3.2.2. Influence of the variables on the dative choice

3.2.2. Influence of the variables on the dative choice

Results

In order to find out which factors impact on the dative choice, I will now examine the individual items and variables. In doing so, I will concentrate on the most striking effects. The following chart provides details on each of the 15 items. The nine variables are listed in bold at the top of columns 2 to 10. Each line shows the forms which a variable assumes for one item.

Under Previous dative3 the variables previous = prepositional and previous = none from the model formula are combined. Apart from the forms under previous dative, I did not determine the forms of the variables myself, but adopted them from a list from Bresnan and Ford which was sent to Dr. Schlüter by Daphne Theijssen. Ford and Bresnan used the factor verb in their 100 split-task. As their calculations on the influence of different verbs are not available to me, I am making use of the factor verb sense, which they used in their corpus model and for which they provided calculations in another article (Bresnan and Ford 2010: 178). As verb sense is one factor among many and as the difference between the influence of verb sense and verb is not very huge, this should not falsify the results. Like before, the items are numbered from 1 to 15 according to rising corpus probabilities (in the left column). However, they are sorted according to the participants’ ascending ratings now. The three columns on the right present the corpus probabilities and participants’ ratings in percent, and the difference between them in percentage points. Columns containing forms of variables that reduce the probability of a prepositional dative are marked in blue. Those that increase its probability are marked in red. The weighting of the factors according to the model formula is provided in the cells at the very bottom. *1.819 under Log Length Difference means that the logarithm of the length difference shown in the columns is multiplied by this figure.

Table 4: Overview of items, variables, corpus probabilities and ratings

The abbreviations in this table (from left to right) stand for the following terms: p = pronoun; non = nonpronoun; d = definite; in = indefinite; a = animate; in (under animacy recipient) = inanimate; sg = singular; pl = plural, prep = prepositional, double obj. = double object dative. Concerning the verb senses, the abbreviations mean the following: a = abstract; c= communication; t = transfer of possession.

The visualization illustrates the influence of the variables. If the participants are influenced by the variables in the same way as native speakers, the blue columns should appear predominantly at the top as those factors lower the probability of the prepositional dative according to the model formula, and those in red should be found predominantly at the bottom. Indeed, items with a high rating in favour of the prepositional dative combine many factors marked in red (see e.g. items 14 and 15), whereas items with low ratings contain many factors marked in blue (see e.g. items 2 and 1). As for the individual factors, items with a pronominal recipient and a negative length difference (marked in blue) appear at the top, with ratings between only 16.30% and 47.80% in favour of the prepositional dative. Items with an inanimate recipient and a positive length difference (red) appear at the bottom, yielding higher ratings in favour of the prepositional dative (between 47.80% and 95.60%). Items with an indefinite theme (blue) are distributed across the chart. Whilst items with indefinite recipients (red) yield moderate corpus probabilities (20.88% and 62.73%), they now appear at the bottom with relatively high probabilities of 58.30% and 70.10%. Pronominal themes and verbs will be addressed later. The number of the theme will not be taken into account because only one item has a plural theme, rendering it non-representative. The same goes for previous datives, as there is just one instance each of a double object and a prepositional dative.

Discussion

The results already allow for a preliminary interpretation: apparently, German learners of English are sensitive to pronominal recipients, length difference between recipient and theme, indefinite recipients and inanimate recipients. Of those, indefinite recipients seem to have a greater impact on Germans than on native speakers. Indefinite themes apparently have little effect.

3  I am going to write the factors influencing the dative choice in italics in order to facilitate reading.

Created with the Personal Edition of HelpNDoc: Easily create iPhone documentation