Rethinking structuralism:The posthumous publications of Gustave Guillaume (1883–1960) John Hewson 1. Introduction. What is attempted here is a review and discussion of twenty-two volumes of original work of Gustave Guillaume published since his death in 1960, and another half dozen volumes translated into other languages around the globe. G has been described as one of the most important French linguists of the twentieth century (Joly 2004:9), but his work is little known in anglophone circles, for a variety of reasons. First of all, he was viewed unsympathetically by the positivists, behaviorists, and other antimentalists of his day for being an unrepentant mentalist. He, in turn, was often critical of the fashionably mindless positivism of his day, and made his own carefully thought out epistemology for the operation of scientific method in linguistics (Guillaume 1958). It was published in Les Études philosophiques just over a year before his death, and stresses the importance of observation. Observation and reflection, where reflection leads to a more profound observation, and to further cycles of reflection and observation, were the keynotes of his method. The whole is a snowballing experience, in which theory is never at more than one remove from the empirical at every stage. Second, the whole body of his work, except for one translated volume (see below), is inaccessible to scholars who do not read French. Third, like Jakobson, he had wide-ranging interests; he was not just a nuts-and-bolts linguist, but was interested in language as a human phenomenon: for him the mother tongue that a child learned was a mental universe, a system of representation with cosmic dimensions, enabling the child to represent entities in both time and space. Critics with a narrow view of linguistics accused him of dabbling in psychology and philosophy.1 But among those who followed his seminars at the École des Hautes Études in Paris, there were, by contrast, not only linguists, but also a variety of unusual individuals: professionals who attended weekly, year after year, fascinated by the intellectual challenge, some of whom became his treasured 'amiable tormentors' because of the questions they asked, as we know from many passages in the Leçons.2 And last, he has had a quite undeserved reputation of being unreadable, an accusation that may be heard repeated by those who have never read him. Certainly, there are texts of G that are very difficult to read and interpret, but there are also texts, such as the seventeen posthumous volumes of the Leçons de linguistique, that are insightful, entertaining, and sometimes a delight to read. Those who have read him sometimes complain of his neologisms, but the neologisms are mostly additions to a nomenclature [End Page 820] that is entirely traditional, as noted by Stefanini (1973:319). Indeed, it is not the vocabulary that is the difficulty, but the new ideas that go beyond that which is familiar, especially those that run counter to our acquired prejudices. For G language was not a thing, a Ding-an-sich, but an activity requiring a speaker to create discourse (typically sentences) from a previously acquired tongue (without this prior acquisition the creation of sentences would be impossible). In short, there is a means of production, a producer, and a product, and a careful observer will note that the activity of language requires all three. Consequently, if we define a language as 'the totality of utterances that can be made in a speech community', as did Bloomfield (1926, definition 4), following the early Wittgenstein (1919),3 we put ourselves in an abstract limbo, where the imaginary set of ideal sentences is a metaphysical phantom, the idealized final product of possible language activity. If we accept Bloomfield's cartbefore the-horse definition,4 we may have difficulty seeing that the concrete activity of language, as described by G, is to CREATE sentences that express the speaker's cognitive experience, an activity that requires a speaker, whose role cannot be abstracted, as it is in generative models. The consequences of this point are dealt with in §3. 1.1. Guillaume's Scholarly Papers. G bequeathed all his scholarly papers, some 60,000 manuscript pages, to his...