In the framework of multimedia analysis and interaction, speech and language processing plays a major role. Many multimedia documents contain speech from which high level semantic information can be extracted, as in broadcast news or sports videos, with typical applications such as spoken document indexing, topic tracking and summarization. Hence, many multimedia document analysis applications require a collaboration between speech recognition and natural language processing (NLP) techniques. As NLP techniques are traditionally designed for text analysis, this combination can be seen as a mul-timodal fusion issue where the two modalities are audio and text. However, most of the time, both modalities are considered sequentially. A typical approach consists in automatically transcribing the audio track before analyzing the output-here considered as a regular text-with NLP methods. Independently processing the two modalities clearly seems suboptimal. This chapter focuses on recent research work toward a better integration between automatic speech recognition (ASR) and NLP for the analysis of spoken multime-dia documents with the goal of achieving a better transcription of multimedia streams.