The acquisition of postverbal subjects in spanish and its information structure by english-speaking students

THE ACQUISITION OF POSTVERBAL SUBJECTS IN SPANISH AND ITS INFORMATION STRUCTURE BY ENGLISH-SPEAKING STUDENTS TESIS DE DOCTORADO EN ESTUDIOS FILOLÓGICOS María Cornejo Núñez Universidad de Sevilla En co-tutela con: Università degli Studi Roma Tre Octubre 2019 DIRECTOR/TUTOR: CO-DIRECTOR: Ángel Luis Jiménez Fernández Mara Frascarelli DOCTORANDO: María Cornejo Núñez i DEDICATION To my husband, I couldn’t have done it without you, I love you. A mi marido, no lo habría logrado sin ti, te quiero. ii ACKNOWLEDGMENTS When I think about the beginning of this research, the first thought that comes to my mind is my “Trabajo Fin de Máster” (TFM), which is the final project before completing a Master’s degree. In my case, the degree was on teaching Spanish as a foreign language and other modern languages, since my main interest lies on the field of language teaching and learning. The final project of my Master’s degree is closely related to this doctoral thesis. The title of my “TFM” is Estudio longitudinal sobre la adquisición de los patrones V-X-S / X-V-S en español y su estructura informativa en hablantes anglófonos. The development of this project was my first academic experience working on the theory of Information Structure within the field of language acquisition. After that, my PhD “adventure” began. These years have been essential for my professional development and personal growth. During this time, I had the opportunity to study, learn and work really hard in order to achieve my goal. Apart from working on this research, I woke up everyday and went to university to do my job as a teacher. I love my job and I have learnt a lot from this experience, but it has not been easy. For this reason, I have to say thank you. First of all, I would like to thank my tutor, Ángel Jiménez Fernández, for being by my side and giving me the support I needed. I also wish to thank my co-tutor, Mara Frascarelli, for her help during these years, especially during my stay at Roma Tre University. I felt warmly welcomed and thanks to her and her “team”(Valter, Giorgio and Tania) it was a very enriching experience. Thanks to all the informants who completed the experimental tests, providing a fundamental contribution to my work, and thanks to everyone who collaborated in this project in some way. iii Finally, I would like to thank my husband, my family (especially my parents and brother) and my friends, for all the support and understanding. Fede, this could have never been possible without you. Thanks for all your love, your unconditional help and also, your work. All of you were there for me when I needed, making me feel good and encouraging me to continue and complete this long path. Thanks. I thank you with all my heart. ¡Muchísimas gracias por todo! iv TABLE OF CONTENTS Dedication ....................................................................................................................... i Acknowledgments ........................................................................................................ ii Table Of Contents ....................................................................................................... iv List Of Tables ............................................................................................................... ix List Of Figures ............................................................................................................. xii Introduction ................................................................................................................... 1 Introduction ................................................................................................................. 2 Chapter 1. Theoretical background ........................................................................ 5 1.1. General Notions of Information Structure .................................................... 6 1.1.1. Theoretical approach .................................................................................. 6 1.1.2. Focus, Topic and Givenness .................................................................... 11 1.1.3. Given information and new information ............................................... 15 1.2. Organization of constituents ......................................................................... 21 1.2.1. Approaching Word-Order ....................................................................... 21 1.2.2. Information and Syntactic Structures in Spanish ................................. 25 1.3. The concepts of Topic and Focus .................................................................. 31 1.3.1. Types of Topic ............................................................................................ 36 1.3.1.1. Aboutness-Shift Topic. ........................................................................ 36 1.3.1.2. Contrastive Topic. ............................................................................... 36 1.3.1.3. Familiar Topic. ..................................................................................... 37 1.3.2. Types of Foci .............................................................................................. 38 v 1.3.2.1. Mirative Focus ...................................................................................... 40 1.3.2.2. Information Focus ............................................................................... 41 1.3.2.3. Corrective/Contrastive Focus ............................................................ 42 1.3.2.4. Broad Focus .......................................................................................... 44 1.4. Subjects and their information structure ..................................................... 46 1.4.1. Subject Inversion ....................................................................................... 46 1.4.2. Pro-drop languages and the Null Subject Parameter .......................... 51 1.4.2.1. Tradition and literature ...................................................................... 51 1.4.2.2. Agreement in Null Subject languages .............................................. 53 1.4.3. Null Subject Parameter in L2 Spanish acquisition ............................... 56 1.4.3.1. Linguistic approach ............................................................................. 56 1.4.3.2. L2 postverbal subject acquisition in Spanish ................................... 59 Chapter 2. The Linguistic Experiment ................................................................. 62 2.1. The study: a general introduction of the experimental tests. ................... 63 2.1.1. The Smurfs Test: L1 children ................................................................... 64 2.1.2. The Online Test: L1 and L2 adults .......................................................... 64 2.2. Methodology .................................................................................................... 67 2.2.1. Participants ................................................................................................. 67 2.2.1.1. Preparation and distribution of the experimental tests ................. 67 2.2.1.2. Groups of Informants .......................................................................... 68 2.2.1.2.1. The Smurfs Test - Spanish L1 (children) .................................... 68 2.2.1.2.2. The Online Test –Spanish L1: Control Group (adults) ............. 69 2.2.1.2.3. The Online Test – Spanish L2: Target Group (adults) .............. 71 vi 2.2.2. The structure of experimental tests ........................................................ 74 2.2.2.1. The Smurfs Test ................................................................................... 74 2.2.2.2. The Online Test .................................................................................... 79 2.2.2.3. Different methodology for each experimental test ......................... 84 2.3. Data analysis .................................................................................................... 85 2.3.1. Analysis Methods ...................................................................................... 85 2.3.2. Chi-Square Test for Independence ......................................................... 86 2.3.3. Anova Test .................................................................................................. 87 Chapter 3. Discussion of Results .......................................................................... 90 3.1. The Smurfs Test ............................................................................................... 91 3.1.1. Descriptive Statistics ................................................................................. 91 3.1.1.1. Differences across Focus types .......................................................... 92 3.1.1.1.1. Information Focus .......................................................................... 92 3.1.1.1.2. Corrective Focus ............................................................................ 94 3.1.1.1.3. Broad Focus .................................................................................... 97 3.1.1.1.4. Association between strategies and focus types ....................... 99 3.1.3. Statistical Significance: Chi-square Test ............................................... 101 3.2. The Online Test (Control Group: Spanish L1) .......................................... 104 3.2.1. Statistics Analysis .................................................................................... 104 3.2.1.1. Effect of the main factors and their interactions ........................... 104 3.2.1.2. Effect of the position factor .............................................................. 106 3.2.1.3. Effect of the factors focus - position ................................................ 108 3.2.1.4. Effect of the factors verb-position ................................................... 111 vii 3.2.1.5. Effect of the factors focus-verb-position ......................................... 114 3.3. The Online Test (Target Group: Spanish L2) ............................................ 122 3.3.1. Statistics Analysis .................................................................................... 122 3.3.1.1. Effect of the main factors and their interactions ........................... 122 3.3.1.2. Effect of the position factor .............................................................. 124 3.3.1.3. Effect of the factors focus - position ................................................ 126 3.3.1.4. Effect of the factors verb-position ................................................... 129 3.3.1.5. Effect of the factors focus-verb-position ......................................... 132 3.4. Overview analysis l1 vs l2 ............................................................................ 141 3.4.1. Statistics Analysis .................................................................................... 141 3.4.1.1. Effect of the main factors and their interaction ............................. 141 3.4.1.2. Effect of the position language ........................................................ 143 3.4.1.3. Effect of the factors language - position ......................................... 143 3.4.1.4. Effect of the factors language -focus - position ............................. 145 3.4.1.5. Effect of the factors language -verb - position ............................... 147 3.4.1.6. Effect of the factors language -focus - verb - position .................. 148 Chapter 4. Conclusions ......................................................................................... 151 4.1. Research goals and preliminary results ..................................................... 152 4.2. Follow-ups and future research .................................................................. 157 References .................................................................................................................. 158 ANNEX I: Informative flyer for participants ...................................................... 166 ANNEX II: Answer sheet “The Smurfs Test” ..................................................... 168 ANNEX III: “The Smurfs Test” ............................................................................. 171 1 INTRODUCTION 2 INTRODUCTION The aim of this dissertation is to research the interaction between word order and information structure in Spanish. In particular, this study has been developed based on preceding theories that deal with information structure in pro-drop languages and how it affects the acquisition of postverbal subjects in Spanish in order to accomplish a descriptive and experimental research on the interaction between the different levels of analysis. This study concerns the acquisition of postverbal subjects by L1 and L2 English speakers of Spanish from an experimental and descriptive perspective, using the main theoretical concepts assumed in the literature on information structure and argument structure. However, the analysis carried out is not entirely theoretical; the theoretical framework is needed in order to present and develop the main studies on the topic, but this dissertation stands on the statistical and descriptive analysis of two linguistic experimental tests. The general hypothesis is that the realization of postverbal subjects may be sensitive to specific focus types. In line with this, we support that: a) the acquisition of subject production and distribution in Spanish (null/explicit or postverbal subjects) might be related one another and share some of the main properties from a syntax and discourse viewpoint; b) from an evolutionary perspective, postverbal subject strategies can be considered as a late acquisition for narrow focus constructions; c) from a second language acquisition perspective, the existence of differences between L1 and L2 speakers of Spanish in relation to the acceptance of postverbal subject structures is relevant enough to carry out an experimental analysis that allows us to develop some didactic proposals for the classroom in the future. Therefore, this research is intended to: 3 (i) study the realization of subjects (specifically in postverbal position) in broad and narrow constructions from an acquisitional and experimental perspective. For this reason, experimental tests have been carried out by children and adults. (ii) analyze the results obtained following the appropriate statistical methods according to the design of each test, and thus, according to the type of answer required. A chi-squared test was needed to carry out the open-(free) answer test type (children), whereas an ANOVA test was the proper statistical analysis to carry out the Likert scale answer test type (adults). (iii) conduct a comparative (statistical) analysis on the acceptability of postverbal subjects by making use of the results obtained from the L1 and L2 speakers of Spanish (adults) and determine where the differences found are significant. The realization of this study has provided interesting and relevant data for the objectives pursued in this research. However, and more importantly, the data obtained amply justifies further study of the different discourse-syntax phenomena from different perspectives in the near future. The results leave some questions unanswered and open to debate and to be analyzed in detail, focusing not only on the experimental and statistics analysis, but also on the interface perspective to examine in depth some syntax and discourse issues. These issues have been proposed for discussion in literature from previous studies. The present dissertation is organized in four chapters, which are divided as described below: Chapter 1 introduces the theoretical background on which this experimental dissertation is based. In this chapter, the previous theories and 4 background research are presented, describing the preceding literature as the basis for our experimental purposes. The focus of chapter 2 is to present and explain the linguistic experiment that has been developed from an organizational and structural point of view. Throughout the chapter, a very general introduction of this experiment is given as well as the participants’ profile and the creative process and methodology followed. The statistical methods chosen to carry out the analysis of data are also explained in this chapter. Chapter 3 looks at the statistical methodology that has been applied to analyze the results . In addition, the results of this analysis are discussed and the significant differences in data collected from each group are reported. Therefore, the statistical analysis of data which we have based this study on will be explained in detail. Chapter 4 outlines the main conclusions of this study and describes some of the future studies that we intend to conduct in the near future, involving the syntax-discourse interface and its influence on the field of second language acquisition. Theoretical background 5 CHAPTER 1. THEORETICAL BACKGROUND Theoretical background 6 1.1. GENERAL NOTIONS OF INFORMATION STRUCTURE One of the essential aspects that we need to deal with in this study is the importance of Information Structure and its relationship with both language and communication. In general, it has been assumed that information structure is a way of packaging information from a statement with the purpose of responding to the communicative needs of interlocutors in a communicative exchange, as in Chafe (1976), Vallduví & Vilkuna (1998), Villalba (2018), among others. Many theoretical studies about Information Structure are based on the work of Chafe (1976) and, with that perspective in mind, other authors have discussed the basic notions of Information Structure within the field of communication. In this section some of these crucial concepts are mentioned from a theoretical point of view. 1.1.1. Theoretical approach Before discussing the Information Structure approach in greater depth, we will refer to the term “syntax”. In a language, the term syntax is related to the area of linguistics that deals with the connection and combination of words in groups of words, denominated“constituents”, leading to the construction of a statement. The construction of a statement is established by the specific properties of the syntactic structure. Thus, everything pertinent to syntax is also related to terms such as “structure” (Frascarelli, Ramaglia, & Corpina, 2012) As it is known, languages around the world are different from one another. However, in many cases they share similarities that allow us to understand and communicate with each other despite speaking different languages (this is, for instance, the case of Italian and Spanish). Even languages that seem totally unrelated, are not as different as they may appear at first sight. According to one of the several lines of thought on this, the reason why languages Theoretical background 7 present common features is that they share the same purpose: achieving effective communication through a communicative system. In contrast to this point of view, languages may share certain features since they develop in relation to the cognitive abilities of human beings, specifically the capacity of acquisition and “linguistic competence”. Thus, if we consider these abilities to be “innate”, it can be stated that a Universal Grammar is shared by humankind, using a series of instructions regarding how language is organized with the purpose of building a system. Following this theoretical framework, this study is based on the field of Generative Grammar, initiated by Noam Chomsky in the 50’s and developed by numerous linguistic studies over the years. Within the field of Generative Grammar, the principles which rule the language functions are considered to be innate to the human mind, so the concept of language acquisition is closely related to this interpretation (Frascarelli, Ramaglia & Corpina, 2012). Universal Grammar is composed by universal language principles and parameters. The general properties of language refer to the concept of principle, for example the Structure Dependence Principle, related to the fact that the elements of a sentence (words and constituents) do not depend on its linear order, but on the syntactic struture of the sentence itself. In relation to this, another example of principle is the one that establishes that every sentence must have a subject, although in some languages such as Spanish or Italian the subject does not have to be explicit. As for the concept of parameter, it refers to the fact that “languages can vary by “choosing” among a limited number of possible alternatives” (Frascarelli, Ramaglia & Corpina, 2012, p. 15, my own translation). Regarding the previous example of the principle about the obligatory presence of a subject in a sentence, it should be stated that some languages allow the subject not to be explicit, but rather, implied. When this happens, even if the subject is not pronounced explicitly, it remains present within the syntactic Theoretical background 8 structure. A clear example of languages that contemplate this possibility is the following: (1) Sto partendo `*(I) am leaving’ (Frascarelli, Ramaglia & Corpina, 2012, p. 16) This issue is an essential part of the present study, since the Parameter of Null Subject (Chomsky 1982; Rizzi 1986) and how it relates to the acquisition of postverbal subjects is one of the issues included in this dissertation. According to this parameter, some languages allow the option of having a non explicit or implied subject in a sentence, such as Spanish, while others such as English do not allow this option. Thus, this aspect makes reference to the possibility of variation among languages that we mentioned above. This aspect is essential for the existence of the parameter we are discussing. In line with this, it is important to point out that Frascarelli (2007, 2014) and Jiménez-Fernández (2014) “have shown that in null subject languages such as Italian and Spanish (but also Portuguese and Polish) the distribution of NSs is crucially determined by discourse factors”(Camacho-Taboada, Jiménez-Fernández, & López-Rueda, 2014, p. 311). The present study takes this proposal with the objective of discussing the relationship between the Null Subject Parameter and the acquisition of postverbal subjects from an experimental point of view, since languages which allow the subject to appear in postverbal positon are in all probability null subject languages. With respect to the levels of analysis and interfaces, Frascarelli, Ramaglia & Corpina (2012) describe human language as a complex and articulated system, separable at different levels of analysis: phonology, morphology, syntax, Theoretical background 9 semantics, pragmatics. These levels are not dependent on one another. However, they interact with each other at the same time and work within the system to reach a proper functioning. In order to explain the connection and relationship among the diverse levels of grammar, the concept of interface must be explained. This term refers to the fact that inside the system the different levels of analysis are connected to each other. Examples of this connection are the Syntax- Morphology interface or the Syntax-Semantics interface, and so on. Since language is a complex system, as previously mentioned, the elements of grammar should not be considered as independent parts, so it is necessary to introduce an interface analysis taking into account the complex interaction among such parts (Frascarelli, Ramaglia & Corpina, 2012). As for the notion of interface specifically, it has been stated that despite being essentially independent from one another, the diverse levels of grammar interact and work together within the language system. With this perspective of integrated system, these levels do not interact freely and without restriction. According to the theory of Generative Grammar, their relationship follows an order in the process of building a sentence, especially, syntax has a fundamental role in the architecture of grammar. As we know, syntax deals with the formal properties of a language, using rules in order to build complex “entities” such as sentences by combining simple elements. Words are combined to build “groups of words”, called constituents, which are combined in a more complex unit to build the sentences. Then, following the authors’ words: […] syntax represents the generative level of grammar since it is built by an operative system that generates all infinite sentences of a language. In fact, following the Generative Grammar model, human language is formed by lexicon on the one hand, that is, a Theoretical background 16 tends to precede new information” (p. 4, my own translation). 4 This refers to the “Information Flow Principle”, proposed by Lambrecht (1994). Taking the work of Prince (1981) on given vs new information and the three levels of givenness, a summary of this concept and three definitions that refer to this notion are presented by Corpina (2009):  Givenness in relation to predictability. In terms of the interaction between speaker and hearer, the speaker may assume that the hearer is able to predict that a linguistic element can be a part of a statement, located in a specific position as well. Regarding this topic, linguists such as Halliday (1967), for instance, support that elements taking part in a statement are considered old information when they are predicted from a previous context. On the contrary, if this information or element cannot be predicted or assumed from a previous linguistic context, the information is considered new and unpredictable.  Givenness in reference to saliency. The speaker assumes that the hearer may be aware of a particular notion or have some information at the time of interacting. According to this description, it is claimed that given (old) information represents what the speaker considers to be present in the awareness of the hearer in the time of interlocution, while new information would be what the speaker introduces or communicates to the hearer’s awareness at that specific time of linguistic interaction. This definition is accepted by Chafe (1976).  Givenness in terms of shared knowledge. In this case, the speaker believes that the hearer knows or might infer a precise part of information. Following this definition, it can be said that given information is known 4 We refer to the non-marked order in the section about word order in Spanish. Theoretical background 17 and believed by the hearer (or at least, that is what the speaker thinks), while new information is not yet known by the hearer (according to the speaker’s assumption). According to Corpina (2009) and based on the work of Prince (1981), these definitions are not independent; on the contrary, they are related to each other since there is a connection between all three categories: if the speaker assumes the hearer may predict that a constituent is in a specific position inside the statement (predictability), the speaker might presuppose the hearer is aware of that particular notion (saliency). If the interpretation of these facts is right, the speaker must suppose that the hearer should be able to make assumptions and thus draw conclusions. Regarding the different levels mentioned above, it should be noted that Chafe (1987) discussed this topic and suggested the existence of three categories depending on the activation state that a concept is in. These categories are labelled as active, semi-active and inactive concepts (Chafe, 1987): Active concept refers to “a concept in a person’s focus of consciousness” (Chafe, 1987, p. 25). Semi-active concept corresponds to “a concept of which a person has a background awareness, but which is not being directly focused on” (Chafe, 1987, p. 25) Inactive is defined as a concept that “is currently in a person’s long term memory, neither focally nor peripherally active” (Chafe, 1987, p. 25) or, likewise, “what the speaker does not assume to be in the hearer’s consciousness” (Geluykens, 1994, p. 28) This triple distinction corresponds to the concepts of given information, accessible information and new information (Chafe, 1976, 1987), respectively. In Theoretical background 18 relation to this division of concepts, it is assumed that “these activation states are a direct reflection of cognitive activity” (Geluykens, 1994, p. 28) and try to develop this argument. What Geluykens (1994) remarks on here is the importance of interactional dimension in real conversation, since he claims that the cognitive activity is dependent on interaction, adding to Chafe’s view, a matter to be taken into account. Geluykens (1994) elaborates on this dependence by suggesting two stages: Firstly, regarding conversational discourse, the information structure and what it is considered given or new information is highly dependent on speakerhearer communication; this theoretical dimension cannot be ignored. Secondly, another interactional dimension regards the fact that information transmission is a possible aim of communication, but not the only one. This activity could also lead to a more important goal, a request, for instance. Following Geluykens (1994), the aim of communication is “to effect changes in the activation states of concepts in the consciousness of the hearer” (p.30). It can be said that depending on what is or is not in the hearer’s and speaker’s awareness, the active-inactive categories of concepts (or given - new) will have a cognitive base and thus, a psychological effect. This information issue is generally considered as a cognitive evolutionary process that develops through time. The role of interaction is, according to Geluykens (1994) an important fact ignored by Chafe. As he states, “the cognitive activity expressed through discourse is always, in essence, also an interactional phenomenon”(Geluykens, 1994, p. 30). Therefore, the cognitive dimension together with the interactional dimension are dependent on and directly connected to one another. Information flow is a cognitive process as well as an interactional aspect, both closely related to each other. Theoretical background 19 Because of this, the fact that the information flow depends on speakerassumptions and on what is supposed to be in the hearer’s awareness (Chafe, 1987) would make the actual data very difficult to analyze, according to Geluykens (1994). Due to the limitations this interpretation offers, this author proposes the necessity of a more practical and useful identification of the information flow. In relation to the distinction between given information and new information, Geluykens (1994) takes Halliday’s (1967) definition of new information, which is basically defined as non-predictable, following Halliday’s words: “not in the sense that it cannot have been previously mentioned, (…) but in the sense that the speaker presents it as not being recoverable from the preceding discourse” (Halliday, 1967, p. 204). As reported by Geluykens (1994), “given information, then, is information which is presented as being recoverable from the preceding discourse” (p.30). Regarding the notion of predictability used by Prince (1981) and presented at the beginning of this section, Geluykens (1994) considers it to be confusing, “since it suggests that, for an item to be given, it has to be predictable at a specific place in the utterance” (Geluykens, 1994, p. 31). He prefers to use the alternative term “recoverability” in order to replace the traditional notions of given-new information, since, from his point of view, such terms are “over-used and confusing”. Following this idea, he proposes that information can be either “recoverable” or “irrecoverable” as in Geluykens (1992) . In line with this approach, according to Zubizarreta (1998), the notions of new information and old (or given) information are closely related to discourse and have a “very superficial relation to the grammar” (p. 159). A similar view about the distinction between new vs old information is given by Suñer (1982). In agreement with this interpretation, “new information” should be considered as the Theoretical background 20 information that is not shared by the speaker and the hearer at a specific time when interacting, whereas “given information” should be associated to the assumptions made about the shared knowledge between speaker and hearer and in general (what is present in the context). In other words, the distinction between given vs new distinguishes the information the speaker considers to be familiar to the hearer from the information that was not previously present in the hearer’s consciousness. Thus, there is a difference between what is presented as presupposed from what is provided as new or an unfamiliar piece of information in the statement (Leonetti, 2014). As previously seen, different insights into the notions of given and new information have been proposed by diverse authors. Together with the concepts of focus and topic, some basic notions of Information Structure have been introduced. This phenomenon of information packaging responds to the communicative needs of interlocutors, according to Chafe’s definition (1976). It should be stated that the Information Structure deals with the way information is organized in a sentence depending on the communicative needs. The information must be properly included in the context so that the interlocutors have access to an appropriate interpretation and achieve effective communication. . Theoretical background 21 1.2. ORGANIZATION OF CONSTITUENTS 1.2.1. Approaching Word-Order The most significant elements in a sentence are subject, verb and object 5 . Every sentence has a nucleus that is related to the rest of the elements in different ways. Also, every sentence has two basic constituents: a noun phrase and a verb phrase. The Spanish language allows the organization of the elements in groups of words (sentences) and divides the sentence in separate parts. Therefore, it can be said that a basic order exists to organize the elements in a sentence, although it may be subject to change (Fernández Soriano, 1993) The different combinations that can be made when distributing the elements S, V, O are possible in languages with a basic order, so the information can be organized according to what we mean or want to say by changing the distribution of the constituents. These combinations can be: SVO, SOV, VSO, VOS, OVS, OSV. Some languages can be associated with one of these types of basic order. However, other languages can include more than just one possibility. In general, this fact depends on the position of the nucleus, that is, if the nucleus is located at the beginning or at the end of the sentence within the verbal group. In Spanish, the head of the verbal group precedes the complement, so the head is initial and consequently, the word order is SVO (Fernández Soriano, 1993). Taking into account the mentioned concept, there are languages that follow their corresponding combination in a precise way, while others are allowed to modify their basic order in some cases. That is why another division between fixed order languages and free order languages is possible. This 5 Throughout this study, the “O” label in SVO, OVS, VSO and VOS abbreviations refers to any type of object (not only Direct Object) including adjuncts such as PP (Juan thinks about his girlfriend all day, etc). In the case of patterns with unaccusative and unergative verbs, the “O” does not refer to “object”, but indirect objects or adjuncts. Theoretical background 22 depends on the appropriate organization of the sentence, in other words, if the different roles such as agent, patient, subject, etc. are clearly expressed by using the right markers such as declinations, prepositions, concordance, pronouns, etc. within the word group they belong to. This is what allows objects to be expressed correctly without the necessity of following the established system, that is, despite using a different order in the distribution of constituents, it is not difficult to comprehend each part of the sentence (Fernández Soriano, 1993). In relation to the distribution of constituents, it is important to point out that, in general, languages have a basic order (also called unmarked) and several marked orders. The unmarked order in Spanish is SVO; it corresponds to the most frequent distribution of elements, this is, the most neutral ordering since “it is compatible with the highest number of possible contexts of use” (Leonetti, 2014, p. 3, my own translation). In line with this, it should be stated that Spanish is a basic-order language that allows the possible reorganization of elements depending on the communicative needs. At the same time, it is also considered a language with a less limited word order than others, such as English. According to Leonetti (2014), “Spanish is not (…) a free order language, but it is relatively flexible compared to others, even within the limited field of Romanic languages” (p. 3, my own translation). Although word order is closely related to pragmatics, some aspects of word order are dependent on grammar exclusively 6 , and apparently they cannot be explained in pragmatic terms. (Leonetti, 2014). 6 According to Leonetti (2014), discourse consequences are effective when the way of organizing information depends on the speaker’s choice, showing a contrast with other alternative options determined by the system, however, they are not effective when a specific construction is established by internal principles of the system. For example, in the case of partial interrogatives in Spanish, subject inversion is required by grammar principles. Theoretical background 23 The structure of elements in a sentence is not an easy issue in the case of a language such as Spanish, since it allows the shift of its components as appropriate. These possible changes in the distribution of elements may influence their grammatical, syntactic or morphological role, among others. It is known that the way in which words and sentences are organized is not the same in every language, but this syntactic structure is conducted by following some determined sentence patterns and not arbitrarily. According to this, The fact that some languages use different syntactic structures to express a different packaging of information has long been known and numerous studies have been devoted to investigating the appropriate articulation of this packaging, the informational primitives involved, and the syntactic correlations of different information structures in different languages (Casielles-Suárez, 2004, p. 213) In relation to this, it should be noted that packing of information is a basic aspect of communication in general. Since the way of packaging information is not common to every language and neither are the syntactic structures. It is important to take this into account when studying a second language. The study of some aspects such as grammar, lexicon or phonetics is essential to achieve of effective communication. Therefore, the word order of a language, together with its information structure, should be considered as a grammatical aspect to be studied in a Spanish L2 classroom. This could help when dealing with linguistic transfer Theoretical background 24 problems, for instance. As Fernández Soriano (1993) claims, “the word order in a sentence is certainly one of the fundamental typological parameters” (p.114, my own translation). Theoretical background 25 1.2.2. Information and Syntactic Structures in Spanish As mentioned earlier, word order issues are closely related to the distribution of information. In fact, the position of certain elements is associated with aspects such as contrast, emphasis, focus, presupposition and others that do not correspond directly to a “grammatical” category. In addition, word order is generally considered as restricted, and those restrictions can be studied within the grammar field. Some properties of the syntactic structures are associated with the distribution of constituents. In relation to this, terms such as “focalization” and “topicalization” will be explained in this section (Fernández Soriano, 1993, Jiménez-Fernández, 2005 ). Since the notions of focus 7 and topic are closely related to the way of packaging information, some issues related to the behavior of these elements in the sentence must be discussed. According to the theory of CG, previously introduced, a distinction between CG content and CG management is made. In relation to this distinction, two different uses of focus can be distinguished: the use of focus that relate to the communicative needs of the interlocutors is associated to the CG management, while the use of focus that relate to the factual information corresponds to the CG content. An example of the use of focus in reference to the CG management (pragmatic use) 8 is to emphasize the part of an answer that correlates to the whpart of a question. As pointed out by Krifka (2008), a question changes the CG in order to indicate the communicative goal of the speaker, considering the question as a set of propositions. Thus, the question and the possible alternatives of the answers belong to the CG management. Another pragmatic use of focus is to confirm information, for instance. In this case, a context must have been proposed so that this leads the speaker to a confirmative interpretation, excluding other 7 In the next section, the concepts of topic and focus are explained in detail. 8 More uses of pragmatic focus (CG management) such as corrections, parallels, delimitations, etc are mentioned by Krifka (2008), but not inlcuded in this section. Theoretical background 32 in which the given information is considered to be the starting point is called “Topic” (Corpina, 2009). Take the following example from Italian (see Zubizarreta (1998) for Spanish): (20) Cosa sta facendo tua sorella? `*What is doing your sister?’ (Mia sorella) sta parlando al telefono. `(My sister) is talking on the phone’ (Corpina, 2009, p. 9) In this case, the given information is my sister and thus, what we identify as Topic. The topic could be omitted in case of being available from the linguistic context or related to the shared information by speaker and hearer. The verbal group “is talking on the phone” is the new information, what we name Comment, and the information that cannot be omitted. Besides this, the verbal group can also be the given information and the nominal group, on the contrary, is the new information, as we can see in the following example from Corpina (2009): (21) Cosa stai mangiando? `What are you eating?’ (Sto mangiando) la pasta `(I am eating) pasta’ (Corpina, 2009, p. 7) Theoretical background 33 In such case, when the nominal group corresponds to the new information, this group is known as Focus, while the verbal group is named Presupposition (assumption), which can be omitted as in the case of Topic. Following this description, Corpina (2009) presents the following examples: Topic – Comment (22) Hai comprato il giornale? `Have you bought the newspaper?’ Il giornale, l’ho comprato stamattina all’alba. `*The newspaper, I have bought it this morning’ (Corpina, 2009, p. 7) Focus – Presupposition (23) Chi è arrivato? `Who has arrived?’ É arrivato MIRKO `*Has arrived MIRKO’ (Corpina, 2009, p. 7) In line with this, Leonetti (2014) distinguishes the concepts of topic and comment, describing topic as the “starting point” (in agreement with Chafe (1987) among others) which is the link to introduce the rest of information in the sentence. Topics express given information that is connected to the previous speech and they are normally placed in initial position, although it is possible to Theoretical background 34 find them in final position (dislocated to the right). This distinction between topic and comment establishes a division of the sentence structure in two information sections. As for the distinction between focus and background (also called presupposition), it is commonly accepted that the focus is considered new and important information that needs to be remarked. The background or presupposition is the information that we find explicitly in the sentence and thus, assumed (Leonetti, 2014). The different levels of analysis are interrelated and thus, when referring to discourse grammar we cannot forget that information structure is associated to syntactic structure and the different levels mentioned above: semantic, phonological, logical and pragmatic level, as proposed by Chomsky in the 60’s (Corpina, 2009). Therefore, the relationship between the double articulation of the sentence in terms of discourse (Rizzi 1997) presented above (Topic-Comment and Focus-Pressupposition) is carefully connected to the diverse levels of linguistic analysis as in the following table from Puglielli e Frascarelli (2008) : LEVELS OF ANALYSIS AND INFORMATION STRUCTURE PRAGMATICS GIVEN NEW DISCOURSE GRAMMAR TOPIC PRESUPPOSITION FOCUS COMMENT SYNTACMATIC CATEGORY NP VP NP VP SYNTACTIC FUNCTION Related to SUBJECT/OBJECT SUBJECT PREDICATE SEMANTIC VALUE Related to ARGUMENT DISCOURSE Table 1 Levels of analysis and information structure. Adapted from P&F (2008) As we can see in Table 1, this distribution of information in a sentence identifies two main parts: given information (Topic) and new information (Focus), as stated in Zubizarreta (1999). Theoretical background 35 At the beginning of this chapter we introduced the concepts of topic and focus from a very general perspective, mainly following Krifka’s view (2008). In this section, we have concentrated on the description of these concepts based on the insights of some experts such us Chafe (1976, 1987), Krifka (2008), Corpina (2009), Puglielli e Frascarelli (2008), Rizzi (1997), Zubizarreta (1999) and Leonetti (2014). Let us now focus on the types of topic and focus. Theoretical background 36 1.3.1. Types of Topic 1.3.1.1. Aboutness-Shift Topic. As stated in Frascarelli (2007, p. 3) “the Aboutness-shift Topic has the discourse function of introducing a new topic (or proposing a topic-shift) in the discourse”. An example of this is illustrated as follows: (24) Has estado hablando de Juan durante horas… Me han dicho que él no sabe nada de los resultados del examen. ‘You have been talking about John for hours… (as for him), I’ve been told that he doesn’t know anything about the exam results.’ (Camacho-Taboada, Jiménez-Fernández & López-Rueda, p. 314) According to Bianchi and Frascarelli (2010), Aboutness-Shift Topics implement a conversational move and they belong to the dimension of CG management (Camacho-Taboada, Jiménez-Fernández & López-Rueda, 2014) 1.3.1.2. Contrastive Topic. Contrast-marking in topics is used to “split a complex proposition into a conjunction of simpler ones in which a predicate applies separately to each member or a salient set” (Camacho-Taboada, Jiménez-Fernández & López- Rueda, 2014, p. 314). An example of this is illustrated in (25): (25) A: ¿Cómo nos organizamos para preparar la fiesta? ‘How shall we do to organize the party?’ Theoretical background 37 B: Yo me encargo de la compra, tú puedes enviar las invitaciones. ‘I will do the shopping, you can send the invitations’ (Camacho-Taboada, Jiménez-Fernández & López-Rueda, 2014, p. 314) 1.3.1.3. Familiar Topic. Following the description given by Frascarelli (2007): “The relevant familiar topic is used to refer to background information, for topic continuity (Givón 1983) or as an “afterthought”(p.7). This is illustrated in (26): (26) A: ¿Tomamos gazpacho para cenar? ‘Shall we have gazpacho for dinner?’ B: ¡No, por Dios! El gazpacho, lo tomo todas las noches! ‘No, please, I have gazpacho every evening!’ (Camacho-Taboada, Jiménez- Fernándes & López-Rueda, 2014, p. 314) As for pronominal subjects, when they are Familiar G-Topics a null subject shows up, as we can see in this example by Camacho-Taboada, Jiménez- Fernández & López-Rueda, (2014, p. 315): (27) No he visto a María desde mayo. *Ella/okpro debe estar muy ocupada. ‘I haven’t seen María since May. She must be very busy’. Theoretical background 38 1.3.2. Types of Foci In relation to the notions of focus previously mentioned and according to the definition given by Rooth (1985, 1992), it should be stated that the concept of focus refers to the possibility of having different alternatives, since “focus indicates the presence of alternatives that are relevant for the interpretation of linguistic expressions”. (Krifka, 2008, p. 18). An example of this is shown by Jiménez-Fernández (2015a, p. 51) […] for a question such as “what does Mary want?” there is a number of propositions which vary in the content provided by the focused direct object (Mary wants COKE, Mary wants ICECREAM,…) […] the whole set of propositions make up what is known as congruent answers to the question In the Structured Meaning approach by Krifka (2006), instead “the proposition is divided in two parts, namely a Focus (e.g., “ice-cream” in the example above) and a background ((…) i.e., the property of being something that Mary wants)”. (Jiménez-Fernández, 2015a, p. 51) In line with this, it is assumed that there are different types of foci (and topics, as we have seen in section 1.3.1) which may behave differently depending on “the relative position of subject and verb both across languages and within a single-language”, as stated in Jiménez-Fernández (2015a, p. 50). Normally, the term focus is used to make reference to two different discourse functions (Kiss (1998), Zubizarreta (1998)). As Jiménez-Fernández claims, Theoretical background 39 (a) to introduce new information, which is known as Information Focus (IF), and (b) to introduce a contrast with respect to a previous assertion by denying one part and proposing another part (…), typically referred to as Contrastive Focus (CF) (2015a, pp. 50-51) A clear distinction between these discourse categories has been supported by diverse experts in the field, such as Kiss (1998) and Zubizarreta (1998). These differences are based on morphological, syntactic, phonological and discourse factors. A distinction between different types of Focus is supported by crosslinguistic evidence (Molnár (2006), Bianchi & Bocci (2012), Cruschina (2012) Jiménez-Fernández (2015b)). The syntactic position of focus, specifically IF and CF, may change depending on the language. To illustrate, the position of CF in Romance languages is typically in the left periphery, whereas IF commonly occurs in postverbal position, as shown by Zubizarreta (1998): (28) A: ¿Qué compró Pedro? `What did Peter buy’? B: Pedro compró manzanas B’: #Manzanas compró Pedro (Jiménez-Fernández, 2015a, p. 51) The different possibilities regarding the position of focus do not work in every case, but in general, it does for Standard Spanish and for some varieties of Theoretical background 40 Spanish. In fact, the possibility of a left-periphery position for IF is available according to Jiménez-Fernández (2015a). Apart from the syntactic properties, the different types of foci are differently interpreted at the interfaces regarding discourse and prosodic properties (Frascarelli & Ramaglia, 2013). With respect to the different “sizes” of constituents within a sentence, terms such as “Broad Focus” (BF) and “Narrow Focus” (NF) are used (Lambrecht, 1994). The main difference between these two aspects is that NF refers to focus on single words or constituents, while BF refers to complex constituents, VP for instance, being all parts newly introduced into the discourse at the moment of utterance (Krifka, 2008). In particular, BF is used to refer to all-focus sentences, where the whole content of a proposition is new. Within NF, diverse subtypes can be distinguished, such as Information Focus, Contrastive 12 /Corrective Focus or Mirative Focus, among others (Krifka (2008), Cruschina (2012)). For the purposes of the present study we will concentrate on two of these subtypes (Information Focus and Contrastive/Corrective Focus) along with Broad Focus later on, referring to Mirative Focus from a merely descriptive point of view. 1.3.2.1. Mirative Focus Based on Cruschina (2012), Mirative Focus is considered as not merely informative. “It provides new information and, based on the speaker’s knowledge of what the hearer’s expects, indicates that such information will be unexpected” (Jiménez-Fernández, 2015a, p. 52) 12 Notice that no specific distinction is made between contrastive/corrective focus. Both terms are related to the rejection of an alternative Theoretical background 41 (29) ¡No puedo creerme eso de María! ELLA ha terminado sus estudios de doctorado! ‘I can’t believe that about Mary! She has finished her doctorate studies!’ (Camacho-Taboada, Jiménez-Fernández & López-Rueda., 2014, p. 315) Due to the unexpected information typically related to Mirative Focus, the set of alternatives is very large. Besides, this type of focus does not depend on a question-answer context. According to Bianchi (2012), within the shared knowledge of the participants a contrast must be established and it can be described as a ”proposal to negotiate a shared evaluation” from a semantic point of view (Jiménez-Fernández, 2015a, p. 52) 1.3.2.2. Information Focus IF refers to new information purely (Zubizarreta, 1998). Following the Structure Meaning approach by Krifka (2006), the Focus in the answer should refer to the interrogative phrase of the question (wh-phrase), which is known as “question-answer congruence” (Jiménez-Fernández, 2015b, p. 122). An example of this is illustrated in (30): (30) A: ¿A quién viste en la playa? ‘Who did you see at the beach?’ B: Vi a Marta ‘I saw Marta’ (Jiménez-Fernández, 2015b, p. 122) Theoretical background 48 Interrogative and exclamative sentences: (46) ¿A quién ha visto Juan? (p. 122) *¿A quién Juan ha visto? ‘Who has Juan seen?’ (47) No sé a quién ha visto Juan (p. 122) *No sé a quién Juan ha visto ‘I don’t know who Juan has seen’ (48) ¡Qué bonito coche tiene Juan! (p. 122) *¡Qué bonito coche Juan tiene! ‘What a nice car Juan has got! ‘ (49) Llegaron niños (p. 124) *Niños llegaron ‘Children arrived’ In cases such as example (49), more elements can be included in the noun phrase in order to make the sentence acceptable. This facilitates the use of the preverbal subject. In coordination, as in (51), we find the same property: (50) Hombres de todo el mundo se han concentrado en Nueva York en señal de protesta (p. 125) ‘Men from all over the world have come together in New York in protest ‘ Theoretical background 49 (51) Hombres, mujeres y niños se han concentrado en señal de protesta (p. 125) ‘Men, women and children have come together in protest’ As observed in Fernández Soriano (1993), when the subject is not accompanied by other complements within the noun phrase, it must be placed in postverbal position necessarily, as shown in example (49). In addition, there are specific types of intransitive verbs that allow this kind of construction in Spanish, such as arrive (llegar) or grow up (crecer), as in examples (52) and (53). For the rest of verbs, another element must appear in initial position in the noun phrase, as observed in example (54). (52) Han llegado niños (Fernández Soriano, 1993, p. 125) ‘*Have arrived children’ (53) Han crecido flores (Fernández Soriano, 1993, p. 125) ‘*Have grown flowers’ (54) Aquí han comido niños (Fernández Soriano, 1993, p. 126) ‘*Here have eaten children’ In line with this, it is important to mention that there is a contrast between transitive and intransitive verbs. Throughout the experimental part of this study, we concentrate on transitive and intransitive verbs for the development of the linguistic experiment (structure of the experimental tests), and within intransitive verbs we have specifically distinguished between unaccusative and Theoretical background 50 unergative verbs. In reference to intransitive verbs, some of them have particular characteristics that distinguish them from other types of verbs and because of this, they have been studied in different languages such as French or Italian (see Burzio (1986) for Italian). With regard to Spanish, Fernández Soriano states that the “basic” position of the subject is postverbal in certain structures. As a matter of fact, in languages such as English or French, which do not allow postverbal subjects in unmarked situations, with unaccusative verbs they are accepted: Il est venue un homme (*Il a parlé un homme) There came a man (*There spoke a man) […] in fact, that is the “natural” organization, based on interpretative properties, but the important thing is that sometimes this is the only option […] with some of those verbs, such as happen (suceder), take place (ocurrir), miss (faltar)…” (Fernández Soriano, 1993, p. 129, my own translation). (55) Falta café/el café/Juan *Café/el café/Juan falta ‘Coffee/Juan is missing’ (Fernández Soriano, 1993, p. 129) (56) Sucedió un incidente *Un incidente sucedió ‘An incident happened' (Fernández Soriano, 1993, p. 129) Theoretical background 51 1.4.2. Pro-drop languages and the Null Subject Parameter 1.4.2.1. Tradition and literature As mentioned earlier, some languages allow the realization of null arguments in certain situations, while others do not. This is the case of languages such as Italian or Spanish, where “the occurrence of null subjects correlates with a set of other syntactic properties, thus suggesting that the same abstract property is responsible for apparently unrelated syntactic phenomena” (Barbosa, 2011, p. 551). However, this possibility is not feasible in languages such as English. An example of this is illustrated in (57): (57) a. Chegaram [Portuguese] ‘They have arrived’ ‘b. *Arrived (Barbosa, 2011, p. 551)’ This phenomenon is known as pro-drop and it has been a focus of study over the years. In Frascarelli & Jiménez-Fernández’s words (2019), “in its original formulation (Perlmutter 1971), the pro-drop parameter was aimed at capturing the empirical observation that in some languages a definite, referential, pronominal subject must be expressed in all finite clauses” (p. 1) Later on, in the 80s, this idea was elaborated and extended in the theory of Government and Binding (Chomsky, 1981) and the Extended Projection Principle (Chomsky, 1982). Since then, many authors have studied the properties of the pro-drop parameter and several analyses from different points of view have been proposed regarding this matter. Some of the authors who have worked on this issue in recent years are Frascarelli (2007, 2017), Jaeggli (1982), Jaeggli & Safir (1989), Rizzi (1982, 1986), among others. Theoretical background 52 According to Chomsky’s approach to pro-drop, “a phonetically null pronominal argument (pro) is an inherently unspecified nominal projection whose features are supplied contextually” (Barbosa, 2011, p. 551). Conforming to the Principles and Parameters approach to the theory of grammar, a null argument is considered a pronominal category (pro), hence languages that permit the realization of NSs are known as pro-drop languages or null subjects languages (NSLs). In this type of languages, we can find examples where the subject is apparently missing, however, from a syntactic perspective the subject is present in the sentence since “it can bind an anaphor in the object position just like any overt subject” (Barbosa, 2011, p. 551). (58) Feriam-se a si mesmos [Portuguese] `They hurt themselves’ Besides, within the category of NSLs there can be found two types of null subject constructions determined by the subject position, this is, constructions in which the subject is referential (57) and those in which it is not (59). As illustrated in the following example from Barbosa (2011), this difference can be observed: (59) a. Chove (2011, p. 552) [Portuguese] ‘b. It rains’ Also, when the subject is not referential, non-NSL such as English have an expletive subject, as it can be seen in (59). Theoretical background 53 1.4.2.2. Agreement in Null Subject languages It is clear that pro-drop is not a simple issue and thus, there are a series of factors that influence this phenomenon. With regard to the main aspects of NSL and this research approach, it is important to highlight one of these factors: “Languages with rich subject agreement morphology (…), such as Italian, Spanish, Portuguese, Hungarian, Greek among many others. In this type of language, subjects are freely dropped under the appropriate discourse conditions” (Barbosa, 2011, p. 552). Regarding the discovery of this phenomenon, it was proposed that the possibility of subject drop in this kind of languages is correlated with some other syntactic properties, consequently, a theory of parameter regarding language acquisition and language variation emerged. Within the theory of Government and Binding (GB) (Chomsky, 1981), the Extended Projection Principle (EPP) (Chomsky, 1982) “stated that all clauses must have a structural subject” (Barbosa, 2011, p. 553). Based on the observation that rich subject agreement languages presumably have the possibility of dropping an argument, a distinction between NSLs and non-NSLs was proposed by authors such us Chomsky (1981), Jaeggli (1982) and Rizzi (1982). In line with this, some relevant properties of the rich agreement NSLs are discussed in Barbosa (2011). The first property, crucially related to this research, is the SV/VS order alternation or “free-inversion”. An example of this for Portuguese is illustrated in Barbosa (2011, p. 556): (60) a. O Joao telefonou *the Joao called ‘John called’ Theoretical background 54 b. Telefonou o Joao *called the Joao ‘John called’ As mentioned previously, languages such as Spanish or Italian belong to the category of rich agreement NSLs and in these cases, just like in Portuguese, order alternation is also allowed. Another property shown in Barbosa (2011) is the lack of that-trace effects: subject extraction is from post-verbal position. An illustration of this is shown in the following contrast between English and Spanish: (61) *Who did you say that bought a computer? ¿Quién dices que compró un ordenador? [Spanish] Who say-2SG that bought a computer `Who do you say bought a computer?’ (Barbosa, 2011, p. 556) This lack of that-trace property is generalized in all Romance NSLs. As reported in Barbosa (2011): Rizzi (1982) and Jaeggli (1984) pursue an account on this contrast based on the claim that the subject is extracted not from the pre-verbal position but rather from post-verbal position. Burzio (1986, p. 165), however, notes that a stronger statement is needed: it is not simply the case that subjects in Italian can be extracted from post-verbal position; in fact, they must be (p. 556) Theoretical background 55 Regarding order alternation (SV/VS) in Romance NSLs (and specifically Spanish), an aspect which is related to this property and a matter of interest among many authors such as Contreras (1991), Ordoñez (1998), Valduvì (1992) (for Catalan) or Zubizarreta (1998), is that “preverbal subjects tend to be topics, whereas post-verbal subjects tend to be foci 13 ” (Barbosa, 2011, p. 557). In line with this, another feature that needs to be mentioned is the association between object Clitic Doubling and post-verbal subject constructions in Spanish. According to Ordóñez and Treviño (1999), “both cases pattern similarly in relation to the determination of binding in certain cases of mismatches in person between the doubling DP and the object clitic or between the post-verbal subject and Agr” (Barbosa, 2011, p. 557). In this connection, the actual issue about the thematic subject position in NS sentences is its association with pronominal clitics in general. According to Villa-García (2012), apart from purely syntactic factors, other aspects regarding the discourse-pragmatic interface or the lexicon-syntax interface affect the existence and distribution of subjects. In any case, Barbosa (2011) states that some of these NSLs features, among others, have not been accurately analyzed and understood in the literature over the years. As mentioned before, the pro-drop or NS parameter is not a simple matter. Diverse and various theories about rich agreement NSLs have been proposed over the years, becoming a matter of interest in the linguistic field and an issue to deal with in this research, as it seems to be connected to post-verbal subject structures along with some general aspects of Information Structure. In line with this, Frascarelli (2007) and Frascarelli & Jiménez-Fernández (2019) deal 13 But see Leonetti & Escandell (2017) for the possibility that a postverbal subject is not focus. Theoretical background 56 with an IS strategy and discuss about the interpretation of a referential NS. According to them, there is a relationship of agreement between pro and a specific type of Topic, the Aboutness-shift Topic (A-Topic) (see Jiménez- Fernández 2016 for a similar analysis in Spanish). As stated in Frascarelli & Jiménez-Fernández (2019): The A-Topic combines the [aboutness] feature with the [shift] feature, which is the property of A-Topics to move to conversation from one topic to another, updating the discourse context (Bianchi & Frascarelli, 2010). This means that A-Topics are present in every predicative sentence to establish “what the sentence is about” and shift the topic across sentences when overtly realized (2019, p. 3). 1.4.3. Null Subject Parameter in L2 Spanish acquisition 1.4.3.1. Linguistic approach Within the field of linguistic research, and particularly in the relevant literature on syntax and acquisition, the Null-Subject Parameter has been a focus of discussion over the years since it was first discussed by Chomsky (1981), Jaeggli (1982), and Rizzi (1982, 1986), as mentioned above. In addition, this topic has generated a hot debate concerning NSP acquisition in adult L2 studies, resulting in controversy and lack of understanding, especially in relation to Universal Grammar (UG) role in L2 acquisition (Rothman & Iverson, 2007). The concept of Universal Grammar (UG) assumes that children are born with the ability of learning the given language they are exposed to, being able to acquire “a complete native grammar that reflects more knowledge that can be derived directly from the input they receive”. (Rothman & Iverson, 2007, p. 329). In fact, humans are supposedly born with a specific mental sub-system intended to the development of the language skill, so when we are born, we do not need Theoretical background 57 to learn all aspects of the native language. In line with this, some authors such as Bley-Vroman (1990), Hawkins and Chan (1997), Tsimpli and Roussou (1991), argue that a potentially influential factor regarding L1 and adult L2 acquisition may be the differences in UG-accessibility (Rothman & Iverson, 2007). Some of these variations can be related to specific learning problems. According to White (1985), some examples of differences in UG-accessibility are issues such as mapping problems, transfer of the L1 or problems in interface-conditioned properties (e.g. at the syntax/phonology, syntax/semantics and syntax/pragmatic interfaces), being the latter one of the main aspects of discussion in this study. As stated in Rothman (2009), following Sorace (2005) and Valenzuela (2006): […] interfaces are precisely the source of L2 non-convergence. Sorace and others contend that interfaces (e.g. the syntaxpragmatics, syntax-semantics interfaces) are especially vulnerable for adults and therefore subject to greater difficulty and delays, if not the principle loci for inevitable fossilization in adult grammars (p. 952) In fact, according to Sorace (2005), in the field of adult L2 acquisition some frequent problems are the ones related to the interpretation of features, since this is relevant to the syntax-discourse interface. Regarding the pragmatics-discourse interface, studies have demonstrated that it is not unusual to find adult monolingual variability and developmental L1 delays with properties, as well as cross-linguistic interference and optionality of properties for bilinguals at this interface. Because of this, it has been pointed out The Linguistic Experiment 64 word order variation and aimed at investigating the acquisition of postverbal subject in Spanish. The organization of the two experimental tests is divided as follows: EXPERIMENTAL TESTS a) The Smurfs Test: Spanish native-speakers (children) b) The Online Test: - Spanish native-speakers (adults) - English-speaking students of Spanish as a second language (adults) 2.1.1. The Smurfs Test: L1 children “The Smurfs Test” is an experimental test taken by children in which we tested language production of postverbal subjects depending on different situational contexts (93 children). The following example shows one of the contexts presented to the children in Spanish: The Smurfs Test (Spanish L1) Los Pitufos tienen que escribir una carta muy importante, la está escribiendo Papá Pitufo, ¿no? The Smurfs have to write a very important letter, Papa Smurf is writing it, right? [they are shown the picture of Smurfette writing the letter] [they are shown the picture of Smurfette writing the (corrective focus question type) 2.1.2. The Online Test: L1 and L2 adults “The Online Test” is an experimental test distributed to adults in which we test the acceptability of postverbal subject depending on different situational The Linguistic Experiment 65 contexts. This test is intended to L1 and L2 speakers of Spanish. The control group is composed of 137 Spanish native-speakers, while the target group is composed of 34 English-speaking students of Spanish as a second language. This is an example of a dialogue from the online test in Spanish: The Online Test (Spanish L1/L2) Esta mañana había un gato en el jardín, pero ya no está ¿Quién lo ha cogido? This morning there was a cat in the garden and now I can’t see it anymore. Who took the cat (information focus question type) The Online Test (Spanish L1/L2) El armario está todo desordenado, ¿qué ha pasado? The wardrobe is a mess… what happened? (broad focus question type) The fact that both groups of adults completed the exact same online test is the key for a real comparative analysis of data from a linguistic-methodological point of view, as we will see further on in section 3.4. In this way, the group of L1 speakers of Spanish is labelled Control Group while the group of L2 speakers is named Target Group. It is essential to mention that the methodology followed is different for children, therefore, we will use the Control Group as a “reference” group when discussing the children’s responses in an attempt to do so from an acquisitional and evolutionary perspective. The Linguistic Experiment 66 So far in chapter 2, what we have presented is the main focus of this study along with a very general summary of the experimental tests that we designed. The aim of this linguistic experiment was to collect the necessary information for the development of this research. In the subsequent sections of this chapter, the aim will be to describe the creative process of this experiment by explaining in detail the methodology followed and providing specific information in relation to the participants who collaborated in the methodological part of the study. The Linguistic Experiment 67 2.2. METHODOLOGY 2.2.1. Participants In this section, the profile of the participants who collaborated in this study is described. These participants have been classified in two groups: in the first place, the academic colleagues who took part in the research by helping mostly in the experimental process and, in the second place, the groups of informants (children and adults) who were asked to complete the experimental tests. Thanks to all of them, we were able to collect the data needed for developing and analyzing this written dissertation. 2.2.1.1. Preparation and distribution of the experimental tests First of all, it is essential to mention that this study has been carried out under joint supervision by the universities of Seville and Roma Tre. The collaborative work and joint effort between teams have proved essential for the preparation and distribution of the experimental tests. Secondly, we must take into account that this research and specifically, the elaboration and distribution of the experimental tests, have been conducted in both languages Spanish and Italian. In this study, we deal with the experimental tests in Spanish, and thus, the description of the tests and the discussion of results refer to the surveys written in Spanish. The process that led to the creation of the different experimental tests and the distribution procedure that helped us collect the data are the basis of this thesis from a methodological point of view. The written test intended for adult participants was distributed online, while the test intended for children was carried out “in situ” at Kindergarten and Primary schools. Through the development of this experimental method we aimed to analyze the validity of our initial hypothesis. The Linguistic Experiment 68 2.2.1.2. Groups of Informants 2.2.1.2.1. The Smurfs Test - Spanish L1 (children) The Smurfs Test was designed to be completed by Kindergarten and Primary students. In the case of Spanish, the test was completed by a total of 93 children. All of them were students at Colegio de Educación Infantil y Primaria San Isidoro, Seville, Spain. The students come from the geographical and linguistic area of Seville or close-by areas (same variety of Spanish). …. Figure 1 Children age ranges. The Smurfs Test As we can see in the figures, our students range from 4 years old (II Kindergarten School) to 9 15 years old (IV Primary School). Thus, it was necessary to divide children into two groups in order to provide a more accurate analysis of data from an evolutionary perspective . We separated participants into two groups: o Children from 4 to 6 years old, corresponding to the classes of II, III Kindergarten School and I Primary School o Children from 7 to 9 years old, corresponding to the classes of II, III and IV Primary School. 15 By the time the test was conducted some of the children were already 10 years old, but most of them were 9 y.o. This difference is not relevant, so 10 y.o. children have been included in the 7-9 y.o. group. Group 4-6 Y.O. II Kindergarten (4-5 y.o.) III Kindergarten (5-6 y.o.) I Primary School (6-7 y.o.) Group 7-9 Y.O.. II Primary School (7-8 y.o.) III Primary School (8-9 y.o.) IV Primary School (9-10 y.o.) The Linguistic Experiment 69 GROUP PARTICIPANTS GENDER CLASS GENDER GROUP 4-6 Y.O. 38 MALE 17 II Kindergarten School 9 M 3 F 6 III Kindergarten School 12 M 6 FEMALE 21 F 6 I Primary School 17 M 8 F 9 GROUP 7-9 Y.O. 55 MALE 32 II Primary School 14 M 8 F 6 III Primary School 21 M 16 FEMALE 23 F 5 IV Primary School 20 M 8 F 12 Table 2 The Smurfs Test; Participants data The 4-6 years old group is composed of 38 participants in total, of whom 17 are male and 21 female. The 7-9 years old group is composed of 55 participants in total, of whom 32 are male and 23 female. In Table 2, we can see the number of participants per class and the number of male and female children within each class. 2.2.1.2.2. The Online Test –Spanish L1: Control Group (adults) The Online Test is for speakers of Spanish as a first and as a second language. Firstly, we shared the test with L1 speakers since we needed a control group of native informants in Spanish in order to carry out a proper analysis of the L2 responses. The test was distributed online by sending the link through social networking channels such as Facebook or Twitter and also by email contact. Due to the fact that the distribution of this test was online, it was completed by speakers of more languages other than Spanish, so we omitted those who were not speakers of Spanish as a first language. The informants had access to a survey by clicking the link previously received by one of these channels and then they could enter on the online platform in order to complete the survey. In Figure 2 we can see the website screen to the survey: The Linguistic Experiment 70 Figure 2 Access to the online survey This test was completed by a total of 157 informants. However, as mentioned earlier, some of them were omitted because they did not respond to the test’s needs, so the final number of informants is 137. The information about the participants profile 16 is registered in the first part of the survey. The participants’ identity is anonymous but there are some questions regarding the sociolinguistic and geographic profile of these informants that we would like to highlight: Figure 3 Sociolinguistic & Geographic Information Regarding the participants’ gender, a total of 100 female informants completed the test, while a number of 37 were male. As for the age, there are 67 16 Personal information about the participants (gender, age, geographical area, etc) has not been relevant for the statistical analysis, however, we were interested in having a similar profile of informants to make it as accurate as possible) The Linguistic Experiment 71 informants who are under 29 years old and 70 who are older than 29 years old. In relation to the geographical origin of the informants, all of them were Spanish speakers from the area of Spain. 2.2.1.2.3. The Online Test – Spanish L2: Target Group (adults) The Online Test, apart from being intended for our control group of Spanish native speakers, as we have seen above, is also intended for our target group of English-speaking students who study Spanish as a second language. The test was distributed online by sending the link through social networking channels such as Facebook or Twitter and also by email contact. We also designed and distributed a flyer written in English with a summary of the study and the instructions to the survey to make it more accessible. See annex I. However, unlike the control group, the target group of informants was not simple to get in contact with, since we needed a specific profile of participants and not just any native speaker of English. The profile 17 of these students was quite general regarding the study program or degree, but less general in terms of age and language knowledge: most of them are university students between 18 and 25 years old with upperintermediate or advanced level of Spanish (B2-C1). The target group consists of 34 informants who speak English as a first language and whose level of Spanish is upper-intermediate or advanced. This specific feature allowed us to carry out a much more accurate analysis of data, as we are working with two groups of native speakers (Spanish in the case of the control group, English in the case of the target group) and their behavior as regards understanding postverbal subjects in Spanish. 17 See note 16. The Linguistic Experiment 72 The website to the survey was the one previously used for the control group. The students could enter on the online platform and complete the test by clicking the link received through one of the channels mentioned previously, including the flyer. In the next figure we can see the website screen to the survey: Figure 4 Access to the survey Just like in the case of the control group, the specific information about the participants is registered in the first part of the survey. The participants’ identity is anonymous but several questions about the sociolinguistic and geographic profile of the participants are illustrated in the following figure: Figure 5 Personal Information Regarding participants’ gender, a total of 25 female informants completed the test, while only 9 were male. As for the age, there are 19 informants who are under 24 years old and 15 who are older than 24 years old (not significantly older The Linguistic Experiment 73 anyway). In relation to the geographical origin of the informants, half of them (17) are originally from The United States and the other half (17) are from The United Kingdom. As we mentioned earlier, the 34 informants speak English as a first language and Spanish as a second language. The Linguistic Experiment 80 - Lo ha cogido el niño, el gato (It took the boy, the cat) VSO - El niño ha cogido el gato (The boy took the cat) SVO - Ha cogido el gato el niño (It took the cat the boy) VOS - El gato lo ha cogido el niño (The cat it took the boy) OVS o Corrective Focus A: Esta mañana había un gato en el jardín, pero ya no está. Estoy segura de que lo ha cogido esa niño que lo miraba tanto… B: Te equivocas... A: This morning there was a cat in the garden and now I can’t see it anymore. I’m sure that it has been taken by that girl who was looking at it. B: You are wrong…) - Te equivocas, lo ha cogido el niño, el gato VSO - Te equivocas, el niño ha cogido el gato SVO - Te equivocas, ha cogido el gato el niño VOS - Te equivocas, el gato lo ha cogido el niño. OVS o Broad Focus A: Te veo nerviosa... ¿Qué ha pasado?... A: You seem nervous… what happened? - Que hoy lo ha cogido mi hijo, un gato, y se lo ha traído a casa! VSO - ¡Que hoy mi hijo ha cogido un gato y se lo ha traído a casa! SVO - ¡Que hoy ha cogido un gato mi hijo, y se lo ha traído a casa! VOS - ¡Que hoy un gato lo ha cogido mi hijo, y se lo ha traído a casa! OVS The Linguistic Experiment 81 Figure 6 Information Focus question simple The online survey is divided into two main parts: sociolinguistic and geographic information (previously exposed in section 2.2.1.2) and the list of 21 questions, including 3 distractors, so students answered 24 questions which were randomized to avoid any priming deviance. See annex IV. This experimental test is the same version for both the control group of Spanish native-speakers and the target group of English-speaking students. In this way, we were able to collect results for the same test from two different informants so that the analysis of data is comparable. The Linguistic Experiment 82 The Online Test Structure (adults) A - INFORMATION FOCUS B - CORRECTIVE FOCUS C - BROAD FOCUS TRANSITIVE VERBS UNACCUSATIVE VERBS UNERGATIVE VERBS TRANSITIVE VERBS UNACCUSATIVE VERBS UNERGATIVE VERBS TRANSITIVE VERBS UNACCUSATIVE VERBS UNERGATIVE VERBS 1 -TRANSITIVE OD [+ANIMATED] 2 -TRANSITIVE OD [- ANIMATED] 3 - MOTION VERB 4 - INHERENTLY REFLEXIVE VERB 5 - CHANGE OF STATE 6 - MONOARGUMENTAL 7 - BIARGUMENTAL 1 -TRANSITIVE OD [+ANIMATED] 2 -TRANSITIVE OD [- ANIMATED] 3 - MOTION VERB 4 - INHERENTLY REFLEXIVE VERB 5 - CHANGE OF STATE 6 - MONOARGUMENTAL 7 - BIARGUMENTAL 1 -TRANSITIVE OD [+ANIMATED] 2 -TRANSITIVE OD [- ANIMATED] 3 - MOTION VERB 4 - INHERENTLY REFLEXIVE VERB 5 - CHANGE OF STATE 6 - MONOARGUMENTAL 7 - BIARGUMENTAL 1 - A 2 - A 3 - A 4 - A 5 - A 6 - A 7 - A 1 – B 2 – B 3 - B 4 - B 5 - B 6 - B 7 - B 1 - C 2 - C 3 - C 4 - C 5 - C 6 - C 7 - C Table 5 The Online Test Structure (adults) The Online Test (adults): Answer Structure 0 Least Acceptable 1 2 3 4 - Most Acceptable VSO SVO VOS OVS Table 6 The Online Test (adults): Answer Structure The Linguistic Experiment 83 Figure 7 Survey questions from 1 to 12 Figure 8 Survey questions from 13 to 24 The Linguistic Experiment 84 2.2.2.3. Different methodology for each experimental test It is essential to highlight the fact that the way in which questions are posed in each test is different, depending on the structure of the test itself, the profile of the participants and the purposes from a linguistic point of view in each case. Note that The Online Test, intended for adults, includes a Likert scale from 0 to 4 (4 meaning that they would definitely use it in the context provided, and 0 meaning that they would never use it) to show the level of “acceptability” according to the context. All options are grammatical though. On the contrary, The Smurfs Test, intended for children, consists of a series of sentences/dialogues set in different audiovisual contexts and a question that motivates the child’s spontaneous answer, instead of the selection of alternatives, as previously mentioned. Both tests have the acquisition of Postverbal Subjects as the main focus of interest, but from a methodological perspective we considered the spontaneous answer option more suitable for carrying out the children’s tests “in-situ”, while the structure of The Online Test and the fact that it was distributed online and intended for adults made the option of a Likert scale more appropriate and practical. This is a crucial factor when it comes to deciding the statistical method for the analysis of each test. The Linguistic Experiment 85 2.3. DATA ANALYSIS 2.3.1. Analysis Methods The tests that have been discussed in the previous chapter create two different types of answers: - For the Smurfs Test, the informant generates a “qualitative” answer based on the production of one of four different patterns (VSOSVO– OVS-VOS) - For the Online Test, the informant generates a quantitative answer where he rates the acceptability of the postverbal subject position. This rate is done through the Likert Scale, which measures the attitude of the informant towards a specific answer. The way these data should be analyzed are debated in a paper by (Carifio & Perla, 2008) where the two major hypothesis are discussed (“ordinal view” vs “interval view”). This thesis follows the “interval view” which states “that the Likert response format produces empirically interval data” (i.e. Carifio (1976), (1978); Vickers (1999)) and thus, a parametric analysis can be performed. As discussed in the previous paragraph, the type of answer determines the type of methodology that should be applied to analyze these answers: - The qualitative answers received in The Smurfs Test lead us to apply a chi-square test for independence to evaluate the null hypothesis (no significant difference among the groups). - The “interval view” of the Likert Scale gives us the opportunity to apply a parametric analysis to individuate significant differences among the answers. The analysis has been carried out by applying a Multi-Factor ANOVA. The Linguistic Experiment 86 2.3.2. Chi-Square Test for Independence The Chi-Square Test is a non-parametric statistic test that is used to determine whether there is a significant relationship between the expected frequency and the observed frequency of two variables. While the observed frequencies are derived directly from the experiment (i.e., you flip a coin ten times and you count how many times appeared heads), the observed frequencies are calculated applying the statistical theory (i.e., flipping a coin you would expect 50% chances to appear heads). The formula for expected cell frequencies is: !",$ =&'∙&) * Where: Ei,j= Expected frequency for the cell in the ith row and jth column; Ti= Total number of answers in the ith row; Tj= Total number of answers in the jth row; N= Total number of answers in the whole table. In this case the variables are: - The age group of the informants (II Kindergarten – IV Primary). - The subject position (VSO -OVS- VOS – SVO) The statistic test is based on the comparison of two hypothesis: - The “Null Hipothesis”: the two variables are independent; - The “Alternative Hipothesis”: the two variables are not independent; To calculate the chi-square value the following formula has been applied: The Linguistic Experiment 87 +2= --. /0,1 −!0,1 3 2 !0,1 4 1=1 6 0=1 Where: +27 = Chi – square test for independence; Oi ,j= observed value of two nominal variables; Ei,j = expected value of two nominal variables Once the chi-square value has been calculated, we need to determine whether to accept or reject the “Null hypothesis”. The null hypothesis states that the two variables are independent, in other words, that the answer distribution is the same independently from the age group. For this analysis a significant value of 0,05 has been set; this means that the p-value has to be below this number to reject the null hypothesis (in other words, to prove that the answer distribution is different). To calculate the p-value we need to provide two values: one is the chisquare value previously calculated, and the other is the degree of freedom of the system, which is calculated as (i-1)*(j-1), where i and j represent the number of rows and the number of columns of the analysis. In the subsequent sections the analysis that has been done in The Smurf Test (L1 children, reference group) is reported according to this methodology. 2.3.3. Anova Test According to the “interval view” of the Lickert scale (illustrated in section 2.2.2.2), a parametric test can be used to investigate the outcome. For The Online Test (L1 and L2 adults, control and target groups) we have used the ANOVA (ANalysis Of VAriance) test. The Linguistic Experiment 88 By applying the Multi-Factor ANOVA analysis it has been possible to define the significant factors in a multi factors model, using the analysis of variance to determine, on the one hand, the amount/type of data influenced by each factor taken into examination and, on the other, the amount of data influenced by a casual error. Specifically, the factors analyzed are the following: - Position, with 4 possible identities: VSO, SVO, VOS, OVS. - Focus, with 3 possible identities: Information Focus, Corrective Focus, Broad Focus. - Verb, with 3 possible identities: Transitive, Unaccusative, Unergative. Within the same group (L1 or L2) the 4 combinations of variables have been analyzed (Position, Focus Vs Position, Verb Vs Position, Focus Vs Verb Vs Position) in order to determine what factor or combination of them is “significant” (i.e., which one generates a significant difference in the outcome). In each combination the factor “position” has to be present, since this is the key element of this study. In this analysis a significant value of 0,05 has been established, which means that when the p-value is lower the two means are different (and thus, the null hypothesis is rejected). As previously mentioned, although the ANOVA test allows us to individuate if and what factor (position, focus or verb) or combination is significant, it does not tell us where this difference lies (i.e., comparing the VSO with the other orders within the factor “position”, there could be a significant difference for the pairs VSO-SVO and VSO-OVS, but not for VSO-VOS). In order The Linguistic Experiment 89 to determine which specific combination of variable means are different from one another, a Tukey HSD 24 test has been carried out. Finally, a comparison between the two groups (L1 and L2 participants) has been conducted. The ANOVA test has been used in order to analyze if and what factor (or combination of them) generates a significant variation among the answers given for the same type of variables identified for the 2 groups. In this case, a new factor is generated, the “language” factor, with two possible identities: L1-Spanish and L2 – English speaking students). 24 Tukey HSD “honestly significant difference” or “honest significant difference” test is a statistical test which is used to determine whether the difference between two means (influenced by three or more variables) is greater than the standard error, and therefore, significantly different between each other. Discussion of Results 96 Figure 11 CF trend by class group Figure 12 CF trend by age group If we analyze the trend to assess the evolution from a different perspective, we can establish that there is an increase in the VS order (OVS, VOS) and, on the other hand, a decrease in SVO, although it is largely used in all ages. Therefore, the use of VS strategy starts to emerge earlier for CF than for IF. The DP strategy is not very relevant in CF contexts, while Cleft sentences are constant over the analysis (9%). 0,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% 70,0% 80,0% 90,0% 100,0% II Kindergarten III Kindergarten I Primary II Primary III Primary IV Primary A - VSO Bove - SVO Bnul - SVO C - VOS D - OVS E - Cleft F - DP G - Other 0,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% 70,0% 80,0% 90,0% 100,0% 4 - 6 Years Old 7 - 9 Years Old A - VSO Bove - SVO Bnul - SVO C - VOS D - OVS E - Cleft F - DP G - Other Discussion of Results 97 3.1.1.1.3. Broad Focus Table 11 Differences across focus types BF Let us now examine the data concerning BF constructions: As expected, SVO is definitely the preferred strategy to realize BF constructions at any age. The VS strategy has a very limited relevance, generally restricted to unaccusative verbs. The SVO (null-subjects) answers 26 are more frequent for BF, in particular in younger children, while for IF and CF in most SVO answers the subject is overt. Also, reduced strategies are not an option for BF. The “other” category for BF presents higher percentages in children from 4 to 7 y.o, since kindergarten children have sometimes provided free answers not related to the context in this case. 26 The reason for this might be the fact that the referents of subjects constituents were “shown” in the picture and they might be considered as “given” elements by children. Discussion of Results 98 Although the VS strategy has very limited relevance for BF, the use of the OVS order is especially interesting, being not completely immaterial and very similar at all ages, as illustrated in Table 12. Table 12 Answering strategies for BF in Spanish children In particular, the use of VS constructions in BF is limited to unaccusative (especially motion) verbs, in agreement with Frascarelli & Stortini (2019) for Italian, as observed in Table 13 below: Broad Focus: Unaccusative Verbs Order Motion Result change Progressive change 4 - 6 Years Old 7 - 9 Years Old 4 - 6 Years Old 7 - 9 Years Old 4 - 6 Years Old 7 - 9 Years Old SV 15,79% 14,55% 89,47% 83,64% 92,11% 98,18% VS 73,68% 85,45% 10,53% 16,36% 0,00% 1,82% Other 10,53% 0,00% 0,00% 0,00% 7,89% 0,00% Table 13 VS strategy in BF (unaccusative verbs) Figure 13 BF trend by class group 0,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% 70,0% 80,0% 90,0% 100,0% II Kindergarten III Kindergarten I Primary II Primary III Primary IV Primary A - VSO Bove - SVO Bnul - SVO C - VOS D - OVS E - Cleft F - DP G - Other Discussion of Results 99 Figure 14 BF trend by age group As seen in Figure 13 and Figure 14, the trend shows that SVO is definitely the preferred strategy at any age for BF. In this case, the VS strategy has a very limited relevance, even if the OVS order is the second option. This result is far from the SVO preference, but still interesting. 3.1.1.1.4. Association between strategies and focus types The results obtained with this experiment allow for some general considerations on the development of Focus strategies. As is shown, the SVO order strategy is the most frequent for all the foci examined until the age of 6-7 y.o. The VS strategy, on the contrary, shows a constant increase emerging at the age of 7 (even earlier for CF, 6 y.o.) which affects specially CF and IF. The OVS order becomes the preferred strategy in children from 7 to 9 y.o. In BF sentences, the SVO is more frequent for all age groups, whereas VS is more frequent with specific types of verbs (i.e. unaccusative verbs) after the age of 6-7 y.o. The VSO order does not seem an option in Spanish, where VOS is preferred. Regarding the differences among verb types, both object types and adjuncts were always mentioned in the context preceding the questions, so any variation is not related to an experimental bias. 0,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% 70,0% 80,0% 90,0% 100,0% 4 - 6 Years Old 7 - 9 Years Old A - VSO Bove - SVO Bnul - SVO C - VOS D - OVS E - Cleft F - DP G - Other Discussion of Results 100 These results support the claim that the VS strategy can be considered a late acquisition emerging at the age of 6 y.o., becoming a significant alternative to SVO order for NF constructions in the children’s competence after that age. The evolution of answering strategies is illustrated in Figure 15. Figure 15 Evolution of answering strategies (SV-VS) by age As seen in Figure 15, the blue line shows that there is an emerging trend in reducing the use of the SV strategy as children grow up. On the other hand, the orange line illustrates a clear increasing of VS strategies (considering the union of VOS, VSO and OVS) with age. The difference between the use of SV and VS is higher in Kindergarten and I Primary children, while this difference is less pronounced as they grow up, as observed in this figure. As for reduced strategies, they are largely used by kindergarten children in alternation with SVO. However, its frequency shows a decrease with school education becoming almost immaterial. 0,0% 5,0% 10,0% 15,0% 20,0% 25,0% 30,0% 35,0% 40,0% 45,0% 50,0% 55,0% 60,0% 65,0% II Kindergarten III Kindergarten I Primary II Primary III Primary IV Primary Evolution of answering strategies by age SV VS Cleft DP Other Discussion of Results 101 3.1.3. Statistical Significance: Chi-square Test The statistical analysis has been carried out to determine the existence of significant differences between answering strategies in the two age groups. The test used is the Chi-square Test, a non-parametric statistical test described in section 2.3.2. In particular we are interested in verifying whether there is a significant difference from an evolutionary perspective. Hence, each column in Table 14 indicates the statistical difference attested in the answering strategies used in the two groups (4-6 and 7-9 y.o. children), where the lower the value is, the higher the significance. The outcome of this test is shown as follows: Chi-Square Test Results (Conf. Level 0,05) Information Focus Corrective Focus Broad Focus Transitive <0,001 <0,001 0,133 Transitive (+anm obj.) 0,001 0,005 0,270 Transitive (-anm obj.) <0,001 0,018 0,213 Unaccusative <0,001 <0,001 <0,001 Unaccusative (motion) <0,001 0,003 0,001 Unaccusative (result change) <0,001 0,323 0,588 Unaccusative (progressive change) <0,001 <0,001 0,031 Unergative <0,001 0,001 0,039 Unergative (mono) <0,001 0,228 0,047 Unergative (bi) 0,001 0,003 0,483 Table 14 Chi-Square test results In Table 14, colors represent what follows: - Red: values marked in red show a significant difference - Black: values marked in black do not show a significant difference In order to correlate the descriptive statistics with the statistical significance the following summary is presented: Discussion of Results 102 As illustrated in Table 14, the use of answering strategies shows differences that are statistically significant in most cases for IF and CF (applying a 5% confidence level). Regarding IF, Table 14 shows that the answering distribution changes with age (groups 4-6 y.o. and 7-9 y.o.). This means that the hypothesis about an evolution in the answers depending on the age is feasible, since the answers vary and go from choosing SVO in kindergarten levels to producing VS structures while they grow in age. The answers provided in IF are significantly different (<0,001) in all cases. In this case, the general trend is clear and all the differences attested in the descriptive analysis are significant. Although children in general prefer SVO order in IF, the decreasing of the preference for SVO with age is significant according to the chi-square test, as well as the increasing of the VS strategy with age, specially VOS and OVS, since all the possible answers show a <0,001. The significance of this result is in line with the late acquisition of VS strategies that we support. As for CF, important differences are attested regarding the three groups of verbs: transitives (<0,001), unaccusatives (<0,001), unergatives (0,001), as shown in Table 14. In particular, for unaccusative verbs important differences are observed in motion and progressive change verbs, (0,003) and (<0,001) respectively, while for unergatives, significant differences are observed for biargumental (0,003). Only in the case of result verbs (unaccusatives) and monoargumental (unergatives) differences are not significant. This means that the evolution of the answers is different in relation to most verb types, showing that the learning process generates a variation in the production of the answers given by children. However, this is not the case of result verbs (unaccusatives) and mono-argumental (unergatives). Discussion of Results 103 In general, for BF it emerges that the answers given do not show an evolution with age except for the unaccusative motion and progressive change, as well as unergative mono-argumental (see Table 14). While for unaccusative motion the significant difference is clear (0,001), for unaccusative progessive change (0,031) and unergative mono-argumental (0,047) the p-values obtained are closed to the limit of significance, being this fact a possible object of research in further studies from a statistical perspective. It should be noted that in the case of unaccusative verbs the results are particularly interesting. On the one hand, the evolution of the responses in motion and progressive change verbs shows a significant difference regardless the type of focus (IF, CF and BF). On the other hand, as for the unaccusative (result change) verb, the answers do not show a significant difference with age in the cases of CF and BF, while they do for IF. Discussion of Results 104 3.2. THE ONLINE TEST (CONTROL GROUP: SPANISH L1) 3.2.1. Statistics Analysis The Online Test has been carried out by a control group (137 speakers of Spanish as a first language). The answers given by the participants have been collected in order to: - explore the effect of the different factors on the acceptability of each answer; - use these data as reference values to compare them with those of English-speaking students of Spanish as a second language. The acceptability judgments obtained in the experimental test have been analyzed to determine whether there is a significant difference within the three factors involved regarding the acceptability variation between the different contexts. In this case, the statistical method used for this is the ANOVA (ANalysis Of VAriance) test, described in section 2.3.3. 3.2.1.1. Effect of the main factors and their interactions Table 15 L1 ANOVA test, significance of the factors In the previous table the following information is reported: Discussion of Results 105 - The first column on the left corresponds to the factors being examined: the target is focused on the factors that contain “position”, as we are interested in analyzing the acceptability of different orders. - SS (sum of square): it is the sum of the squares of the deviations of all the answers xi, it represents the sum of the squared differences from their mean value 8 9. - Degrees of Freedom (DF): the degrees of freedom (from now on called m) are calculated as the number of variables for each factor (3 different Focus types, 3 different Verb types, 4 different positions) minus one 27 . - MS (`Mean Sum of Square’): it is calculated as the Sum of Square divided by the DF and it is a representation of population variance. - F represents the F- statistic, and it is calculated by dividing the MS value by the MS error. It helps to verify whether the variance between the means of two populations is significantly different. - Finally, the last column corresponds to the p-value; the p-value is the probability of obtaining a result at least as extreme as the one observed; the p-value represents the most important output of this table as it gives us the smallest significance level at which the null hypothesis should be rejected. If the pvalue is less than 0,05 the effect of the relevant factor or interaction is considered significant (all significant values are marked in red in this study). 27 Where a combination of factors is analyzed, the degrees of freedom of the groups are calculated as (a-1)*(b-1) (i.e. for Focus – Position the degrees of freedom are (4-1)*(3-1) = 3*2 = 6). The degrees of freedom for the Error row are the n data collected minus the product of all the factors (7 questions * 3 foci * 4 positions * 137 subjects – 4 position * 3 verbs * 3 foci). For a detailed illustration and explanation of relevant notions, the interested reader is referred to The ANOVA table (SS, df, MS, F), GraphPad, 2014. Discussion of Results 112 Figure 18 L1 Verb – Position means In Table 21 and Figure 18 the average Likert acceptability of each pair is shown. In line with the previous Focus – Position analysis, SVO is the most frequently selected option for all verb types. In this case, the level of acceptability for this order does not depend on the type of verb, since there is no significant difference among them: transtive and unaccusative verbs (Table 20, cell (2;6) = 0,118), transitive and unergative verbs (cell (2;10) = 0,872) and between unaccusatives and unergatives (cell (6;10) = 0,994). As for OVS, in two out of three cases (unergative and transitive) this is the second preferred order (even though it does not show a significant difference in unergative verbs compared to VOS, see Table 20, cell (11;12) = 0,757). Its level of acceptability decreases with unaccusative verbs, being the least acceptable option (see Figure 18) . 0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50 2,75 3,00 3,25 3,50 3,75 4,00 Transitive Unaccusative Unergative Verb - Position OVS VOS SVO VSO Discussion of Results 113 Similarly to the Focus – Position analysis, the VOS option shows a considerable constant level of acceptability, although the only no significant difference is between transitive and unergative verbs (Table 20, cell (3;11) = 0,183). The VSO construction is the least acceptable option. It shows no relation at all between VSO – TR/UNERG/UNACC or with any other position or verb type (for all significant differences across all types of orders and verbs, see Table 20, row 1 and 5). To summarize the results of this analysis, it is shown that: - SVO order is the preferred option and this does not depend on the verb type - For unergative verbs, there is no significant distinction between VOS and OVS strategies. - OVS varies considerably depending on the type of verb; - Although VOS constructions vary, they have a considerable constant level of acceptability among all verb types. - The VSO strategy is generally the least acceptable order. Discussion of Results 114 3.2.1.5. Effect of the factors focus-verb-position This section shows the effects and the means for each output: L1 TUKEY HSD Test: Focus - Verb - Position; subtable Information Focus Cell Row IF-TRANS-VSO IF-TRANS-SVO IF-TRANS-VOS IF-TRANS-OVS IF-UNACC-VSO IF-UNACC-SVO IF-UNACC-VOS IF-UNACC-OVS IF-UNERG-VSO IF-UNERG-SVO IF-UNERG-VOS IF-UNERG-OVS Cell Column 1 2 3 4 5 6 7 8 9 10 11 12 IF-TRANS-VSO 1 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.959 0.649 <0.001 <0.001 <0.001 IF-TRANS-SVO 2 <0.001 0.404 <0.001 0.998 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 IF-TRANS-VOS 3 <0.001 0.992 <0.001 <0.001 0.010 0.348 <0.001 <0.001 <0.001 IF-TRANS-OVS 4 <0.001 <0.001 0.143 <0.001 <0.001 0.979 <0.001 0.321 IF-UNACC-VSO 5 <0.001 <0.001 <0.001 <0.001 <0.001 0.018 <0.001 IF-UNACC-SVO 6 <0.001 <0.001 <0.001 0.583 <0.001 <0.001 IF-UNACC-VOS 7 <0.001 <0.001 <0.001 <0.001 1.000 IF-UNACC-OVS 8 1.000 <0.001 <0.001 <0.001 IF-UNERG-VSO 9 <0.001 <0.001 <0.001 IF-UNERG-SVO 10 <0.001 <0.001 IF-UNERG-VOS 11 0.002 IF-UNERG-OVS 12 Table 22 L1 Tukey HSD Test: Focus - Verb – Position; subtable Information Focus Cell No. FOCUS VERB POSITION MEAN 1 Information Focus TRANSITIVE VSO 1.20 2 Information Focus TRANSITIVE SVO 3.44 3 Information Focus TRANSITIVE VOS 1.79 4 Information Focus TRANSITIVE OVS 3.13 5 Information Focus UNACCUSATIVE VSO 1.97 6 Information Focus UNACCUSATIVE SVO 3.60 7 Information Focus UNACCUSATIVE VOS 2.81 8 Information Focus UNACCUSATIVE OVS 1.40 9 Information Focus UNERGATIVE VSO 1.48 10 Information Focus UNERGATIVE SVO 3.34 11 Information Focus UNERGATIVE VOS 2.35 12 Information Focus UNERGATIVE OVS 2.81 Table 23 L1 HSD Test, Focus - Verb - Position Means; subtable Information Focus Discussion of Results 115 Figure 19 L1 Information Focus: Focus - Verb – Position means As illustrated in Figure 19, in the case of IF answers, the SVO order is preferred for all verb types, and is not influenced by the verb, since there is no significant difference (Table 22, cell (2;6) = 0,998, cell (2;10) = 1,000, cell (6;10) = 0,583) . The values for this type of answer are, in all cases, between 3 and 4 in the Likert scale. 30 In the case of transitive verbs, no significant difference has been found for SVO and OVS strategies (Table 22, cell (2;4) = 0,404. As for the OVS order, (postverbal subject associated with left-dislocation in Spanish) it is shown that for transitive and unergative verbs OVS is quite an acceptable option, between 2,75 and 3,25 in the Likert scale (and no significant difference has been found between these two verbs, as shown in Table 22, cell (4;12) = 0,321), while this order can be considered almost unacceptable for unaccusative verbs, with values under 1,50 in the Likert scale. 31 In relation to the VOS order, (postverbal subjects associated with “in situ” objects), low values are shown (1,79, Table 23) for transitive verbs. As for the rest 30 This may mean that the subject has been moved to a position in the left periphery. As argued by Jiménez-Fernández (2015b) , IF can be constructed by moving the relevant constitutent to the left. 31 This shows that left-dislocation is not used for the objects of unaccusatives, while this is frequent for transitives (especially) and unergatives. 0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50 2,75 3,00 3,25 3,50 3,75 4,00 Transitive Unaccusative Unergative Information Focus OVS VOS SVO VSO Discussion of Results 116 of the verbs, in general VOS shows values around 2,5 in the Likert scale (see Table 23). The VOS strategy scores different values for different verbs types in the case of IF, highlighting the importance of the influence of the verb factor (see Figure 19). As mentioned earlier, the VSO order for IF shows very low values in general (Table 23). As with VOS, the VSO strategy shows different values for different verb types. Both OVS and VOS orders are acceptable options when the postverbal subject is focused, after the preferred SVO order. It should be noted that in the case of transitive and unergartive verbs, the OVS order is the preferred option after SVO, while VOS is the preferred option in the case of unaccusative verbs, followed by VSO and lastly OVS. L1 TUKEY HSD Test: Focus - Verb - Position; subtable Corrective Focus Cell Row CF-TRANS-VSO CF-TRANS-SVO CF-TRANS-VOS CF-TRANS-OVS CF-UNACC-VSO CF-UNACC-SVO CF-UNACC-VOS CF-UNACC-OVS CF-UNERG-VSO CF-UNERG-SVO CF-UNERG-VOS CF-UNERG-OVS Cell Column 13 14 15 16 17 18 19 20 21 22 23 24 CF-TRANS-VSO 13 <0.001 <0.001 <0.001 0.016 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 CF-TRANS-SVO 14 <0.001 0.089 <0.001 0.221 <0.001 <0.001 <0.001 0.010 <0.001 1.000 CF-TRANS-VOS 15 <0.001 0.236 <0.001 <0.001 0.068 0.999 <0.001 <0.001 <0.001 CF-TRANS-OVS 16 <0.001 1.000 <0.001 <0.001 <0.001 1.000 <0.001 0.433 CF-UNACC-VSO 17 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 <0.001 CF-UNACC-SVO 18 <0.001 <0.001 <0.001 1.000 <0.001 0.746 CF-UNACC-VOS 19 <0.001 <0.001 <0.001 0.034 <0.001 CF-UNACC-OVS 20 <0.001 <0.001 0.959 <0.001 CF-UNERG-VSO 21 <0.001 <0.001 <0.001 CF-UNERG-SVO 22 <0.001 0.099 CF-UNERG-VOS 23 <0.001 CF-UNERG-OVS 24 Table 24 L1 Tukey HSD Test: Focus - Verb - Position; subtable Corrective Focus Discussion of Results 117 Cell No. FOCUS VERB POSITION MEAN 13 Corrective Focus TRANSITIVE VSO 1.12 14 Corrective Focus TRANSITIVE SVO 3.16 15 Corrective Focus TRANSITIVE VOS 1.80 16 Corrective Focus TRANSITIVE OVS 3.53 17 Corrective Focus UNACCUSATIVE VSO 1.49 18 Corrective Focus UNACCUSATIVE SVO 3.46 19 Corrective Focus UNACCUSATIVE VOS 2.70 20 Corrective Focus UNACCUSATIVE OVS 2.14 21 Corrective Focus UNERGATIVE VSO 1.62 22 Corrective Focus UNERGATIVE SVO 3.58 23 Corrective Focus UNERGATIVE VOS 2.34 24 Corrective Focus UNERGATIVE OVS 3.22 Table 25 L1 HSD Test, Focus - Verb - Position Means; subtable Corrective Focus Figure 20 L1 Corrective Focus: Focus - Verb – Position means As we can see in Table 25 and Figure 20 the preferred options in the case of CF are SVO and OVS. Postverbal subject in the OVS order is extremely frequent with transitive and unergative verbs (relevant values in the Likert scale are over 3 in both cases), while for unaccusative verbs, not only SVO but also the VOS order is preferred over OVS. 32 In Table 24, no significant difference between OVS transitive and unergative verbs (cell (16;24) = 0,433) emerges, while both of them are different with unaccusative OVS (cell (16;20) = <0,001 and cell (20;24), = <0,001). An 32 As mentioned earlier, left dislocation is not associated to unaccusative verbs. 0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50 2,75 3,00 3,25 3,50 3,75 4,00 Transitive Unaccusative Unergative Corrective Focus OVS VOS SVO VSO Discussion of Results 118 important feature of the OVS strategies for CF is that not only is there no significant difference among them, but also there is no significant difference between OVS transitive and SVO transitive (cell (14;16) = 0,089), between OVS transitive and SVO unergative (cell (16;22) = 1,000), between OVS unergative and SVO unergative (cell (22,24) = 0,099). This means that, in transitive and unergative verbs, there is no significant difference for OVS and SVO answers, the latter being the first option and the former, the second. SVO is once again the preferred option 33 . While SVO unaccusative does not reflect a significant difference between SVO transitive (cell (14;18) = 0,221) and SVO unergative (cell (18;22) = 1,000), there exists a significant difference between SVO transitive and unergative (cell (14;22) = 0,010). The VOS order is the third preferred option in the case of transitive and unergative verbs, after SVO and OVS. However, the this order shows a higher level of acceptability than OVS in the case of unaccusative verbs. The difference between VOS unaccusative and the rest of the variables is significant (Table 24, row and column 19); no relation has been found between any VOS order in the Corrective Focus (Table 24, cell (15;19) = <0,001, cell (15;23) = <0,001, cell (19;23) = 0,034). The VSO order is the last option in all cases, showing values that are located far from the preferred options, thus showing a significant difference (<0,05) among all the other variables (Table 24, row 13 and 17), except for VSO unaccusative and VSO unaccusative, where the difference is not significant (cell (17;21) = 1,000) 33 It would be very interesting to contrast this preference for SVO in a non-written test, that is, by offering the possibility of spontaneous answers to the same questions. In this way, we could check whether the option in most SVO answers is influenced by the fact that this is an online test completed by adults who think through and take time to mark the “correct” answer. Even if they are said that all options are grammatically correct, they tend to do this unconsciously. Discussion of Results 119 L1 TUKEY HSD Test: Focus - Verb - Position; subtable Broad Focus Cell Row BF -TRANS-VSO BF -TRANS-SVO BF -TRANS-VOS BF -TRANS-OVS BF -UNACC-VSO BF -UNACC-SVO BF -UNACC-VOS BF -UNACC-OVS BF -UNERG-VSO BF -UNERG-SVO BF -UNERG-VOS BF -UNERG-OVS Cell Column 25 26 27 28 29 30 31 32 33 34 35 36 BF -TRANS-VSO 25 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.981 <0.001 <0.001 <0.001 <0.001 BF -TRANS-SVO 26 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 BF -TRANS-VOS 27 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 BF -TRANS-OVS 28 <0.001 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 1.000 BF -UNACC-VSO 29 <0.001 0.999 <0.001 <0.001 <0.001 1.000 <0.001 BF -UNACC-SVO 30 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 BF -UNACC-VOS 31 <0.001 <0.001 <0.001 0.998 <0.001 BF -UNACC-OVS 32 <0.001 <0.001 <0.001 <0.001 BF -UNERG-VSO 33 <0.001 <0.001 0.764 BF -UNERG-SVO 34 <0.001 <0.001 BF -UNERG-VOS 35 <0.001 BF -UNERG-OVS 36 Table 26 L1 Tukey HSD Test: Focus - Verb - Position; subtable Broad Focus Cell No. FOCUS VERB POSITION MEAN 25 Broad Focus TRANSITIVE VSO 0.56 26 Broad Focus TRANSITIVE SVO 3.85 27 Broad Focus TRANSITIVE VOS 2.89 28 Broad Focus TRANSITIVE OVS 1.39 29 Broad Focus UNACCUSATIVE VSO 2.30 30 Broad Focus UNACCUSATIVE SVO 3.84 31 Broad Focus UNACCUSATIVE VOS 2.44 32 Broad Focus UNACCUSATIVE OVS 0.75 33 Broad Focus UNERGATIVE VSO 1.52 34 Broad Focus UNERGATIVE SVO 3.81 35 Broad Focus UNERGATIVE VOS 2.28 36 Broad Focus UNERGATIVE OVS 1.26 Table 27 L1 HSD Test, Focus - Verb - Position Means; subtable Corrective Focus Discussion of Results 120 Figure 21 L1 Broad Focus: Focus - Verb – Position means Regarding BF, the SVO order is definitely associated with this type of focus, so the numbers in this case meet the expectations (see Table 27 and Figure 21). The values for the preferred option (SVO) are higher than in any other case, there are no SVO options under 3,75 in the Likert scale. It can therefore be claimed that the connection between SVO and BF is appropiate. The SVO preference in BF is not influenced by the verb type, as there is no significant difference between any of the SVO answers (Table 26, cell (26;30) = 1,000, cell (26;34) =1,000 and cell (30;34) =1,000). Despite that, it is necessary to highlight the relevance of VOS order in all BF cases, since the values for this option are mostly over 2 on the Likert scale and is the informants‘ second preferred option. There is no significant difference in the acceptability of VOS answers between unaccusative and unergative (Table 26, cell (31;35) = 0,998), whereas there is significant difference between transitive and unergative (cell 27;35) and transitive and unaccusative verbs (cell 27;31). Regarding the results for VSO order, transitive verbs show really low values (Table 27, 0,56), while unaccusative (2,30) and unergative (1,52) verbs show higher values. It is important to notice how, for unergative verbs, there is no significant difference between VSO and OVS (Table 26, (cell 33;36) = 0,764), 0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50 2,75 3,00 3,25 3,50 3,75 4,00 4,25 Transitive Unaccusative Unergative Broad Focus OVS VOS SVO VSO Discussion of Results 121 while, for unaccusative verbs, there is no significant difference between VSO and VOS (Table 26, cell (29;31) = 0,999). While for CF and IF the OVS strategy was a valid answer, for BF it is shown that this order is not an option. In all verb types the acceptability of this answer is below 1,5 (see Table 27) with a lower mean for unaccusative verbs and a grade of acceptability of 0,75. There is no significant difference in the answers given for transitive and unergative verbs Table 26, cell (28;36) =1,000). The results of this analysis can be summarized as follows: - SVO is generally the most acceptable answer, and this is not influenced by the type of verb (except for CF between transitive and unergative verbs). - The preference for OVS answers is strongly influenced by the type of focus: it is widely accepted in IF and CF, but it is not considered as a valid order for BF. Generally, OVS answer strategies are best linked with transitive verbs, with slightly less preference for unergative verbs, while unaccusative verbs are considered a less acceptable option. - The VOS strategies are generally accepted with all focus types and verbs; for IF and CF the least preferred VOS order is linked with transitive verbs, while in BF most preference is given to this option. - The VSO strategy is are the least acceptable option. The only value over 2 can be found in BF for unaccusative verbs. Discussion of Results 128 by the verb type, as there is no significant difference between each VSO pair compared to each other (Table 31, cell (1;5) = 1,000, cell (1;9) = 1,000 and cell (5;9) = 0,945). While the VSO order does not show significant difference with VOS for IF and BF, in the case of CF, the VSO order does not show any significant difference with the OVS answers (Table 31, cell (5;8) = 1,000). The OVS order is the least acceptable option in all cases, although in the case of CF there is no significant difference with VSO. As with previous cases, OVS does not show any significant difference between IF and CF (Table 31, cell (4;8) = 0,709) and between IF and BF (cell (4;12) = 0,583), while there is a significant difference between BF and CF regarding the OVS order (cell (8;12) = 0,002). According to these results, the following conclusions are reached: - Overall, the grade of acceptability of the orders is not influenced by the focus type in the case of L2 informants (the only exception is CF – OVS vs BF – OVS); - The SVO order is the most acceptable option, followed by VOS, VSO. The least acceptable order is OVS. - There is no significant difference between VOS and VSO, thus both can be considered as the second preferred option in the cases of BF and IF. - In the case of CF, there is no significant difference between VSO and OVS, the latter being the last option and the former, the third. Discussion of Results 129 3.3.1.4. Effect of the factors verb-position In the following table, the significant relation between the factors verb and position (without differentiating by focus) is illustrated: L2 TUKEY HSD Test: Verb - Position Cell Row TRANS-VSO TRANS-SVO TRANS-VOS TRANS-OVS UNACC-VSO UNACC-SVO UNACC-VOS UNACC-OVS UNERG-VSO UNERG-SVO UNERG-VOS UNERG-OVS Cell Column 1 2 3 4 5 6 7 8 9 10 11 12 TRANS-VSO 1 <0.001 <0.001 0.203 <0.001 <0.001 <0.001 1.000 0.002 <0.001 <0.001 0.994 TRANS-SVO 2 <0.001 <0.001 <0.001 0.996 <0.001 <0.001 <0.001 0.909 <0.001 <0.001 TRANS-VOS 3 0.026 1.000 <0.001 0.986 <0.001 0.651 <0.001 0.953 <0.001 TRANS-OVS 4 0.005 <0.001 <0.001 0.008 0.962 <0.001 0.678 0.894 UNACC-VSO 5 <0.001 0.982 <0.001 0.437 <0.001 0.875 <0.001 UNACC-SVO 6 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 UNACC-VOS 7 <0.001 0.024 <0.001 0.168 <0.001 UNACC-OVS 8 <0.001 <0.001 <0.001 0.699 UNERG-VSO 9 <0.001 1.000 0.094 UNERG-SVO 10 <0.001 <0.001 UNERG-VOS 11 0.013 UNERG-OVS 12 Table 33 L2 Tukey HSD Test: Verb Position Cell No. VERB POSITION MEAN 1 TRANSITIVE VSO 1.53 2 TRANSITIVE SVO 3.56 3 TRANSITIVE VOS 2.33 4 TRANSITIVE OVS 1.88 5 UNACCUSATIVE VSO 2.34 6 UNACCUSATIVE SVO 3.44 7 UNACCUSATIVE VOS 2.47 8 UNACCUSATIVE OVS 1.44 9 UNERGATIVE VSO 2.06 10 UNERGATIVE SVO 3.36 11 UNERGATIVE VOS 2.14 12 UNERGATIVE OVS 1.67 Table 34 L2 Tukey HSD Test, Verb - Position Means Discussion of Results 130 Figure 24 L2 Verb – Position means As expected, SVO is the preferred order for all verbs. In Table 34 it is observed that values are around 3,4 on the Likert scale. The level of acceptability for this order does not depend on the type of verb, since there is no significant difference in any of the cases: between transitive and unaccusative verbs (Table 33, cell (2;6) = 0,996), transitive and unergative verbs (cell (2;10) = 0,909) and unaccusatives and unergatives (cell (6;10) = 1,000). This result is similar to the output obtained for L1, as there is no significant difference among the verb types. The VOS order is the second most acceptable option for all verbs. As observed in Table 34, its means are over 2 points in all cases and it does not depend on the verb type since there is no significant difference among them (Table 33, cell (3;7) = 0,986, cell (3;11) = 0,953 and cell (7;11) = 0,168). Additionally, it is important to highlight that there is no significant difference between the VOS and VSO orders for unaccusative verbs (Table 33; cell (5;7) = 0,982) and unergative verbs (cell (9;11) = 0,982), in both cases its mean values are over 2 on the Likert scale. As mentioned above, VSO is generally the third preferred option, although it is the least acceptable option for transitive verbs (as previously mentioned, there is no significant difference between VSO and VOS for unergative and unaccusative verbs). The level of acceptability of VSO is not Discussion of Results 131 influenced by the verb type in the case of unaccusative and unergative verbs, since there is no significant difference between them (Table 33, cell (5;9) = 0,437), while both of them show significant differences with transitive verbs (cell (1;5) = <0,001 and cell (1;9) = 0,002), which is not considered as an acceptable option (Table 34: 1,53). The VSO and OVS options do not show any significant difference between them in transitive (cell (1;4) = 0,203) and unergative verbs (cell (9;12) = 0,094). The OVS order is the least acceptable option, except for transitive verbs. There is no significant difference between unergative and unaccusative verbs (Table 33, cell (8;12) = 0,699) and between transitive and unergative verbs (cell (4;12) = 0,894), while there is a significant difference between unaccusative and transitive verbs (cell (4;8) = 0,008) According to these results, the following conclusions are reached: - The SVO order is the most acceptable option and this does not depend on the verb type; - The VOS order does not depend on the focus, since there is no significant difference among the focus types. This option shows a similar level of acceptance to VSO in unaccusative and unergative verbs. - The acceptability of the VSO order is not influenced by the focus type in relation to unaccusative and unergative verbs (relatively acceptable answer), while it is not considered such an acceptable option in the case of transitive verbs. - OVS is generally considered the least acceptable option, except in the case of transitive verbs. Discussion of Results 132 3.3.1.5. Effect of the factors focus-verb-position This section shows the effects and the means for each output: L2 TUKEY HSD Test: Focus - Verb - Position; subtable Information Focus Cell Row IF-TRANS-VSO IF-TRANS-SVO IF-TRANS-VOS IF-TRANS-OVS IF-UNACC-VSO IF-UNACC-SVO IF-UNACC-VOS IF-UNACC-OVS IF-UNERG-VSO IF-UNERG-SVO IF-UNERG-VOS IF-UNERG-OVS Cell Column 1 2 3 4 5 6 7 8 9 10 11 12 IF-TRANS-VSO 1 <0.001 0.040 0.913 0.002 <0.001 <0.001 1.000 0.011 <0.001 0.382 0.999 IF-TRANS-SVO 2 <0.001 <0.001 <0.001 1.000 <0.001 <0.001 <0.001 0.937 <0.001 <0.001 IF-TRANS-VOS 3 1.000 1.000 <0.001 1.000 0.016 1.000 0.032 1.000 0.970 IF-TRANS-OVS 4 0.992 <0.001 0.216 0.862 0.998 <0.001 1.000 1.000 IF-UNACC-VSO 5 <0.001 1.000 <0.001 1.000 0.027 1.000 0.697 IF-UNACC-SVO 6 0.001 <0.001 <0.001 0.999 <0.001 <0.001 IF-UNACC-VOS 7 <0.001 1.000 0.753 0.819 0.021 IF-UNACC-OVS 8 0.003 <0.001 0.259 0.999 IF-UNERG-VSO 9 0.097 1.000 0.848 IF-UNERG-SVO 10 0.001 <0.001 IF-UNERG-VOS 11 1.000 IF-UNERG-OVS 12 Table 35 L2 Tukey HSD Test: Focus - Verb - Position; subtable Information Focus Cell No. FOCUS VERB POSITION MEAN 1 Information Focus TRANSITIVE VSO 1.37 2 Information Focus TRANSITIVE SVO 3.63 3 Information Focus TRANSITIVE VOS 2.24 4 Information Focus TRANSITIVE OVS 1.90 5 Information Focus UNACCUSATIVE VSO 2.30 6 Information Focus UNACCUSATIVE SVO 3.47 7 Information Focus UNACCUSATIVE VOS 2.58 8 Information Focus UNACCUSATIVE OVS 1.39 9 Information Focus UNERGATIVE VSO 2.31 10 Information Focus UNERGATIVE SVO 3.12 11 Information Focus UNERGATIVE VOS 2.06 12 Information Focus UNERGATIVE OVS 1.75 Table 36 L2 Tukey HSD Test, Focus - Verb - Position Means; subtable Information Focus Discussion of Results 133 Figure 25 L2 Information Focus: Focus - Verb – Position means As illustrated in Figure 25 the three-factor analysis shows that SVO is the most acceptable order. Its acceptability within IF does not depend on the verb, as there is no significant difference between transitive and unaccusative verbs (Table 35, cell (2;6) = 1,000) transitive and unergative verbs (cell (2;10) = 0,937) and unaccusatives and unergatives (cell (6;10) = 0,999). The VOS order is the second most acceptable option, and its level of acceptability is not influenced by the verb, as there is no significant difference between transitive and unaccusative verbs (Table 35, (cell (3;7) = 1,000), transitive and unergative verbs (cell (3;12) = 1,000) and unaccusatives and unergatives (cell (7;11) = 0,819). In all cases, the VOS order shows no significant difference with VSO (cell 5;7) = 1,000) for unergative and unaccusative verbs (cell (9;11) = 1,000) and with OVS for transitive verbs (cell (3;4) = 1,000). In relation to VSO, there is a high level of discrepancy regarding acceptability since it is the second preferred option in the case of unergative verbs, but the least preferred option for transitive verbs. While for unaccusative and unergative verbs the acceptability of VSO is not influenced by the type of verb (as there is no significant difference, Table 35, cell (5;9) = 1,000), in the case of transitive verbs there is (cell (1;5) = 0,002 and cell (1;9) = 0,011). As for 0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50 2,75 3,00 3,25 3,50 3,75 4,00 Transitive Unaccusative Unergative Information Focus OVS VOS SVO VSO Discussion of Results 134 unergative verbs, the VSO order does not show any significant difference with the other orders (row 9); in the case of unaccusative verbs, it shows no significant difference with VOS (see Table 35, cell (5;7)=1000) while in transitive verbs the VSO order does not show any significant difference with OVS (Table 35, row 1). The OVS order is, in general, the least acceptable order. It is not influenced by the verb type, as observed in Table 35 (cell (4;8) = 0,862, cell (4;12) = 1,000; cell (8;12) = 0,999). This order does not show a significant difference between transitive verbs and unergative verbs, with VSO (cell (1;4) = 0,913 and cell (9;12) = 0,848 respectively) and VOS (cell (3;4) = 1,000 and cell (11;12) = 1,000). According to these results, the following conclusions are reached: - The SVO order is the most acceptable option and this does not depend on the verb type; - The acceptability of VOS order is not influenced by the focus type and is similar to VSO in unaccusative and unergative verbs. - The VSO option is not influenced by the focus type regarding unaccusative and unergative verbs (acceptable answers), while it is not considered such an acceptable option in the case of transitive verbs. - The OVS order is considered the least acceptable option, although it is not significantly different with both VSO and VOS in the case of unergative and transitive verbs. Discussion of Results 135 L2 TUKEY HSD Test: Focus - Verb - Position; subtable Corrective Focus Cell Row CF-TRANS-VSO CF-TRANS-SVO CF-TRANS-VOS CF-TRANS-OVS CF-UNACC-VSO CF-UNACC-SVO CF-UNACC-VOS CF-UNACC-OVS CF-UNERG-VSO CF-UNERG-SVO CF-UNERG-VOS CF-UNERG-OVS Cell Column 13 14 15 16 17 18 19 20 21 22 23 24 CF-TRANS-VSO 13 <0.001 0.848 0.913 0.999 <0.001 0.013 1.000 0.999 <0.001 0.546 1.000 CF-TRANS-SVO 14 <0.001 <0.001 <0.001 1.000 0.003 <0.001 <0.001 1.000 <0.001 <0.001 CF-TRANS-VOS 15 1.000 1.000 <0.001 1.000 0.434 1.000 <0.001 1.000 1.000 CF-TRANS-OVS 16 1.000 <0.001 1.000 0.556 1.000 <0.001 1.000 1.000 CF-UNACC-VSO 17 <0.001 0.677 0.951 1.000 <0.001 1.000 1.000 CF-UNACC-SVO 18 0.001 <0.001 <0.001 1.000 <0.001 <0.001 CF-UNACC-VOS 19 <0.001 0.936 0.017 1.000 0.494 CF-UNACC-OVS 20 0.972 <0.001 0.156 1.000 CF-UNERG-VSO 21 <0.001 1.000 1.000 CF-UNERG-SVO 22 0.002 <0.001 CF-UNERG-VOS 23 0.999 CF-UNERG-OVS 24 Table 37 Tukey HSD Test: Focus - Verb - Position; subtable Corrective Focus Cell No. FOCUS VERB POSITION MEAN 13 Corrective Focus TRANSITIVE VSO 1.68 14 Corrective Focus TRANSITIVE SVO 3.46 15 Corrective Focus TRANSITIVE VOS 2.24 16 Corrective Focus TRANSITIVE OVS 2.21 17 Corrective Focus UNACCUSATIVE VSO 2.03 18 Corrective Focus UNACCUSATIVE SVO 3.42 19 Corrective Focus UNACCUSATIVE VOS 2.53 20 Corrective Focus UNACCUSATIVE OVS 1.62 21 Corrective Focus UNERGATIVE VSO 2.06 22 Corrective Focus UNERGATIVE SVO 3.37 23 Corrective Focus UNERGATIVE VOS 2.32 24 Corrective Focus UNERGATIVE OVS 1.93 Table 38 L2 Tukey HSD Test, Focus - Verb - Position Means; subtable Corrective Focus Figure 26 L2 Corrective Focus: Focus - Verb – Position means 0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50 2,75 3,00 3,25 3,50 3,75 Transitive Unaccusative Unergative Corrective Focus OVS VOS SVO VSO Discussion of Results 136 As expected, SVO is once again the most acceptable order by far. It does not show any significant difference between transitive and unaccusative verbs (Table 37 cell (14;18) = 1,000) transitive and unergative verbs (cell (14;22) = 1,000) and unaccusatives and unergatives (cell (18;22) = 1,000); in addition, SVO answers are statistically different compared to any other answer (Table 37, row 14 and row 22). The VOS order is the second most acceptable option. It is not influenced by the verb type since there is no significant difference between transitive and unaccusative verbs (Table 37, cell (15;19) = 1,000), transitive and unergative verbs (cell (15;23) = 1,000) and unaccusative and unergative verbs (cell (19;23) = 1,000). The VSO is the third preferred option for unaccusative and unergative verbs, and as with SVO and VOS, there is no significant difference between transitive and unaccusative verbs (Table 37, cell (13;17) = 0,999) transitive and unergative verbs (cell (13;21) = 0,999) and unaccusatives and unergatives (cell (17;21) = 1,000). Regarding OVS, this order is not influenced by the verb type either, since there is no significant difference between transitive and unaccusative verbs (Table 37, cell (16;20) = 0,556), transitive and unergative verbs (cell (16;24) = 1,000) and finally, unaccusative and unergatives (cell (20;24) = 1,000). In the case of transitive and unergative verbs, apart from SVO, there is no significant difference among the three other orders (transitive verbs: VOS-VSO cell (13;15) = 0,848, VOS-OVS cell (15;16) = 1,000, and VSO-OVS cell (13;16) = 0,913. Unergative verbs: VOS-VSO cell (21;23) = 1,000, VOS-OVS cell (23;24) = 0,999, and VSO-OVS cell (21;24) = 1,000). As for unaccusative verbs, without including SVO, VSO does not show any significant difference with VOS ( cell (17;19) = 0,677) and with OVS ( cell Discussion of Results 137 (17;20) = 0,951), while between VOS and OVS there is a significant difference (cell (19;20) = <0,001). According to these results, the following conclusions are reached: - In the case of CF, the acceptability of all orders is not influenced by the verb type. - While SVO is by far the most acceptable option for any type of verb, there is no significant difference between the other three orders in transitive and unergative verbs. However, in unaccusative verbs there is no significant difference between VSO and VOS/OVS, but VOS and OVS show a significant difference compared to each other.