The distribution of null subjects in non-native grammars

The distribution of null subjects in non-native grammars: syntactic markedness and interface vulnerability Juana M. Liceras Anahí Alba de la Fuente Cristina Martínez Sanz University of Ottawa

“Illicit” null subjects 1. Pragmatically illicit null subjects: Heritage speakers Elicited production: to test spontaneous production of subjects Entró el abuelito con el perro y pro cortó el estómago del lobo y pro sacó a la abuelita y a el Caperucita Roja. Mientras estaba dormido el lobo *pro le llenaro. N el estómago con piedras, y la abuelita estaba lista pa coserle el estómago. […] “The grandfather came in with the dog and pro cut the wolf’s stomach and took out the grandmother and Little Red riding Hood. While was sleeping the wolf, *pro filled the stomach with stones, and the grandmother was ready to sew him the stomach. ” […. ] WHO FILLED THE WOLF’S STOMACH WITH STONES?

“Illicit” null subjects … Despertó el lobo cosido, lleno de piedras, y el abuelito, la abuelita y la caperucita roja estaban riendo de él. Al fin todo salió bien y *pro se fue a casa con el abuelito. “Woke up the wolf sewn, full of stones, and the grandfather, the grandmother and the Little Red Riding Hood were laughing at him. In the end everything went well and *pro went home with the grandfather. ” [HS #205, intermediate (Montrul 2004 (30): 133)] WHO WENT HOME WITH THE GRANDFATHER?

“Illicit” null subjects Entonces la Caperucita roja encontró, pro fue a ver quién estaba en la cama, entonces pro encontró que era el lobo. ? pro estaba corriendo del lobo y entonces pro salió fuera, o *pro se la comiÓ, *pro se comiÓ a la abuelita y a la Caperucita. “So, the Little Red Riding Hood found, pro went to see who was in the bed, so pro found that was the wolf. ? Pro was running away from the wolf and the pro went out, or *pro ate her up, *pro ate up the grandmother and the Red Riding Hood”. [HS #209, intermediate (Montrul 2004 (32): 133)] NO AMBIGUITY BUT CHANGE OF REFERENT MAKES THE PARAGRAPH PRAGMATICALLY INCOHERENT

“Illicit” null subjects 2. Pragmatically illicit null subjects: Non-native grammars Elicited production: to test spontaneous production of subjects Evidence against the unidirectionality of pragmatic deficits (Sorace 2004). [Montrul and Rodríguez-Louro 2006: Table 3]

“Illicit” null subjects versus “agreement errors”

“Illicit” null subjects Grammaticality judgments: to test choice of topic versus non-topic connected subject A: Hola John “Hi, John” B. Hola Ana “Hi, Ann” A: ¿Te gustaría almorzar conmigo? “Would you like to eat lunch with me? B: Si, me gustaría. ¿Puede venir Beth también? “Yes, I would. Can Beth come too? A: Seguro, ¿a qué hora quieres ir? “Sure, when do you want to go? ” B: 1) Bueno, está en clase ahora. ¿Está bien a las 12: 30? “Is in class now. Is 12: 30 ok? ” 2) Bueno, Beth está en clase ahora. ¿Está bien a las 12: 30? “Beth is in class now. Is 12: 30 ok? ” [Lafond et al. 2001: 126]

“Illicit” null subjects • In terms of anaphoric relations, in languages such as Italian and Spanish, null and overt subjects seem to respect Carminati’s (2002) “Position of Antecedent Hypothesis (PAH)”: “The null pronoun prefers an antecedent which is in the Spec IP position, while the overt pronoun prefers an antecedent which is not in the Spec IP position”. [Carminati 2002: 33] • Therefore, a null subject whose choice of antecedent violates the PAH is also an “illicit” null subject.

“Illicit” null subjects Picture Verification task: to test PAH • Non-native speakers behave like native speakers when it comes to the choice of antecedent for null subjects: DO NOT produce pragmatically “illicit” null subjects with forward (9) or backward (10) anaphora (they respect the PAH). [Sorace and Filiaci 2006: 352 ]

“Illicit” null subjects Story telling: to test spontaneous production of subjects Picture verification task: to test the interpretation of null and pronominal subjects (PAH) • Null pronominal subjects were produced at a comparable rate in spontaneous production by both the near native and the control groups. [Belleti et al. 2007: 671] • There were significant differences between the near-native and the native speakers in the interpretation of pronominal subjects BUT NOT in the interpretation of null subjects. [Belleti et al. 2007: 674]

“Illicit” null subjects 3. Native grammars Entonces cuando el gigante lo vio a él, David le dijo a él: tú vienes a mí con espada y jabalina, yo vengo contra ti en el nombre de Jehovah y cogió una honda. Puso una piedrecita así. Pero el gigante tenía todas esas cosas puestas y ahí mismo él agarró la honda, le tiró la piedra, y ahí mismo lo mató. Le dio ahí y cuando *pro cayó, cogió la misma espada de él y le mochó la cabeza. Then, when the giant saw him, David told him: you come to me with sword and javelin, I come against you in the name of Jehovah and (he) took a sling. (He) put a little stone like that. But the giant was wearing all those things and right there he grabbed the sling, (he) threw the stone against him, and right there (he) killed him. (He) hit him there and when *pro fell, (he) took his own sword of him and cut off his head. [Martínez Sanz (forthcoming): Eliser, 8 UW]

Null subjects in Romance-derived Creoles According to Lipski (1999) some null subjects are possible in Philipinie Chabacano Spanish, Mauritian Creole, Papiamento and Palenquero but they are the exception. (i) The majority of null subjects are produced in main clauses (ii) Instances of null subjects in embedded clauses which are coreferential with subject in matrix (double null subject configurations) are rare. Lipski (1999) argues that these Creole null subjects are “null constants” (Lasnik and Stowell 1991). These Creoles do not exhibit pro null subjects of the Spanish and Italian type because they do not have the resources to license and identify them.

Licensing and identification of null subjects: GB Null subjects have to be licensed and identified (Rizzi 1986). In languages such as Italian and Spanish, they are licensed via a [+strong] feature in INFL and identified via the phi-features in AGR, as in (2) [Liceras et al. 1998: 264]

Null subjects: Minimalist accounts Alexiadou & Anagnastopoulous (1998). The set of phifeatures of I is interpretable. Agr is a referential pronoun, therefore there is no need for pro. Following Rohrbacher (1992) and Speas (1994) these authors propose that strong morphemes (the Spanish agreement markers) have individual lexical entries in the numeration and that in this type of language EPP is checked via merge. Holmberg (2005), Manzini and Roussou (1999), Platzack (2003, 2004).

EPP checking: Merge [Liceras et al. : forthcoming] • The Spanish agreement markers are clitic pronouns Roberts (2001): a unmarked operation of core grammar because it DOES NOT create a new layer of structure.

EPP checking: Move [Liceras et al. : forthcoming] Roberts (2001): a marked operation of core grammar because it DOES create another layer of structure.

Spanish overt pronouns [Liceras et al. : forthcoming]

Where are we? Learners of Spanish have to master two different sets of subject pronouns: (i) strong pronouns, which, by default, have the feature [+topic shift]; and (ii) weak pronouns which do not seem to have such a feature. Merge is the operation which leads to the incorporation of the weak pronouns in the structure. These bound morphemes are interpretable and part of the numeration. The strong pronouns are adjoined and occupy a focus position.

Where are we? Both sets of pronouns are present in non-native Spanish grammars from the early stages of development. There seems to be a correlation between the pragmatically deviant use of strong pronouns and the production of errors with weak pronouns—agreement errors (Montrul & Rodriguez-Louro 2006). This correlation does not show with respect to null pronouns. Non-native speakers intuitions are not as clear-cut as native speakers intuitions when judging pragmatically deviant null subjects (Lafond et al. 2001), namely when besides the agreement marker an overt subject is needed. Both native and non-native speakers produce and accept some pragmatically deviant null subjects. These null subjects are seldom ambiguous in terms of establishing a discourse referent.

Research questions Will native and non-native speakers written narratives show different patterns in terms of the relationship between the use of overt (strong) and bound (weak) pronouns? Will null subjects with switch reference produced by native and non-native speakers in written narratives be different in terms of both overall quantity and use of ambiguous agreement markers?

Hypotheses • Null subjects will not be problematic for non-native speakers because the EPP checking via Merge is an unmarked operation of core grammar. But… • Native and non-native speakers of Spanish may differ in terms of how they resolve ambiguity. CONSEQUENTLY… #1 Non-native speakers of Spanish will produce less instances of agreement subjects (bound pronouns) with switch reference than native speakers because they may have more problems with bound morphemes. # 2 Non-native and native speakers will differ with respect to the patterns of identification of ambiguous bound morphemes. # 3 Given the findings of previous studies, the number of pragmatically deviant (“illicit”) null subjects—instances of unresolved ambiguity—will be small both in the case of native and non-native speakers.

Hypotheses • These non-native grammars will not show the pattern of distribution of null subjects which characterizes Romance-based Creoles because bound pronominals (agreement markers) are part of these grammars from the early stages. CONSEQUENTLY… # 4 Null subjects with switch reference will occur in both main and subordinate clauses. # 5 Instances of null subjects in embedded clauses which are coreferential with null subjects in the matrix clauses will not be rare.

The study • Data Narratives: (i) Non-native (NN) participants were asked to write freely about the character that would result from choosing from a list of randomized characteristics regarding age, marital status, profession, hobbies, place of residence, etc. Each narrative had an average of 500 words. (ii) Native participants were asked to give their impressions about a short film they were shown. • Participants -15 intermediate and 15 advanced (L 1 English) non-native speakers of Spanish from two North American universities. -15 native speakers of peninsular Spanish. University students from various faculties at a Spanish university in Spain.

Codification of data: categories • Impersonal subjects, quirky subjects and subject relatives were not counted

Codification of data: identification / ambiguity Bound morphemes as pronominal subjects: AGR Ambiguity: “Discourse” • comiÓ [two or more candidates: he, she, you-formal(s) ate] • comiero. N [two or more “plural” candidates: they, you-formal(p) ate] Ambiguity: “FLEX” • era, sería, fuera [1 sts, 3 rds, 2 nds (usted)] I, she, you-formal(s) use to be/was, would be / had been

Results: Null and overt subjects 80. 00% 70. 00% 60. 00% 50. 00% 40. 00% SW OVERT 30. 00% SW NULL 20. 00% 10. 00% NN Intermediate NN Advanced Native

Results: Null and overt subjects The results of a Two-way ANOVA on the use of null and overt subjects showed a significant effect of Group 1 (F(2, 42)=8. 855, p= 0. 001). A post-hoc analysis, using the Bonferroni Correction, shows that the NN Intermediate group differs significantly from both the NN Advanced and the Native groups, both in the use of null and overt subjects (p . 004 in all cases). The Native and NN Advanced groups do not differ significantly in the use of null (p= 1. 00) and overt subjects (p= 1. 00). 1 The variable “Group” includes NN Intermediate, NN Advanced and Native.

Non-ambiguous versus ambiguous bound pronominal subjects 90. 00% 80. 00% 70. 00% 60. 00% 50. 00% NN Intermediate 40. 00% NN Advanced Native 30. 00% 20. 00% 10. 00% AGR AMB-FLEX AMB-DIS

Non-ambiguous versus ambiguous bound pronominal subjects In terms of “Identification” 1 of null subjects, the results of a Threeway ANOVA showed a significant effect of Group for AGR (p= 0. 025), but not for A-Flex (p=. 087) and A-DISC (p=. 086). A post-hoc analysis, using the Bonferroni Correction, shows that the Intermediate group differs significantly from the Native group with respect to the use of AGR (p=. 027). None of the other comparisons yielded any significant differences (p. 08 in all cases). 1 The label “Identification” includes the variables AGR, A-FLEX and A-DISC.

“Illicit” null subjects

“Illicit” null subjects: NN Intermediate Group (1) Si (el) ballet no *estuve [fuera] parte de mi vida no *sabe [sabría] como * estoy[estar]: (el) ballet es mi y *pro (yo) soy (el) ballet [502 -W 081 -BUC] If ballet were not part of my life I would not know how to be: ballet is me and I am ballet. (2) Queremos dar a nuestros hijos la vida que mis padres me dieron. Ahora *pro sabe (usted? ? ? ) un poco acerca de Vicente. [511 -W 081 -HAW] We want to give our children the life that my parents gave me. Now you (singular-formal? ? ? ) know a little bit about Vicente. (3) Pero, si *pro (uno/se) quiere ser una psiquiatra buena, la primera cosa que *pro (uno/se) tiene que aprender es como hablar con muchas personas diferentes y tipos. [512 -W 081 -HOO] But, if (you) want to be a good psychiatrist, the first thing that (you) have to learn is how to speak with many different people and types (of people).

“Illicit” null subjects: NN Advanced Group (1) Hace poco cumplí 33 años y empecé a tener dudas que había un lugar para mi o que *pro (yo? ) podía cambiar el rumbo rápido de mi vida. [602 -W 071 -BLA] Not too long ago I turned 33 and began to have doubts that there would be a place for me or that (I? /the rapid course of my life? ) could change. (2) Espero que *pro (ustedes? /ellos? ) comprendan y que *pro (ustedes? /ellos? ) me lo perdonen. [612 -W 071 -GUA] I hope that (you-plural-formal? /they? ) understand that (you-plural-fromal? /they? ) forgive me for that. (3) Luc y yo vivimos en una mansión con piscina y una cancha de tenis. Me encanta nuestra casa, nuestro vecindario, y nuestra piscina… Aquí *pro (usted/él) tiene mi primer secreto: a veces me siento sola. [613 -W 071 -HIG] Luc and I live in a mansion with a pool and a tennis court. I love our house, our neighborhood and our pool… Here you (you-singular-formal/he) have my first secret: sometimes I feel lonely.

“Illicit” null subjects: Native group (1) En mi caso particular, la adolescencia se caracterizó por dar una vida excesiva a los papeles. Ante la indiferencia, y los complejos que creaba *pro (yo? /la adolescencia? ) construía mundos. [C 06 -JUF] In my particular case, adolescence characterized itself by giving an excessive life to papers. Confronted with the indifference, and the complexes that it created (I? / adolescence? ) built worlds. (2) No era un conjunto de batallitas de diario o de conquistas de lo que *pro (yo? /the diary’s author? ) presumía. El plagio o la mentira no eran sus líneas. [C 06 -JUF] It was not about the set of little diary stories or achievements that (I? /it? ) was bragging about. Plagiarism or lies were not (its/the diary’s author? ) lines. (3) Me miró sorprendido y me sonrió. En el folio *pro (él/yo) llevaba las respuestas del examen y una nota. Cuando termines rompe el papel. [C 10 -SOL] He looked at me surprised and smiled. On the sheet (he/I) was carrying the answers to the exam and a note. When you finish destroy the paper.

Percentage of null subjects in Main and Subordinate clauses 80. 00% 70. 00% 60. 00% 50. 00% Null Main 40. 00% Null Subordinate 30. 00% 20. 00% 10. 00% NN Intermediate NN Advanced Native

Coreferential null subjects in embedded and subordinate clauses • Bound morphemes have clear pronominal status

Use of Strong Subject Pronouns 80. 00% 70. 00% 60. 00% 50. 00% NN Intermediate 40. 00% NN Advanced Native 30. 00% 20. 00% 10. 00% "pp" / overt "pp" / null "pp"/ total SU

Use of Strong Subject Pronouns The results of a One-way ANOVA indicate that there are significant differences among the groups. A posthoc analysis using the Bonferroni Correction shows no significant differences between the Native and Advanced groups. There are significant differences, however, between the Intermediate group and both the Advanced (p<. 001) and the Native groups (p=. 009).

Conclusions Our first research question was: Will native and non-native speakers written narratives show different patterns in terms of the relationship between the use of overt (strong) and bound (weak) pronouns? The answer is: Yes, they do with respect to the overall production of overt personal pronouns versus bound pronominals (agreement markers) but only in the case of the Intermediate group.

Conclusions Our second research question was: Will null subjects with switch reference produced by native and non-native speakers in written narratives be different in terms of both overall quantity and use of ambiguous agreement markers? The answer is: the NN Intermediate group is significantly different from the NN Advanced and the native group in terms of the overall quantity of null subjects they produce. The NN Intermediate group is also significantly different from the Native group in terms of the use of ambiguous agreement markers (bound pronouns).

Conclusions Our hypotheses were: • Null subjects will not be problematic for non-native speakers because the EPP checking via Merge is an unmarked operation of core grammar. However, native and non-native speakers of Spanish may differ in terms of how they resolve ambiguity. CONSEQUENTLY… #1. Non-native speakers of Spanish will produce less instances of agreement subjects (bound pronouns) with switch reference than native speakers because they may have more problems with bound morphemes. This hypothesis was confirmed but the results were only significant in the case of the NN Intermediate group.

Conclusions #2. Non-native and native speakers will differ with respect to the patterns of identification of ambiguous bound morphemes. This hypothesis was confirmed but again only for the NN Intermediate group who used less discourse and person ambiguous bound pronouns (agreement markers) than the Advanced and the Native groups. #3. Given the findings of previous studies, the number of pragmatically deviant (“illicit”) null subjects—instances of unresolved ambiguity—will be small both in the case of native and non-native speakers. This hypothesis was confirmed. The total number of “illicit” null subjects was very low and was similar for all three groups. In terms of the [+topic shift] feature, our data shows (contra Sorace 2000) that, at least in written narratives, native and non-native Spanish bound pronominals can bear a [+topic shift] feature.

Conclusions We did not expect to find similarities between non-native grammars and Romance-based Creoles with respect to the pattern of distribution of null subjects because bound pronominals (agreement markers) are part of nonnative Spanish grammars from the early stages. CONSEQUENTLY… # 4 Null subjects with switch reference will occur in both main and subordinate clauses. In fact, our data showed that null subjects with switch reference occur in both main and subordinate clauses. # 5 Instances of null subjects in embedded clauses which are coreferential with null subjects in the matrix clauses will not be rare. This is in fact the case with our data, since our subjets systematically produced sequences of “double null subjects” as the ones listed in slide 35.

Further research • We would like to suggest that: (1) The referential status of bound pronominals –agreement markers- has to be taken into consideration when investigating so-called null subjects. (2) Bound pronominals—as suggested by Lozano (2008)—be analzyed in relation to the specific person (first, second, third) and number (singular or plural) they encode. Lozano (2008) has shown that vulnerability —mainly with respect to overproduction of overt pronouns—affects pronouns which carry [3 rd person] and [+animate] features but not those which carry [1 st person] [-animate] features. (3) A distinction should be made between “illicit” null subjects which create ambiguity that cannot be resolved via an overt pronoun but requires an overt DP subject from “illicit” null subjects which bear contrastive focus or violate Carminati’s Position of Antecent Hypothesis.

Acknowledgments • Institutions University of Ottawa University of Barcelona [Data from Research project “El desarrollo del repertorio lingüístico en hablantes no nativos de castellano y catalán" (MEC-SEJ 2006 -11083), 2006 -2009; Principal Investigator: Joan Perera]. University of Alabama [Prof. Diana Carter] • Graduate students G. Boudreau, J. La. Montagne, P. López-Morelos, L. Walsh

Selected References • Alexiadou, A. and E. Anagnostopoulou, E. 1998. Parametrizing AGR: word order, Vmovement and EPP-checking. Natural Language and Linguistic Theory 16: 491 -539. • Belletti, A. , Bennati, E. and A. Sorace. 2007. Theoretical and developmental issues in the syntax of subjects: Evidence from near-native Italian. Natural Language and Linguist Theory 25: 657– 689. • Hulk, A. and N. Müller. 2000. Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition 3(3): 227 -244. • Lafond, L. , R. Hayes and R. Bahatt. 2001. Constraint demotion and null subjects in Spanish L 2 acquisition. In J. Camps and C. Wiltshire (eds. ), Romance Syntax, Semantics and L 2 Acquisition (pp. 121 -135). Amsterdam: John Benjamins. Lasnik, H. and T. Stowell. 1991. Weakest crossover. Linguistic Inquiry 22: 687 -720.

Selected References • Lipski, J. 1999. Null subjects in (Romance-derived) creoles: routes of evolution. Paper presented at the Annual Meeting of the Society for Pidgin and Creole Linguistics, Los Angeles, January 8, 1999. • Maranz, A. 1995. The minimalist program. In G. Webelhuth (ed. ). Government and binding theory and the Minimalist Program (pp. 351 -382). Oxford: Blackwell. • Montrul, S. 2004. Subject and object expression in Spanish heritage speakers: a case of morpho-syntactic convergence. Bilingualism: Language and Cognition 7 (2): 125 -142. • Montrul, S. & C. Rodríguez Louro. 2006. Beyond the syntax of the null subject parameter. A look at the discourse-pragmatic distribution of null and overt subjects by L 2 learners of Spanish (pp. 401 -418). In L. Escobar & V. Torrens (eds. ), The Acquisition of Syntax in Romance Languages. Amsterdam: John Benjamins.

Selected References • Rizzi, L. 1994. Early null subjects and root null subjects. In T. Hoekstra and B. D. Schwartz (eds. ). Language acquisition studies in generative grammar (pp. 151 -176). Amsterdam: John Benjamins. • Rohrbacher, B. (1992) English AUX-NEG, Mainland Scandinavian NEG-AUX and theory of V-to-I raising. Proceedings of the 22 nd Western Conference on Linguistics (WECOL 92). • Sorace, A. 2004. Native language attrition and developmental instability at the syntaxdiscourse interface: data interpretations and methods. Bilingualism: Language and Cognition 7 (2): 143 -145. • Sorace, A. and F. Filiaci. 2006. Anaphora resolution in near-native speakers of Italian. Second Language Research 22(3): 339 -368.