Íslenskt mál og almenn málfræði - 01.01.2023, Blaðsíða 186

examples which do not fit the broad semantic patterns just described include
gleyma ‘forget’ and loka ‘close’, which have dative objects that are neither applica-
tives nor motion themes. Icelandic therefore seems to have, in contrast with e.g.
German, “truly direct object datives” (Wood 2015:129) which are not derived
from the semantic associations just described (or the structural distributions dis-
cussed in subsection 1.1.2).
These exceptions of course do not erase the patterns just described, which
also have been highlighted in the literature (e.g. Van Valin 1991, Jónsson 1997–
1998, 2000, 2003, 2005 and 2013, Maling 2002, Svenonius 2002, Thráinsson 2007,
Barðdal 2008). These patterns, as is further discussed in Chapter 2, are particu-
larly interesting in the context of how children can use form to acquire meaning
and ongoing debates about the use of possibly universal cues (e.g. number of
arguments) and/or language-specific cues (e.g. morphological case) in this
respect (Lidz et al. 2003, Göksun et al. 2008, Matsuo et al. 2012, Leischner et al.
2016). But despite the clear semantic relationship between e.g. dative and recipi-
ents/beneficiaries/goals in Germanic languages with a solid dative-accusative
distinction in general, “the correlations between cases and semantic or thematic
roles are notoriously approximate” (McFadden 2020:295), with pairs of verbs
with close semantics but different case frames being taken “to show that the rela-
tionship between cases and semantic/thematic roles cannot be direct” (ibid). This
is emphasized by e.g. Sigurðsson (2012:324), who uses Icelandic verb pairs such
as aðstoða ‘assist’ (accusative object) and hjálpa ‘help’ (dative object) to argue
against a relationship between case and semantics.
Maling (2002) furthermore states that although “the correlation between the-
matic role and morphological case is not one-to-one, there is clear evidence of
semantic generalizations at work in the language” (p. 96). In a similar vein,
Thráinsson (2007) observes that “although it is possible to find some relationship
between thematic roles of arguments [...] and morphological case in Icelandic, the
relationship is many-to-one and one-to-many” (p. 200). He adds that “the case-
marking possibilities are to some extent restricted by the grammatical function
(subject, object, indirect object) involved” (p. 200) and later that “no predictions
can be made about the thematic role of a subject given its case” (p. 206). Similarly,
as an argument against adopting Schäfer’s (2008) account on Icelandic datives,
Wood (2015) explains that “the semantic correlations are not exact” (p. 137) and
in another instance states that “it remains true that one cannot predict from event
semantics which case any given verb will assign, though there are some semantic
regularities” (p. 135).4 In the approach taken in this dissertation, I make a case for
exceptions: Even though event structure does not fully determine case marking
patterns, it is clear that case can indicate certain types of events. Semantic corre-
lations do not need to be exact for the learner to discover and productively use
Iris Edda Nowenstein186
4 Although the literature discussed here consists of recent approaches to case in Ice -
landic, the fact that the alignment of thematic roles and case/syntactic roles is notoriously
approximate was highlighted much earlier (e.g. Rosen 1984).