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

An atlas of the Galactic plane along with the molecular clouds in Orion, o Oph, ([4¡ .7 \ b \ 4¡ .7), and Taurus-Auriga, has been produced at 60 and 100 km from IRAS data. The atlas consists of resolution-enhanced co-added images with 1@È2@ resolution and co-added images at the native IRAS resolution. The IRAS Galaxy Atlas, together with the Dominion Radio Astrophysical Observatory H I line/21 cm continuum and FCRAO CO (1È0) Galactic plane surveys, which both have similar (D1@) resolution to the IRAS atlas, provides a powerful tool for studying the interstellar medium, star formation, and large-scale structure in our Galaxy. This paper documents the production and characteristics of the atlas.

No abstract

No abstract