In null subject languages, an embedded overt pronoun in subject position cannot be
bound by a quantified expression (Montalbetti, 1984, 1986). Consider (5a) and (5b) in
isolation, where the overt pronoun él/aftos ‘he’ and the null pronoun pro can be in
free alternation. However, the preceding context disallows él/aftos to take the bound
variable reading, i.e., the overt pronoun cannot be bound by a quantified determiner
phrase (QDP) cada estudiante/o kathe mathitis ‘each student’. Only pro can act as a
bound variable here. Note that in the English equivalent (5c) the QDP each student
can indeed bind the overt pronoun he as there is no overt/null alternation in English
(i.e., pro is not allowed in English).
(5)
Context: The government has published a report about students’ financial
situation. The report concludes that...
a. cada estudiantei dice que {*éli/proi} tiene poco dinero. (Spanish)
b. o kathe mathitisi lei {*aftosi/proi} ehi liga lefta.
(Greek)
(English)
c. each studenti says that {hei/*proi} has little money.
The OPC phenomenon is well attested not only in Spanish (e.g., Luján (1999),
Montalbetti (1984, 1986)), but also in typologically-related languages, such as
Portuguese, Italian and Greek (Montalbetti, 1984), Catalan (Picallo, 1994) and in
typologically-unrelated languages like Chinese (Xu, 1986), Korean (Kanno, 1997,
1998), Arabic-Tarifit (Ouhalla, 1988). For expository purposes, we will not present
evidence from these languages. These studies argue (and some only suggest) that the
OPC is an invariant principle of UG4.
Further evidence for the claimed universality of the OPC comes from learnability
theory. OPC constructions show a very low frequency in the input5. They represent a
typical case of a ‘poverty of the stimulus’ phenomenon, since the ungrammatical
construction [QDPi … *overti] is not present in the Spanish input. It is widely
acknowledged that input in the form of positive evidence alone does not contain
ungrammatical expressions. Therefore, OPC knowledge must be part of UG principles
(see Pérez-Leroux & Glass (1997) and Schwartz (2000) for discussion).
The universality of the OPC is therefore a testing-ground in non-native language
acquisition for the claim that UG constrains adult L2 grammatical mental
representations.
2.2. The null pronoun constraint (NPC)
In contast to (5), the context in (6) biases for an interpretation where the overt
pronoun él/aftos ‘he’ is correferential with one of the previous discourse antecedents
(Mr López), and not with the QDP (cada estudiante/o kathe mathitis).
(6)
Context: Mr Lópezj and Ms Garcíak work at the university and at a famous
publishers. However...
a. cada estudiantei dice que {élj/*proj} tiene poco dinero. (Spanish)
(Greek)
b. o kathe mathitisi lei {aftosj/*proj} ehi liga lefta.
c. each studenti says that {hej/*proj} has little money.
(English)
The inclusion of two referential antecedents (which stand in contrast in the discourse)
requires one of them to be focused contrastively, hence the obligatory presence of the
overt pronoun él/aftos ‘he’ (or ella/afti ‘she’, depending on whether we want to
contrastively focus on Mr López or Ms García). From an informational point of view,
overt pronouns are feature specified as [+focus] in NPC contexts, whereas null
pronouns are [–focus].
There is disagreement in the literature about the structure and locus of a focus
projection (for a discussion, see Kiss, 1995). However, all proposals suggest that a
Focus Phrase (FP) is projected above TP. López (1999) suggests that FP is headed by
F, which is specified as [+focus]. This feature triggers movement of the focalised
element for checking purposes.
In the case of Spanish, a contrastively focused overt pronoun like él raises from [Spec,
TP] to [Spec, FP] to check its interpretable [+focus] feature with the uninterpretable
[+focus] counterpart in F6. However, null pronouns in Spanish are specified for [–
focus], which amounts to saying that a null focused pronoun pro is ungrammatical in
[Spec, FP], given that it cannot check its features there, as shown in (7).
FP
Spec
él/*pro
[+focus]/[-focus]
F’
F
[+focus]
TP
Spec
T’
T
VP
tiene
poco dinero
(7)
The linguistic literature does not report NPC cases as being part of a universally
invariant principle, in contrast to OPC cases. The use of an overt (vs null) pronoun in
contrastive environments is (i) language-specific and amenable to parameterisation
(different languages parameterise this contrast with different mechanisms (PérezLeroux & Glass, 1997, 1999; Luján, 1987) and (ii) discourse-dependent, as its
appearance depends on the status of the discourse referents.
To summarise our discussion so far, OPC contexts disallow an overt pronoun (8a) but
allow a null pronoun (8b), whereas the opposite holds true for NPC contexts
(9a,b).
(8)
OPC contexts:
a. [QDPi … *overti]
b. [QDPi … nulli]
(9)
NPC contexts:
a. [QDPi … overtj]
b. [QDPi … *nullj]
3. Research questions
As pointed earlier, there are two expected outcomes with respect to the OPC and
NPC.
(i)
(ii)
Research question 1: where universal principles like the OPC are involved,
learners are expected to show native-like knowledge by behaving similarly to
natives. This is irrespective of the learners’ L1, as knowledge of universal
principles is claimed to be constrained by UG (Montalbetti, 1984, 1986;
Marsden, 1998; Pérez-Leroux & Glass, 1997, 1999). If this is the case, L1
transfer can be discarded as the possible source of OPC knowledge.
Research question 2: where language-specific constrains like the NPC are
involved, learners are expected to show sensitivity to them providing there is
no parametric differences between the learner’s L1 and L2 with respect to the
functional features under investigation. If the L1-L2 pairing does not match,
we expect learners to show representational deficits (i.e., to diverge from
natives’ knowledge and show near-native mental representations. If this is the
case, L1 can be regarded as the possible source of L1 transfer.
Hawkins (2000) argues that representational deficits (i.e., native-like representations)
in end-states appear whenever these three conditions are met simultaneously:
1.
2.
3.
if the L1 does not instantiate the uninterpretable functional feature present in
the L2;
if the learner has been exposed to the L2 after puberty;
if the learner has received abundant input in the form positive evidence
regarding the construction under investigation.
To test the predictions (i and ii) under Hawkins (2000) conditions (1 to 3), we
conducted an experimental study involving three groups of subjects.
4. Subjects
The first group (control group) consisted of 11 Spanish native speakers. The second
group (experimental group) consisted of 20 English natives learning Spanish as an L2.
The third group were 20 Greek natives learning Spanish as an L3 (they had previously
learnt English as an L2).
Subjects were administered a proficiency test7. Only learners with a proficiency level
of ≥80% (advanced) were included in the study.
5. Method
An acceptability judgement test (AJT) was used. Subjects had to judge whether a
given sentence was more or less acceptable (as opposed to grammatical). Each
stimulus consisted of a contextualising sentence (10) and two target sentences, (10a)
and (10b).8 Each target sentence was accompanied by a 5-point Likert rating scale.
Value +2 corresponded to completely acceptable and value –2 completely
unacceptable.
(10)
El señor López y la señora García trabajan en la universidad y en una famosa
editorial. No obstante…
(a) cada estudiante dice que él tiene poco dinero. –2 –1 0 +1 +2
(b) cada estudiante dice que tiene poco dinero. –2 –1 0 +1 +2
The test also controlled for other unwanted variables9
The completed and usable tests were finally coded in the statistical package Excel
(version 97) and analysed in SPSS (version 9.0).
6. Results
The one-sample Kolmogorov-Smirnov test revealed that each groups is normally
distributed (p>0.05 for each group in each condition). Inferential test were
subsequently applied (multiple analysis of variance with post-hoc LSD tests).
6.1.OPC results
The bar chart in Figure 1 shows the acceptance rates on the Likert scale (y axis) of the
three groups (x axis) for the grammatical condition (light bars) and ungrammatical
condition (dark bars).
Judgement scale (mean acceptance rate)
+2
+1
0
-1
*[QDPi...OVERTi]
-2
[QDPi...NULLi]
English
Greek
Spanish
Group
Figure 1: OPC results: acceptance rates of overt/null pronoun
Within-group analyses reveal that each grup clearly distinguishes between the
grammatical dondition [QDPi … nulli], which is highly accepted, and the
ungrammatical condition [QDPi … overti], which is rejected. A manova with post-hoc
LSD confirms this, given that each pairing (grammatical vs. ungrammatical) is
statistically significant for each group (p<0.01). This suggests that Spanish natives as
well as learners discriminate between the grammatical and the ungrammatical
constructions, since all groups disfavour an overt pronoun interpretation (hence the
negative values) and prefer a null pronoun interpretation (positive values), as Error!
Reference source not found. shows. This is what the OPC theory predicts.
In the between-group comparisons, each group of learners was compared against the
Spanish native performance. Results revealed the following:
(i)
For the grammatical construction (null pronoun), there are no significant
differences between each group of learners and the Spanish natives (p>0.05
for all comparisons).
(ii)
For the ungrammatical construction (overt pronoun), there are no significant
differences between each group of learners and the Spanish natives:
English = Spanish
(p=0.738)
Greek = Spanish
(p=0.847)
These between-group comparisons indicate that neither the English speakers nor the
Greek speakers statistically differ from Spanish natives. This suggests that all learners
are sensitive to the OPC, whatever their L1.
6.2.CFC results
In the NSC bar chart (Figure 2), the dark bars now represent the grammatical
condition with the overt pronoun, whereas the light bars represent the ungrammatical
null pronoun.
Judgement scale (mean acceptance rate)
+2
+1
0
-1
[QDPi...OVERTj]
-2
*[QDPi...NULLj]
English
Greek
Spanish
Group
Figure 2: NPC results: acceptance rates of overt/null pronoun
A within-group analysis reveals that each group distinguishes the grammatical vs.
ungrammatical condition, following the Spanish pattern (manova with post-hoc
Tukey, p<0.01 for each comparison).
Between-group comparisons were performed. Each group of learners was compared
against the Spanish natives’ performance. Comparisons revealed the following:
(i)
In the grammatical construction, (overt pronoun), there were no significant
differences between groups (p>0.05 for all comparisons).
(ii)
In the ungrammatical construction, (null pronoun), there was a significant
difference between English-Spanish but no significant difference between
Greek-Spanish:
English ≠ Spanish
(p=0.011)
Greek = Spanish
(p=0.844)
7. Discussion
The OPC results show that both Greek and English advanced learners of Spanish are
sensitive to the OPC to the same statistical extent as Spanish native are. Learners
correctly reject as ungrammatical the use of an overt pronoun as a bound variable,
[QDPi … *éli] and they correctly accept the grammatical null counterpart [QDPi …
proi]. The source of knowledge to this phenomenon could derive from:
(i)
L1 transfer in the case of the Greek speakers, as Spanish and Greek do not
differ parametrically with respect to instantiations of the OPC. But, crucially,
this explanation is untenable in the case of the English speakers as OPC
phenomena are not instantiated in native English (Kanno, 1997, 1998;
Marsden, 2001; Pérez-Leroux & Glass, 1997, 1999).
(ii)
(iii)
positive evidence in the input. However, as we argued above, this is not the
case as ungrammatical constructions [QDPi …*éli] are not present in the input
the learners are exposed to (Schwartz, 2000).
classroom instruction. However, OPC constructions are neither explained in
the classroom nor explained in textbooks (Marsden, 1998; Kanno, 1997,
1998).
It seems plausible to propose that the source of knowledge in cases where universal
principles are involved stems from UG. This finding answers question i posed in the
introduction: learners are sensitive to universal constructions like the OPC, even
though they are absent in the input and (sometimes) not instantiated in their L1. Other
studies also support this proposal (e.g., Kanno, 1997, 1998; Marsden, 1998; PérezLeroux & Glass, 1997; 1999).
The NPC results indicate that both Greek and English speakers behave similarly to
Spanish natives. Learners are sensitive to the grammaticality of an overt pronoun
bound by an antecedent in the previous discourse, thus accepting [QDi … élj] and
rejecting [QDi … *proj]. But, crucially, English natives statistically differ from
Spanish natives in the ungrammatical condition, i.e., some English speakers accept a
null pronoun in contrastively focused environments. This interpretation, [QPi …
*proj], is infelicitous in native Spanish, as explained earlier.
NPC contexts are cases where an overt pronoun is needed for contrastive focus
purposes. An overt pronoun like él can be licensed in a focus phrase [FP] above TP as
it is specified for [+focus]. However, a null pronoun cannot be in a focus phrase [FP]
as it is specified for [–focus]. (11) schematises the discussion so far.
(11) Pronominal mental representation in NPC contexts
Phonetic Informational Phonetic
Group
status
status
realisation
(a) [+overt] [+focus]
élj
Spanish, Greek, English
(b) [–overt] [–focus]
proj
Spanish, Greek, English
(c) [–overt] [+focus]
*proj
English
In (11a,b), we can observe that learners’ mental representation of overt and null
pronominals are identical to Spanish natives’. This is expected at advanced levels of
proficiency. However, as the contrast in (11b,c) shows, the grammar of English
natives is characterised by optionality10 in the informational status of null pronouns,
which can be (correctly) interpreted as unfocused and, in some cases, (incorrectly)
interpreted as focused11. This finding answers question ii in the introduction: some
learners are not sensitive to constructions that abound in the input. This is the result of
L1 influence. Question iii also receives a straightforward answer: linguistic areas like
functional features can be amenable to representational deficits if the L1-L2 differ
parametrically.
Future research will need to address the issue of informational and discursive
constraints in L2 acquisition.
8. Conclusion
This study has examined a pronominal construction in post-childhood Spanish
acquisition by English and Greek natives. The first construction (OPC) is constrained
by universal principles and the second construction (NPC) is constrained by
discursive, language-specific constraints. Learners’ mental representational are
identical to Spanish natives’ with respect to universal properties, but differ when L1L2 functional features mismatch.
9. References
Allan, D. (1992). Oxford Placement Test. Oxford: Oxford University Press.
Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press.
Cowart, W. (1997). Experimental Syntax: Applying Objective Methods to Sentence
Judgments. Thousand Oaks, CA.: SAGE.
Franceschina, F. (2001). Morphological or syntactic deficit in near-native speakers?
An assessment of some current proposals. Second Language Research 17. 213147.
González, M., Kettle, L., and Placencia, M. E. (1995). Sueños World Spanish.
London: BBC books.
Hawkins, R. (2000). Persistent selective fossilisation in second language acquisition
and the optimal design of the language faculty. Essex Research Reports in
Linguistics, 34. 75-90.
Hawkins, R. (2001a). Implications of the ‘Minimalist Program’ for theories of postchildhood SLA. Paper presented at EUROSLA 11,University of Paderborn
(Germany).
Hawkins, R. (2001b). The theoretical significance of Universal Grammar in second
language acquisition. Second Language Research, 17(4): 345-367.
Hawkins, R. and Chan (1997). The partial availability of Universal Grammar in
second language acquisition: the ‘failed functional features hypothesis’. Second
Language Research 13. 187-226.
Kanno, K. (1997). The acquisition of null and overt pronominals in Japanese by
English speakers. Second Language Research 13. 265-287.
Kanno, K. (1998). Consistency and variation in second language acquisition. Second
Language Research 14. 376-388.
Kiss, K. E. (1995, ed). Discourse configurational languages. Cambridge: CUP.
Liceras, J. M. (1989): On some properties of the “pro-drop” parameter: looking for
missing subjects in non-native Spanish. In S. M. Gass & J. Schchter (eds).
Linguistic Perspectives on Second Language Acquisition. Cambridge: CUP.
Lozano, C. (forthcoming 2002a). Knowledge of expletive and pronominal subjects by
learners of Spanish. ITL Review of Applied Linguistics, 135-136.
Lozano, C. (forthcoming 2002b). The interpretation of overt and null pronouns in
non-native Spanish. Durham Working Papers in Linguistics, 8.
Luján, M. (1987). Los pronombres implícitos y explícitos del español. Revista
Argentina de Lingüística, 3:19-54.
Luján, M. (1999). Expresión in omisión del pronombre personal. In I. Bosque and V.
Demonte (eds). Gramática descriptiva de la lengua española. Madrid: Espasa,
pp. 1275-1315.
López, L. (1999). The syntax of contrastive focus: evidence from VP ellipsis. In
Gutierrez-Rexach, J. and Martínez-Gil, F. (eds). Advances in Hispanic
Linguistics. Somerville, MASS: Cascadilla Press.
Marsden, H. (1998). A study of L1 influence in the L2 acquisition of Japanese with
respect to a ‘poverty of the stimulus’ phenomenon. Unpublished MA
dissertation. University of Durham.
Marsden, H. (2001). L1 transfer and UG in the L2 acquisition of Japanese. Paper
presented at the Fourth Durham Postgraduate Conference in Theoretical and
Applied Linguistics.
Montalbetti, M. (1984). After binding. PhD dissertation, MIT.
Montalbetti, M. (1986). How pro is it? In Jaeggli, O. and Silva-Corvalán, C. (eds.),
Studies in Romance linguistics. Dordrecht: Foris, 137-152.
Núñez del Prado, Z., Foley, C., Proman, R., Lust, B. (1994). Subordinate CP and prodrop: evidence for degree-n learnability from an experimental study of Spanish
and English. Proceedings of the North East Linguistic Society (NELS) 2. 443460.
Ouhalla, J. (1988). A note on bound pronouns. Linguistic Inquiry, 19: 485-454.
Parodi, T. (2001). Optionality in developing grammars: clitics in L2 acquisition.
Paper presented at EUROSLA 11, University of Paderborn, Germany,
September 2001.
Pérez-Leroux, A. T. and Glass, W. R. (1997). OPC effects on the L2 acquisition of
Spanish, pp. 149-165. In Pérez-Leroux, A. T. and Glass, W. R. (eds.),
Contemporary perspectives on the acquisition of Spanish, volume 1: Developing
grammars. Somervile, MA: Cascadilla Press.
Pérez-Leroux, A. T. and Glass, W. R. (1999). Null anaphora in Spanish second
language acquisition: probabilistic versus generative approaches. Second
Language Research 15. 220-249.
Phinney, M. (1987). The pro-drop parameter in second language acquisition. In
Roeper, T. and Williams, E. (eds.), Parametric setting. Dordrecht: Reidel.
Picallo, M. C. (1994). Catalan possessive pronouns: the Avoid Pronoun Principle
revisited. Natural Language and Linguistic Theory, 12:259-299.
Rizzi, L. (1997). A parametric approach to comparative syntax. In Haegemand, L.
(ed.), The New Comparative Syntax. London: Longman.
Sorace, A. (1993). Incomplete vs. divergent representations of unaccusativity in nonnative grammars of Italian. Second Language Research 9. 22-47.
Schwartz, B. D. (2000). The L1=L2 hypothesis revisited. Paper presented at the
Second NWCL second language acquisition and language teaching
workshop, University of Salford, 9-10 April 2000.
University of Wisconsin (1998). The University of Wisconsin College-level Placement
Tests (Form 96M). Madison, Wisconsin: University of Wisconsin Press.
Xu, L. (1986). Free empty category. Linguistic Inquiry 11. 75-93.
Notes
1
Licensing refers to the grammatical mechanisms that allow a null pronominal subject, pro, to occur.
[AGR]reement is the phonological realisation of phi-features, which consist of person, number,
gender (Chomsky, 1995:35). For Rizzi (1997) [AGR] consists of person and number only.
3
It has been known for some time (e.g., Liceras, 1988; Lozano, forthcoming 2002; Phinney, 1987) that
learners of Spanish do license null pronouns from the earliest stages of acquisition.
4
There are, certainly, many counterexamples in these studies where the OPC applies partially (i.e., in
cases where the overt/null alternation does not obtain, as with possessive pronouns, pronouns as the
complement of a preposition, etc). These exceptions will not be considered any further since no
researcher disputes the fact that the OPC applies whenever a matrix QDP c-commands an embedded
pronominal subject. This is true for all languages providing the alternation overt/null obtains.
5
At least, this must be true of L2A since there is empirical evidence that OPC constructions are never
explained in textbooks (Kanno, 1997; Marsden, 1998, 2001; Pérez-Leroux & Glass, 1997, 1999). We
also conducted searches on Spanish corpora (Real Academia Española, corpus CREA y CORDE). The
searches generated no hits on OPC constructions. Therefore, instruction or positive evidence can be
discarded as the source of knowledge of OPC.
2
6
Chomsky (1995:277ff) distinguishes between interpretable features (readable by the conceptualintentional component) from uninterpretable features (unreadable by the conceptual-intentional
component. The [+focus] feature of the overt pronoun él is [+interpretable] since it affects the
interpretation speakers give to the pronoun (it contrasts Mr López vs. Ms García). The [+focus] feature
of T is, however, [–interpretable] since it does not affect interpretation (it is there so that él can move to
the TP configuration for checking purposes).
7
The Spanish placement test used was the University of Wisconsin Placement Test, Form 96M
(University of Wisconsin, 1998). An extra placement test in English, the Oxford Placement test (Allan,
1992), was administered to Greek natives. The threshold was ≥80.
8
Note that if the contextualising sentence is not provided, both target sentences (a and b) are
grammatical in adult Spanish. Therefore, in order to bias towards one interpretation (overt or null
pronoun in the embedded clause), a context is needed.
9 The AJT test consisted of twelve target stimuli (6 OPC, 6 NPC) and twelve distractors. Two training
stimuli were placed at the beginning of the test plus two stimuli at the end were used to control for
tiredness effects on learners.
Target stimuli were constructed following a factorial design (Table Error! Main Document Only.) to
ensure that subjects were not reacting to OPC or NPC constructions randomly. While an overt pronoun
is ungrammatical in OPC cases, it is grammatical in NPC cases. In this way, it was ensured that
subjects were not randomly rejecting overt pronouns in all contexts.
Table Error! Main Document Only.: Test design (target stimuli)
indices
(joint=OPC/disjoint=NPC)
pronoun
*QDPi…OVERTi QDPi…OVERTj
(overt/null) QDPi…NULLi
*QDPi…NULLj
The type of pronominal subject in the embedded clause was either an overt third person pronoun (50%
of the time él ‘he’ and 50% ella ‘she’), or a null pronoun pro. The type of binding between the matrix
QDP and the subordinate overt/null pronominal subject was of two types: joint coreference (OPC
constructions) and disjoint coreference (CFC constructions). Three universal quantifiers were used for
each condition, namely, todo el mundo ‘everybody’, cada X ‘each X’ and ningún X ‘no X’. Each of
these appeared twice in each condition.
In order to avoid extraneous variables, several measures were taken. Presentational effects were
avoided by using (i) overt pronouns 50% of the time in the a sentence, and 50% of the time in the b
sentence and (ii) two versions of the test with the same sentences but different sequential order.
Sentences were also randomised in each version of the test, following Cowart’s (1997) ‘blocking’
procedure. Vocabulary was also controlled, including beginners’ vocabulary only (González et al.,
1995) so that learners could clearly understand the sentences. The target sentence length was also
controlled. It never exceeded nine words.
10
Optionality is a well attested phenomenon in the grammars of advanced learners (Parodi, 2001).
11
An alternative explanation concerning the acceptability of null pronouns in NPC contexts is offered
in Lozano (forthcoming, 2002b).