Abstract :
[en] A conspicuous characteristic of Germanic languages is that they have two morphological strategies at their disposal to express the preterite. The oldest strategy is called the ‘strong’ inflection, and derives from the Proto-Indo-European aspectual ablaut system (English sing ~ sang). The ‘weak’ inflection, by contrast, is a diachronically innovative strategy in Germanic, and uses a dental suffix (English work ~ worked). Most verbs take either the strong or the weak inflection, but the distribution is historically (as well as socio- and dialectally) in flux: over time, many verbs have shifted from strong to weak, or – less commonly – from weak to strong, or shift from one strong ablaut class to another, or from one weak class to another. The Germanic strong and weak preterite formation has been the subject of numerous studies which have approached the issue from different angles (philological, acquisitional, comparative …) (see, among others, Van Haeringen 1940, De Vriendt 1965, Seebold 1970, Tops 1974, Bybee Slobin 1982, Bammesberger 1986, Pinker Prince 1988, Van Coetsem 1990, Van Santen Lalleman 1994, Hare Elman 1995, Van Santen 1997, Kühne 1999, Nübling 2000, Pinker Ulman 2002, Albright Hayes 2003, Salverda 2006, Mailhammer 2006, 2007a,b, Nowak 2010, Vosters 2012, Heinzle 2013, Knooihuizen Strik 2014, Strik 2014). The availability of large corpora in more recent approaches have allowed scholars to scale up the earlier findings. See, for instance, Lieberman et al. (2007) and Carroll et al. (2012) on the evolutionary dynamics of English and German verbs respectively, or Anderwald (2012a,b,c) and Cuskley et al. (2014), who set out to investigate the (ir)regularization of English verbs working with the 400M COHA Corpus. Second, new methods are currently being developed, which hold the potential to shed new light on the class-shifting whims of Germanic verbs. Agent-based modelling, an in-silico simulation of languages processes (Shoham Layton-Brown 2009, Steels 2011), is currently applied to Germanic strong verbs, both in proof-of-concept stage (Pijpops et al. 2014) and in a more advanced fashion (Pugliese et al. 2014). Still, there is a significant gap between, on the one hand, the philological scrutiny of more ‘traditional’ scholars (past and present), who often work with limited datasets, but explore them in profound detail, and, on the other hand, the bird’s-eye perspective in the more recent ‘big data’ approaches, which have access to quantitatively more robust datasets, but perhaps inevitably gloss over many of the philological intricacies that haunt the verbal preterite morphology of Germanic. To date, the big data approaches have, for instance, not been overly concerned with the type frequency effect of ablaut class membership (see the critique in Carroll et al. 2012 on Lieberman et al. 2007), and often rely on a very crude distinction between regular weak inflection (work ~ worked) and irregular verbs, i.e. all the rest, both strong and irregular weak (think, have etc.). The time has come to bring researchers together to exchange methods and results that can help us to bridge the gap between different approaches.