Automatic verb-noun collocation extraction is an important natural language processing task, the results of which can be applied in various spheres including machine translation, language teaching, summarization, information extraction, disambiguation, etc. The paper describes a set of experiments the aim of which is to compare several approaches to automatic verb-noun collocation extraction. The main subjects under observation are the impact of span size and POS-filtering on the quality of collocation extraction. The experiments have shown that collocations lists extracted by means of POS-filtering are significantly more precise than those obtained without POS-filtering, whereas the extension of a span size has an ambiguous effect. On the one hand, it enables the extraction of distant collocates, but on the other hand it results in erroneous collocates, which leads therefore to consider the use of syntax-based approach for verb-noun collocation extraction