Concepts and Definitions

The questions that Lipset and Rokkan raised in 1967 about the nature of political divisions and their origins have produced some of the best comparative politics scholarship of the last four decades. Scholarly interest in the socio-cultural underpinnings of party politics has even increased over time, but the field has fragmented into many separate and often isolated debates. As a result, despite the conceptual innovations and refined measurement techniques, we still have little systematic understanding of how cleavages emerge, are maintained, or how they collapse. Indeed there remains little consensus even on the usage of basic terms.

Despite a few commonly cited conceptual frameworks—Bartolini and Mair (1990) and Knutsen and Scarbrough (1995)—scholars still differ quite dramatically in their understandings of “cleavage.”  This workshop seeks to find common ground for understanding 1) what combinations of socio-structural, attitudinal, organizational and behavioural elements can be defined as cleavage and 2) how the list of cleavage categories derived from the tradition of Lipset and Rokkan must be altered to apply to cases outside of Western Europe and to fit the new employment and communication structures of industrial democracies.

Debate: "Concepts that Hinder Understanding"

In the summer 2009 newsletter of the Organized Section in Comparative Politics of the American Political Science Association, 2007 Helsinki workshop participant Oddbjorn Knutsen offered his thoughts on how to rescue the troubled concept of "Cleavage."

Debate: Conceptualizing Cleavage

Heather Stoll (HS):Reconciling our basic definitional typologies (note that I use “cleavage” and “division” synonymously throughout) [Kevin Deegan-Krause (KDK):Here one of the few sharp differences we have, but not one that I want to deal with until I’ve talked about some of the other smaller but substantively more significant differences.  With the concepts will come the names] [Heather Stoll HS:Fair enough.  I continue using division/cleavage interchangeably for the moment, in fact often writing them together as I just did, with the understanding that we’ll settle the terminology after we’ve nailed down the conceptual framework.]:

• HS:Kevin’s “structural difference”, a division with what he calls “structure” but absent both “attitudes” and “institutions” corresponds to my “latent” or “exogenous” cleavage.  These are nascent fault lines within society that may or may not be developed by political entrepreneurs into groups with a sense of “groupness” (“attitudes”) and may or may not ultimately give rise to (or, perhaps more accurately, be supplied with) political parties etc. (“institutions”).
  • HS:These nascent (or latent) fault lines may be ascriptive sociological traits such as skin color, gender, and language.  [KDK:Yes, I think the important thing about these is that they are difficult to change.  I say this though I fully agree that an individual’s self-understanding may be greater or smaller, and that the sense of groupness can vary wildly] In my work, I have pushed hard for also including here objective characteristics of the society that do not relate to such individual traits, although these are more difficult to pin down.  For example, I have been very concerned with what I call (following Lijphart 1999 etc.) a foreign policy cleavage, divisions amongst the citizenry about a country’s place in the world etc.  [KDK:I can understand this but I wonder if it is the same kind of thing.  It is in the sense of “enduring dividing line” but not necessarily in the sense of involuntary positioning] [HS:I will grant that there is less prima facie involuntary positioning here.  However, borrowing from the constructivists, even the ascriptive sociological traits that we view as relatively immutable are, in fact, mutable.  For example, we have a fair amount of evidence that the specific ethnic group with which individuals identify (as well as the strength of that identification) varies as circumstances such as political institutions, the party system, etc. change:  i.e. that ethnicity, like other aspects of demography, is endogenous, and that positioning regarding ascriptive traits like these is often chosen.  Without pushing things too far, the constructivist position does incline me to think that both are equivalent “nascent fault lines”, the exact positioning of individuals around which depends upon a variety of factors, even if the positioning choice set is more constrained when we’re dealing with ascriptive traits.]  In this light, characteristics of the country such as its past involvement in international organizations, wars, trade disputes, etc. might provide comparable fault lines to, say, race for political entrepreneurs to capitalize upon [KDK:This is excellent, though I would suggest that there is a major difference between this and, say, language to the extent that while the dividing line may be in some ways exogenous, an individual’s position on this dividing line may not be, or if it is, it is usually tied either to one of the ascriptive traits defined above (region or language or socioeconomic position) or to some kind of underlying group/network that is not easily defined in economic terms.  [HS:I basically agree with some caveats---see above.]  Interestingly, this is the kind of division that I say is ultimately responsible for differences between Slovakia and the Czech Republic—because of their place in the Austro-Hungarian Empire and Czechoslovakia, Slovaks had a history of being divided about relationships with Czechs and other foreigners whereas Czechs did not.  That longstanding (exogenous) division created differences within the Slovak population that (since it was not rooted in the structure of ethnicity, and the side chosen by any particular individual was undetermined [and, by the demographic polling data, it literally was undetermined]) I called an issue divide]  Interesting that you are dealing with the same thing but seeing it as something more akin to a full cleavage.  This suggests—and I think I will say this later as well—that what we are really both onto is a question of depth or persistence and that what seems to have happened is that in the absence of easy ways to talk about depth or endurance of a particular difference, we’ve used sociostructural anchors as surrogates.  The problem is that not all enduring cleavages have their basis in the sociostructural characteristics measured in polls, and furthermore, that is not the only enduring type of division since there are others in which the dividing line is enduring but not necessarily any individual’s membership on a particular side.  [HS:Very interesting.  Yes, I’m viewing these non-sociological structural factors as equivalent “anchors” for divisions/cleavages.  However, I admit to the possibility that the cleavages anchored in ascriptive, sociological traits might be more persistent/enduring.  By accepting both types of anchors as possibilities, though, we leave the matter of relative persistence/endurance to hypothesis, and hence subject to empirical investigation, instead of resolving the matter by definitional fiat.  So… I agree that both of us are interested in persistence/depth/endurance of divisions, and that this is a useful lens through which to view the structural aspect of cleavages.  But for me it’s also about exogeneity:  about being able to predict what politics will look like, based upon the raw materials with which political entrepreneurs work, as discussed below. If you want to predict when foreign policy-type issues will appear on the political agenda, it then makes sense to me to look for these kind of structural anchors such as wars, trade disputes, etc.]
  • HS:This type of cleavage is very important in the literature that I know best, the party systems literature, because it is by definition exogenous to both political competition and political institutions.  That is, if we want to explain the number of parties at least partially in terms of the number/type of divisions within society, and partially in terms of political institutions, we need those divisions to be exogenous to both the dependent variable of political competition and other independent variables such as political institutions.  And once what Kevin calls “attitudes” and “institutions” become involved, it’s hard to argue that any such division is truly exogenous. [KDK:I now see where you’re coming from on here.  In part we differ on this to the extent that I am not as interested in explaining numbers of parties and don’t need to clear the attitudes/institutions ground in order to make those comparisons.  But from different directions and for different reasons we’ve identified a phenomenon: an enduring conflict not located in basic structural categories]  [HS:Agreed!  Although, again, I also admit to the possibility that both types of divisions/cleavages could be short-lived, and (as discussed above) find the issue of relative persistence/endurance to be an interesting research question.]
• HS:It becomes a little messier from here on out, though.  The remaining two types of cleavages that I identify, which I call “political” and “particized”, correspond to divisions characterized by Kevin’s “structure” combined with both his “attitudes” and “institutions”.  The reason why my two types of cleavages both map onto one of his types (what he calls a “full cleavage”) has to do with our understanding of the institutional component, which differs significantly.  I differentiate between political parties and all other (non-party) types of institutions, such as churches, bowling leagues, interest groups, etc.  Kevin rolls all of these institutions together into one category.  Hence, my “political” cleavage corresponds to divisions that possess his “structure”, “attitudes”, and non-party “institutions” attributes, whereas my “particized” cleavage corresponds to divisions that possess his “structure”, “attitudes”, non-party and party “institutions” attributes.  [KDK:So far so good.  Since particized requires “non party and party” attributes,

  it would not be too difficult to map on to my own framework, but I have a feeling that it would not be very useful to do so and that there is another framework that I feel (but cannot yet articulate) that is emerging from this.  One question, though: do you identify cleavages that are divided among party but not non-party? Or must a particized cleavage include the non-party institutions]

   [HS:I previously always viewed the progress as linear:  that first a politicized cleavage is developed, and from this a particized one might emerge.  Our conversation has made me realize that this might be more restrictive than I would really like the typology to be, as I discussed below:  that we should admit to the possibility of party institutions sans non-party ones (i.e., that particized may or may not go with politicized).  However, as discussed below, this seems to me to be an undemocratic situation, i.e. a failure of political representation that should not endure (i.e., not represent an equilibrium outcome) if the democracy is a healthy one.  But that doesn’t mean that it can’t exist, and moreover it may be the case that representational failures are more common than I’d like to think.  So I’m willing to open it up here, as discussed below.]
  • HS:I believe that distinguishing between non-party institutional/organizational attributes of a division, and the division in some way having formal political representation (either through a “single issue” party dedicated solely to pursuing the interests of groups formed along the division, or through “multi-issue” parties pursuing those interests along with others) is important for a simple reason:  only politicians can make public policy.  (Excepting, of course, countries where ballot measures play a prominent role in policy-making, but let’s sweep this under the rug for the time being.)  Hence, if we want to either make normative assessments of political representation or track the impact of divisions upon public policy, the distinction seems critical.  (For example, perhaps even though the division is “particized”, i.e., takes on party “institutional” attributes, public policy is not altered due to the country’s political institutional features, etc.  But we might draw different conclusions about political representation in this situation than in the situation where the division is never particized at all.)  [KDK:Here I’d like to get a few more specific details.  Can you talk about some examples or give a more systematic sense of the kinds of issue/country units in which you find politicization but not particization?  Do these have long duration?  I accept this as a potentially important category, but as we’ve discussed in the past, it seems that most issues become particized soon after they are politicized, perhaps because there are so many parties that nearly all issues get taken up by somebody sometime.] [HS:This may indeed eventually happen, but I think that there are notable cases where a set of issues is kept off of the political agenda (i.e., non-particized) for a significant period of time.  I would argue that the ethnic cleavage between Jews of Ashkenazic and Sephardic origins in Israel was one example (although it has now been particized); it also seems to me that the black-white ethnic/racial cleavage in the US is another (off of the agenda for decades until the civil rights movement after a blip of being on the federal agenda around the time of the civil war; also in its earlier form---slavery---it was kept off of the federal agenda for the US’s first fifty-sixty years or so by confining it to the state level---I’m stealing from Riker here, of course).  I’m very interested in these “gaps” from a representational perspective.]
• HS:Accordingly, I’m sympathetic to the idea of identifying different attributes that cleavages might have, such as a structural basis (“structure”); a sense of groupness (“attitudes”);  [KDK:these two things are actually not quite the same and that the distinction you make is one that points to the advantage of your schema.  I have not built groupness into my model, but the more I think about it, the more the self-conscious collective identification is an important part, at least to some kinds of cleavage.  I am wondering if at least at first it would not be profitable to start to look at various characteristics of hypothesized cleavages (for the moment not worrying about what to call them, though that will be an issue later) to see what’s there and what isn’t]  [HS:Ah.  I guess that I was interpreting your attitudinal attribute as groupness plus, but maybe it makes sense to break this attribute down into groupness on the one hand and attitudes/values on the other.  By looking at various characteristics, do you mean to identify a list of potential attributes (such as groupness, attitudes, etc.)?  If so, that sounds like a fruitful way to proceed.]  and an institutional component (“institutions”).  However, I would like to break down the latter to distinguish between non-party and party institutional attributes, where the former give self-aware groups the capacity for collective action and the latter is as defined above.
• HS:Moreover, I’m sympathetic to then combining the different attributes to define different types of cleavages, with fewer attributes corresponding to more general, less developed divisions and more attributes corresponding to less general (more restrictive), more developed divisions.  This yields my “latent”, “politicized”, and “particized” typology, and Kevin’s “difference”, “divide”, and “full” typology.  However, whereas my typology maps particular attributes to each step of the ladder of generality, Kevin’s allows any combination of attributes, with the number of attributes alone responsible for movement up the ladder.  My ladder is thus a little less “general” than his:  I have in mind both the generality/restrictiveness of the concept, as well as the relationship of the concept to the political process, with moving up the ladder being associated with the cleavage becoming more politically engaged.  Effectively, I rule out some of the combinations that he admits.  Further discussion follows. 
  • HS:First, with respect to the division with a single attribute, the “difference”, does it make sense to speak of divisions that are “attitudinal” without being “structural”, “institutional” without being “attitudinal”, etc.?  I can see how the latter might be possible, but I have a harder time envisioning the former.  Here’s why.  (Note that it makes perfect sense to me to speak of a division that is “structural” but neither “attitudinal” nor “institutional”:  this is my “latent” cleavage discussed above.) 
    • HS:With respect to structures and attitudes:  whether or not attitudes can exist without a structural basis turns on our understanding of the structural component.  If we define structure to consist solely of ascriptive, sociological traits, they certainly can.  This effectively speaks to the distinction drawn by Allardt and Pesonen (1967) between structural and non-structural cleavages, and allows us to explore whether or not structural cleavages (“attitudes” plus “structure) are indeed more persistent than non-structural ones (“attitudes” alone), etc.  However, if we adopt the more expansive definition of structure that I have pushed for (as discussed above), then it doesn’t make as much sense, although it still might be possible.  There is something very intuitive about forcing a connection between the two:  grounding our attitudes in hard facts about ourselves and our society suggests a la the sub-altern literature that they are “authentic”, not just elite creations.  However, what’s unattractive about forcing such a connection is that structure is still a very slippery concept, and it seems to overly reify facets of identity that are ultimately socially constructed (I think that constructionists are by and large correct about this, particularly in the long run).  In sum, this strikes me as a very tricky issue turning on the definition of a very difficult concept.  And I don’t know what the answer is.  [Nor do I, but I think it is helpful to identify two different types or aspects of structure, one that involves enduring individual aspects and the other which involves enduring tension but in which individuals may or may not be fixed.  This one, as you note, is harder to operationalize, but the simplicity of the “structured individual” approach is actually harder to define than we might want to admit because of the degree to which identities are somewhat fluid and to which categories themselves are fluid.]  [HS:Agree---see my comments above.  I should have read all the way down, I guess, before beginning commenting; here you anticipate my earlier comments, rendering them superfluous…  ;) ]
    • HS:With respect to non-party institutions and attitudes:  it might be the case that political entrepreneurs first develop non-party institutions, such as churches or interest groups, which they then use to create “groupness” (i.e., “attitudes”).  So we might at a given slice of time see a division with an institutional but not an attitudinal attribute.  However, this should not be a stable, i.e. equilibrium, situation.  From the institutions, the attitudes should follow, or the institutions should collapse (barring really dedicated entrepreneurs who delight in lost causes).   Note that I view the possibility of this kind of disjuncture to be so small, and for the most part uninteresting vis-à-vis the larger phenomena I’m interested in exploring, that I effectively group “attitudes” and “non-party institutions” together; they both give rise to self-aware groups capable of collective action.  Others may find the possibility of disjuncture here more interesting than I do, though, so it may be beneficial to not foreclose the possibility by definition, as I currently have done.  [KDK:I agree with you about the possibility of disjunction.  I do tend to see a possibility, however, of a stable set of institutions representing a stable structural group without much of an attitudinal basis, or, perhaps better conceptualized, with an attitudinal basis that does not extend much beyond the rather basic attitude that “we should support our group.”  I think here of African American political leaders in many US cities whose leadership has more to do with identity than with a variety of other politically salient characteristics (cultural values, even economic values).  The point of this is not to disagree but perhaps to talk about the degree to which members of a group and their leaders share opinions with each other and the extent/range of opinions that are shared within the group.  Again this pushes us out of the range of dichotomy and nomenclature and into the realm of ranges and scales, but perhaps we can do that.] [HS:I think that we’re on the same page here:  that there’s a useful distinction to be made between “groupness” and “attitudes/values”, where there will often, but not always, be a reasonably close connection between the two.  I also think that the kind of structure minus attitude (beyond basic “groupness”) situation that you describe might be particularly likely in less democratically consolidated places, where elections are often referenda on whose group is the largest---i.e., not ideological.  I need to give this further thought, though.  But for the advanced industrial democracies that much of the literature deals with, it strikes me that usually the twain shall meet.]
    • HS:With respect to party institutions and attitudes:  similarly, politicians (elites) may be divided (hence the division having a party institutional attribute), even if there are no corresponding attitudinal and/or non-party institutional divisions within society at large.  Again, however, if democracy/representation is truly working as we believe it should be working, this should not be sustainable in the long run.  Now, it may be the case that democracy often doesn’t work as we expect it to, so I concede that it should be helpful to allow for this disjuncture; I actually have had exploring just this kind of disjuncture in mind from the beginning, but I think that my definitional typology somewhat unintentionally forecloses the possibility.  So I’d like to open it up a bit here. [KDK:I think you are right, though, that this is not a sustainable equilibrium position.  It is useful only as a part of the broader question.  The whole notion of a “difference” is really primarily useful as a building block of  the overall structure.  This conversation is reinforcing my second thoughts about whether these are the only building blocks or whether I am interpreting and measuring them in the most useful way. 

   

  • HS:Second and related, regarding the two-attribute “divide”, can we have both “structures and institutions” without “attitudes”?  I find the combination problematic for the reasons laid out brilliantly by Achen in his article “Breaking the Iron Triangle of Voting Research”.  As he argues, demographics (“structure”) cannot be a causal factor of voting behavior etc.; causality only enters when “attitudes” are built on top of the demographics.  It is for this reason that I find much political behavior research uninteresting, which I know sets me apart from a great many political scientists.  (Let me reiterate that I know that it’s me who’s the outlier here.)  I just don’t see how Horowitz’s “census” elections cannot, in fact, have an attitudinal basis:  there must be a sense of “groupness” there, an “us vs. them” mentality, which to me is at the core of the “attitudinal” attribute, even if it doesn’t translate into different opinions about various issues.  As discussed above, it’s possible that this combination could exist in the short run, by both understandings of institutions, but I don’t view it as a long run, equilibrium situation. [KDK:I share your hesitation on this and it is something that I have wrestled with in my own work.  Yet I also believe that there is something different about the group loyalty of a census election than there is in an issue/ideology election like those of Western Europe.  I wonder if there is a way of looking at the “number of questions” on which partisans agree (as Kitschelt et al do) or in degree of homogeneity on a wide range of questions.  Of course all this might mean is that ethnic attitudes are the only ones that are salient (overwhelmingly so) and so we are back to full cleavage since the political division has both structural and attitudinal components.  [HS:Could you elaborate on the Kitschelt et al. etc. methods and what this will get us? I’m not quite sure that I know where you’re going here.]

     

    • HS:In summary, then, I’d like to see us move somewhere between our two existing typologies.  Mine can be loosened up a bit to allow for other combinations of attributes, as described above, but I’d like to at least discuss how certain of the combinations that your “all possible” approach admits are unlikely to occur.[KDK:Point absolutely taken.  More about this below. 

       

      I also introduce three additional definitional parameters in my own work:  the persistence of the division or cleavage (restricting the analysis to. long term divisions, or also including short term divisions; I have taken the latter approach, contra much of the literature); [KDK:Here I /absolutely/ agree.  Just because a division is short term does not mean that it is insignificant.  There are more voices supporting this position of late, but it still lacks a strong and oft-cited chamption] whether attention is confined to divisions arising from sociological traits or not (as discussed above); and whether the subject matter is divisions that are ideological in nature, or divisions that are issues.  My work has focused solely on ideological divisions, but many others are interested in issue divisions. [KDK:Here I have tended to conflate the two, but mainly by devaluing issues or more precisely by elevating it to a rather high level of generality.  In fact what I call an “issue divide” (as with Lijphart and his “issue dimension” is almost always the combination of multiple specific issues]  [HS:I thought that we were actually on the same page here, too, but I wasn’t quite sure.  But now I know that we are!  This is probably worth working into the schema somewhere, though, since there are those who are interested in the less general, single issues, and this is another area where people can easily talk past each other---particularly when a term involving “issue” is used for something multi-issue, i.e. ideological.  But again, terminology can fall out later.]

       

    [KDK:In an effort at summary, I really appreciate your perspectives on this.  They fundamentally challenge those assumptions on which I have always felt most uneasy (and they also challenge some assumptions on which I’ve been more confident, but perhaps not rightly so).  Our discussion seems to be pointing to is a potentially useful disaggregation of aspects—more than just the three that I use—accompanied by an exploration of their interrelationships that could begin at a theoretical/definitional level but that must ultimately be rooted in empirical tests. ]  [HS:I’m enjoying it too!  And realizing, indeed, some unfortunate/unanticipated consequences of my typology, as discussed above.  So this is all very good.]]

    I want at this point to do some diagram like the tripod that I use in the book and the chapter, but I’m feeling a bit too uncertain to do that in any meaningful way.  It is possible, though, to identify certain elements that need some attention]   [HS:I agree that this would be good to aim for.  Let’s work on this after doing a little more thinking; I need some time to fully digest everything that we’ve discussed.  Perhaps one of us can---after a little more digestion---generate a draft, and we can then take it from there.]

    I think that scholarship in this field is fairly solid at what I call the “issue divide” level,  where it’s possible to figure out what “issues” and “attitudes” politics is really about, but even here I think that scholarship has not gone far enough in coordinating all of the tools at our disposal:  the standard relationship between party and voter opinion (most simply done with polls and crosstabs but now enhanced by factor and discriminant analysis), but also with elite interviews, manifesto and media statement analysis, voting patterns in parliaments).  It would be nice to create multifaceted pictures of cleavage according to these multiple indicators for particular cases.  At a certain point, we could then begin to talk intelligently about what particular indicators are more likely to see than others.  [HS:I agree to a large extent, but worry that much of this knowledge is case-specific.  At least, it seemed to me when I was working on my dissertation that there’s very little truly large-scale cross-national work to this effect.  Of the cross-national work that I could find, most was relatively limited in scope (to, say, ten EU countries).  So I think that there is a lot of work still to be done, either in gathering cross-national data or in performing some kind of meta-analysis to combine all of the individual, small-N studies’ findings.  Also, I fear that a good deal of the literature falls into the Achen “Iron Triangle” trap of assuming that structure implies attitudes, which makes me uncomfortable; I think that there’s more work to be done on the attitudinal side to flesh the picture out.  Keep in mind, though, that the last is my admittedly extremist views shining through! ]

    When we move away from this relatively simple dyad (though as the previous paragraph suggests, the simplicity is only relative), things tend to get a bit fuzzier.  I the notion of politicization with particization is useful but I have a hard time knowing how to how to quantify it (though I know you’ve done work on that).  [HS:For me, quantifying particization involves identifying what parties/elites think politics is about, which can be done using multiple methods, from analyzing manifestos to surveying elites.  As mentioned above, I’ve always effectively assumed that where there’s particization, there’s also politicization (the whole linear progression discussed above).  But our conversation has made me see that this might not be the case.  Leaving this aside, though, for me, a particular cleavage is particized if parties care about the given set of issues, so I think it’s actually not so bad to measure.]  Similarly, I see difficulties dealing with the question of “groupness” both in terms of operationalization and relationship to the allied but distinct notions of “attitudes” and “structures” (for which there is no subjective standard of mutual recognition).  [HS:Agreed.]  Finally, there is the tension between common notions of structure and your extremely useful notion of enduring-but-non-ascriptive structural conflicts which you have done a better job than anyone else at operationalizing.

    And in the process of resolving these kinds of questions, we probably need to be conscious of the question of nomenclature.  [HS:Agreed here, too.  But I think that your initial point that it makes sense to deal with the terminology last makes a lot of sense.]  There are strong partisans of a very high standard (high intention, low extension) for “cleavage.” I think this definition has sometimes diverted scholarship into looking for (or proving the existence of) this rare bird, and yet it also is defensible for the degree of clarity that it brings.  I’m afraid, however, that in my own attempts to respect the clarity, I’ve had to introduce a variety of new terms that are , perhaps, unnecessary and simply confuse conversation even more.

      

    Zsolt Enyedi:

    Let me comment Heather’s and Kevin’s (fascinating) discussion on four points: latency, endurance, particization and groupness. I feel that inserting my comments into the discussion would disturb the dialogue, so I rather copy them here:

      

    Latency

    I agree that non-structural differences may form the basis for enduring political divides. But if virtually anything can be politicized, and if we don’t have a technique to set apart those factors that will engender such divides from those that won’t, I am not sure how useful the concept of ‘latent cleavage’ remains. We can, of course, point out post hoc what was the raw material of the cleavages, but that doesn’t help us in predicting. Obviously, history has produced only a limited number of deep political divides and therefore we can come up with a manageable list of factors that tend to play a divisive role. But I would be inclined of categorizing these factors as “sources of cleavages” and not cleavages with adjective (i.e. latent cleavages), because they have no structuring/dividing role prior to politicization. 

     

    Endurance

     

     

    It was a bit confusing that in some parts of the discussion both Heather and Kevin emphasize endurance while in other parts both allow for short-term divisions. I think endurance should be kept as part of the definition because otherwise the concept will loose its status as the opposite of ephemeral differences/conflicts. I wouldn’t be rigid at this point, though: if a conflict polarizes the political system for e.g. 4-5 years, especially if this happens in a new democracy, I could live with calling it a cleavage.  

     

      

    Particization

    I fully agree with the idea of separating “particized” and “non-particicized” cleavages because salient divides may have no partisan channels and because you may have divides that have virtually only partisan manifestations. More integrated the various sub-systems of the society are, less likely it is to obtain these patterns. But recently the tendency is rather towards disintegration, and therefore it becomes increasingly relevant to monitor these developments. (They may even form equilibrium situation, like it is implied in the ‘cartel thesis’!) The non-particized cleavages may be nevertheless expressed in party politics via sub-sections of certain parties.

     

    Groups

    I also agree that attitudes and group-awareness should be separated from each other, simply because they do not necessarily move together. Bartolini and Mair and many others regard identity or self-awareness as important component of the definition, but since they allow all components to vary in strength, there is no reason to assume crystallized identities with sharp borders behind every single cleavage. At the same time I would be inclined to insert some sort of  “group-building project” into the definition. Without some, more or less consciously designed, mechanisms aimed at group-creation (common symbols, language, identification of the enemy, etc.) one is simply left with abstract dimensions of political space.