Most organisational theories and learning theories alike represent linear, time-framed conceptions of work succession and coordination. This includes, in my opinion, also some present, vocal representations of action research theories addressing the making of learning systems at work. Taking two of these research traditions as my point of departure - on the one hand the so called Helsinki school of activity theory, among others represented by Engeström's conception of "co-configuration" (cf. Engeström 1998; 2001; 2003; Engeström, Puonti & Seppänen, 2003), and on the other hand the Norwegian action research tradition directed at enterprise development by means of "broad participation" based on social dialogue arrangements (cf. Gustavsen 2001; Pålshaugen 2001; Pålshaugen 2002) - I shall in this paper argue for an alternative model of heterogeneous, interrelated spatiality of learning. I shall call it the spatiality of in-between learning. While trying to elaborate some of the prerogatives of the former, that is, connected to the discursive mediation of work place development as forwarded by the Norwegians and the ecological, task-oriented approach represented by the model of "expansive learning" of Engeström's, I hope to sketch out a learning model that is more complex and less mechanic; infinitely open and persistently close to practice.

The creative power of heterogeneous, in-between learning, I shall advocate, originates in extended differentiations and modest politicisations of working practice, techniques of learning, that is, that are supplementing - replacing whilst recycling - the plurality of current technologies at work.The empirical base for my argument stems from an action research project in the micro-electronic industry in the Eastern region of Norway directed at enhancing work-based learning practice. This project quasi ended when it was only "halfway" in regard of the accomplishment aims of the project progressively attained due to inner and outer intrusions and disturbances of various kinds affecting the working climate all through the process, thus rather making a network constellation of partly connected enterprises than a project organisation of traditional type. This misfortune, though, opened our eyes for the web of partial connections of disparate discourse practices each enterprise consisted in or / and was partaking in. The individual enterprise was no longer to be considered one, a unified whole. To the contrary, the "core" of the project as well as that of each individual enterprise turned out to be nothing but the intersection of various interface connections between the firms and the environment. True to the overall goal of promoting work-based learning the trajectories and characteristics of each firm "decomposing" interferences then tentatively were worked out, with intent so remaking a partial coherent learning system.

In this particular case we found seven distinctive discursive regularities - as distinct from one-to-one causal connections - of significance in interplay or /and collision: a discourse of work-and-learning design or a model discourse of learning, a discourse of industrial logic, a discourse of market logic, a discourse of administrative logic, a discourse of global standardisation, a discourse of social partnership and, eventually, a discourse of domestic affiliations.

In the paper these regularities shall be described more closely.To focus as we do here upon the importance of pointing out discursive regularities in firms re-conceptualise the socio-technical unity of enterprise most often taken for granted - as something undifferentiated, an interdependent "whole", when not split into two, a social dialogue part and an industrial-technical part. Addressing the multiple practices of divergent discursive regularities makes the foundation of a learning system directed at learning rather than that of politics (for the most part tantamount to the making of a "coherent" or "unified" conception of market or organisational strategies of firms), that is a system focussing on the coexistence of connections and distinctions rather than the options of simple variations, adaptations or unifications. In opposition to activity theory there are no simple transparency (or aim) to hold on to and as distinct from action research theories based on social dialogues here the "voices" in play are multiplied onto the whole organisation - so, among others, also transgressing the distinction often made in action research separating the politics of the "developmental organisation" from the causal facts (or "nature") of the "work organisation".

Theoretically, my advocacy for the existence of multiple voices in play or manifold of discursive regularities at work is, broadly speaking, basically inspired by the so called "interpretative school" of organisational studies. Hence, my argument of discursive regularities takes inspiration from the French sociologist, Laurent Thévenot, especially his conception of disparate "pragmatic regimes" (Thévenot, 2001; 2002).

I shall also be drawing on various theories on discursive spaces (e.g. Appandurai, 1990, Knorr Cetina & Preda, 2001) as well as some conceptions made by Barbara Czarniawska of the narrations of (the work) organisations, including her metaphors of organisational "bazaars" (Czarniawska, 1999) and organisational infrastructure "palimpsests" (Czarniawska, 1998).

Finally, in addition shall my argument make use of the relational conceptions built-in in the notion of "boundary objects" offered by Susan Leigh Star (Star & Griesemer, 1989; Star, 1989, Bowker & Star, 2000) as well as that of "co-genetic logic" presented by one of the ancestors in socio-technical theory, Philip Herbst (1989). Going theoretically "up-streams", so to speak, I hope to reactivate a socio-technical tradition still of great importance, I believe, for our understanding and conceptualisation of learning at work.