Query expansion (QE) aims at improving information retrieval (IR) effectiveness by enhancing the query formulation. Because users' queries are generally short and because of the language ambiguity, some information needs are difficult to answer. Query reformulation and QE methods have been developed to face this issue. Relevance feedback (RF) is one of the most popular QE techniques. In its manual version, the system uses the information on the relevance -manually judged- of retrieved documents in order to expand the initial query. Rather than using users' judgment on the document relevance, blind RF considers the first retrieved documents as relevant. Generally speaking, RF methods consider the terms that cooccur with query terms within positive feedback documents as candidates for the expansion. Rather than considering feedback documents in their all, it is possible to analyze local information. This paper presents a new method that uses local context from feedback documents for QE. The method uses POS information as well as the remoteness from query terms within feedback documents. We show that the method significantly improves precision on TREC collections.