Naturalness

About: Naturalness is a research topic. Over the lifetime, 1305 publications have been published within this topic receiving 31737 citations.

The Naturalness Differential Hypothesis: Cross-linguistic Influence and Universal Preferences in Interlanguage Phonology and Morphology

Abstract: The role of the mother tongue has been a major topic of second language acquisition research over the last few decades, but despite the overwhelming empirical evidence of cross-linguistic influence in learner language a number Of questions still remain to be answered: what and how much is transferred when, how and why? This study explores the extent to which a theory of linguistic naturalness — the conceptual opposite of markedness, as formulated in Natural Phonology and Natural Morphology — might provide some insights into the issue of cross-linguistic influence. As an alternative to Eckman's (1977) 'Markedness Differential Hypothesis', a more detailed 'Naturalness Differential Hypothesis' is formulated in terms of phonological processes and morphological preferences. Unlike typological markedness, which may be regarded as a mere research tool, the notion of naturalness offers an explicit functional explanation of the observed learning difficulty, mainly in terms of ease of production and perception. Empirical support for these Claims will be drawn from research on the spontaneous acquisition of Italian as a second language.

How to Live Naturally and not be Bothered by Economy

Abstract: Markedness plays a central role within Natural Morphology and, more generally, in any functionalist approach to language This is not intended to deny that other forces, more or less conflicting or competing with markedness, may also play a role in shaping a natural language However, the main concern of this paper is that identifying naturalness, namely "what promotes the true nature of a thing", amounts to grasping the teleology of (any module of) a language grammar Concepts such as frequency or economy do not provide by themselves any deep insight into the essence of language, unless they are taken in the right perspective of being in a way symptoms of naturalness In particular, economy must be related to markedness reduction in order to capture its role within the architecture of grammar

Methodologies for the User Evaluation of the Motion of Virtual Humans

Abstract: Virtual humans are employed in many interactive applications, including (serious) games. Their motion should be natural and allow interaction with its surroundings and other (virtual) humans in real time. Physical controllers offer physical realism and (physical) interaction with the environment. Because they typically act on a selected set of joints, it is hard to evaluate their naturalness in isolation. We propose to augment the motion steered by such a controller with motion capture, using a mixed paradigm animation that creates coherent full body motion. A user evaluation of this resulting motion assesses the naturalness of the controller. Methods from Signal Detection Theory provide us with evaluation metrics that can be compared among different test setups, observers and motions. We demonstrate our approach by evaluating the naturalness of a balance controller. We compare different test paradigms, assessing their efficiency and sensitivity.

The naturalness criterion for supersymmetric scenarios

Abstract: Supersymmetry (SUSY) has been considered since long ago the leading paradigm of beyond the Standard Model (SM) physics as a framework that tackles the SM hierarchy problem, provides gauge coupling unification and a well-behaved cold dark matter (DM) candidate. Nevertheless, current experimental searches for new physics seem to have cornered the minimal versions of these models in unnatural regions of their parameter space, according to the standard Natural SUSY scenario. With the aim of formulating a general naturalness criterion for minimal supersymmetric scenarios, we have carefully re-examined the standard fine-tuning measure, which only deals with the cancellations needed to obtain the electroweak symmetry breaking (EWSB) scale, introducing several improvements such as the mixing of the fine-tuning conditions and the dependence on the low and high-energy (HE) scales. Furthermore, we have outlined a method that allow to straightforwardly derive naturalness bounds on the initial parameters and mass spectrum of any minimal supersymmetric standard model (MSSM) defined at any HE scale. We have applied this method to specific scenarios to compute a complete set of naturalness bounds. and also employed it to find the most natural gauge-mediated SUSY breaking model. Contrary to what was expected, we show that Natural SUSY, in general, does not demand light stops. The most stringent upper bound from naturalness is that of the gluino mass, which typically sets the level of fine-tuning, but strongly depends on the HE scale. The most robust result of Natural SUSY is by far that Higgsinos should be rather light. Besides, we have investigated other potential sources of fine cancellations in the MSSM, that if present, must be combined with that of the EWSB scale. The most important being the tuning to obtain the experimental Higgs mass and that required to reproduce the correct DM relic abundance. We have quantified them with p-value like measures that allow us to multiplicatively combine them with the electroweak (EW) fine-tuning. Regarding DM, we have considered the lightest neutralino as the DM particle and explored the various possibilities for its mass, composition and interactions that could give rise to accurate arrangements of the initial parameters to achieve the observed DM relic density. Finally, to illustrate the utility of the above stated criteria to estimate the global degree of naturalness, we have applied all of them to a specific model that features low-mass neutralinos and sleptons at low-energy. We find that these scenarios are rather unnatural when taking into account the aforementioned sources of tuning, which would have gone unnoticed, if we have only considered the EW fine-tuning.

Cross-Lingual Dialogue Dataset Creation via Outline-Based Generation

Abstract: Multilingual task-oriented dialogue (ToD) facilitates access to services and information for many (communities of) speakers. Nevertheless, its potential is not fully realized, as current multilingual ToD datasets—both for modular and end-to-end modeling—suffer from severe limitations. 1) When created from scratch, they are usually small in scale and fail to cover many possible dialogue flows. 2) Translation-based ToD datasets might lack naturalness and cultural specificity in the target language. In this work, to tackle these limitations we propose a novel outline-based annotation process for multilingual ToD datasets, where domain-specific abstract schemata of dialogue are mapped into natural language outlines. These in turn guide the target language annotators in writing dialogues by providing instructions about each turn’s intents and slots. Through this process we annotate a new large-scale dataset for evaluation of multilingual and cross-lingual ToD systems. Our Cross-lingual Outline-based Dialogue dataset (cod) enables natural language understanding, dialogue state tracking, and end-to-end dialogue evaluation in 4 diverse languages: Arabic, Indonesian, Russian, and Kiswahili. Qualitative and quantitative analyses of cod versus an equivalent translation-based dataset demonstrate improvements in data quality, unlocked by the outline-based approach. Finally, we benchmark a series of state-of-the-art systems for cross-lingual ToD, setting reference scores for future work and demonstrating that cod prevents over-inflated performance, typically met with prior translation-based ToD datasets.