Íslenskt mál og almenn málfræði - 01.01.2023, Blaðsíða 191

of subjects and objects play a role, with non-nominative on a subject forcing an
earlier reinterpretation (perhaps of the ‘privileged’ link between agents and sub-
jects mentioned by Hartshorne et al. 2015, see also Becker 2014). In general, the
results indicate that the properties of Icelandic dative subjects could prove impor-
tant in research on the acquisition of psych verbs (e.g. Hartshorne et al. 2015 and
2016, Harrigan et al. 2019 and Shablack et al. 2020) but also in other wider con-
texts. For example, one interpretation of the results could be that non-agents in
subject position are somehow more salient than non-patients in object position,
a finding which is somewhat in line with the work of Rissman and Majid (2019)
and Ünal et al. (2021) where agentivity and causativity are more salient cognitive-
ly and linguistically.
Chapter 2 tackles case in comprehension, but under the premise that gener-
alizations which guide verb learning are also relevant in production: Icelandic
dative productivity provides a scenario where exceptions, or non-default pat-
terns, can be structured into productive rules and generalized systematically
based on distinctions which arise in comprehension. Although the same form-
meaning patterns found in the linguistic input are underlying, I assume that the
processes of linking together form and meaning in comprehension (word learn-
ing) on the one hand, and production (with acquired verb roots) on the other
hand, are fundamentally distinct in nature. When acquiring the meaning of
verbs, the directionality of the hypothesis formation goes from form to meaning.
When producing language with acquired verbs, the rules are formed in the meaning-
form direction, or even form-form in some cases. This has implications for the
association patterns we expect to find in comprehension versus production
experiments, as is further explored in Chapter 3, where the focus is on finding
the semantic contexts in which children acquiring Icelandic form productive
dative case marking rules.

3. Producing non-default case
In Chapter 3, I show how the generalizations which guide verb learning are also
relevant in production, further investigating the patterns which can be derived
from the distribution of the dative and how they are learned. This is done with a
production task featuring existing and novel verbs (N = 101, children aged 3–
13), targeting the contexts where dative productivity can be found. Additionally,
I explore whether measures of the participants’ receptive and productive vocab-
ulary can predict their use of the dative at an individual level. I argue that study-
ing the productivity of Icelandic datives also entails addressing fundamental
questions about rule formation and productivity. In fact, rule formation and pro-
ductivity are at the heart of current but long-standing debates within language
acquisition research, e.g. in discussions about stored exemplars versus abstrac-
tions (e.g. Ambridge 2020 and responses to his article), rule-based accounts of
acquisition or single-/dual-route models (e.g. Yang 2016, Blything et al. 2018)
and different approaches to (over)generalization more broadly (e.g. O’Donnell
Project rationale and core ideas 191