We take a close look at a recent dataset of TED-talks annotated with the questions they implicitly evoke, TED-Q (Westera et al., 2020). We test to what extent the relation between a discourse and... Show moreWe take a close look at a recent dataset of TED-talks annotated with the questions they implicitly evoke, TED-Q (Westera et al., 2020). We test to what extent the relation between a discourse and the questions it evokes is merely one of similarity or association, as opposed to deeper semantic/pragmatic interpretation. We do so by turning the TED-Q dataset into a binary classification task, constructing an analogous task from explicit questions we extract from the BookCorpus (Zhu et al., 2015), and fitting a BERT-based classifier alongside models based on different notions of similarity. The BERT-based classifier, achieving close to human performance, outperforms all similarity-based models, suggesting that there is more to identifying true evoked questions than plain similarity. Show less
We present a new dataset of TED-talks annotated with the questions they evoke and, where available, the answers to these questions. Evoked questions represent a hitherto mostly unexplored type of... Show moreWe present a new dataset of TED-talks annotated with the questions they evoke and, where available, the answers to these questions. Evoked questions represent a hitherto mostly unexplored type of linguistic data, which promises to open up important new lines of research, especially related to the Question Under Discussion (QUD)-based approach to discourse structure. In this paper we introduce the method and open the first installment of our data to the public. We summarize and explore the current dataset, illustrate its potential by providing new evidence for the relation between predictability and implicitness {--} capitalizing on the already existing PDTB-style annotations for the texts we use {--} and outline its potential for future research. The dataset should be of interest, at its current scale, to researchers on formal and experimental pragmatics, discourse coherence, information structure, discourse expectations and processing. Our data-gathering procedure is designed to scale up, relying on crowdsourcing by non-expert annotators, with its utility for Natural Language Processing in mind (e.g., dialogue systems, conversational question answering). Show less
MATISSE, the VLTI 2nd generation spectro-interferometric L, M and N bands imager, has been commissioned from March 2018 to March 2020. It is open to the General User since April 2019. A complete... Show moreMATISSE, the VLTI 2nd generation spectro-interferometric L, M and N bands imager, has been commissioned from March 2018 to March 2020. It is open to the General User since April 2019. A complete analysis of its performances is given in this paper for MATISSE standalone (with UTs and ATs) and for the GRAVITY for MATISSE (GRA4MAT) mode (with ATs) where the GRAVITY fringe tracker is used to stabilize the fringes in MATISSE and hence improve its sensitivity and spectral coverage at high spectral resolution. This paper presents the key operation parameters of MATISSE and decomposes its performances in fundamental precision per spectral channel for all measurements and in broad band calibration errors on the accuracy of visibility and closure phase. It is intended to give the user a full description of the different errors that must be considered and weighted in the model fitting and image reconstruction. The first image reconstructions achieved by MATISSE are discussed. The performances demonstrated here in the full very broad spectral domain of MATISSE open a very large domain of scientific applications that includes but strongly expands quantitatively and qualitatively the initial science program of the first generation instrument MIDI and, combined with GRAVITY, offers an extremely powerful tool to characterize the temperature and composition of dusty and molecular components of YSOs, AGNs and evolved stars. Show less
This paper presents results of a corpus study on the South-Cushitic language Gorwaa which investigated the ‘enigmatic’ -oo/-(h)ee suffix (Mous & Qorro 2010:47, Harvey 2018). Various contexts in... Show moreThis paper presents results of a corpus study on the South-Cushitic language Gorwaa which investigated the ‘enigmatic’ -oo/-(h)ee suffix (Mous & Qorro 2010:47, Harvey 2018). Various contexts in which this nominal suffix occurs are identified, including negation, polar questions, universal quantification, adverbials, the object of comparison, and locatives. I characterise these as non-specific contexts and frame-setting topics. I compare these contexts to those in which the augment (nominal pre-prefix) is dropped in Bantu languages and, based on the empirical similarities, I discuss whether analyses of the Bantu augment can account for the Gorwaa cases. One analysis proposes that the suffix marks (non-)specificity, which I show is not fully satisfactory. Instead, I propose that what truly conditions the appearance of the suffix is the syntactic position of the nominal with respect to the verb phrase. Show less
In forensisch zaakonderzoek komt steeds vaker spraakmateriaal in verschillende talen voor. Dit roept de vraag op of er taalonafhankelijke spreker-specifieke kenmerken zijn. De bilabiale nasaal /m/... Show moreIn forensisch zaakonderzoek komt steeds vaker spraakmateriaal in verschillende talen voor. Dit roept de vraag op of er taalonafhankelijke spreker-specifieke kenmerken zijn. De bilabiale nasaal /m/ is één van de meest spreker-specifieke segmenten, wat wordt toegeschreven aan de rigiditeit van de neusholte [1]. Tegelijkertijd is de mondholte ook betrokken bij de productie en heeft de tong daarbij geen vaste positie [2]. Hierdoor is er ruimte voor binnen-sprekervariatie, die mogelijk taalafhankelijk is. Wij onderzochten in hoeverre de realisatie van /m/ verschilt tussen de eerste (L1) en tweede taal (L2) van meertalige sprekers.Er zijn monologen gebruikt van 53 vrouwelijke sprekers uit D-LUCEA [3], in hun L1 Nederlands en L2 Engels. De sprekers waren eerstejaarsstudenten van University College Utrecht en hadden een bovengemiddelde beheersing van het Engels. De nasalen werden gesegmenteerd in Praat en geanalyseerd op verschillende akoestische kenmerken.De resultaten laten zien dat de verschillen tussen de realisaties in de L1 en L2 minimaal zijn. Alleen de tweede nasale formant (N2) liet een taalverschil zien: hoger in de L2 dan in de L1. Sprekers verschilden in de mate waarin ze deze verschuiving vertoonden en voor sommigen was het resultaat in tegengestelde richting.Hoewel de gevonden L1−L2 verschillen in de uitspraak van de /m/ klein zijn, lijkt de bilabiale nasaal niet geheel taalonafhankelijk. De N2 wordt gerelateerd aan de mond- en neusholte [4], wat duidt op een aanpassing in de productie. Rekening houdend met deze aanpassing zou de /m/ bruikbaar kunnen zijn in meertalige sprekervergelijkingen. Vervolgstappen zijn om dit te onderzoeken met sprekerclassificatie. Show less
Background: It is controversial whether patients with gallbladder cancer (GBC) presenting with jaundice benefit from resection. This study re-evaluates the impact of jaundice on resectability and... Show moreBackground: It is controversial whether patients with gallbladder cancer (GBC) presenting with jaundice benefit from resection. This study re-evaluates the impact of jaundice on resectability and survival.Methods: Data was collected on surgically explored GBC patients in all Dutch academic hospitals from 2000 to 2018. Survival and prognostic factors were assessed.Results: In total 202 patients underwent exploration and 148 were resected; 124 non-jaundiced patients (104 resected) and 75 jaundiced patients (44 resected). Jaundiced patients had significantly (P 0.05) more pT3/T4 tumors, extended ( 3 segments) liverand organ resections, major postoperative complications and margin-positive resection. 90-day mortality was higher in jaundiced patients (14% vs. 0%, P < 0.001). Median overall survival (OS) was 7.7 months in jaundiced patients (2-year survival 17%) vs. 26.1 months in non-jaundiced patients (2-year survival 39%, P < 0.001). In multivariate analysis, jaundice (HR1.89) was a poor prognostic factor for OS in surgically explored but not in resected patients. Six jaundiced patients did not develop a recurrence; none had liveror common bile duct (CBD) invasion on imaging.Conclusion: Jaundice is associated with poor survival. However, jaundice is not an independent adverse prognostic factor in resected patients. Surgery should be considered in patients with limited disease and no CBD invasion on imaging. Show less