Though state-of-the-art sentence representation models can perform tasks requiring significant knowledge of grammar, it is an open question how best to evaluate their grammatical knowledge. We explore five experimental methods inspired by prior work evaluating pretrained sentence representation models. We use a single linguistic phenomenon, negative polarity item (NPI) licensing in English, as a case study for our experiments. NPIs like any are grammatical only if they appear in a licensing environment like negation (Sue doesn't have any cats vs. *Sue has any cats). This phenomenon is challenging because of the variety of NPI licensing environments that exist. We introduce an artificially generated dataset that manipulates key features of NPI licensing for the experiments. We find that BERT has significant knowledge of these features, but its success varies widely across different experimental methods. We conclude that a variety of methods is necessary to reveal all relevant aspects of a model's grammatical knowledge in a given domain. 1 Other prominent theories of NPI licensing are based on notions of non-veridicality (

No abstract

BackgroundAluminum-made settings are widely used in healthcare, schools, public facilities and transit systems. Frequently-touched surfaces of those settings are likely to harbour bacteria and be a potential source of infection. One method to utilize the effectiveness of copper (Cu) in eliminating pathogens for these surfaces would be to coat the aluminum (Al) items with a Cu coating. However, such a combination of Cu and Al metals is susceptible to galvanic corrosion because of their different electrochemical potentials.MethodsIn this work, a new approach was proposed in which electrolytic plasma oxidation (EPO) of Al was used to form an oxide surface layer followed by electroplating of Cu metal on the top of the oxide layer. The oxide was designed to function as a corrosion protective and biocompatible layer, and the Cu in the form of dots was utilized as an antibacterial material. The antibacterial property enhanced by superhydrophilicity of the Cu-dotted oxide coating was evaluated.ResultsA superhydrophilic surface was successfully prepared using electrolytic plasma oxidation of aluminum (Al) followed by electroplating of copper (Cu) in a Cu-dotted form. Both Cu plate and Cu-dotted oxide surfaces had excellent antimicrobial activities against E. coli ATCC 25922, methicillin-resistant Staphylococcus aureus (MRSA) ATCC 43300 and vancomycin-resistant Enterococcus faecium (VRE) ATCC 51299. However, its Cu-dotted surface morphology allowed the Cu-dotted oxide surface to be more antibacterial than the smooth Cu plate surface. The enhanced antibacterial property was attributed to the superhydrophilic behaviour of the Cu-dotted oxide surface that allowed the bacteria to have a more effective killing contact with Cu due to spreading of the bacterial suspension media.ConclusionThe superhydrophilic Cu-dotted oxide coating surface provided an effective method of controlling bacterial growth and survival on contact surfaces and thus reduces the risk of infection and spread of bacteria-related diseases particularly in moist or wet environments.

Causative morphology has been associated with either the introduction of an event of causation or the introduction of a causer argument. However, morphological causatives are mono-eventive, casting doubt on the notion that causatives fundamentally add a causing event. On the other hand, in some languages the causative morpheme is closer to the verb root than would be expected if the causative head is responsible for introducing the causer. Drawing on evidence primarily from Tagalog and Halkomelem, I argue that the syntactic configuration for morphological causatives involves Voice over Voice, and that languages differ in whether their ‘causative marker’ spells out the higher Voice, the lower Voice or both.

No abstract