We invite you to Arqus Workshop "Methodology questions in Experimental linguistics"
Location: Seminarų kambarys B3
Tuesday, April 26
13:30 - 13:45: Introduction
13:45 - 14:30: Alexandre Cremers, Lea Fricke, Edgar Onea: "The importance of being earnest: How truth and evidence affect statement acceptability"
14:45 - 15:30: Maya Cortez-Espinoza, Lea Fricke "Testing German einige: How tense and cardinality affect a scalar term"
16:30 - 17:15: Lea Fricke, Edgar Onea, "The pragmatics of exhaustivity in embedded questions: A crosslinguistic comparison of German and English"
17:15 - 17:30: Open discussion and closing remarks
Thursday, April 28
10:00 - 10:15: Introduction
10:15 - 11:45: Alexandre Cremers: "Why statistical models should be informed by theoretical considerations when analysing semantic/pragmatic data".
13:30: "Bring your own dataset" interactive session (language: R).
18:00: end of workshop, social gathering
Alexandre Cremers, Lea Fricke, Edgar Onea: "The importance of being earnest: How truth and evidence affect statement acceptability"
Truth-value judgments are one of the most common measures in experimental semantics and pragmatics, yet there is no standardized way to elicit such judgments. Despite anecdotal remarks on how proper choice of prompts or response options could help disentangle pragmatic from semantic effects, little is known regarding the relation between parameters of the task and what it actually measures. We tested a range of prompts and two response options for their sensitivity to truth of the target sentence, prior evidence, and the interaction between these two factors. Our results confirm that prompts vary wildly in their sensitivity to pragmatic factors, and should allow researchers to make an informed choice depending on what they want to test. Response options were shown not to have much effect, with a Likert scale being systematically preferable to binary choice. To conclude, we discuss some theoretical consequences of our results for pragmatics, philosophy of language, and social psychology.
Maya Cortez-Espinoza, Lea Fricke
Testing German einige: How tense and cardinality affect a scalar term
When interpreting a scalar term like the German quantifier einige ‘some’, listeners will take a number of factors into consideration. We present an experiment that tested two of these factors. The first factor we investigated is the size of the set einige N refers to. Previous experiments on English some (van Tiel & Geurts 2013, Degen & Tanenhaus 2015) suggest that higher cardinalities are regarded as more typical representatives of a phrase like some N than smaller cardinalities (≤ 3). Our experimental results suggest that this is the case for German einige, too. The second factor is tense. We hypothesized that for a past-tense sentence like Some participants danced the scalar implicature some, but not all is more frequently drawn than when the sentence is in future tense (see Chierchia et al. 2000 who found a similar effect regarding the exclusivity inference of or). Our results support this hypothesis: SI-violations are more strongly penalized in past tense than in future tense. We argue that this is due to a difference in the immediate question under discussion (Roberts, 2012) between past and future tense.
Lea Fricke, Edgar Onea (joint work with Emilie Destruel & Malte Zimmermann)
The pragmatics of exhaustivity in embedded questions: A crosslinguistic comparison of German and English
We present a cross-linguistic experimental study investigating the exhaustivity properties of questions embedded under wissen/to know and vorhersagen/to predict in German and English. For the longest time, a very influential position in the theoretical literature was that such embedded questions should only come with a strongly exhaustive (SE) interpretation. Recent experimental research has shown, however, that the intermediate exhaustive (IE) reading is the dominant interpretation. We explore the puzzle of this discrepancy between introspection and experimental findings by investigating the pragmatic likelihood of the different readings. For wissen/to know, we found SE readings to have the highest probability in both languages, in line with the judgements in the early theoretical literature; IE readings were attested too, but we did not replicate the results of previous experiments. For German vorhersagen, the IE-reading clearly had the highest probability, whereas for English to predict the preference for IE over SE was only slight. This cross-linguistic difference correlates with independent evidence from corpus data which indicate that German vorhersagen and English to predict are not fully equivalent in lexical meaning.