Abstract :
[en] This study presents a systematic, data-intensive way to uncover semantics differences between constructions by iteratively digging deeper into corpus variation. The case study at hand is the directional alternation in Dutch. This directional alternation can be found with a number of verbs, which may realize their arguments in both the Strictly Transitive Construction (STC) and the Intransitive Motion Construction, as in (1) vs. (2) (Goldberg 1995: 3–4, 117–119). (1) Deze schutter heeft heel bewust slachtoffers gezocht met een andere huidskleur. This rifleman has very deliberately victims searched with a different skin_color (Sonar corpus, Oostdijk et al. 2013, WR-P-P-G-0000186016.p.1.s.4) ‘This rifleman has very deliberately searched for victims with a different skin color.’ (2) Reddingswerkers hebben de hele dag tussen het puin naar slachtoffers gezocht. Rescue_workers have the whole day between the rubble to victims searched (Sonar corpus, Oostdijk et al. 2013, WS-U-E-A-0000065564.p.1.s.3) ‘Rescue workers have searched for victims between the rubble the whole day.’ The proposed method consists of three steps. The first is to compile a list of all verbs exhibiting this alternation. This yielded 13 verbs, of which 6 displayed sufficient variation to be put under further scrutiny, namely bellen ‘to phone’, grijpen ‘to grap’, happen ‘to snap’, peilen ‘to gauge’, verlangen ‘to desire’ and zoeken ‘to search’. In the second step, we investigate the alternation for each verb separately, a.o. by executing a distinctive collexeme analysis on the object slot of the verb (Gries and Stefanowitsch 2004). Such an analysis still leaves some variation to be accounted for, however. For instance, the object slachtoffer ‘victim’ of the verb zoeken ‘to search’ is shown to be a distinctive collexeme of the IMC, yet a great deal of its occurrences still exhibit the STC. What, then, makes the language user opt for the STC or the IMC in (1) and (2)? To answer such questions, the third step digs another level deeper and keeps both the verb and the object constant. For zoeken and slachtoffer, this reveals a distinction that can be described as Lack of Affectedness (see a.o. Broccias 2001; Perek 2015: 91–92; Bernolet and Colleman 2016: 179–180). That is, an aggressor looking for a victim, as in (1), tends to be encoded in a Strictly Transitive Construction. Here, the object will undergo a change of state to become a victim. Conversely, a helper searching for a victim, as in (2), tends to be encoded in an Intransitive Motion Construction. Here, the object has already become a victim – only his or her location still needs to be determined. However, this distinction in terms of Lack of Affectedness does not seem to hold true for other objects of zoeken, such as oplossing ‘solution’. That is, even within one and the same verb, the directional alternation appears to be employed to express various contrasts in meaning. This calls us to adjust our view on constructional semantics, as also argued by Perek for the English conative alternation (Perek 2015: 90–103).
