Good and bad reasons for using and not using quantitative methods in linguistic research: a personal view

According to Gries (2013:4), linguistics in all its sub-disciplines is currently 'experiencing a paradigm shift towards more empirical methods' which is seen in the steady increase in studies employing quantitative research methods (cf. Janda 2013:5). However, quantitative methods have not traditionally been part of most linguists’ training and even today, few UG and PG linguistics curricula include more than a token treatment of quantitative research methods, including statistical methods. What is more, many linguists (myself included) do not feel a natural affinity to number crunching and advanced statistics. Should linguists nevertheless now bite the bullet, give up their old ways and fully embrace the quantitative revolution? It will be argued that there are good reasons not to use quantitative methods in a given situation and bad reasons for using them in general. Among those discussed are situations when a construct cannot be sensibly or satisfactorily operationalized (i.e. measured), an overly mechanistic view of language or a reductionist view of complex phenomena. Equally, however, there are bad reasons for not using quantitative methods and excellent reasons for using them. The former often include a lack of necessary knowledge or confidence as well as a lack of access to (sympathetic) expert advice while the latter are manifold and include the opportunity to investigate research questions that would otherwise remain elusive and in many cases a chance to elevate the degree of intersubjectivity and therefore reliability of results obtained. These aspects are discussed using examples taken primarily from the area of historical linguistics, in particular work on the speed of linguistic change. The conclusion is offered that the challenge is to use the good reasons and avoid the bad which, given the advantages that quantitative methods frequently offer, is likely to entail a further increase in the use of those methods in linguistics and the putting in place of the necessary support systems, but also a healthy dose of skepticism towards and critique of applications of quantitative methods where they are in danger of overcomplicating or obscuring rather than shedding light on linguistic matters. Gries, S. Th. (2013). Statistics for linguistics with R: A practical introduction (2nd ed). Berlin: De Gruyter. Janda, L. A. (2013). Quantitative methods in cognitive linguistics: An introduction. In L. A. Janda (Ed.), Cognitive linguistics - the quantitative turn: the essential reader (pp. 1-32). Berlin: De Gruyter