Merete Anderssen & Marit Westergaard – University of Copenhagen

Danish voices > WILA8 2017 > Workshop programme > Merete Anderssen & Mar...

Complexity, Frequency and Cross-linguistic Influence in Heritage Language: Subject Shift and Object Shift in Norwegian

Merete Anderssen & Marit Westergaard, University of Tromsø (Thursdag 10:00-10:30)

This study investigates two word order phenomena in Norwegian Heritage Language (HL) in the US, compares this to non-heritage Norwegian, and discusses which factors may account for differences between the two. Norwegian exhibits vari¬ation with respect to the position of subject and object pronouns relative to negation, so-called subject shift (SS) and object shift (OS). SS is found in non-subject-initial clauses with V2: In main clauses, pronominal subjects typically occur in front of negation (1a) and DPs after (1b). In embedded clauses, both pronouns and DPs mainly appear in the shifted position. In non-heritage Norwegian, the percentage of shifted subjects is 85-90% in both clause types (Westergaard 2011). OS is found in contexts in which the verb moves out of the VP. DP objects always appear after negation (2a); pronominal objects with nominal antecedents precede negation (2b), while pronominal objects with non-nominal antecedents follow (2c) (Anderssen & Bentzen 2012). Pronominal objects with nomi¬nal and non-nominal antecedents shift at different rates, 87% vs. 5% (Bentzen et al. 2013).

HL typically has “simplified,” non-standard characteristics (Scontras et al. 2015). However, economy of movement tends not to be a crucial factor (Westergaard & Anderssen 2015). Instead, frequency has been considered to be more important, protecting against attrition. Furthermore, while structural similarity has often been found to be a factor in bilingual acquisition, structural difference seems to play a role in HL (Kupisch 2014).

If structures involving syntactic movement are vulnerable to attrition (pace Westergaard & Anderssen 2015), we expect both SS and OS to be affected. If frequency plays a major role, OS and SS in embedded clauses should be more vulnerable than SS in main clauses. And since both SS and OS involve V2, there is little overlap with English, and cross-linguistic influence should play no role (cf. Müller & Hulk 2001).

We investigate the CANS-corpus (Johannessen 2015) and find that the heritage speakers behave like Norwegian adults with regard to SS in subordinate clauses, shifting at 86.1% (Table 1). In main clauses, however, subjects are shifted only 59.3%, and non-target-consistent examples are attested (3). An examination of OS reveals that the heritage speakers shift pronominal objects with nominal and non-nominal antecedents at 61% and 11% (Table 2), clearly distinguishing between the two. Surprisingly, the most vulnerable structure is SS in main clauses, particularly in yes-/no-questions or tags (4).

Consequently, complexity and/or frequency are not important factors. We therefore consider cross-linguistic similarity/difference. There are two exceptions to the lack of overlap between Norwegian and English: SS in embedded clauses (the two languages are identical) and in questions with auxiliaries or be, where English has residual V2 and subject-auxiliary inversion. Thus, both English and Norwegian display word order variation in these questions – with preferences that go in opposite directions, with S-Neg being preferred in Norwegian and Neg-S in English (5).

Cross-linguistic influence thus seems to be responsible for our results, causing SS to be unproblematic in embedded clauses (positive transfer) and vulnerable in questions with aux/be (negative transfer).

  1. a. I går spiste han ikke (?han) middag. yesterday ate he not dinner
    b. I går spiste (Jon) ikke (Jon) middag.
    yesterday ate John not John dinner
  2. a. Peter så ikke bilen.
    Peter saw not car.def
    ‘Peter didn’t see the car.’
    b. Peter så den ikke.
    Peter saw it not
    ‘Peter didn’t see it.’
    c. Mari synes den er fin, men Peter synes ikke det. (det = ‘that it is nice’)
    Mari thinks it is nice but Peter thinks not it
    ‘Mari thinks it is nice, but Peter doesn’t think so.’
  3. nei jeg veit da vi begynte på skolen så # kunne ikke vi # snakke engelsk at all
    no I know when we started at school so could not we speak English at all
    ‘No, I know that when we started school, we couldn’t speak English at all.’
  4. ja ## er rart hvor fort disse åra har gått forbi er ikke det ?
    yes is strange how fast these years have gone bye is not it
  5. a. Isn’t he clever?/Is he not clever?
    b. What didn’t he like?/What did he not like?
S-Neg Neg-S
Main clauses 59.3% (86/145) 40.7% (59/145)
Embedded clauses 86.1% (31/36) 13.9% (5/36)


Table 1. Pronominal subjects in CANS (n=50)

O-Neg Neg-O
Nominal antecedents  61% (25/41) 39% (16/41)
Non-nominal antecedents 11% (11/100) 89% (89/100)

 Table 2. Pronominal objects in CANS (n=50)

Selected references

  • Bentzen, Anderssen & Waldmann. 2013. Object Shift in Mainland Scandinavian: A corpus study of Danish, Norwegian and Swedish.
  • Johannessen. 2015. The Corpus of American Norwegian Speech (CANS).
  • Kupisch. 2014. Adjective placement in simul¬tane¬ous bilinguals (German-Italian) and the concept of cross-linguistic overcorrection.
  • Müller & Hulk. 2001. Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient languages.
  • Westergaard. 2011. Subject positions and information structure: The effect of frequency on acquisition and change.