ABSTRACT This article focuses on Arne Naess's work in the philosophy of language, which he began in the mid-1930s and continued into the 1960s. This aspect of his work is nowadays relatively neglected, but it deserves to be revisited. Firstly, it is intrinsically interesting to the history of analytic philosophy in the twentieth century, because Naess questioned some of the established philosophical methodologies and assumptions of his day. Secondly, it suggests a compelling but unacknowledged intellectual pedigree for some recent developments (...) in linguistics. Naess's philosophy of language developed from his reaction against logical positivism, in particular against what he saw as its unempirical assumptions about language. He went on to establish ?empirical semantics?, in which the study of language was based on real-life linguistic data, drawn primarily from questionnaires issued to philosophically naïve subjects. He also experimented with methods for ?occurrence analysis?, but concluded that the collection and analysis of sufficiently large bodies of naturally-occurring data was impractical. Empirical semantics was not well received by Naess's philosophical contemporaries. It was also seen as being at odds with contemporary trends in linguistics. However, some present-day branches of linguistics have striking resonances with Naess's work from as much as seventy years ago. In sociolinguistics, questionnaires have become an established means of collecting linguistic data. In corpus linguistics, advances in technology have made Naess's unobtainable ideal of ?occurrence analysis? a viable methodology. Some of the principal conclusions reached as a result of this methodology are strikingly similar to Naess's own findings. (shrink)