Does word order affect attention to changes in complement clauses? Testing a semantic hypothesis experimentally.

In modern spoken Danish, we find a high degree of variation between so-called ’main
clause’ and ’subordinate clause’ word order in complement clauses as determined by the
relative ordering of finite verb and sentence adverbials: In short, main clause word order is
Verb > Adverb (1), and subordinate clause word order is Adverb > Verb (2) (cf. Diderichsen
1946 and later versions).
1. så vidste jeg jo bare at der var ikke noget mellem ham og hende
then knew I JO just that there was not anything between him and her
―then of course I just knew that there was nothing between him and her‖
2. så vidste jeg jo bare at der ikke var noget mellem ham og hende
then knew I JO just that there not was anything between him and her
―then of course I just knew that there was nothing between him and her‖
The use of Main Clause Phenomena (MCP) in complement clauses has been studied for
a range of languages and from diverse theoretical perspectives, and there is wide agreement
that the distribution of MCP to a large extent is conditioned by semantic or discoursepragmatic factors (e.g. Hooper and Thompson 1973; Meinunger 2006; Wiklund, Bentzen et
al. 2009; Aelbrecht, Haegeman et al. 2012). One line of research hypothesizes that main
clause word order in subordinate clauses signals that the more important information,
sometimes called the ‖main point of the utterance‖ or the ‖foreground‖ is to be found in the
subclause, rather than in the main clause (e.g. Boye and Harder 2007; Simons 2007; Wiklund,
Bentzen et al. 2009; Jensen and Christensen 2013).
Previous observational studies of spoken language data in Danish have shown a range
of statistically significant, linguistic predictors for the use of Verb > Adverb word order in
complement clauses (Jensen and Christensen 2013). Two of the most important predictors
are presence or absence of a subordinator (typically at ―that‖) and the type of matrix clause
predicate. SUBORDINATOR ABSENCE clearly favours Verb > Adverb (i.e. main clause) word
order (cf. Thompson and Mulac 1991); and while COMMUNICATIVE and COGNITIVE
predicates also favour the use of Verb > Adverb word order, FACTIVE (i.e., presuppositiongenerating) predicates disfavour them (Kiparsky and Kiparsky 1971). Semi-factive
predicates allows for two readings (cf. Simons 2007) and occupy the middle ground
regarding the distribution of the two word orders (Jensen and Christensen 2013).
49
A reasonable extension of the foregrounding hypothesis is that a foregrounded clause
will attract more attention from listeners or readers (in psycholinguistic terms:
comprehenders) than a backgrounded clause. In order to test whether comprehenders are
in fact more attentive to complement clauses with main clause word order than to those
with regular subordinate clause word order, we have performed an experiment under the
Text Change-paradigm. This paradigm assumes that comprehenders are disinclined to
notice change made to already processed discourse, so-called change blindness. However,
the degree to which change blindness occurs is affected by a number of linguistic and
cognitive factors, including the age of the comprehender (Price 2008) and well as sentential
complexity (Sanford, Sanford et al. 2005). Linguistic factors that capture the attention of the
comprehender are found to reduce change blindness. This suggests that discourse under
the scope of attention capturing devices is processed fuller and more deeply than discourse
which is not (Sanford, Sanford et al. 2006), which makes the testing paradigm relevant for
the hypothesis of the semantico-pragmatic function of word order in subordinate clauses. If
Verb > Adverb is a foregrounding signal, we would expect foregrounded clauses to cause
less change blindness than non-foregrunded clauses.
In a first round of experiments using a text change-design, 38 students each read 24
complex sentence constructions twice. Each construction contained a complement clause
with either Verb > Adverb or Adverb > Verb order. Half of the complement clauses were
governed by a semifactive predicate (open to both word orders) and the other half by what
we call a semantically secondary predicate (favouring Verb > Adverb word order). All
complement clauses were initiated by the subordinator at ‖that‖ to ensure that
comprehenders read the relevant clause as a complement rather than as a main clause.
Attention to the complement clause was tested by measuring how disinclined they
were to notice change of words in the complement clause when re-reading the same
construction after a minimal break (see an English example in 3).
3. First presentation: then she realised that I did not have any sort of plan
Second presentation: then she realized that I did not have any kind of plan
A mixed-models statistical analysis showed significantly less change blindness with
Verb > Adverb complements than with Adverb > Verb complements. We consider this a
preliminary confirmation of the semantico-pragmatic hypothesis that main clause word
order functions as a foregrounding signal for comprehenders.
We will discuss possible objections to the design, among them the fact that the stimuli
were presented in written form, whereas the foregrounding hypothesis relates to spoken
language.

References
Aelbrecht, et al. (2012). Main Clause Phenomena. John Benjamins.
Boye and Harder (2007). Complement-taking predicates. Studies in Language 31(3).
Diderichsen (1946). Elementær Dansk Grammatik. Gyldendal.
Hooper and Thompson (1973). On the Applicability of Root Transformations. Linguistic
Inquiry 4(4).
Jensen and Christensen (2013). Promoting the demoted. Lingua 137.
Kiparsky and Kiparsky (1971). Fact. Steinberg and Jakobovits (eds.) Semantics. CUP.
Meinunger (2006). On the discourse impact of subordinate clauses. Molnár and Winkler
(eds.) The Architecture of Focus. Mouton de Gruyter.
Price (2008). The use of Focus Cues in Healthy Ageing. Univ. Glasgow.
50
Sanford et al. (2005). Depth of lexical-semantic processing and sentential load. Journal of
Memory and Language 53(3).
Sanford et al. (2006). Shallow processing and attention capture in written and spoken
discourse. Discourse Processes 42(2).
Simons (2007). Observations on embedding verbs, evidentiality, and presupposition. Lingua
117(6).
Thompson and Mulac (1991). The discourse conditions for the use of the complementizer
that in conversational English. Journal of Pragmatics 15.
Wiklund, Bentzen et al. (2009). On the distribution and illocution of V2 in Scandinavian
that-clauses. Lingua(119)