Misgivings about universal developmental sequences in nonnative acquisition of syntax

Abstract

Several recent international publications (e.g. Platzack 1996, 2001; Håkansson, Pienemann & Sayehli 2002; Pienemann 1998; Pienemann & Håkansson 1999) have claimed that learners take the same developmental route in the acquisition of syntax of a second or foreign language, irrespective of their first language (L1). Targetlike verb placement in a Germanic V2 language like German, Dutch or Swedish (finite verb in second position in main clauses), is seen as an exceptionally difficult phenomenon/constraint/parameter to acquire. Even if both the L1 and L2 are V2 languages, learners are said to acquire V2 late, only partially, or never.

As empirical support for these claims, cross-sectional data have been adduced from Swedish children learning German at school and German adults learning Swedish at university. Violations of V2 by learners with a V2 mother tongue are taken as support for the resurrection of a hypothesised universal developmental path in L2 German verb placement – essentially the developmental stages Clahsen & Muysken 1986 proposed on the basis of Romance L2ers.

Explanations propounded in the literature are (1) the so-called “vulnerability” of the C-domain (the left periphery of the clause in Minimalist syntactic theory terms, Platzack 2001), (2) SVX being a more basic word order (e.g. Platzack 1996, building on Kayne 1994), and (3) “canonical” SVX word order and XSV word order (i.e. V3, with a fronted element before the subject) being “easier to process” than XVS (i.e. V2). Easier processability of SVX and XSV is argued for by e.g. Håkansson 2001, Håkansson, Pienemann & Sayehli 2002, Pienemann 1998, Pienemann & Håkansson 1999 and Sayehli 2001. (For quite a different view, see e.g. Eubank 1993, Bohnacker 2003.) Vulnerable C-domain, processability, and the idea of a universal developmental path have also found their way into the linguistics and teacher training courses and reading lists at a number of institutes of higher education in Sweden and are not seldom uncritically presented as axioms and as the basis of classroom teaching methods. This, however, may be premature, and it may unduly skew the ideas our teachers-to-be entertain about language acquisition.



The present paper aims to show (a) that the above studies suffer from some methodological flaws, (b) that other studies of learners of a V2 language (German) with a V2 L1 (Swedish) yield strikingly different results, and therefore (c) that the empirical underpinnings for a proposed universal developmental path in L2 verb placement are shaky at best.

I will present quantitative (and some qualitative) data from 2 completed and 3 ongoing studies of the acquisition of German verb placement by tutored post-puberty learners. The data come from several oral and written corpora, all of which are new, by teenage school pupils, students at university-level, and old-age pensioners attending German classes at a community centre.



Cross-sectional data:

• 23 16-year-old pupils, elementary-intermediate (exposure: 3 yrs classroom German):

23 recordings of oral productions (narrative); 10 written productions (essay/letter)

• 100 university students, advanced (exposure: 5-8 yrs classroom German): 200 written productions

• 6 adult academics, highly advanced (exposure: 4 yrs classroom German plus 3 yrs immersion in Germany), oral and written production (Bohnacker 2003)



Quasi-longitudinal data:

• 6 adults, old age pensioners, ab initio learners, oral productions (narrative, dialogue), recorded after 4 months and 9 months of exposure to classroom German

• monolingual German native speaker control

Oral data was transcribed, oral and written data tagged by hand and sorted.



Findings for the cross-sectional data:

(I) Irrespectively of length of exposure and proficiency level, learners produce a high number of subject-initial and of non-subject initial clauses, with no strong preponderance of SVX. Non-subject initial clauses constitute on average ca. 40%, subject-initial clauses ca. 60% of all main clauses (with a range of 20%-70%, depending on the individual learner).

(II) Across all learners, virtually all non-subject initial main clauses show targetlike verb second. V2 violations are exceedingly rare, ranging from 0% to 2% between individuals.

The claim that V2 is universally hard to acquire must therefore be rejected.



Yet how can these findings be reconciled with those of Håkansson 2001, Håkansson, Pienemann & Sayehli 2002 and Sayehli 2001, who attest V2 violations? Are the learners in the present study simply too advanced? Or are other factors at work?

Note that Håkansson, Pienemann and Sayehli only look at Swedes learning German as a third language, after substantial exposure to English (3-6yrs). But in this case, before attributing V2 violations to a universal developmental route, possible influence of English (non-V2) on L3 German would need to be ruled out first.

In order to do this, I carried out a microcomparative study of truly ab initio learners of German:

a. 3 adult Swedish learners of ab initio German (with no English), versus

b. 3 matched learners (with L2 English).



Findings for the ab initio learners:

(i) Similarity: Both a.- and b.-learners produce many subject-initial clauses (67%), but also non-subject-initial clauses (ca. 30%). SVX never is the exclusive word order, neither at 4 nor 9 months.

(ii) Categorial difference: a.-learners don’t violate V2 in their interlanguage German (0%-0.1%), but b.-learners do (20%-50%). This is a crucial finding,

(iii) Similarity: Both a.- and b.-learners productively use V2 after 4 months of exposure, (b.-learners in 65% of all non-subject-initial main clauses). This is an important result too.

(iv) Interlanguage syntax in other areas can be nontargetlike for a.- and b.- learners, e.g. early head-initial VPs (VO) giving way to German-style head-final VPs (OV) over time.



The paper will present more detailed quantification of the bulked and individual data, alongside qualitative results (which type(s) of clause-initial non-subject elements; comparison with native control, etc.).

The findings suggest that ab initio learners do not necessarily start out with SVX (see (i)). If there exists a “canonical word order” at all, it is certainly not exclusive, contra current claims in the literature.

The findings also clearly show that Germanic V2 is not hard to acquire per se, contra e.g. Håkansson, Pienemann & Sayehli 2002: For Swedish ab initio learners of German (a.-group), V2 is targetlike already after 4 months (see (ii)). Knowledge of a non-V2 language (English) can make it initially harder to acquire another V2 (see (ii)), but at the same time, V2-L1 syntax is also made use of to some extent (iii).



Combining these results with studies of other learners (Romance, Turkish, Korean, English,) we find divergent L2 developmental routes for groups of learners with typologically distinct L1s (e.g. Schwartz & Sprouse 1994, Young-Scholten & Vainikka 1994). Therefore, theories that postulate a vulnerable C-domain and universal SVX and XSV are misguided, and should be abandoned in favour of L1 transfer approaches (e.g. Schwartz & Sprouse 1996). Methodologically sound comparative empirical studies of interlanguage will need to be carried out for a larger variety of L1/L2 language combinations, and before results from these are in, extreme caution is advised when propounding “universals” for the acquisition of syntax.

• author: Bohnacker, Ute
• organization: Division of Swedish, Danish, Icelandic, Public Service Interpreting and Translation
• publishing date: 2005
• type: Chapter in Book/Report/Conference proceeding
• publication status: published
• subject: Languages and Literature
• host publication: Forskning om undervisning i främmande språk. Rapport från workshop i Växjö 10-11 juni 2004. Acta Wexionensia 58/2005.
• editor: Larsson-Ringqvist, Eva and Valfridsson, Ingela
• volume: 58
• pages: 46 - 56
• publisher: Växjö University Press
• ISSN: 1404-4307
• ISBN: 91-7636-450-X
• language: English
• LU publication?: yes
• additional info: The information about affiliations in this record was updated in December 2015. The record was previously connected to the following departments: Swedish (015011001)
• id: bed566b8-36ae-4718-ac4f-7307ac9712b1 (old id 536845)
• date added to LUP: 2016-04-01 15:54:05
• date last changed: 2018-11-21 20:37:14