1We have previously published in the BMS (de Saint Leger and van Meter, 2005) an initial analysis of all the abstracts of presentations given at the first congress in 2004 of the Association Française de Sociologie (AFS). That analysis was done with a method based on a hiarchically-ascending classification of the co-occurence of keywords (“co-word analysis”) called Calliope and developed by one of the authors (de Saint Leger, 1997). That article in the number 85 issue of the BMS was accompanied by analyses of the same corpus of data by three other methods: Trideux factoral correspondence analysis; Alceste hierarchically-decending classification analysis (Boudesseul, 2005); Prospéro text-based analysis of controversies (Trabal, 2005). These four analyses were developed further in a recently-published work, Analyses textuelles en sociologie – Logiciels, méthodes, usages (Demazière et al., 2006).
2In the preceding issue of the BMS, Gérard Boudesseul (2006) presented an initial analysis of the abstracts of the 2006 AFS congress in Bordeaux, again using the Trideux factoral correspondance method and the Alceste hierarchically-decending classification method. In this research note, we report our initial resultats based on the analysis of the same corpus with Calliope.
3In our 2005 and 2006 analyses of the 2004 AFS abstracts, we noted the intertwined relationship between the keywords Travail (Work) and Femme (Women). Women was clearly the most central keyword and the cluster associated with that keyword had more external ties than any other cluster (see the “Strategic Diagram” on page 128 in Demazière et al., 2006). Nonetheless, the key work and its cluster were far from being the most powerful structuring forces in the corpus. This was made evident by re-analyzing the sub-set of all abstracts which included the key work Women and producing a new “Strategic Diagram” of this sub-set after removing the key work Woman (see the “Strategic Diagram” on page 131 in Demazière et al., 2006). Indeed, the keyword Work occupied the same position as Women had previously in the diagram and the structure of the other keywords and their clusters in the four quadrants of that two-dimensional diagram remained very much the same. In short, the key work Women was very frequent and often tied to other keywords, but it was the keyword Work that was largely responsible for structuring the corpus.
4One of our major questions in examining the 2006 AFS abstracts was to see if this complex intertwined relationship involving Women and Work continued to play a major role in structuring this new corpus. This answer is “yes”, but in an even more complex and intertwined relationship that is almost the mirror image of the 2004 AFS abstracts.
5Below is the Strategic Diagram for the 2006 AFS abstracts with its first axis of centrality (number of external ties between clusters) and its second axis of density (number of internal ties within clusters). This, of course, produces the well-known four quadrants of “mainstream science” (the first quadrant which is both central and dense), “ivory-tower science” (the second quadrant which is dense but not central), “unstructured science” (the third quadrant which is neither dense, nor central), and “bandwagon science” (the fourth quadrant which is central but not dense).
Figure 1: Strategic Diagram of all 2006 AFS abstracts
6This diagram can be compared with that of all 2004 AFS abstracts (Demazière et al., 2006: 128) where Women is in the top right corner of quadrant one. Here, for 2006, it’s Work that is in that same position. But besides that substitution, the two Strategic Diagrams look very much alike with many of the same clusters in more-or-less the same position. Indeed, the clusters in both diagrams are drawn out between the dominant cluster (Women or Work) in the top right (with a few other clusters distributed throughout quadrant one) and a concentration of clusters at the bottom and to the left (distributed throughout quadrant three and part of quadrant four). Several of the clusters closest to the dominant cluster are the same in both diagrams: Etablissement scolaire / Enseignement (schools / teaching), Medecine, Immigration, Domaine juridique (legal domain).
7This similarity in structure reflects the similiarty in keywords since there is very little difference between the established thesarus of keywords resulting from the analysis of the 2004 AFS abstracts and the new set of keywords found in the 1,172 abstracts of the 2006 AFS congress. The variations are in the slight differences in terminology (Enseignement [teaching] / Enseignant [teacher]) and a bit more in frequencies of occurrence and co-occurrence of keywords. This also helps to explain how we can still have rather similar results in both cases when Women and Work have replaced eachother. Indeed, the slight variation in keyword frequencies results in more clusters closer to the dominant 2006 cluster (Work) and in or near the first quadrant than in the 2004 Strategic Diagram. Also, this slight variation results in Women becoming part of the cluster Work (in the 2006 case) instead of the inverse (as in the 2004 case).
8Let us look more closely at the 2004 and 2006 clusters formed around Women and Work. Figure 2 below represents the 2006 Work cluster which is the dominant structuring cluster of the 2006 Strategic Diagram. The 2004 Work cluster can be found on page 132.
Figure 2: The Work cluster from the 2006 AFS abstracts
9Both clusters contained ten keywords other than Work. Out of those ten keywords, seven of them were the same in both cases: Market, Organization, Employment, Survey, Business Milieu, Sociology, and Practices (English translations). Moreover, these seven keywords accounted for the vast majority of internal ties in both the 2004 and the 2006 cluster. In short, their structure seems to be solid and stable. But this result can only be considered relative to the fact that the exact framework of the 2004 Work cluster was actually constructted in the sub-set of all 2004 abstracts containing the keyword Women (592 out of a total of 1035, and for which the keyword Women was removed to make the analysis possible). But as we mentioned previously, this vast sub-set was a rather clear reproduction of the general corpus of all 2004 AFS abstracts.
10A quick look at the 2004 and the 2006 Women clusters confirms this continuity of structure. Figure 3 below represents the 2004 Women cluster and Figure 4, the 2006 Women cluster.
Figure 3: The Women cluster for the 2004 AFS abstracts
Figure 4: The Women cluster for the 2006 AFS abstracts
11As with the Work clusters, there is a marked similiarity in the two cases, not only concerning the keywords present, but also the distribution or strength of internal ties. The 2004 Women cluster contains 13 keywords, including Work. The 2006 Woman cluster contains 10 keywords and does not contain Work since this cluster was constructed in the sub-set of all 2006 AFS abstracts containing the key work Work (like the 2004 Work cluster above was constructed in the sub-set of abstracts containing the key work Women and for which the keyword Women was removed to make the analysis possible). Most (12) keywords of the 2004 Women cluster are contained in the 2006 Women cluster. Moreover, the strongest internal ties in both cases are between the same keywords. Thus we have almost the same situation as with the 2004 and 2006 Work clusters.
12The term Women is included in the Work cluster, but we can look at its external ties which are co-occurences of key word ties with terms included in other clusters. These external ties are interpreted as “see also” associations of the term Women.
Table 1: External ties of Women in the 2006 AFS corpus
13The percentage of ties for each cluster indicates the percentage of all external ties for each cluster with the key word Women. One can note that in 2006 Women’s external ties where with the Espace feminin (Women’s milieu). In that cluster, we find the same terms as in the 2004 Women cluster, with one exception; the term Feminisme which according to Calliope is an “emerging term” in 2006 (see “Key-Word Strength of Attratcion é Emerging Terms” below).
14In 2004, it was Work that was included in the cluster Women and Work’s external ties were with the Emploi (employment), Monde de l’entreprise (Business milieu) and Activité professionnelle (Professional activity) clusters. Here again, in Table 2, the percentage for each cluster indicate percentage of all external ties for each cluster with the keyword Women.
Table 2: External ties of Work in the 2006 AFS corpus
15If we consider the 2006 sub-set of 943 abstracts (out of the total of 1,172) associated with the cluster Work and re-analyze that sub-set, we obtain the following Strategic Diagram (Figure 5) where – as in 2004 – the Women cluster dominants in the top right-hand corner of the fist quadrant.
Figure 5: Strategic Diagram of 2006 abstracts associated with the Work cluster
16The similarity between Figure 5 and the general 2004 Strategic Diagram are striking: Women in the same position and a very similar distribution of the other clusters, many of which are grouped together in the lower left-hand of the diagram. This similarity reflects that of the general 2006 Strategic Diagram compared with the Strategic Diagram of the 2004 sub-set of abstracts associated with the keyword Women (Demazière et al., 2006: 132). For these two latter diagrams with the cluster Work in the upper right-hand corner, there is again the group of clusters in the lower left-hand corner, but there are also other clusters closer to Work in the first quadrant.
17Returning to the 2006 sub-set of abstracts associated with the Work cluster and diagramed in Figure 5, since the keyword Work has been removed for the purposes of the analysis, the keyword Monde de l’entreprise (Business milieu) seems to have “filled the gap” left by the absence of Work. See Figure 6 below.
Figure 6: The cluster Monde de l’entreprise (Business milieu) within the Work sub-set of 2006 abstracts
18By comparing the strength of a keyword’s internal ties (within a cluster) and external ties (outside its cluster), in relation to the totality of ties in the corpus, Calliope can calculate the “strength of attraction” for each keyword in forming a cluster. With two corpuses such as the 2004 and 2006 AFS abstracts, Calliope can calculate a “global thematic tie” (see de Saint Leger, 1997) and classify keywords in four categories: stable, emerging, fluctuating and declining.
19With the AFS 2004 and 2006 abstracts, the global thematic tie is 0.924 for a possible maximum of 1.0 (when two corpuses are identical). This means the 2004 and 2006 thematic keywords are very similar, and the 461 keyword enterning into the analysis can be classified with 222 stable terms, 118 emerging terms and 121 declining terms. Out of the 118 emerging terms, the 13 most important (on the basis of the global thematic tie) are Action sociale, Activité physique (Physical activity), Activité de recherche (Research activity), Altermondialisation (Alterglobalization), Apprentissage (Training), Banlieue (Suburbs), Blog, Clivage (Social gaps), Coexistensivité (Social coexistence), Communisme, Féminisme, Mixité social (Gender mixture) and Mondialisation (Globalization). Figure 7 below shows the increase in the “strength of attraction” of these 13 emerging keywords.
Figure 7: Increase in the “strength of attraction” of 13 emerging keywords
20Eight of the 13 emerging keywords figure in the following eight clusters.
21One should note that most of these clusters are “stars” formations; that is clusters with ties only to the central keyword and few or no ties between other keywords in the cluster. Such “stars” are rather unstable structures since the disappearance of the central keyword means there is no tie structure at all. With these clusters, it would be more appropirate to speak of “bubbling activity” instead of emerging stable structures.
22In conclusion, we can note the stable but complex and intertwined role of Work and Women in structuring both the 2004 and the 2006 corpuses. The two corpuses are very similar in relation to their set of keywords and also in relation to the ties between these keywords, although there are is movement as we noted with certain emerging terms. These results are rather similar to those found by Gérard Boudesseul (2006) in his preliminary analysis of the 2006 AFS abstracts.