Jan Svanlund, Stockholm University, The Department of Swedish Language and Multilingualism (formerly The Department for Scandinavian Languages)

The main aim of this project was to investigate how new Swedish compounds become established, i.e. conventionalized, and especially how they receive their meanings. What factors contribute mostly, and how do they interplay?

I also had the opportunity to enhance the study with simple tests on informants. This turned out to become very valuable. The informants had to report how they conceived of the meanings of the words (or guess, if they were unsure or had never heard them before), and if these were familiar to them.

The project developed into a number of case studies, where the conventionalization of nine new compounds of various kinds were analyzed in great detail: användargrupp (user group), bidragsmiljon (grant million), bokstavsbarn (letter child), curlingförälder (curling parent), flytträtt (transfer right), nollalternativ (zero alternative), riksintag (national intake), skurkstat (rogue state), and stafettläkare (relay doctor).

All articles where the analyzed words were used were extracted from two big full-text archives: Mediearkivet and Presstext. This meant that their usage contexts could be analyzed in great detail. The Swedish Language Bank was also used, mainly to establish which kinds of word formation patterns the words normally display.

Results

- The meaning was generally very stable for all of the words during the analyzed conventionalization period. No large fluctuations were recorded for any of them. All of the words displayed one dominant, established main sense, except användargrupp (user group), which turned out to be strongly polysemous right from the beginning. For some of the words, there were a few marginal citations with a radically deviant meaning. Flytträtt ('transfer right', where the first part in Swedish means 'to move (something/oneself') conventionally denotes the right to transfer (pension) savings from one bank or insurance company to another. But the word also shows up in the meaning 'right to move wherever one wishes' (for certain groups in society). The presence of such deviant meanings can be interpreted in several ways. Several of the compounds have a broad theoretical meaning potential, i.e. the parts of the compounds can activate several kinds of meaning associations and meaning combinations. The conventionalization of the very word form might then lead the word in itself to become more easily activated when needed, even to express meanings rather different from the conventionalized one. Some of the observed semantic stability could be due to the written sources being used. Perhaps they are not suitable enough to catch the very earliest stages in conventionalization. Some of the words could have been already conventionalized in some more specific speech community, and/or in speech, when they turn up in newspapers for the first time.

- Metacomments (like explanations or statements about usage and the spreading of the words) and metasignals (like citation marks and "so called"-hedges) are striking contextual features for several of the compounds. They also seem relevant to the conventionalization status: The most metacommented and metasignalled words were also the most familiar ones to the informants. Considering that very few words were included in the study, it is difficult to know how general this pattern actually is. It is also difficult to decide between cause and effect - are metasignalling a result of the words being noticed for some independent reasons, or are they especially noticeable due to the metasignalling? On the other hand, some of these new words are used completely without any metamarking or metacomments. Contrary to the metasignalled words, these are treated as being totally unproblematic, in spite the fact that some of them are very new and not very "transparent" at all. One important research question for the future is therefor to analyze these differences in usage between new words treated as problematic and those treated as unproblematic. (This is partly covered in a running project financed from Vetenskapsrådet called "Metacommunication about words", which I am conducting.)

- Meanings are often motivated by various kinds of patterns on various levels, like specific word formation patterns (in compounds, for instance, round figures are often associated with amounts of money, as in bidragsmiljon, 'grant million'), various contextual patterns (for instance which people are using the words, which exemplifications of the concept are abundant, which values are associated with them, which semantic roles are generally specified, which collocations and constructions do they participate in, and to what extent are the compounds used anaphorically). The contributions from these different patterns and factors vary from word to word. It is often hard to decide which are the most dominant patterns for a certain word, but sometimes the connection to a strong pattern can explain why many users renders the meaning rather self-evident, in spite the fact that it apparently is not, when tested on informants who have never heard the word.

This study also led to an in-depth discussion of the very notion of conventionalization, which is less clear cut than often imagined (Svanlund 2009, 227 f). Various types of conventionalization criteria were discussed, where corpus data could be compared to data from informants. One candidate indicator concerns the amount of contextual metacomments and metasignals. Contrary to the original assumption, these metatraits did not decline in frequency during the period studied (with one exception). For some of the words (mainly skurkstat 'rogue state', and stafettläkare 'relay doctor') it seems like the metasignals have rather become part of the standard phraseology; i.e. it is almost conventionalized to introduce these words with a hedge like "so called".

The original intention was to study compound conventionalization also in spoken contexts, which can be said to exemplify more local conventionalization processes within a specific dialogue. This could then be compared to the more global processes reflected in written media like newspapers and magazines. This turned out to be more difficult to study than expected. One problem was getting hold of relevant data from transcribed discourse of reasonable size. Another problem was that in the data that was finally obtained, there were virtually no indisputably novel non-conventionalized compounds. It is sometimes said that the creation of novel compounds in Swedish is a phenomenon primarily connected to written language. This can be said to have been confirmed by this study.

Partial results from the project have been presented at
- the conference Svenskans beskrivning 2005 (Svanlund 2006)
- a seminar at The Department for Scandinavian Languages, Stockholm University 2007
- a seminar at The Department for Linguistics, Stockholm University 2009
- the conference Corpus Linguistics, Liverpool 2009

Publications

Svanlund, Jan 2009. Lexikal etablering. En korpusundersökning av hur nya sammansättningar konventionaliseras och får sin betydelse. Monografi. Acta Universitatis Stockholmiensis. Stockholm Studies in Scandinavian Philology. New Series 52. Vällingby: Elanders
Svanlund, Jan 2006. Hur man etablerar sig som curlingförälder och stafettläkare. I: Svenskans beskrivning 28, s 359-368.