Data Collection Advice

Henrik Rosenkvist Henrik.Rosenkvist at
Wed Oct 26 07:09:22 UTC 2011


There are two basic ways of elicating quanttitative judgements: absolute 
judgements (on different scales) or magnitude estimation. The trendiest 
method is probably magnitude estimation (see for instaince Heycock et al 
2010, but that requires advanced statistics. Bader & Häussler, comparing 
the two methods, concludes that "Practical matters aside, we do not 
think that there are strong reasons to favor one method over the other." 
(Bader & Häussler 2009:322).

In the ScanDiaSyn field work, we have been using a 5-degree scale, and 
in my latest questionnarie study I formulated the alternatives as below:

Grade the sentences in this leaflet on a scale 1–5.
    1 = not at all in accordance with general language use (you never 
say so)
    2 = possibly in accordance with general language use (you may 
perhaps say so in certain
    3 = in partial accordance with general language use (sometimes you 
can say so)
    4 = in accordance with general language use (you can say so)
    5 = in full accordance with general language use (this is what you 
normally say)

I think that acceptable/non-acceptable leads the informant to grade 
sentences according to sociolinguistic norms.

It is worth pointing out that in a large study with many informants, the 
method chosen is the only factor determining reliability. If the 
researcher has trained a group of informants and has had regular 
sessions with them, the situation is different (see a good paper by 
Alison Henry 2005 about this).

A starting point is also Cornips & Poletto 2005, and, of course, Schütze 


Henrik R.

s.t. bischoff skrev:
> Hello all,
> Some colleagues and I are working on a survey to find  out what types of
> code-switching  might be acceptable (in terms of traditional grammaticallity
> judgments) for constructions like the following.
> (1) The toy verde was my favorite
> (2) The casa big está a la venta
> (3) Leamos juntos el book interesante
> (4) No me gusta the cuchillo sharp
> We want use the data to develop some psycholinguistic experiments regarding
> code-switching.  At the moment we are trying to determine a good way to
> collect responses. We intend to provide an online survey with examples like
> 1 through 4, but haven't worked out how to collect the responses. We've seen
> different ways in which this has been done, with linguists often simply
> providing an "acceptable" or "not acceptable"  choice and psychologist
> providing at times up to seven different options. . Any thoughts or advice
> on what has been useful and what hasn't been useful for folks would be
> welcome. We want the survey to be short and simple and we aware of the many
> pitfalls of grammaticallity judgements.
> Thanks,
> Shannon

Henrik Rosenkvist
docent, nordiska språk
Språk- och litteraturcentrum
Lunds universitet
Box 201
221 00 Lund
tel: 046-222 87 04
e-post: Henrik.Rosenkvist at

Henrik Rosenkvist
Associate Professor, Scandinavian Languages
Dept. of Languages and Literature
Lund University
P. O. Box 201, SE-221 00 Lund, SWEDEN
Tel.: +46 46 222 87 04
E-mail: Henrik.Rosenkvist at

More information about the Funknet mailing list