1.3 Preliminaries
1.3.3 Empirical Perspectives
1.3.3.2 Experimental Methodology
Experimental methods in linguistics are a well-established and great tool to test pre- dictions our theory makes. The scientific method of observing, questioning, forming a hypothesis and finally testing the predictions that can be derived from the hypothesis is at the core of the research program I am following in this thesis. I will briefly comment on the experimental paradigms I have been using for my study in Chapter 3. I con- ducted a forced-choice experiment, an acceptability rating experiment, a reading time (RT) experiment and an eye-tracking experiment (the latter in cooperation with Fabian Schlotterbeck and Oliver Bott).
12To be found here: https://www.herzen.spb.ru/en
In my first pilot study, Pilot 1, I used a forced-choice task to figure out a pref- erence for a disambiguated sentence when presented without context. In Pilot 2, I used an acceptability rating study to enforce a certain reading by providing a context in advance. Pilot 2 was also thought of as a pilot for the self-paced RT-experiment in order to establish whether the contexts did their job, i.e. disambiguated a sentence towards one or the other reading.
Reaction times are a good window into semantic processing. I used a self-paced reading studywith a moving-window technique. According to Rayner & Clifton (2002), such a self-paced reading test is more precise as compared to only measuring overall read- ing times. One can control the amount of text and the size of the segment that the subject can see at once. This was of importance for my investigation since I am interested in exactly where and how fast the ambiguity is resolved. The comparison of one reading being quicker than the other at a certain point is helpful to establish processing difficul- ties and thus complexity of one over the other structure. Longer reaction times are a sign of higher processing cost.
Finally, I also applied eye tracking. The ambiguity that I was interested in, cf. example (13), turned out to be a subtle one when it comes to testing it experimentally. Thus, a fine-grained method was called for in order to better test my prediction of one reading being more complex than the other in several respects. As Rayner & Clifton (2002) put it:
Eye movements have been utilized to study a variety of language com- prehension processes, and data gleaned from eye-tracking studies have been found to reflect moment-to-moment cognitive processes. One early finding was that where readers look and how long they look there is directly related to the ease or difficulty of cognitive processing [...]
(Rayner & Clifton 2002: 265) We measured first pass time, first pass regression ratios, regression path duration and total times. However, only in the first pass regression ratios did we find a significant interaction that was predicted. I will describe and discuss all the experiments in detail in Chapter 3.
Variation in the Standard Argument
The precise semantics of the comparative operator is of course the object of much debate [...]
(Beck 2011: 1342) Strong arguments have been brought forth for the view that the comparative constructions of different languages employ different compositional mecha- nisms. The semantics of the comparative operator used thus has to be deter- mined on a case-by-case basis. [...] Thus, we are faced with the problem of determining, for a given comparative construction, which comparative oper- ator and corresponding path of composition this construction uses.
(Beck, Hohaus & Tiemann 2012: 147) Inspired by this line of research, this chapter provides a cross-linguistic discussion of comparative operators. In addition to what is already known from the literature about other languages like English, German, Greek and Hindi, I also investigate Russian and Tundra Nenets.
The overarching research question of this chapter is Q1, namely:
Q1: What is the range of available degree operators cross-linguistically? In particular: a. Do we see evidence for both phrasal operators described in the literature (compHeim1985
and compKennedy1997)?
b. Or do we only need the stronger one (compHeim1985) that covers a wider range of
The hypotheses repeated in short from 1.1 for Russian (H1R), German (H1G) and Nenets
(H1N) are the following:
H1R: Genitive-marked synthetic comparatives in Russian use compKennedy. H1N: Nenets uses compHeim
H1G: German only has compclausal
The research contribution of this chapter is that I provide novel analyses of compari- son constructions in Russian and Tundra Nenets thereby contributing to the question of the exact inventory of comparative operators. In this chapter, I also provide a little roadmap, summarized in a flowchart, of how to analyze the standard argument when faced with a new research language. Russian genitive-marked comparatives are ana- lyzed as involving a phrasal comparative operator that goes back to Kennedy (1997), compKennedy, while čem-clauses (a wh-word marked by instrumental case) are best an-
alyzed as using the clausal comparative operator compclausal. Nenets comparatives are
analyzed as using a phrasal operator going back to Heim (1985), compHeim. The main
distinction between the two operators is their scopal mobility.
In order to deal with the overarching question Q1, I will also think about (i) what the different compositional routes to the same meaning there are, and (ii) which evi- dence in favor of these different operators we have in the languages under discussion. In the following theoretical background, I will first discuss the difference between the so- called “phrasal” and “clausal ” analyses in subsection 2.1.1 and then discuss the different operators that are on the market in subsection 2.1.2, briefly introduce the Bhatt & Taka- hashi diagnostics in subsection 2.1.3 and finally elaborate on the inventory of different operators cross-linguistically in subsection 2.1.4. I will subsequently turn to comparison operators in Russian in 2.21, then Nenets in in 2.3 and finally German in 2.4 and apply
the diagnostics introduced in section 2.1.4.