Full Title: More Than Just Noise: Detecting Patterns in Acceptability Judgment Data (DGfS 2026 Workshop)

Date: 25-Feb-2026 - 27-Feb-2026
Location: Trier, Germany
Meeting Email: [email protected]

Linguistic Field(s): General Linguistics; Neurolinguistics; Psycholinguistics

Call Deadline: 15-Sep-2025

Call for Papers:

AG7 of the Annual Meeting of the German Linguistic Society: https://www.uni-trier.de/universitaet/fachbereiche-faecher/fachbereich-ii/forschung-und-zentren/dgfs2026
Organized by: Jana Häussler (Uni Bielefeld), Thomas Weskott (Uni Göttingen), Sarah Zobel (Uni Hannover / HU Berlin)

Linguistic acceptability is one of the major tools to detect patterns in language: our intuitions about whether a sentence is "good" or "bad" are a source of evidence that is readily accessible and easy to communicate. Since the advent of more rigorous measurement of linguistic acceptability in the 1990s (Cowart 1997, Schütze 1996), the acceptability judgment task (AJT) has been used in controlled experiments employing factorial designs to collect judgments from samples of multiple participants for samples of multiple items; statistical regression methods are utilized to separate information about the underlying linguistic patterns from the "noise", i.e., the variance generated by repeated measurements. Subdisciplines that are traditionally more theoretically inclined, like syntax, semantics, and pragmatics, have hugely profited from the empirical progress that this experimental approach has engendered, as witnessed by the publication of handbooks like, e.g., Cummins and Katsos (eds., 2019), and Goodall (ed., 2021).

While the usefulness of AJT data to inform us about linguistic patterns is undebatable, we think that the true potential of the method as a source of insights about language patterns is underestimated. This, we argue, is due to an erroneous assumptions about the AJT underlying current experimental practice: the variance generated by interindividual differences between participants (think, e.g., of dialect, literacy, proficiency (including L2), or age), as well as the variance that comes with testing multiple items, and the possible interactions of these two sources of variance, are treated as "random"; and any possible information these variances might contain---beyond parameter estimation---is discarded by the statistical procedures usually employed. This practice loses information that is potentially informative about the complex way in which humans react to linguistic stimuli---what Barr (2018) has called "encounters", and which he proposes to consider as the unit of generalization, rather than populations of speakers and/or items. This loss of information is mostly due to the way in which studies employing the AJT are designed: they are usually focused on the effect of interest, disregarding how it might be related to other systematic properties buried in the "random units". A further loss of information is due to the schematic use of statistical procedures like mixed model regression, which usually focuses on the difference between (a set of) means, while disregarding the potentially informative properties of the underlying distributions (cf. Kneib et al. 2023).

This workshop aims at addressing these two problems by discussing contributions that propose to go beyond the current methodological standard practice. We invite:

- contributions attempting to address the systematic effects of participant-level properties in acceptability judgments; examples are working memory, dialect/sociolect, age, proficiency, literacy; more generally, contributions investigating just any property that potentially affects the AJT in a systematic fashion are welcome

- contributions addressing item-level properties like complexity, register, markedness/frequency, context sensitivity, etc.

- contributions that investigate the effect of item-level properties on the behavior of acceptability measures, like different types of benchmarking techniques, satiation effects, "squishing" and "stretching" of scales, etc.

- contributions that employ novel statistical methods---on actual or modelled data, and not necessarily limited to AJT data---that seek to establish effects over and above mere differences between means like, e.g., distributional regression (Bayesian or "classical");

- contributions that relate data from the AJT systematically to other dependent variables, like truth value/felicity judgments, reading time/eye tracking, or EEG data.

By bringing together researchers working on these different aspects of the problem, we hope to initiate a discussion within the workshop, and possibly beyond that, of how our use of acceptability measures in the detection of linguistic patterns can be improved.

Invited Speakers:
- Anne Mette Nyvad (Aarhus University)
- TBA

Submission Details:
- Abstract length: a single page abstract of max. 350 words, plus up to two pages containing details of experimental design, materials, and statistic analyses (incl. graphs); note that the 350 words abstract will be published in the conference booklet and thus should be comprehensible without the additional materials; the abstracts with the additional two pages will be published on OSF. Please note that each participant may appear as the first author and as presenter on only one submission.

Deadline for submission: Monday, September 15th, 23:59 (EXTENDED)
Please send your abstract as an anonymous PDF to [email protected]
Please add the names and affiliations of all authors in the body of the email.

Important Dates:
- Call for papers opens: 7 July 2025
- Submission deadline: 15 September 2025 (EXTENDED)
- Notification of acceptance: beginning of October
- Workshop: 25-27 February 2026

References
Barr, Dale J. (2018). Generalizing over encounters: Statistical and theoretical considerations. In: Rueschemeyer, Shirley-Ann & M. Gareth Gaskell (eds.). Oxford Handbook of Psycholinguistics. Oxford, UK: OUP.
Cowart, Wayne (1997). Experimental Syntax. Applying Objective Methods to Sentence Judgments. Thousand Oaks: SAGE.
Cummins, Chris, & Napoleon Katsos (eds., 2019). The Oxford Handbook of Experimental Semantics and Pragmatics. Oxford, UK: OUP.
Kneib, Thomas, Alexander Silbersdorff, & Benjamin Säfken (2023). Rage Against the Mean – A Review of Distributional Regression Approaches. Econometrics and Statistics, Vol. 26, 99-123.
Schütze, Carston (2016). The empirical base of linguistics: Grammaticality judgments and linguistic methodology. Berlin: LSP.