Extended Processability Theory and Its Application to L2- Swedish

Extended Processability Theory and its Application to L2- Swedish

Leo Conroy The University of Melbourne

In this article, I discuss canonicity, iconicity and unification (of grammatical information) from the perspective of Extended P[rocessability] T[heory] (Pienemann 2005) and will illustrate my operationalisation of those notions in the analysis of the longitudinal data representing the developing L2-Swedish interlanguage of one informant. My analysis suggests that canonical and/or iconic form-function mappings play a role in L2-acquisition. The lexically driven progression from canonical to non-canonical mapping may materialise or fail to materialise in different ways. Some canonical/iconic mappings appear to facilitate the unification of grammatical features in a systematic, interlanguage-based and/or target-like manner, without impeding the development of the interlanguage. Other purely interlanguage-idiosyncratic mappings may be transitional, giving way to target-like forms, while others may become entrenched and block the process of convergence toward the target language.

Keywords: L2-acquisition, processability, lexicalist orientation, canonicity, iconicity, syntagmatic juxtaposition, unification, target language, interlanguage, bad choice hypothesis, continuity assumption, generative entrenchment

1 Introduction

The objective of this article is not so much to investigate (Extended) PT’s claims concerning the sequence of acquisition of specific linguistic structures. Instead, I intend to examine L2 phenomena captured by such concepts as canonicity which have been accorded a more prominent and theoretically more strongly motivated position within the framework of Extended PT. I will nevertheless broadly characterise my informant’s production of L2-Swedish in terms of the sequential stages of (Extended) PT’s hierarchy of processability. The notions of canonicity and unification are both central in Extended PT’s conceptualisation of language acquisition and production/comprehension, while the notion of iconicity, although not explicitly invoked by proponents of Extended PT, nevertheless overlaps conceptually with that of canonicity to a considerable degree and, in my own view, can be profitably exploited to further enrich the explanatory scope of canonicity. A further construct, that of syntagmatic juxtaposition (of morphophonologically harmonious forms), represents my own attempt to conceptualise my intuition that the notion of unification, if applied uncritically to structural phenomena exhibiting an apparent

65 LEO CONROY exchange of grammatical information, may conceal other, perhaps phonologically rather than morphosyntactically based processes involved in second language production.

2 The Explanatory and Predictive Scope of Extended PT

2.1 Canonicity – Its Universal and Language-particular Dimensions

Extended PT predicts that L2-learners will initially map conceptual structure onto surface structure canonically according to the principles of the “Unmarked Alignment Hypothesis”1 which defines canonicity as a mapping process of conceptual structure onto surface structure characterised by a “linear relationship between a-structure, f-structure and c-structure” (Pienemann et al. 225-226; 229). These structures represent “prominence hierarchies” (Choi 24) relating to thematic/syntactic roles (a- structure), abstract grammatical functions (f-structure) and the ordering of surface lexico- grammatical forms (c-structure).2 Hence, canonical or linear mapping, which corresponds to Levelt’s notion of “congruent grammatical encoding” (266), entails the assignment of the most prominent thematic role (AGENT) to the most prominent grammatical function (SUBJ) and its surface realisation in the most prominent c-structural position (first position in its clause). Particularly in the earliest stages of language acquisition, learners are hypothesised to be guided by “th[is] architecture of syntax with its three levels of structure” (Pienemann et al. 231). However, it is important to note that, with regard to the acquisition of c-structural configurations, the L2-learner is viewed as reconstructing the canonical patterns of the target language on the basis of L2-input. Pienemann et al. refer to “the default [c-structural patterns] the learner creates earlier” which may conflict with later, non-default “mapping processes” (202). In order for the L2-learner to establish the default, c-structural configurations of the L2, s/he must experience and “[identify] c-structure regularities” (236), as s/he “does not know in advance what the relevant canonical mapping of the TL will be” (209). This claim is consonant with Pienemann’s long- held assumption that L2-acquisition is very much a bottom-up process of reconstruction (Pienemann 1998: 80-83) which, within a lexicalist framework such as (Extended) PT, is logically conceptualised

1 This hypothesis is derived from Lexical-Functional Grammar’s “Lexical Mapping Theory” (Bresnan 302-20) and Optimality Theory’s notions of “Harmonic Alignment” (Lee qtd. in Pienemann et al. 224) and of the “emergence of the unmarked” (McCarthy 212-3). Pienemann et al. point out that “the notion of direct [canonical] mapping goes back a long way” and they refer in particular to the work of Slobin (1985) and Pinker in formulating this notion (227-8). 2 In Extended PT, a-structure and f-structure are construed as universal hierarchies of semantico-syntactic and grammar-functional representation respectively. The realisation of c-structure, on the other hand, is largely language-particular, although, even at this surface level of lexico-grammatical representation, there are dominant tendencies cross-linguistically, such as the relative ordering of “subject” and “object”, viz. subject > object.

66 EXTENDED PROCESSABILITY THEORY as proceeding, to a greater or lesser extent, on a lexeme-by-lexeme basis (Pienemann 1998: 63; 72; 83; 86; 92; 106; 109). Pienemann et al. appear to employ the term “canonical” somewhat ambiguously. They use it to refer to phenomena involving the abstract, universal levels of thematic/syntactic representation (a-structure and f-structure). However, they also use it to refer both to “universal [i.e. canonical] aspects of c-structure” (230) and to the canonical or unmarked surface- structure configurations specific to each language (209). Their assertion that L2-learners do not know a priori the canonical patterns of the target language would seem to imply that L2-learners are programmed to induce and assimilate canonical patterns. The example below illustrates canonical correspondence relationships of clausal constituents at the three levels of representation which entail “the computationally least costly manner of organising L2 syntax” (Pienemann et al. 230).

1) Hon läser svenska… agent pred. theme → a-structure SUBJ PRED OBJ → f-structure NPsubj Vfin NPobj → c-structure (“She is studying Swedish…”)

Pienemann, citing Andersen, also assumes that, at a basic surface-structural level, early interlanguage may well be characterised by (idiosyncratic) “canonical (i.e. one-to-one)” (Pienemann et al. 202) form-function mappings (Pienemann 1998: 159). Givón’s principle of iconicity implies, likewise, a “one-to-one correlation between form and meaning” (189). Indeed, as Pienemann et al. assert, “one-to-one correspondence as the natural default follows from a processing perspective as well as from an epistemological perspective” (224). From the discussion so far it will be apparent that the term “canonical” is employed by Pienemann (et al.) to refer to a range of phenomena which can be characterised as entailing basic (and, in some cases, universal), linear one-to-one abstract function-function and more concrete form-function correspondences.

2.2 The Processability Hierarchy – Processing Procedures and Non-linearity

The successive stages in the acquisition of an L2 involve the sequential development or activation of implicationally and hierarchically ordered, highly task-specific processing procedures – both lexical and syntactic (the latter functioning as grammatical memory stores) – which enable the learner’s stage-wise transition from the non-language-particular encoding of conceptual structure as surface structure according to the canonical principle of one-to-one correspondence – between a-structure, f- structure and c-structure – to a language-particular, surface encoding of conceptual structure

67 LEO CONROY involving increasingly greater degrees of “non-linearity”. Non-linear processing arises as a consequence of the “discontinuity [imposed on] the relationship between conceptualising and articulation” by language-particular “syntactic constraints” (Pienemann 1998: 56-7). Such discontinuity requires the L2-learner to temporarily store in grammatical memory (i.e. syntactic processing procedures) grammatical features which need to be exchanged or unified between constituents where one constituent (or conjunction of constituents) constitutes the source of an agreement phenomenon and the other constituent(s) the destination (Pienemann 1998: 73). The following represents the simplified processability hierarchy for L2-Swedish: stage 1 – unanalysed words and holophrases; stage 2 – lexical/categorial procedure (the assignment of words to lexical categories and the suppliance of those words with morphological features expressing such diacritic parameters as tense on verbs and number and definiteness on nouns; canonical word order - SVO); stage 3 – phrasal procedure (exchange of grammatical information between the constituents of NPs and complex VPs; clause-initial adverbials followed by SVO) ; stage 4 – S(entence) procedure (agreement between SUBJ and predicative adjective; inversion of SUBJ and finite V in interrogatives and in declarative main clauses after clause-initial focused constituents – [XP]VSO); stage 5 – interclausal procedure (distinction between main and subordinate clause word order). The notion of non-linearity refers to mappings – between the three levels of linguistic representation (i.e. a-, f-, and c-structure) – entailing varying degrees of deviation from the principle of canonical one-to-one correspondence as defined in 2.1. The degree of deviation is also reflected in the degree of “cost in terms of processing” (Pienemann et al. 201) entailed by non-linear, non- canonical mappings. The construct of “unmarked alignment” refers principally to the canonical mapping between a-structure and f-structure and more specifically to the intrinsic predicate-argument structure of individual verbs (as exemplified in 2.1).3 Beginning L2-learners are hypothesised to assign arguments to predicates according to the unmarked alignment hypothesis. The acquisition of the target-like, intrinsically non-canonical predicate-argument structure of “exceptional lexical entries” (e.g. “receive”) and “non-default verb forms” (e.g. passive lexical correlates of active verbs) is predicted in Extended PT to take place later than that of intrinsically canonical predicators (Pienemann et al. 223).

3 Within a lexicalist framework such as Extended PT, the target-like acquisition of the valence of individual predicators constitutes a language-particular learning task of pre-eminent importance because, metaphorically speaking, each verb is conceived of as “knowing” and “communicating” with its arguments (Pienemann 1998: 63; Pienemann et al. 223).

68 EXTENDED PROCESSABILITY THEORY

The “Unmarked Alignment Hypothesis” and the “Topic Hypothesis”4 are the two concepts which most strongly differentiate Extended PT from its immediate precursor (PT – 1998).

3 Methodology

3.1 The Data

In what follows, I will focus primarily on the longitudinal written data of one informant, to whom I will refer using the pseudonym Paul. He was taking a course in Swedish as part of his undergraduate studies and had no knowledge of Swedish prior to the commencement of the course. He had received instruction in Swedish for six weeks when the first set of data was collected from him. The data in general consist of two basic types – 1. assignments written as homework and 2. tasks performed in test conditions – and represent the informant’s production of L2-Swedish over a period of approximately three and a half semesters (i.e. ¾ of Beginners’ Swedish and all of Intermediate Swedish), comprising 29 points in time and a corpus of around 6,500 words. At the end of Paul’s two-year course, he had received approximately 280 hours of class-room instruction in L2-Swedish.

3.2 The Analytical Perspective Recommended by Pienemann

It should be borne in mind that (Extended) Processability Theory has gradually evolved out of research into the naturalistic acquisition of L2 German (cf. Meisel et al.; Clahsen et al.). However, studies of instructed SLA have generally confirmed that formal and naturalistic “acquisition processes are constrained by the same principles” (Pienemann 1987: 110) and that instructed and non-instructed L2 learners alike will only acquire a certain morphosyntactic structure when they are developmentally ready to do so (Ellis 1989: 307; 2002: 232; Lightbown 445).

4 The “Topic Hypothesis” (Pienemann et al. 232ff.), which I do not discuss in any great detail in this paper, refers to the mapping between c-structure and f-structure. The learner-controlled dislocation (Pienemann et al. 225) of the default topic (“subject”) from the c-structurally most prominent position (in essentially configurational languages, such as English, German and Swedish) entails an increasingly greater degree of non-canonicity/non-linearity according as the focused constituent represents a non-argument function (e.g. an adverbial adjunct = lesser degree of non-canonicity) or an argument function (e.g. direct object = greater degree of non-canonicity).

69 LEO CONROY

Pienemann5 states quite explicitly his adherence to an interlanguage-centred approach to the characterisation of each learner’s approximation of the target language (Pienemann 1998: 162-4; 2005: 77). Indeed, he emphasises that the systematic form-function mappings idiosyncratic to each learner’s interlanguage6 should form the basis on which to determine the developmental dynamics of that interlanguage, as each interlanguage is to be regarded as “a system entirely in its own right” (Pienemann 1998: 163; cf. also Bley-Vroman 4, and Klein). The interlanguage perspective assumes that, at any one stage of development, each individual interlanguage will be characterised, to a greater or lesser extent, by systematic form-function relationships which may or may not correspond to the norms of the target language. Pienemann also subscribes to “the assumption that the basic architecture of cognitive processes is comparable between individuals” (1998: 132). He therefore assumes that, while there may be inter-individual differences in the rate at which stages of the PT processability hierarchy are traversed, the sequence in which the processing procedures located at successive stages are acquired as “automatic processes” (1998: 84) will be the same for all L2-learners.7 Pienemann’s approach to the study of the “developing interlanguage system”, which conceptualises language acquisition in terms of the sequential progression through a series of stages characterised by an increasingly greater degree of “processing cost” (Pienemann et al. 227) seems to imply that “successful” language acquisition is a process entailing the gradual convergence of the (successful) L2-learner’s interlanguage toward the norms of the target language. This is clearly apparent in Pienemann’s discussion of the “bad choice hypothesis” (1998: 324ff.), the potential for the “generative entrenchment” (1998: 316ff.) of such bad choices and the “developmental dead end[s]” (1998: 325) characteristic of the (relatively unsuccessful) L2-learner’s fossilised interlanguage.

5 It is important to point out that Pienemann states quite clearly that “spontaneous oral language production” should be used to test the predictions of (Extended) PT (1998: xvi). Therefore, my use of written data limits the validity of my findings in charting the chronology of the stage-wise “emergence” of the procedural skills whose operation one may infer on the basis of the production of stage-specific linguistic structures. However, my primary objective in this article is to characterise Paul’s developing interlanguage in terms of the principles of canonicity and iconicity and to investigate to what extent such principles may facilitate or impede interlanguage development. Nonetheless, my tentative diagnosis of the stage attained by Paul in the various domains of L2-Swedish morphosyntax is based on the predictions of Extended PT. 6 Pienemann stresses the importance of using “individualised data” (2005: 77) and of “atomis[ing] linguistic contexts” (1998: 139) when investigating the process of interlanguage development. 7 The measurement of interlanguage development is to be carried out by means of the emergence criterion, according to which, in theory, one productive occurrence of a stage-specific syntactic structure constitutes sufficient evidence to credit the L2-learner with having attained that developmental stage (Pienemann 1998: 133). However, in the domain of morphology, a finely-grained distributional analysis of the learner’s production is required in order to establish which (interlanguage) forms perform which functions (144).

70 EXTENDED PROCESSABILITY THEORY

In this article, I will discuss aspects of Paul’s interlanguage which illustrate how the canonical/iconic, systematic one-to-one mapping of forms and functions can, indeed, lead to developmental dead ends. Such evidence indicates that there is a tension between the interlanguage perspective and the target language perspective and Pienemann’s characterisation of fossilised interlanguages suggests that he ultimately resolves this tension by opting for a target language perspective.

3.3 The Nature of the Learning Task and the Typology of the L2 - Swedish

In their exposition of Extended PT, Pienemann et al. refer to the L2-learner’s “[identification] of c- structure regularities” (236). This suggests that (lexically mediated) positive evidence is considered to be the source of the learner’s knowledge of language-particular structures (Pienemann 1998: 92). The “non-conscious” (Pienemann 1998: 41) extraction of such regularities from the L2-input will be influenced by two main factors: the (possibly innate) tendency to organise linguistic elements canonically and the degree of morphosyntactic complexity of the L2. Swedish is an inflectional language displaying many of the features commonly encountered among languages of its typological class such as the polyfunctionality and, in some cases, functional indeterminacy of individual forms and the formally heterogeneous realisation of individual functions (Cf. Bolinger 11). Two important factors facilitating learnability have been defined as “conceptual motivation” and “formal homogeneity” (Schwarze 150). It is therefore easy to predict that the specific typology of Swedish will render its target-like acquisition (by learners postulated to be guided by such abstract universal principles as canonicity) somewhat less than straightforward. Crucially, this clash between the relative complexity of certain aspects of Swedish morphosyntax and the canonical, iconic principles guiding the learner of L2-Swedish will often result in “decomplexifying”, and possibly idiosyncratic interlanguage phenomena which, as I will demonstrate on the basis of examples from my own data, may become “generatively entrenched” (Pienemann 1998: 316ff.) and thereby impede the gradual convergence of the learner’s language toward the target language.

4 Emergence of the “Canonical”

Paul’s early interlanguage realisation of the Swedish equivalent of the possessive pronoun “her” is suggestive of the instantiation of canonical patterns as the following examples8 illustrate:

8 The bracketed abbreviation (e.g. 1-Test) after each example refers to the time of data collection (e.g. 1=first) and the type and/or context of production of each task or set of tasks (e.g. Test=test context).

71 LEO CONROY

2) Hon har många saker i hennes kök She has many things in her kitchen She has many things in her kitchen… (1-Test)

3) R. är lång och hon hår är lockig R. is tall and she hair is curly R. is tall and her hair is curly. (3-Essay)

4) Hon-s barn äter frukost på mitbord …. C. har She-s children eat breakfast at dining table… C. has

lockig hår, tycker om hennes mor... curly hair, likes her mother... Her children are eating breakfast at the dining table….C. has curly hair, likes her mother… (5-Test)

A comparison of examples 2, 3 and 4 suggests that hon/hons (idiosyncratic interlanguage forms based on the subject pronoun hon – “she”) and hennes (the target-concordant non-reflexive possessive pronoun corresponding to “her”) are not simply free variants but occur in complimentary distribution. The form hennes occurs in the same structurally more deeply embedded contexts where one would expect to encounter the object pronoun henne in utterances displaying (Swedish) canonical word order. The forms hon/hons, on the other hand, occur in the c-structurally most prominent clausal position, onto which the subject constituent is canonically mapped. It would therefore seem reasonably plausible to propose that Paul’s differential lexico-grammatical realisation of the Swedish equivalent of the possessive pronoun “hers” is in some way guided by the innate or inferred principles underlying the unmarked alignment of c-structural configurations. As there are no further instances of this idiosyncratic contrast in Paul’s production, one may conclude that, after one semester of studying Swedish, he moves beyond a canonical and toward a more language-particular encoding of this conceptual structure.

4.1 The Canonical Decomposition of Lexicalisations

The decomposition of lexicalisations9 can be accounted for in Extended PT with reference to an initial, more canonical instantiation of categorial prototypicality. In other words, “exceptional lexical entries” will be, to varying degrees, de-exceptionalised. The examples below seem to provide evidence in favour of the notion that language learners subject lexicalisations to a lexico-grammatical

9 One would assume that even relatively idiomatic lexicalisations would be subject to decomposition by the beginning learner, as s/he cannot be assumed to be sensitive, a priori, to the degree of idiomaticity inherent in target-language lexicalisations (Pienemann 1998: 75; Bley-Vroman 13).

72 EXTENDED PROCESSABILITY THEORY analysis, which, in this particular instance, gives rise to an interlanguage construction which, while still involving a non-default mapping of a-structure to f-structure (Falk 109), nevertheless entails a more strongly canonical realisation of the “NP complement” of the VP.

5) Erik ska åka tåg från Stockholm-s station Erik will travel train from Stockholm-GEN10 station Erik is going to travel by train from Stockholm station. (1-Test)

6) Erik ska åka Åsa-s lägenhet Erik will travel Åsa-GEN flat Erik is going to drive [to?] Åsa’s flat. (1-Test)

In an expression such as åka tåg (“to travel by train”), as perceived and employed by a mature native speaker of Swedish, the second element tåg (“train”), which might otherwise be considered prototypically noun-like (Teleman et al. vol. 2: 20), forms part of a lexicalised complex predicate (Teleman et al. vol.3: 179) in which it loses its essentially noun-like properties including its potential to function as the head of a noun phrase and, hence, functions more like a verbal particle (Josefsson 241) than as an object NP. In other words, the vestigially nominal element in a lexicalisation such as åka tåg displays, morphosyntactically, a very high degree of “reduced categoriality” and, semantically, a low level of concrete referentiality (Hopper and Thompson 166-167). Accordingly, Paul’s production of the non-target-like analogous construction represented by example 6 is perhaps most plausibly interpreted as having been in some way pre-conditioned by a canonical decoding of tåg as a discrete, concrete entity which therefore at some level of linguistic representation has been re-invested with the canonical proto-properties typical of representatives of the category of nouns in question and their behaviour in categorially determined canonical syntactic environments (cf. Ackermann and Moore 56).

4.2 The Intrinsically Non-canonical Argument-structure of Passive Verbs

In Extended PT, passive verbs are designated as “exceptional lexical entries” because of their “intrinsically non-canonical argument-structure” (Pienemann et al. 223). The following example may represent an instance of Paul’s preference for a canonical realisation of thematic roles and lends support to Slobin’s (here slightly paraphrased) observations that learners “may well attend to non-

10Abbreviations used in this paper: 3 ‘third person’, ADJ ‘adjective’, ATT ‘attributive’, C ‘common (gender)’, DF ‘definite’, DFDT ‘definite determiner’, DFSX ‘definite suffix’, GEN ‘genitive’, INDF ‘indefinite’, NP ‘noun phrase’, N ‘neuter (gender)’, PASS ‘passive’, PL ‘plural’, PR ‘pronoun’, PRED ‘predicative’, PRES ‘present’, S ‘subject’, SG ‘singular’.

73 LEO CONROY canonical forms, but only to assimilate them to the canonical [function]” and that “new forms first express old functions” (1982: 168).

7) hon sygga-s till och tvätta-s honom she tidy-PRES.PASS up and wash-PRES.PASS him She tidies him up and washes him. (11-Précis)

This utterance affords a very favourable context for the canonical instantiation of the thematic roles of agent and patient as such roles are rather starkly contrasted in the conceptualisation of this “real” world event. Paul clearly perceives this contrast and (almost certainly unconsciously) converts the passive construction to be found in the original text into an active one, although he preserves the passive morphology, with which he may be assumed to have, at best, only a very weak acquaintance.11 (Paul produces no instances of target-like passive constructions throughout his entire production of beginner-level and intermediate-level Swedish. This fact alone suggests that active forms are more basic or canonical than passive forms). His preservation of the passive form is consistent with his less than optimal sensitivity to the featural distribution of Swedish morphology which results in his occasional production of forms displaying a fairly high degree of what Pienemann refers to as “categorial squishiness” (Pienemann 1998 : 50).

5 Suppliance and Unification of Grammatical Information.

One of the central features of Extended PT remains the notion of the suppliance or unification of grammatical information at different, hierarchically ordered syntactic levels as discussed in section 2.2 above. In this section, I will present further data produced by Paul and relate these data to the various stages of the Processability Hierarchy. By the time of the earliest data collections, Paul was producing morphological structures located at stages 2 (lexical morphology) and 3 (phrasal morphology) of the processability hierarchy. In 2-Trans(lation), he produces two (vägg-ar-na – NP:C.PL-DFSX and fönstr-en – NP:N.PL- DFSX) of the only half a dozen or so types of nouns (comprising a total of roughly 20 tokens in a corpus of approximately 6, 500 words) inflected for both plural and definiteness produced throughout the entire data collection period of three and a half semesters. In this area of lexical morphology, it is difficult to discern any real development in his interlanguage. There are only three contexts in which he appears to alternate morphologically complex and simplex forms of one and the same noun

11 At the point in time when Paul produced these “passive” forms, his class had not yet received much formal instruction in the formation of passive constructions.

74 EXTENDED PROCESSABILITY THEORY

(assignments 22-Essay, 23-Test and 29-Test ). In his final composition (29-Test), Paul alternates barn (NP-N.PL.DFSX = children – four tokens) and barn-en (NP-N.PL.DFSX = the children – five tokens). This type (represented by the tokens barn/barnen) is the only one in his entire production which displays a target-like formal contrast involving a noun bearing, contrastively, simple (in fact, zero) and complex affixation. In 27-Essay (a free essay), he resorts to a non-target-like use of the form -a12, and the three tokens of this form each performs a different function. Paul’s linguistic behaviour would seem to be informed by (often erroneous) hypotheses derived from the complexities of Swedish nominal inflection encountered in the input. However, certain other aspects of his production tend to indicate that he is in some way guided by a principle or hypothesis which prevents him from neutralising featural distinctions in contexts where the target language requires such neutralisation. Such an interpretation suggests itself in the case of a number of adjectivally premodified, common gender, definite plural NPs which Paul produces: den vackera öarna (TL: de vackra öarna = the beautiful islands), den gamla frackerna (TL: de gamla frackerna = the old suits) and den stora festerna (TL: de stora festerna = the big parties). In all three cases, the features of “plurality”, “common gender” (den) and “definiteness” would appear to have received overt marking. In the target language, the feature GENDER would be neutralised in these three instances. In particular, the definite determiner den, which bears the features “common gender” and “singular” would not occur in these contexts; instead, the gender- neutral, plural definite determiner de would be used. The inference that den is being used here to express the features “C(ommon gender)” and “D(e)F(inite)” is based on fairly scant evidence; on the one hand, Paul virtually only uses den as the syntactic definite article in common gender NPs (both SG and PL), and, on the other, there is some evidence to suggest that he preserves the same distinctions on the few occasions he attempts to produce complex neuter NPs, as the following example shows:

12 This morpheme is, in fact, used in standard Swedish to express the feature DF on “native” Swedish nouns belonging to the 5th declension, e.g. äpple-n-a (apple-PL-DF = the apples). It is also used to mark PL on INDF adjectives and DF on all adjectives. Furthermore, it marks either DF or PL+INDF on de-adjectivised nominals and PL on many words of Greek and Latin origin.

75 LEO CONROY

8) den vacker-a ö-ar-na, DFDT.C.SG beautiful-DF.C/N.SG/PL island-PL-DF.S(uffi)X.PL

det trevlig-a gaml-a hus-et DFDT.N.SG. nice-DF.C/N.SG/PL old-DF.SG/PL house-DFSX.N.SG

och det varm-t vädr-et and DFDT.N.SG warm-INDF.N.SG weather-DFSX.N.SG The beautiful islands, the nice old houses and the warm weather. (4-Trans)

In Paul’s Swedish interlanguage, the contrast between N.SG and N.PL adjectivally premodified definite NPs would appear to be expressed by means of differential adjectival inflection (i.e. –a for plural and –t for singular). In 23-Test (another translation – produced a year after 4- Trans), he translates the two tokens of the NP “under the telephone booths” by under telefonhus (NP- INDF.N.?SG/?PL) and under telefonhus-et (NP-DFSX.N.SG) respectively. The latter token is representative of his tendency to mark only one function morphologically. The target-like translation is under telefonhus-en. The suffix –en in this case expresses fusionally the features “neuter gender+definite+plural”. However, the same morpheme (or allomorph thereof) is also used to express the features “definite+singular” on common gender nouns and this function is almost certainly more canonical, given that nouns of common gender considerably outnumber those of neuter gender. Paul’s apparent preservation of distinctions which are neutralised in the target language may be interpreted as entailing a more canonical, one-to-one mapping of form and function.

5.1 Unification of Grammatical Information – Or Syntagmatic Juxtaposition?

It is of particular interest to compare Paul’s realisation of (stage 3) adjectivally premodified, indefinite NPs of the following two types: C.SG.ATT (common singular attributive) and N.SG ATT (neuter singular attributive). Although he produces fewer instances of the latter type, he manages to produce this type with a greater degree of consistency than corresponding NPs of the type C.SG.ATT. One possible explanation for this fact is that NPs of the type N.SG ATT (e.g. ett stor-t hus = a big house) are characterised by a much higher degree of “coding coherence” (Givón 210), “phonological harmony” (Kato 282), “formal homogeneity” (Schwarze 150) and, importantly, a concomitant lack of featural “neutralisation” (McCarthy 81), entailing distinct and salient marking, whereas indefinite NPs of the type C.SG.ATT are coded, non-iconically (Croft 104), by an absence of distinct marking (e.g. en stor-θ bil = a big car). In fact, since the C.SG.INDF.ATT form of the adjective simply corresponds to the base form of the adjective which, to a large extent, appears not to be subject to any inherent restrictions in terms of the form it may take, the learner is less likely to feel

76 EXTENDED PROCESSABILITY THEORY constrained when entertaining hypotheses about the formal suitability of a particular adjective s/he encounters and assumes to represent the base form of the adjective. In the case of N.SG adjectivally premodified definite NPs, they are produced fairly consistently according to the iconically/canonically motivated pattern det varmt vädret, which displays exponence of the feature “neuter gender” on all constituents (de-t varm-t vädr-et) and thus deviates from the target-like pattern (de-t varm-a vädr-et) where the feature “gender” is neutralised on the adjective. Adopting an interlanguage-oriented approach to determine well-formedness (in accordance with Pienemann’s own recommendations), one would have to accept as “well-formed” NP structures conforming to the pattern det varmt vädret as they clearly display exchange of grammatical information between phrasal constituents and appear to be the default form in Paul’s interlanguage. The fact that no target-like adjectivally premodified definite plural NPs are produced according to the norms of the target language can be explained, at least partly, by the simple fact that the target-language form de (they/ the-PL), which performs the functions of 3rd person plural subject pronoun and plural definite determiner, has been assigned the exclusive function(s) of 3.SUBJ/OBJ.PL.PR in Paul’s Swedish interlanguage.13 I will discuss this form below in my general discussion of the possible repercussions of unanalysed or inadequately analysed lexical entries on language processing. It is not possible to determine whether Paul attains stage 4 (interphrasal agreement) on the basis of his production of C.SG.PRED adjectives, because, like its indefinite attributive correlate, the form of the C.SG.PRED corresponds to the base form. Therefore, in order to establish whether Paul acquires stage 4 in the domain of morphology, one has to look at his production of N.SG.PRED adjectives. If one examines the percentage figures for the realisation of N.SG.PRED adjectives, one may conclude that they provide sufficient evidence in support of the assumption that Paul acquires this structure to some degree. However, a finely grained distributional analysis of all occurrences of this structure reveals a somewhat more complex picture (Pienemann 1998: 144). Assuming that the pronominal form det is annotated with N.SG specifications in Paul’s interlanguage, the only apparently productive marking of predicative adjectives for N.SG occurs in contexts which are highly favourable for such marking, e.g. in 8-Trans (a translation) which contains no fewer than 16 contexts (both attributive and predicative) for the instantiation of nouns and adjectives marked for neuter

13 For a discussion of the earlier emergence of 3rd person pronouns of forms functioning both as pronouns and as definite articles, see Pfaff (273).

77 LEO CONROY gender.14 In other cases, the N.SG.PRED(icative) adjective occurs in a clause in which the N.SG. NP antecedent is already modified by an appropriately inflected N.SG.ATT(ributive) adjective, or the predicative adjective is followed by a N.SG.DF NP adverbial adjunct (det ska vara varm-t i vattn-et - 15-Test). The fact that neuter singular predicative adjectives typically occur in contexts such as (17- Essay) below suggests that syntagmatic juxtaposition or spatial proximity (possibly in conjunction with “phonological harmony” – a notion I discuss below in section 7) may play a more decisive role in determining the agreement morphology than the operation of the stage 4 processing procedure hypothesised to underlie the production of this structure. In other words, the stage 3 N.SG.INDF.ATT form of the adjective has simply been transferred holistically from one syntactic context to another in adjacent clauses. It should also be noted that, in the second clause, the second adjective “vit” (“white”) has not been inflected to agree with its antecedent (TL: “vitt”).

9) de bodde i stor-t hus. they lived in big-N.SG.ATT.INDF house-N.SG.INDF

Det var stor-t, vit 3.N.SG was big-N.SG.PRED white-BASEFORM they lived in (a) big house. It was big, white…(17-Essay)

A further measure of whether Paul acquires stage 4 morphology involves applying Pienemann’s “version of the continuity assumption”, according to which the alternation of “rule application… with non-application in consecutive [assignments] in the presence of ample contexts for the rule” should be interpreted as indicative of the failure on the part of the learner to have truly acquired that rule in the first place (1998: 147). Paul fails to produce stage 4 morphology in any of the contexts in the last two (sets of) assignments collected from him. The following example shows how strong Paul’s tendency is to juxtapose lexemes bearing allomorphs marking neuter gender and displaying “phonological harmony”.15

14 Not surprisingly, this assignment contains instances of the overmarking of the feature “neuter gender” including NPs of the following type: e-tt vit-t bord-et – DT.N.SG.INDF+ADJ-N.SG.INDF.ATT+NP- N.SG.DFSX. 15 The first line contains randomly ordered words and phrases which the students were required to put together to form grammatically correct Swedish sentences, starting each such sentence with the constituent underlined.

78 EXTENDED PROCESSABILITY THEORY

10) Idag/fortfarande/inflytandet/stort/är/från engelskan (20-Test)

Idag är fortfarande stor-t inflytande-t från Today is still great-N.SG.INDF influence-N.SG.DFSX from

engelskan English

TL: Idag är inflytandet från engelskan fortfarande stort. Today, the influence from English is still great.

The evidence suggests, therefore, that, in the domain of morphology, Paul does not advance beyond stage 3. In the example above, he combines an adjective bearing the features N.SG and INDF with a noun bearing the features N.SG and DF to form an (iconically/phonologically motivated) adjectivally premodified phrase (stage 3), which is not well-formed from a target-language perspective, as it violates “the uniqueness principle” (Bresnan 47), according to which “every attribute has a unique value” (Bresnan). (In this example, the attribute DF has different values for each constituent of the phrase.) However, as I have discussed, from the perspective of Paul’s own interlanguage, the phrase stort inflytandet may be considered marginally well-formed, as it could simply be another instance of the overmarking of the feature “neuter gender” at the phrasal level. In Extended PT, Paul’s construction would be explained in terms of his not having acquired the S- procedure (stage 4). The simple factor of perceived formal congruity (“phonological harmony” – see section 7) would seem to have influenced his decision about the relative positioning of the constituents of this utterance.

6 Level of Attainment in the Domain of Syntax

The stage 4 structure of inversion16 enters Paul’s interlanguage at a relatively early stage but its developmental trajectory is characterised by a fairly high degree of fluctuation in target-like use. At certain points, it seems to disappear (almost) entirely from his interlanguage. However, if one analyses such contexts more closely, one finds that factors other than the non-acquisition of the structure of inversion may condition his failure to produce this structure. The primary factor in such instances would appear to be his use of lexical items which he has not annotated with the full range of target-like syntactic specifications. In other words, the variation in his realisation of “inversion”

16 Inversion refers to the V2-constraint, whereby the subject and the finite verb are inverted in main clauses introduced by a focused element (e.g. XPVSO).

79 LEO CONROY can be attributed to the effect of “immature lexical entries” (Pienemann 1998: 97).17 His failure to adequately annotate such lexical entries may be a further reflection of an initial, canonical one-to-one mapping of form and function in the process of his reconstruction of his L2-Swedish lexicon. While there seems to be sufficient evidence from which one may infer that Paul acquires the syntactic structure located at stage 4, the evidence relating to his production of the stage 5 structure of subordinate clause word order is both scant and ambiguous.

7 Iconicity Motivating Morpholexical Forms

It is possible to explain Paul’s consistent target-like realisation of the NP structure “neuter+singular+indefinite+attributive adj.+noun” (e.g. ett stort hus) in terms of iconicity/canonicity (Givón 189; Pienemann 1998 : 159) defined as the one-to-one correspondence between form and function. A similar factor may underlie his relatively early acquisition and consistently target-like realisation of the verb phrase “ska + infinitive”.18 Typical examples of this construction such as ska åka (“will travel”) and ska äta (“will eat”) illustrate the apparent morphophonemic “symmetry” characterising this combination of auxiliary and infinitive. No other combination of auxiliary/modal verb + infinitive displays such “symmetry” (e.g. kan/måste/vill åka – “can/must/wants to travel”; har/hade åkt – “has/had travelled”) and Paul produces each of them with a (much) lower rate of target-language concordance. Kato argues that “phonological harmony” (282) may play a facilitating role in the acquisition of certain morphosyntactic structures (agreement morphology) in some Romance languages. This phonological link may be broken when “the elements involved are [not] superficially adjacent” (283), resulting in the commission of errors in agreement morphology.19 This phenomenon was demonstrated by Paul’s failure to inflect N. SG predicative adjectives, as discussed in section 5.

17 For instance, in assignment 17-Essay, in which he produces inversion in only two out of nine obligatory contexts, he employs the bi-categorial lexeme “sedan” (temporal conjunction/temporal adverb) in four semantically appropriate contexts. The first context requires the use of “sedan” as a temporal subordinator with subordinate clause, i.e. non-inverted, word order, which Paul produces. However, the remaining three instantiations of “sedan” involve the use of this lexeme as a temporal adverb in declarative main clauses. Paul fails to produce obligatory inversion in any of those contexts. This suggests that his first instantiation of “sedan” activates a certain (appropriate) syntactic procedure, which he overgeneralises to inappropriate contexts. 18 Paul uses ska as a future auxiliary, whereas in standard Swedish it tends to encode a variety of modal meanings, some of which overlap with future reference. 19 Kato refers specifically to the acquisition of Brazilian Portuguese in which “gender can be acquired as a phonological relation since the article and the noun endings are vowel harmonic”, e.g. a casa (“the house”) and o cavalo (“the horse”) (Kato 282).

80 EXTENDED PROCESSABILITY THEORY

8 Evidence for the Lexical Nature of Language Acquisition and Production/Comprehension

A lexical item which causes Paul a good deal of trouble is de when used as the plural syntactic definite article. I indicated above that he had assigned to this lexeme the exclusive function of 3rd person plural pronoun. As the examples below will illustrate, his failure to annotate this lexeme with its full range of target language functions will ultimately lead to a breakdown in language processing. (For an explanation of the task involved, see footnote 14),

11) restaurangerna/ vad/ tyskarna/ svenska/ om/ de/ tycker (5Test)

Paul: Vad tycker om tysk-ar-na What think about/?like German.NP-PL-DFSX

svensk-a restaurang-er-na Swedish-?DF.ADJ.PL restaurants.NP-PL.DFSX

TL: Vad tycker tyskarna om de svenska restaurangerna? What do the Germans think about the Swedish restaurants?

In example 11), Paul obviously finds no use for de and simply omits it. Example 12) below forces him to use de (as part of a complex NP adverbial adjunct).

12) De senaste decennierna/ förändrat/ engelska/ mycket/ svenska språket/ har.

Paul: De senaste decenni-er-na har DFDT.PL last decades-NP-PL.DFSX ?have/?has

mycket förändrat svensk-a much changed Swedish.ADJ-SG.DF.ATT

språk-et engelska languageNP-N.SG.DFSX EnglishNP-C.SG.INDF [During?] the[?] last decades ?have/?has much changed the Swedish language English.20

It is difficult to assign syntactic functions to the various constituents of Paul’s construction. In fact, one can only say that Paul’s sentence is uninterpretable. It is tempting to speculate that the PL.DFDT de in the NP adverbial adjunct is the source of Paul’s confusion, given that, on the one

20 TL: De senaste decennierna har engelska förändrat svenska språket mycket = During the last decades, English has changed Swedish a lot.

81 LEO CONROY hand, he otherwise uses sentence-initial NP adverbial adjuncts with a certain degree of frequency and facility (e.g. den här semester, den tjugonde november – 15-Test), while, on the other hand, he never produces a single instance of de with the function of plural syntactic definite article. Paul’s final encounter with the DFDT de (in example 14 below), or more precisely the demonstrative pronoun de där, occurs in the test at the end of the first semester of intermediate-level Swedish (i.e. almost a year and a half after he started studying Swedish).

13) besöka/ de/ kanske/ vid/ sjutiden/ ska/ oss/ ikväll (23-Test)

Paul: De ska kanske besöka oss vid sjutiden TheyP.3.PL will perhaps visit us at about seven o’clock

ikväll this evening

14) böckerna/ gärna/ vi/ de där/ vill/ nu

Paul: De där köpa vi gärna vill Those [?] buyINF we [really] want to

böck-er-na nu books.NP-PL.DFSX now Those buy we really want to books now.21

These examples reinforce the impression that Paul has relatively few problems constructing sentences using the form de when it has the function of 3rd person plural pronoun. However, when the form de performs the function of syntactic definite article, he may simply omit it. But, regardless of whether he omits or retains this non-personal pronoun de, the sentences he produces are constructed in such a way as to be largely incomprehensible.

9 General Discussion

On the basis of the analysis of Paul’s Swedish interlanguage, one can tentatively conclude that at the end of his second year of studying Swedish, he had attained stage 3 (“the phrasal procedure”) in the domain of NP morphology. However, right throughout the data-collection period, Paul appeared to maintain a very strong tendency of marking only one function on nouns by means of lexical morphology. In other words, he

21 TL: De där böckerna vill vi nu gärna köpa = We really want to buy those books now.

82 EXTENDED PROCESSABILITY THEORY marked plural or definiteness on individual nouns, but very rarely both.22 Such behaviour may be attributed to the relatively high degree of complexity characterising both the inflectional morphology and syntax of Swedish NPs and can be interpreted as illustrating fairly convincingly Pienemann’s observation that the acquisition of target-like morphological structures is “filtered through complex form-function relationships” (Pienemann 1998: 251). While Paul did produce a number of appropriately inflected neuter, singular predicative adjectives (stage 4 morphology), a finely-grained distributional analysis of the contexts of occurrence of those adjectives suggested that their instantiation was attributable to factors other than Paul’s acquisition of the processing procedure underlying their production. Moreover, he failed to produce such adjectives in any of the obligatory contexts at the last two data collection points. As pointed out above, the complete disappearance of a feature from the learner’s interlanguage is considered by Pienemann to indicate that such a feature had not been truly acquired in the first place. In the domain of word order syntax, Paul’s attainment (after two years of studying Swedish) was stage 4 on the Processability Hierarchy. I suggested that the variation in his realisation of “inversion” could be attributed to the effect of immature lexical entries and his failure to distinguish between main and subordinate clauses.

10 Conclusion

This paper highlighted issues such as the following: the role of the universal principle of canonicity; the importance of acquiring a target language lexicon whose entries are appropriately annotated with semantic and syntactic specifications (Pienemann 1998: 75; Pienemann et al. 221); the fact that the acquisition of such a lexicon is both a gradual process (Pienemann 1998: 97) and an essential prerequisite for the further development of the interlanguage, as it plays a mediating role in the acquisition of morphosyntactic processing procedures and in the activation of those procedures in language production (Pienemann 1998: 92; 63; MacWhinney 35); the need to carry out a finely- grained distributional analysis of the interlanguage production of the individual learner in order to establish whether a particular structure has in fact been acquired; and finally, the fact that, even at a relatively advanced stage of development when most of the morphosyntactic processing procedures have been acquired, the occurrence of an immature lexical entry in an utterance may result in a breakdown in the L2-specific morphosyntactic processing23 of that utterance (Pienemann 1998: 83).

22 However, given the relatively frequent occurrence of fusional morphology in Swedish, it was sometimes impossible to determine the function(s) assigned by Paul to typically fusional morphemes. 23 “If diacritic features [of a lexeme] are missing or have no values […] the processor will be inoperable” (Pienemann 1998: 80).

83 LEO CONROY

The criterion of systematicity, which appears to be the factor motivating and justifying the treatment of each learner’s interlanguage as a language in its own right, was shown to be problematic as the learner may systematically (and “canonically”) map one form to one function, with the result that such a rigid interlanguage assignment of form and function may and, in Paul’s case did, become generatively entrenched and lead to a developmental dead end, i.e. fossilisation. This was demonstrated in the case of his assignment of the lexeme de to the exclusive function of 3rd person plural pronoun, which appeared to prevent him from decoding and deploying this lexical item when it performed any of its other target language functions. His failure to demonstrate an unequivocally productive use of neuter singular adjectival morphology in predicative contexts may also have been conditioned by the same tendency to map forms and functions according to the one-to-one principle and may have been further reinforced by the relatively “iconic” realisation of NPs of the type “indefinite article+attributive adjective+neuter singular noun” (e.g. ett stort hus). In other words, because the data suggested that Paul strongly associated adjectives inflected for the features “singular” and “neuter gender” with the attributive, prenominal c-structural position, this strong association, based largely on superficial adjacency, may have impeded his acquisition of the “rule” relating to the production of adjectives with the same set of features (i.e. “singular” and “neuter gender”) in predicative contexts. Such an assumption can hardly be dismissed as mere speculation, considering the developmental trajectory, in terms of functional annotation, of the lexeme de. Furthermore, the fact that Paul generally marked only one function on nouns by means of inflectional morphology may be interpreted as pointing to the same tendency toward decomplexification/“canonicalisation” evident in other areas of his Swedish interlanguage. On the basis of the evidence from Paul’s L2-Swedish, my conclusion is that Pienemann is right in assuming that beginning L2-learners may well create their own paradigms characterised by canonical (but not necessarily target-like) one-to-one form-function relationships. However, it is obvious that Pienemann is equally concerned that the learner’s interlanguage should ultimately converge with the target language and not reach “a developmental dead end”. The latter scenario is conceivable if the learner’s interlanguage development remains overly constrained by principles such as canonicity. This study showed that the principle of canonicity or iconicity in the broader sense of both abstract and surface-structural one-to-one form-function relationships facilitated the learning process for Paul in those cases where the target language itself manifested such symmetries. However, in certain cases where Paul’s canonical, but non-target-like form-function mappings became entrenched, the development of his interlanguage did appear to come to “a dead end”.

84 EXTENDED PROCESSABILITY THEORY

Works Cited Ackermann, Farrell, and John Moore. Proto-properties and Grammatical Encoding – A Correspondence Theory of Argument Selection. Stanford: CSLI Publications, 2001. Andersen, Roger W. “The One-to-one Principle of Interlanguage Construction”. Language Learning 34.4 (1984): 77-95. Bley-Vroman, Robert. “The Comparative Fallacy in Interlanguage Studies: The Case of Systematicity.” Language Learning 33.1 (1983): 1-17. Bolinger, Dwight L. “Meaning and Memory.” Forum Linguisticum 1 (1976): 1-14. Bresnan, Joan. Lexical-functional Syntax. Oxford: Blackwell, 2001. Choi, Hye-Won. “Phrase Structure, Information Structure, and Resolution of Mismatch.” Formal and Empirical Issues in Optimality Theoretic Syntax. Ed. P. Sells. Stanford: CSLI Publications, 2001. 17-62. Clahsen, Harald, Jürgen M. Meisel and Manfred Pienemann. Deutsch als Zweitsprache. Der Spracherwerb Ausländischer Arbeiter. Tübingen: Narr, 1983. Croft, William. Typology and Universals. Cambridge: Cambridge UP, 2003. Ellis, Rod. “Are Classroom and Naturalistic Acquisition the Same? A Study of Classroom Acquisition of German Word Order Rules.” Studies in Second Language Acquisition 11.3 (1989): 303-28. ---. “Does Form-focused Instruction Affect the Acquisition of Implicit Knowledge – A Review of the Research.” Studies in Second Language Acquisition 24.2 (2002): 223-36. Falk, Yehuda N. Lexical-functional Grammar. Stanford: CSLI Publications, 2001. Givón, Talmy. “Iconicity, Isomorphism, and Non-arbitrary Coding in Syntax.” Iconicity in Syntax – Proceedings of a Symposium on Iconicity in Syntax. Ed. J. Haiman. Amsterdam: John Benjamins, 1985. 187-220. Hopper, Paul J. and Sandra A. Thompson. “The Iconicity of the Universal Categories ‘Noun’ and ‘Verb’.” Iconicity in Syntax – Proceedings of a Symposium on Iconicity in Syntax. Ed. J. Haiman. Amsterdam: John Benjamins, 1985. 151-86. Josefsson, Gunlög. Ord. Lund: Studentlitteratur, 2005. Kato, Mary Aizawa. “Child L2 Acquisition – An Insider Account.” (In)vulnerable Domains in Multilingualism. Ed. N. Müller. Amsterdam: John Benjamins, 2003. 271-93. Klein, Wolfgang. “The Contribution of Second Language Acquisition Research.” Language Learning 48.4 (1998): 527-50.

85 LEO CONROY

Lee, Hanjung. “Markedness and Word Order Freezing.” Formal and Empirical Issues in Optimality Theoretic Syntax. Ed. P. Sells. Stanford: CSLI, 2001, 63-127. Levelt, Willem J. M. Speaking: From Intention to Articulation. Cambridge, MA: MIT Press, 1989. Lightbown, Patsy M. “Anniversary Article: Classroom SLA Research and Second Language Teaching.” Applied Linguistics 21.4 (2000): 431-62. MacWhinney, Brian. “Parameters or Cues?” Bilingualism: Language and Cognition 7.1 (2004): 35-6. McCarthy, John J. A Thematic Guide to Optimality Theory. Cambridge: Cambridge UP, 2002. Meisel, Jürgen M., Harald Clahsen, and Manfred Pienemann. “On Determining Developmental Stages in Natural Second Language Acquisition.” Studies in Second Language Acquisition 3.2 (1981): 109-35. Pfaff, Carol W. “The Issue Of Grammaticalization in Early German Second Language.” Studies in Second Language Acquisition 14.3 (1992): 273-96. Pienemann, Manfred. “Determining the Influence of Instruction on L2 Speech Processing.” Australian Review of Applied Linguistics 10.2 (1987): 83-113. ---. Language Processing and Second Language Development: Processability Theory. Amsterdam: John Benjamins, 1998. ---. Ed. Cross-linguistic Aspects of Processability Theory. Amsterdam: John Benjamins, 2005. Pienemann, Manfred, Bruno Di Biase and Satomi Kawaguchi. “Extending Processability Theory.” Cross-linguistic Aspects of Processability Theory. Ed. M. Pienemann. Amsterdam: John Benjamins, 2005. 199-252. Pinker, Steven. Language Learnability and Language Development. Cambridge, MA: Harvard UP, 1984. Schwarze, Christoph. “Representation and Variation: On The Development of Romance Auxiliary Syntax.” Time over Matter – Diachronic Perpectives on Morphosyntax. Ed. M. Butt and T. Holloway King. Stanford: CSLI Publications, 2001. 143-72. Slobin, Dan Isaac. “Universal and Particular in Acquisition.” Language Acquisition: The State of the Art. Ed. E. Wanner and L. Gleitman. Cambridge: Cambridge UP, 1982. 128-70. ---. The Crosslinguistic Study of Language Acquisition. Vols. 1; 2. Ed. Hillsdale, NJ: Lawrence Erlbaum, 1985. Teleman, Ulf, Staffan Hellberg, and Erik Andersson. Svenska Akademiens Grammatik. Vols. 2; 3. Stockholm: Norstedts, 1999.

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Conroy, Leo

Title: Extended processability theory and its application to L2-Swedish

Date: 2007

Citation: Conroy, L. (2007). Extended processability theory and its application to L2-Swedish. In, Proceedings, In Between Wor(l)ds: Transformation and Translation, University of Melbourne.

Publication Status: Published

Persistent Link: http://hdl.handle.net/11343/34874

File Description: Extended processability theory and its application to L2-Swedish