Search

The chapter focuses on V3 patterns in West Flemish in which a subject-initial non-inverted V2 clause is preceded by an adverbial adjunct which modifies temporal or modal coordinates of the associated clause, in apparent violation of the V2 constraint. The pattern is not available in many other varieties of Dutch, including Standard Dutch. The chapter summarizes the main distributional and interpretive properties of the initial constituent, focusing on, among other things, the fact that for its interpretation, the initial adjunct cannot be reconstructed to a clause-internal position. On the basis of the distributional and interpretive properties of the initial constituent, it is argued that these V3 patterns are in line with V2 syntax because the initial constituent is extrasentential. The chapter develops the discourse syntax for main clause external constituents and argues that the micro-variation observed can be captured by the hypothesis that there is micro-variation between Standard Dutch and West Flemish in terms of the derivation of subject-initial V2 root clauses.

Predicates like "coloring-the-star" denote events that have a temporal duration and a culmination point (telos). When combined with perfective aspect (e.g., "Valeria has colored the star"), a culmination inference arises implying that the action has stopped, and the star is fully colored. While the perfective aspect is known to constrain the conceptualization of the event as telic, many reading studies have demonstrated that readers do not make early commitments as to whether the event is bounded or unbounded. A few visual-world studies tested the processing of telic predicates during online sentence processing, demonstrating an early integration of aspectual and temporal cues. By employing the visual-world paradigm, we tested the incremental processing of the perfective aspect in Italian in two eye-tracking studies in which listeners heard durative predicates in the perfective form in a scenario showing a completed and a non-completed event. Differently from previous studies, we compared telic durative predicates such as "coloring-the-star" to punctual predicates such as "lighting-the-candle." While for punctual predicates, the inferences of telicity (the event has a telos) and of culmination (the telos is reached) are lexically encoded in the perfective verb, for durative predicates, the degree of event completion (visually encoded) needs to be integrated with perfective aspect (linguistically encoded) for the culmination inference derivation. By modulating the interaction of visual and linguistic stimuli across the two experiments, we show that the verb's perfective aspect triggers the culmination inference incrementally during sentence processing, offering novel evidence for the continuous integration of linguistic processing with real-world visual information.

We investigate grounded language learning through real-world data, by modelling a teacher-learner dynamics through the natural interactions occurring between users and search engines; in particular, we explore the emergence of semantic generalization from unsupervised dense representations outside of synthetic environments. A grounding domain, a denotation function and a composition function are learned from user data only. We show how the resulting semantics for noun phrases exhibits compositional properties while being fully learnable without any explicit labelling. We benchmark our grounded semantics on compositionality and zero-shot inference tasks, and we show that it provides better results and better generalizations than SOTA non-grounded models, such as word2vec and BERT., Published as a conference paper at NAACL2021

We address the problem of user intent prediction from clickstream data of an e-commerce website via two conceptually different approaches: a hand-crafted feature-based classification and a deep learning-based classification. In both approaches, we deliberately coarse-grain a new clickstream proprietary dataset to produce symbolic trajectories with minimal information. Then, we tackle the problem of trajectory classification of arbitrary length and ultimately, early prediction of limited-length trajectories, both for balanced and unbalanced datasets. Our analysis shows that k-gram statistics with visibility graph motifs produce fast and accurate classifications, highlighting that purchase prediction is reliable even for extremely short observation windows. In the deep learning case, we benchmarked previous state-of-the-art (SOTA) models on the new dataset, and improved classification accuracy over SOTA performances with our proposed LSTM architecture. We conclude with an in-depth error analysis and a careful evaluation of the pros and cons of the two approaches when applied to realistic industry use cases. Borja Requena acknowledges ERC AdG NOQIA, Spanish Ministry MINECO and State Research Agency AEI (FIDEUA PID2019-106901GB-I00/10.13039 / 501100011033, SEVERO OCHOA No. SEV-2015-0522 and CEX2019-000910-S, FPI), European Social Fund, Fundació Cellex, Fundació Mir-Puig, Generalitat de Catalu-nya (AGAUR Grant No. 2017 SGR 1341, CERCA program, QuantumCAT _U16-011424, co-funded by ERDF Operational Program of Catalonia 2014-2020), MINECO-EU QUANTERA MAQS (funded by State Research Agency (AEI) PCI2019-111828-2 / 10.13039/501100011033), EU Horizon 2020 FET-OPEN OPTOLogic (Grant No 899794), and the National Science Centre, Poland-Symfonia Grant No. 2016/20/W/ST4/00314. LL acknowl-edges funding from EPSRC Early Career Fellowship EP/P01660X/1. Finally, authors wish to thank Emily Hunt for giving us her time and English sophisticatio

The empirical focus of this paper is what looks like a verb third (V3) pattern in West Flemish (WF) in which an adverbial modifier (typically a temporal or conditional adjunct) is followed by a non-inverted subject-initial verb second (V2) root clause. This pattern will be referred to as the non-inverted V3 pattern. This paper has two goals. The first and major aim is to document the WF non-inverted V3 pattern, detailing both its syntactic and its semantic properties, in order to make these data available to the research community at large. We also develop an analysis in line with the hypothesis, substantiated in the paper, that WF is a genuine V2 language and that the apparent V3 data in do not jeopardise this assumption. The core hypothesis is that the initial adjunct in V3 patterns is extra-sentential or ‘main clause-external’ (cf. Broekhuis and Corver in Syntax of Dutch. Verbs and verb phrases. Volume 3: Chapter 14: main clause-external elements, Amsterdam University Press, Amsterdam, 2016. http://www.oapen.org/ ), and that it combines with a full-fledged V2 clause by virtue of a discourse-structure (cf. Auer in Pragmatics 6:295–322, 1996). That the non-inverted V3 pattern is not acceptable in all varieties of Dutch is accounted for in terms of microvariation in the syntactic derivation of subject-initial V2 sentences.

The claims that big data holds the key to enterprise successes and that Artificial Intelligence (AI). is going to replace humanity have become increasingly more popular over the past few years, both in academia and in the industry. However, while these claims may indeed capture some truth, they have also been massively oversold, or so we contend here. The goal of this paper is two-fold. First, we provide a qualified defence of the value of less data within the context of AI. This is done by carefully reviewing two distinct problems for big data driven AI, namely a) the limited track record of Deep Learning (DL) in key areas such as Natural Language Processing (NLP), b) the regulatory and business significance of being able to learn from few data points. Second, we briefly sketch what we refer to as a case of “A.I. with humans and for humans”, namely an AI paradigm whereby the systems we build are privacy-oriented and focused on human-machine collaboration, not competition. Combining our claims above, we conclude that when seen through the lens of cognitively inspired A.I., the bright future of the discipline is about less data, not more, and more humans, not less.

This chapter focuses on the use of the pronoun nó as an optional expletive in spoken Vietnamese, a radical pro-drop language. Having provided a detailed description of the distribution and interpretation of this element against the general background of the place of overt expletives in pro-drop languages, we argue that the expletive is inserted not to satisfy purely formal requirements, such as the EPP requirement, but rather to impose a specificity requirement on the eventive content of the clause. The present chapter shows that Vietnamese contributes to a better understanding of the nature of expletives and the conditions of their use cross-linguistically. This chapter also discusses the implications of our findings for the articulation of subject predicate structures and for the well-known distinction between thetic structures and categorical structures.

This paper discusses V3-patterns with a sentence-initial adverbial clause in Standard Dutch (StD) and West-Flemish (WF), which appear to violate the V2 restriction which normally regulates word order in these languages. V3-patterns occur in both languages; they can be interpreted as complying with the V2 constraint provided they are analyzed as the result of merging a regular root clause with V2 order with an extra-sentential adverbial clause. The paper shows that the distribution of V3-patterns is slightly wider in WF than in StD: StD requires that the root clause which combines with the extra-sentential constituent either exhibits subject-verb inversion (XP-V-S) or, in the case of subject initial V2 clauses, that the subject has a distinguished information-structural status such as contrastive focus/topic; WF allows V3-structures more freely, regardless of whether they display subject-verb inversion and regardless of the informational-structural status of the subject. The analysis takes as its point of departure the earlier claim that V2-languages can be symmetric in the sense that the finite verb always leaves the TP domain and occupies the highest head position in the root clause, the complementizer position C in the traditional generative analysis, or asymmetric in the sense that the position of the finite verb varies in that it occupies the C-position when the root clause exhibits subject-verb inversion or a lower TP internal tense position (T) in root clauses without inversion. The hypothesis is that V3-patterns with a sentence-initial adverbial clause are only possible if the initial adverbial clause attains a local relation with the finite verb, and that this requires the finite verb to be in the (higher) C-position. By assuming that StD is an asymmetric V2-language while WF is a symmetric V2-language the variation with respect to the distribution of V3-patterns in these languages can be captured.

In this paper, we test the Cumulative Effect proposed by (Haegeman et al. 2014. Deconstructing the subject condition in terms of cumulative constraint violation. The Linguistic Review 31(1). 73-150). In particular, we focus on the role of factors of external syntax in the modulation of subject opacity to extraction, addressing different constraints presented in (Haegeman et al. 2014. Deconstructing the subject condition in terms of cumulative constraint violation. The Linguistic Review 31(1). 73-150). Two sets of formal acceptability judgments are presented. In the first experiment, we address the opacity of subjects from a broad point of view, in order to assess whether subject DPs are more resistant to extraction than other DPs. The results confirm that subject constituents are more opaque to extraction than object constituents. In the second experiment, we address the impact on the modulation of the Cumulative Effect of three different constraints individually: The Freezing Principle, the Inactivity Condition and the Edge Condition. We did that through the manipulation of two different factors: The position of the extraction site and the predicate type. The results do not confirm the predictions made by the Cumulative Effect, since the interaction between different factors does not appear to be additive and incremental. We discuss the implications of these results for the existing theories of subject islands and address an alternative perspective, recently proposed by (Bianchi, Valentina & Cristiano Chesi. 2014. Subject islands, reconstruction, and the flow of the computation. Linguistic Inquiry 45(4). 525-569), according to which the opacity of subjects is a function of the syntax semantics interface.

© 2015, Springer Science+Business Media Dordrecht. The goal of this article is to explore the utility of experimental syntax techniques in the investigation of syntactic variation. To that end, we applied the factorial definition of island effects made available by experimental syntax (e.g., Sprouse et al. 2012) to four island types (wh/whether, complex NP, subject, and adjunct), two dependency types (wh-interrogative clause dependencies and relative clause dependencies) and two languages (English and Italian). The results of 8 primary experiments suggest that there is indeed variation across dependency types, suggesting that wh-interrogative clause dependencies and relative clause dependencies cannot be identical at every level of analysis; however, the pattern of variation observed in these experiments is not exactly the pattern of variation previously reported in the literature (e.g., Rizzi 1982). We review six major syntactic approaches to the analysis of island effects (Subjacency, CED, Barriers, Relativized Minimality, Structure-building, and Phases) and discuss the implications of these results for these analyses. We also present 4 supplemental experiments testing complex wh-phrases (also called D-linked or lexically restricted wh-phrases) for all four island types using the factorial design in order to tease apart the contribution of dependency type from featural specification. The results of the supplemental experiments confirm that dependency type is the major source of variation, not featural specification, while providing a concrete quantification of what exactly the effect of complex wh-phrases on island effects is.