Íslenskt mál og almenn málfræði - 01.01.2023, Side 193

2015, Yang 2016, Schuler et al. 2016, Yang 2018 and responses to the article,
Ambridge et al. 2018 and Goldberg 2019).
The results of the chapter replicate previous findings with existing verbs (e.g.
H.Þ. Sigurðardóttir 2002, Jónsson and Eythórsson 2003) to some extent and
expand them as well. Starting with subject case, results from verbs with non-
nominative subjects and variation in case marking indicate that children overgen-
eralize both the default nominative and non-default (but ‘inherent’) dative, with
the nominative overgeneralizations decreasing with age while the dative ones
increase. Interestingly, the ‘quirky’ accusative is sometimes (but marginally) cho-
sen with the historically dative finnast ‘find’ as well. For object case, the default
accusative is unsurprisingly extended to dative and genitive objects, but interest-
ingly, dative is also overgeneralized and nominative overgeneralizations appear as
well, mostly in the youngest age group. Dative on indirect objects seems to be
acquired sooner than dative on direct objects. Additionally, although frequency
effects are not the main focus of this chapter, no clear contrasts between high and
low frequency items appear. Overgeneralizations in the case marking of existing
verbs are a clear sign of productivity, but the results with the novel verbs (Figure 2)
provide additional information on the contexts for the non-default productivity
of the dative. In subject case, dative is selected more in the expected experiencer
contexts. Dative selection still proves to be significantly higher with objects than
subjects, both in typical and atypical semantic contexts, pointing towards produc-
tivity that goes beyond the scope of translational equivalents and caused motion
(Jónsson and Thórarinsdóttir 2020) and even Barðdal’s (2008) more diverse schemas
associated with the dative. This indicates that the Icelandic dative competes with
the default accusative in object case.
In general, the results call for a model of rule formation which accounts for
nested non-default productivity and does not hinge on frequency to determine
gradient overgeneralization, since patterns which are not statistically dominant
can become productive and multiple rules can apply to a context simultaneously. I
argue this is most compatible with rule-based accounts (Yang 2002, Albright and
Hayes 2003, O’Donnell 2015, Yang 2016), which also provide the best possible
framework for the different patterns found in comprehension and productivity.
Indeed, rule directionality can be used to account for the fact that dative subjects
are produced less than dative objects but provide better cues in comprehension. At
the individual level, I find that the size of productive, but not receptive, vocabulary
shows a significant relationship to the acquisition of dative productivity. This
might indicate that productivity relies on a child acquiring a sufficient amount of
evidence to derive dative case marking rules, an intuition which is compatible with
the Tolerance Principle (Yang 2016) and is further investigated in Chapter 4.

4. Dative derivability
Thus far, it has been shown that despite known exceptions, non-exact correla-
tions and one-to-many and many-to-all relationships, children do discover asso-
Project rationale and core ideas 193