Constraints – Intersentential / Intrasentential • Codeswitching Is Structured (I.E., Rule-Governed) 1
Total Page:16
File Type:pdf, Size:1020Kb
Programs and Proposals in Codeswitching Research: Unconstraining Theories of Bilingual Language Mixing Jeff MacSwan Arizona State University Overview of the Talk • Central claims – CS research has evolved into two distinct research programs • The Constraint-oriented Program • The Constraint-free Program – The constraint-free program is the more promising path for future research in CS •Outline – The search for a constraint-free solution in CS – The lexicalist solution available in the Minimalist Program • The model • Some sample analyses – A taxonomy of research programs (Lakatos, Feyerabend) – Directions for future research in CS What is Codeswitching? • Codeswitching, or CS for short, is a speech style in which fluent (i.e., simultaneous) bilinguals move in and out of two (or more) languages. • Topics in the literature – Social motivation / grammatical constraints – Intersentential / intrasentential • Codeswitching is structured (i.e., rule-governed) 1. Los estudiantes have seen the Italian movie ‘The students have seen the Italian movie’ 2. *Los estudiantes habían seen the Italian movie ‘The students have seen the Italian movie’ • The research question – What is the underlying structure? What do we mean by “constraint”? • Descriptive sense – Codeswitching is constrained in the descriptive sense – Some patterns are well-formed, others are not 1. My big brother va a comer con nosotros. 2. *He va a comer con nosotros. • Theoretical sense – In the theoretical sense, a constraint is a principle or rule of grammar that bars one or another structure. Constraint-oriented Approach Grammar Codeswitching 2 Grammar1 Constraints Codeswitched Utterances Constraint-free Approach Grammar2 Grammar1 Codeswitched Utterances Constraint-free Approach • Research Agenda – Nothing constrains CS apart from the requirements of the mixed grammars. • No CS-specific mechanisms are permitted – So, CS is the free union of two grammars, with no regulatory grammatical mechanisms permitted. The Search for a Constraint-free Solution in the History of CS • Pfaff (1979) – “It is unnecessary to posit a third grammar to account for the utterances in which the languages are mixed.” • Poplack & Sankoff (1981) – “What is more consistent with the data is simply to allow the possibility that in the uttering of a sentence, the rules used to construct its constituents may be drawn at times from one monolingual grammar and at times from another.” • Woolford (1983) – “There is no need to propose any sort of third, separate code-switching grammar.” • Lipski (1985) – “… gratuitous meta-structures [should] be avoided whenever possible, and … bilingual language behavior [should] be described as much as possible in terms of already existing monolingual grammars.” The Search for a Constraint-free Solution in the History of CS • Di Sciullo, Muysken & Singh (1986) – CS “can be seen as a rather ordinary case of language use, requiring no specific stipulation.” • Clyne (1987) – CS may be “governed by the kinds of structural constraints applying to monolingual performance.” • Belazi, Rubin & Toribio (1994) – CS is “constrained solely by Universal Grammar.” • Mahootian & Santorini (1995) – “… codeswitching sequences are governed by exactly the same principles of phrase structure as monolingual sequences.” Characteristics of Proposed CS Constraints • Researchers directly confront the formal limitations of the grammatical mechanism, then reluctantly but explicitly introduce CS-specific devices. – Sankoff & Poplack (1981) • The constraint or analytic framework is left inexplicit or is not adequately developed. – Woolford (1983) – Mahootian & Santorini (1996) • Researchers develop CS-specific mechanisms but argue that they are vacuously available to monolinguals too. – Di Sciullo, Muysken & Singh (1986) – Belazi, Rubin & Torbio (1994) – Myers-Scotton (1993); Jake, Myers-Scotton & Gross (2006) Sankoff and Poplack’s (1981) Equivalence Constraint • Sankoff and Poplack sought to implement the Equivalence Constraint – Codes will tend to be switched at points where the surface structures of the languages map onto each other. • They noticed that free union of phrase structure grammars would yield ungrammatical results NP → Det Adj N (English PS rule) NP → Det N Adj (Spanish PS rule) • Generates • *The casa white (‘The white house’) • *The house white PS and Lexical Insertion Rules in CS Pick your lexical Resulting insertion rule structures … Pick your PS rule NP Adj → white Adj → blanca N Det Adj Det → the NP Det → la N → house Det N the house White Adj N → casa la casa blanca …. Sankoff and Poplack’s (1981) Equivalence Constraint • The problem of lexical insertion –In the Aspects and early GB period, structures are built from the top down, with lexical insertion postponed until well after the word order had been laid out. – The language contributing the lexical items has strong consequences for the syntactic structure at the onset, but is not identified until the end, at lexical insertion. Sankoff and Poplack (1981) • Sankoff and Poplack’s solution – Superscripting or tagging convention of PS rules •NP → Det Nsp:n Adjsp:adj • the casa blanca – Limitations • Ad hoc, does not follow from otherwise available properties of the grammatical system. • Subscripting mechanism is not made explicit. Woolford (1983) • “Phrase structure rules are drawn freely from both grammars during the construction of constituent structure trees, but the lexicon of each grammar is limited to filling only those terminal nodes created by phrase structure rules drawn from the same language. Nevertheless, in the event that there are phrase structure rules common to both languages, such rules belong simultaneously to both languages. Lexical items can be freely drawn from either language to fill terminal nodes created by phrase structure rules common to both languages.” • But there is no specification of the formal mechanisms! The Government Constraint • Di Sciullo, Muysken & Singh (1986) • Government Constraint –If Lq carrier has index q, then Ymaxq. – In a maximal projection Ymax, the Lq carrier is the lexical element that asymmetrically c-commands the other lexical elements or terminal phrase nodes dominated by Ymax. – Anti-government requirement on CS The Government Constraint • Lexical insertion rule in Di Sciullo, Muysken & Singh (1986) – “All elements inserted into the phase structure tree of a sentence must be drawn from the same lexicon.” • Lexical insertion rule: – If X governs Y, … Xq …Yq … – q indexes a category to a language-particular lexicon • q is a “language index” • CS “can be seen as a rather ordinary case of language use, requiring no specific stipulation” (Di Sciullo, Muysken & Singh, 1986). • However, the “language index” is needed only in the case of CS, and is therefore a CS-specific device. The Quest for a Constraint-free Approach • Researchers in codeswitching have long acknowledged the preference for a constraint-free solution. • The central problem – Late lexical insertion – How do you get the lexical items to line up with the right terminal nodes in a P-marker? • A lexicalist solution – In the Minimalist Program, lexical insertion occurs at the onset, and language-particular properties are carried along with each lexical item into the syntax. – The problem of late lexical insertion disappears • No need to match languages up with their respective phrase structure rules later on The Minimalist Program Lexicon (Rules of Word Formation, Lexical Phonology) Select (CHL) Lexical Array (Numeration) Overt Component (Merge, Move) (CHL) Spell-Out Phonological Covert Component Component (Move) (CHL) (CHL) PF LF Codeswitching in the Minimalist Program Lexicon Lexicon(Lx) Lexicon(Ly) (Rules of Word Formation, (Rules of Word Formation, (Rules of Word Formation, Lexical Phonology) Lexical Phonology) Lexical Phonology) Select (CHL) Select (CHL) Lexical Array Lexical Array (Numeration) Monolingual (Numeration) Bilingual Version Version Overt Component Overt Component (Merge, (Merge, Move/Internal Move/Internal Merge) (CHL) Merge) (CHL) Spell-Out Spell-Out Phonological Covert Phonology(L ) & Covert Component x Component Phonology(Ly) (CHL) Component (Move) (CHL) (CHL) (Move) (CHL) PF LF PF LF Switching in DPs • Asymmetrical switching in Spanish-English DPs –los teachers (‘the teachers’) – *the casa (‘the house’) What formal differences might separate Spanish and English DPs? • Spanish has grammatical gender, English does not. • Suppose that Spanish and English φ-features differ as a set in this way: – Spanish: φ = {person, number, gender} – English: φ = {person, number} Switching in DPs • The CS construction is well-formed if the φ-set of N is included in the φ-set of D – Spanish D, English N •D, φ = {person, number, gender} •N, φ = {person, number} – *English D, Spanish N •D, φ = {person, number} •N, φ = {person, number, gender} Analysis • Accounting for the data – los teachers (‘the teachers’) case •The φ-set of N is able to value the φ-set of D “in one fell swoop” because its members are included in D’s φ-set; the derivation converges. – *the casa (‘the house’) case •The φ-set of N cannot find a corresponding φ-set in D for feature valuation because N’s φ-set is not included in D’s φ-set; the derivation crashes. New Questions Arise! •Data – What are the facts for other language pairs with similar gender systems (identical φ-sets)? – What are the facts for other language pairs dissimilar gender systems (properly included φ-sets)? • Theory – Is feature valuation satisfied under a subset relation rather than