1. Multiple Factors in Second Language Acquisition: The CASP Model Luna Filipović (University of East Anglia) & John A. Hawkins (University of California Davis & University of Cambridge)

2. Introduction Second language acquisition involves the interplay of a number of factors that can either facilitate or impede learning. These include: the typological relationship between L1 and L2 general principles of learning and critical ages general principles of language processing (production and comprehension) social factors involving the general environment for learning, as well as pedagogical factors including teaching methods and materials and types of assessment.

3. Our goal in this talk: to outline a broadly based set of principles for a multi-factor model of learning in second language acquisition. We have developed these principles by drawing on theoretical insights from numerous branches of the language sciences, grammatical theory, typology, language processing, computational linguistics, first language acquisition and second language acquisition. We illustrate our findings using some data-driven studies in the field including our own, as reported in our recent book Criterial Features in L2 English (CUP, Cambridge).

4. The criterial feature concept is a new one that we use to describe the relative stages of second language acquisition. Criterial features are grammatical and lexical properties of an L2 (constructions, words and their meanings, rule types, errors and their frequencies, etc) that are distinctive and characteristic of proficiency at different levels of learning, e.g. at the six levels of proficiency described by the Common European Framework (CEFR).

5. Large electronic corpora of learner English make it possible for us to discover these features. Our 2012 book uses the Cambridge Learner Corpus (CLC), a corpus of 40 million words of examination scripts at levels 2-6 (CEFR A2-C2) from learners of English around the world who speak many different first languages. The CLC has been tagged for parts of speech and parsed using a sophisticated automatic parser (Briscoe et al. 2006) permitting numerous grammatical and lexical searches to be conducted. Between a third and one half has been error-coded using codes devised by researchers at CUP.

6. An analogy for Criterial Features Think of the defining characteristics for recognising faces in a police identikit. You don’t need to see all the features of a person’s face in order to distinguish that person from others, just the important characteristics that capture essential qualities

7. For example, Noun Phrase sequences of Pronoun plus Infinitive are found at level 2 (A2): something to eatlevel 2 nothing to dolevel 2 New features found at level 3 (B1) involve more complex syntax, e.g. an Object Control structure such as: I ordered him [to gather my men to the hall] level 3 This is a criterial construction for level 3 and higher levels and distinguishes them from levels 1 and 2.

8. A new feature for level 4 (B2) includes the so- called “Pseudocleft” structure with an initial what functioning as subject of its verb: What fascinated me was [that I was able to lie on the sea surface] level 4 “Subject-to-Object Raising” constructions with the verb believe appear first at level 5 (C1) and are criterial for this and the next level : I believe her [to be this country’s best representative] level 5

9. A Subject to Object Raising construction with the verb presume is not found before level 6: He presumed work to be the way to live level 6 Some criterial features are negative rather than positive, i.e. incorrect properties or errors that occur with a characteristic frequency for a particular level or levels (the “error bandwidth” for that level).

10. We can now ask: WHY do we see these patterns in the data and why do we see the criterial features changing the way they do at the different levels? In particular, WHAT is it about the features of the higher proficiency levels that makes them late acquired rather than early?

11. It cannot be that learners are simply imitating the words and constructions they are taught in their textbooks. First, because there are many different textbooks and teaching methods around the world. Second, because learners learn more than they are explicitly taught, from their reading materials, papers, magazines, movies, TV, conversations, and so on. I.e. second language learning shares many similarities with first language learning, but not all obviously.

12. We now have enough principles of learning with empirical support for their precise formulation and interaction to propose a model of SLA in the spirit of a complex adaptive system (Gell-Mann 1992) In this model multiple factors interact to produce a range of observable outcomes and different kinds of interlanguages Some of our principles have parallels in first language learning (Slobin 1977, Tomasello 2003, Diessel 2004, MacWhinney 2005). The model is called “CASP”, short for complex adaptive system principles of SLA. Some other recent models of SLA are also multi-factor and interactionist (e.g. Larsen-Freeman and Cameron 2008, Ellis & Larsen- Freeman 2009, MacWhinney 2005). Ours is symbolic rather than connectionist, defined in terms of grammatical and lexical primitives, typologically and psycholinguistically informed, and empirically driven

13. CASP: General Principles (A) Minimize Learning Effort (MiL) Learners of a second language (L2) prefer to minimize learning effort when they learn the grammatical and lexical properties of the L2. Learning effort is minimized when grammatical and lexical properties are shared between L1 and L2 and can be transferred directly into the L2, exploiting pre-existing knowledge from the L1. It is minimized when properties of the L2 are frequently occurring in the L2 input, which increases their exposure to the learner. It is minimized when structural and semantic properties of the L2 are simple rather than complex.

14. (B) Minimize Processing Effort (MiP) Learners of a second language (L2) prefer to minimize processing effort when they use the grammatical and lexical properties of the L2, just as native speakers do. There is a debate in the SLA literature over the extent to which developmental stages in SLA are shaped by ease of processing or by ease of learning. Our view is that there is a need for both, given certain dissociations between them, even though their predictions often overlap.

15. Principles (A) and (B) are principles of least effort. If these were the only principles determining learning and production our learner corpora would reveal increasingly minimal outputs. Clearly, they do not. MLUs (i.e. mean length of utterance figures) increase at higher proficiency levels (cf. Hawkins & Filipović 2012:ch.2.2) as greater use is made of less frequent and more complex structures and meanings. The reason is that learners are trying to increase their expressive power in the L2, and to behave like native speakers, which means learning and using the mix of infrequent and frequent, and complex and simple, linguistic items, just like native speakers do.

16. ( C ) Maximize Expressive Power (MaE) Learners of a second language (L2) prefer to maximize their expressive power, i.e. to formulate in the L2 whatever thoughts they would wish to express in the L1, and to perform the same language functions as L1 users. Successive stages of acquisition reveal more native-like L2 outputs with increasingly complex and less frequent structures for the expression of increasingly complex thoughts, in partial opposition to principles (A) MiL and (B) MiP.

17. (D)Maximize Communicative Efficiency (MaC) Learners of a second language (L2) prefer to maximize their communicative efficiency in relation to the hearer and his/her mental model. Communication is efficient when the message (M) intended by the speaker (S) is calibrated to the hearer's (H) mental model in such a way as to achieve accurate comprehension of M with rapid speed. This requires sometimes more, sometimes less, processing effort, in partial opposition to principle (B) MiP.

18. CASP: Specific Principles (1) Maximize Positive Transfer (MaPT) Properties of the L1 which are also present in the L2 are learned more easily and with less learning effort, and are readily transferred, on account of pre-existing knowledge in L1.

19. Shared L1/L2 properties should result in earlier L2 acquisition, in more of the relevant properties being learned, and in fewer errors, unless these shared properties involve e.g. high complexity and are impacted by other factors. Principle (1) (MaPT) derives from general principle (A) MiL. Positive transfers are also good for processing since processing mechanisms used for production and comprehension in the L1 can be applied directly to the L2, see (B) MiP. Positive transfers also enhance the expressive power of the non- native L2 user, see (C) MaE. Communicative efficiency can also be maximized by positive transfers, see (D) MaC.

20. The next slide shows missing determiner error rates for “the” and “a” at successive proficiency levels for French, German and Spanish learners of English, and for Turkish, Japanese and Russian learners. I.e. errors such as I spoke to President (missing the) I have car (missing a) The first three L1s have an article system similar to English, the latter three do not.. The figures indicate the percentage of errors with respect to the total number of correct uses. E.g. a percentage of 10.0% means a determiner was omitted 1 in every 10 times it should have appeared.

21. Determiner Error Rates in L2 English: Missing “the” Levels 2 3 4 5 6 French4.76 4.675.01 3.11 2.13 German0.00 2.564.11 3.11 1.60 Spanish3.37 3.624.76 3.22 2.21 Levels 2 3 4 5 6 Turkish22.06 20.7521.32 14.447.56 Japanese27.66 25.9118.72 13.809.32 Russian14.63 22.7318.45 14.62 9.57 I.e. generally low error rates for French, German and Spanish, without significant deviation between levels. Turkish, Japanese and Russian show much higher error rates, with significant improvements especially at levels 5 and 6. These figures support principle (1) MaPT.

22. Determiner Error Rates in L2 English: Missing “a” Levels 2 3 4 5 6 French6.60 4.796.56 4.763.41 German0.89 2.903.83 3.62 2.02 Spanish4.52 4.287.91 5.16 3.58 Levels 2 3 4 5 6 Turkish24.29 27.6332.48 23.8911.86 Japanese35.09 34.8024.26 27.4115.56 Russian21.71 30.1726.37 20.8212.69 These figures show a similar pattern to the last slide, and similar support for (1) MaPT.

23. (2) Maximize Frequently Occurring Properties (MaF) Properties of the L2 are learned in proportion to their frequency of occurrence (e.g. as measured in the BNC): more frequent exposure of a property to the learner facilitates its learning and reduces learning effort. I.e. more frequent properties will result in earlier L2 acquisition, more of the relevant properties learned, and fewer errors, in general.

24. Consider the basic sentence types and constructions of English described in terms of “verb co-occurrence frames” (Williams 2007, Hawkins & Filipović 2012). At the earliest level (2) we find simple and frequently occurring intransitive sentences types (he went), transitive types (he loved her), and basic three-place predicate types with a prepositional phrase (she added the flowers to the bouquet). At the higher levels (3, 4 and above) we see more complex and less frequent sentence types, for example different embeddings (he explained how to do it, he asked whether he should come, he told the audience that he was leaving) and gerundive verbs with –ing (I caught him stealing, they worried about him drinking). There is a precise correlation between the order of acquisition and degree of frequency in the input, cf. (2) MaF:

25. Average Token Frequencies in Native English Corpora (including BNC) for New Verb Co-occurrence Frames at the Learner Levels Levels 234/5/6 BNC frequency 1,041,634 38,174 27,615

26. (3) Maximize Structurally and Semantically Simple Properties (MaS) Properties of the L2 are learned in proportion to their structural and semantic simplicity: simplicity means there are fewer properties to be learned and less learning effort is required. I.e. simpler properties will result in earlier L2 acquisition, more of the relevant properties learned, and fewer errors, in general. Cf. Hawkins (2004, 2009) for issues in language complexity.

27. E.g. simpler consonants and consonantal distinctions in a phonological inventory are acquired earlier than more complex ones (see e.g. Eckman 1984). Simpler relative clause constructions are acquired earlier than more complex ones (Hawkins & Filipović 2012). Simpler and more basic meanings for verbs are acquired earlier than more complex and derived extensions in meaning, figurative uses, etc

28. The verb break in its basic physical sense at level 2; break in the sense of INTERRUPT (break the routine) level 3; break an agreement, promise, etc. level 4; break the bank (idiomatic) level 5; break the wall that surrounds him (original figurative) level 6. Cf. Hawkins & Filipović (2012:ch.8)

29. (4) Permit Negative Transfer (PNT) Properties of the L1 which are not present in the L2 can be transferred, resulting in errors, as learners strive to achieve an expressive power and communicative efficiency in L2 comparable to that in their L1 (see principles C and D), while minimizing learning effort (principle A) and/or processing effort (principle B). I.e. when grammatical and lexical properties are shared, transfers from L1 into L2 result in positive or correct properties in the L2. When properties are not shared, and the transfer still takes place, this results in negative or incorrect properties in the L2.

30. Positive transfers are maximized on account of general principles (A)-(D). Negative transfers are different. Sometimes they occur and sometimes they don’t, i.e. they are not maximized. This is one of the big unresolved issues in current SLA. When will they occur and when not? The general and more specific principles of our model can help us understand this. We see negative transfers as motivated by the desire to maximize expressive power (principle (C) MaE) and also to maximize communicative efficiency (principle (D) MaC) in an L2 system that has been incompletely learned, while at the same time minimizing learning effort (principle (A) MiL) and processing effort (principle (B) MiP). the same general forces that result in positive transfers lead to negative transfers as well.

31. But the major difference is that there are severe limitations on expressive power and on communicative efficiency that can be conveyed by linguistic properties that are not part of the L2 and not used by its native speakers. When native speakers communicate with L2 learners they tolerate and compensate for departures from the native language conventions. But when learners depart too radically from these conventions, they are not understood by native speakers. Learners accordingly acquire a sensitivity to the native speaker’s ability to compensate for these violations in conventions of grammar and use. This, we believe, plays a major role in determining whether and when negative transfer can occur.

32. In phonology substitutions of L1 consonants like [t] or [s] or [f] for L2 [θ] in English thin minimize learning and processing effort for learners whose L1s do not have this consonant, generally with communicative success (Lado 1957). In syntax Spanish Pro-Drop (e.g. *is a beautiful country for it is a beautiful country) is often transferred into early L2 English to express the proposition in question and the removal of the subject does not impede communicative success. This structure is simpler than its English counterpart with an overt subject, and transfer is not blocked, as predicted by our principle (3) MaS. Similarly, many article omission errors do not diminish expressive power and communicative success, and at the same time they minimize learning and processing effort through the transfer of L1 structures, see above.

33. By contrast, Chinese prenominal relative clauses do not result in errors whereby the English man whom the woman loves is changed into its Chinese prenominal counterpart *the woman loves whom man This Chinese structure is complex and typologically marked cross- linguistically (Hawkins 2004). Complex or marked structures in an L1 without an L2 equivalent will not generally transfer negatively, on account of principle (3) MaS (cf. Eckman 2011). Similarly complex or less frequent structures and meanings may not transfer from L1 to an L2 even when they are shared, in opposition to principle (1) MaPT. What this means is that principles (2) MaF and (3) MaS can block both positive and negative transfers into L2.

34. We propose principle (5), which derives from the need for communicative efficiency (principle (C) MaC), and which ultimately reflects the sensitivity of learners to their native speaking interlocutors and the latter’s tolerance for errors. (5)Communicative Blocking of Negative Transfer (CBN) The transfer of negative properties from L1 to L2 is filtered in proportion to communicative efficiency (principle D): the more an L1 property impedes efficient communication in L2, the less negative transfer there is.

35. Consider the basic word orders of English and Japanese. These languages have mirror-image patterns, head-initial versus head-final, that are both frequent and productive across languages: [went [to [the cinema]]] versus [[[the cinema] to] went ], Greenberg 1966; Dryer 1992; Hawkins 1983, 1994, 2004. Head-final orders are not transferred into L2 English by Japanese learners because, we argue, that would result in extreme communicative inefficiency: speakers using Japanese word orders in English L2 would simply not be understood! By contrast, head- initial word order variants of Spanish that lack precise counterparts in English (e.g., I read yesterday the book) are negatively transferred into L2 English, since they do not impact efficient communication.

36. We predict that because Japanese is a head-final language, the contrast with the mirror-image word order patterns of English is considerable and transferring head-final patterns into a head-initial language like English, and vice versa, would significantly impair communication. This is why it is imperative for Japanese learners of English, and English learners of Japanese, to acquire correct basic word orders in their L2s early. But speakers of L1 languages with flexible SVO like Spanish do not have the same incentive, because even when they transfer incorrect orders from their L1s into a fundamentally similar head-initial English L2, communication is not significantly impaired.

37. (6)Order of Second Language Acquisition (OSLA) The order of acquisition for properties of the L2 is in accordance with general principles (A)-(D), and with the more specific principles and patterns that are supported empirically. These principles can be incorporated within a multi-factor model of SLA, the CASP model, and used to define possible versus impossible, and likely versus unlikely, interlanguage stages proceeding from a given L1 to a given L2. These principles operate collectively to make constrained predictions for the acquisition of properties of L2 English and of other languages, and for their relative sequencing. Their interaction is complex, because there are several such principles, which sometimes compete and sometimes cooperate, because they are gradient, and because they have different relative strengths.

38. Key sources: Filipović, L. & J.A. Hawkins (forthcoming 2013) ‘Multiple factors in second language acquisition: The CASP model’ Hawkins, J.A. & L. Filipović (2012) Criterial Features in L2 English: Specifying the Reference Levels of the Common European Framework. CUP, Cambridge. References Briscoe, E., J. Carroll and R. Watson (2006) ‘The second release of the RASP system’. In Proceedings of the COLING/ACL 2006 Interactive Presentation Sessions, Sydney, Australia. Council of Europe (2001) Common European Framework of Reference for Languages: Learning, teaching, assessment. CUP, Cambridge. Diessel, H. (2004) The Acquisition of Complex Sentences. CUP, Cambridge. Dryer, M.S (1992) ‘The Greenbergian word order correlations’, Language 68: 81-138. Eckman, F.R. (1984) ‘Universals, typologies, and interlanguage’. In: W.E. Rutherford, ed., Language Universals and Second Language Acquisition, John Benjamins, Amsterdam, 79-105. Eckman, F.R. (2011) ‘Linguistic typology and second language acquisition’. In: J.J. Song, ed., The Oxford Handbook of Linguistic Typology, OUP Oxford, 618-633. Ellis, N.C. & D. Larsen-Freeman (2009) ‘Constructing a second language: Analyses and computational simulations of the emergence of linguistic constructions from usage’. In N.C. Ellis & D. Larsen- Freeman (eds.), Language as a Complex Adaptive System, Wiley-Blackwell, Chichester, 90-125.

39. Gell-Mann, M. (1992) ‘Complexity and complex adaptive systems’. In J.A. Hawkins & M. Gell- Mann, eds., The Evolution of Human Languages, Addison-Wesley, Redwood City, CA. Greenberg, Joseph H. (1966) ‘Some universals of grammar with particular reference to the order of meaningful elements’. In J.H. Greenberg, ed., Universals of Language, MIT Press, Cambridge, Mass., 73-113. Hawkins, J.A. (1983) Word Order Universals. Academic Press, New York. Hawkins, J.A. (1994) A Performance Theory of Order and Constituency. CUP, Cambridge. Hawkins, J.A. (2004) Efficiency and Complexity in Grammars. OUP, Oxford. Hawkins, J.A. (2009) ‘An efficiency theory of complexity and related phenomena’. In D. Gil, G. Sampson & P. Trudgill, eds., Complexity as an Evolving Variable, OUP, Oxford, 252- 268. Lado, R. (1957) Linguistics Across Cultures: Applied Linguistics for Language Teachers. University of Michigan, Ann Arbor, Michigan. Larsen-Freeman, D. & L. Cameron (2008) Complex Systems in Applied Linguistics. OUP, Oxford. MacWhinney, B. (2005) ‘A unified model of language acquisition’. In J.F. Kroll & A.M.B. de Groot, eds., Handbook of Bilingualism: Psycholinguistic Approaches, OUP, Oxford, 49- Slobin, D.I. (1977) ‘Language in childhood and in history’. In J. Macnamara, ed., Language Learning and Thought: Perspectives in Neurolinguistics and Psycholinguistics, University of Maryland Press, Maryland, 185-214. Tomasello, M. (2003) Constructing a Language: A Usage-based Theory of Language Acquisition. Harvard University Press, Cambridge, Mass.

40. Acknowledgments We acknowledge, with gratitude, the following sources of financial support for the research reported in this paper: Funding from Cambridge Assessment and Cambridge University Press to both authors for completion of the 2012 CUP book Criterial Features in L2 English. A Leverhulme and Newton Trust Postdoctoral Research Fellowship to the first author (at the University of Cambridge 2008-2011). Research funds to the second author from the Research Centre for English and Applied Linguistics University of Cambridge (2008-2011), including a teaching buyout at the University of California Davis (2008- 2010), and research funds from UC Davis (2008-2010), including a UC Davis Seed Grant for International Outreach.