The Lexical Semantics of Agency (Part II)12 min read

In a previous post I argued that the reasons why the concept of agency in sociological theory is “curiously abstract” has its roots in the ways theorists conceptualize the notion in particular usage episodes during theoretical argumentation. Particularly, conceptualizing agency as a substance (a “mass noun” like water or heat) continuously distributed in time leads to predictable problems of non-specificity and a lack of direct grounding in experience-near domains.

Yet, I also noted that some theorists, particularly Emirbayer & Mische (1998, p. 963ff; hereafter E&M), do not offer a unitary conceptualization of the notion of agency. Instead, they provide a cluster of distinct (and not necessarily compatible) conceptualizations, some of which are more “curiously abstract” than others. I noted that E&M provide at least three such conceptions: (1) agency as a process distributed in time, (2) agency as a quality or dimension of action, and (3) agency as a force or capacity possessed by persons. Specifically, I noted that the conception of agency as a process inherently embedded in time links up clearly to Giddens’s (1979) earlier definition of agency. This is by far, the most unbearably abstract of all the conceptions, and unfortunately for some, the one that has gone on to be most influential in terms of (usually ritualistic or non-substantive) citations by sociologists (e.g., Hitlin & Elder, 2007).

Note that because the three conceptions of agency are not semantically equivalent, the concept of agency is polysemous in the linguistic sense. As such, it is useful to distinguish them typographically. So in this and the following posts, the process conception will be referred to as agency[p], the dimension conception as agency[d], and the capacity conception as agency[c]. In this post, I continue the examination of agency[p] the most curiously abstract of the concepts. Future posts will provide my take on the lexical semantics of agency[d] and agency[c].


One of the conclusions we reached in the previous treatment is that agency[p] is indeed a “curiously” abstract conceptualization. However, this does not mean that it is semantically empty. Instead, from the perspective of lexical semantics,  agency[p] is an abstract noun, and as such, it is likely to be semantically vague or underspecified (see Goddard & Wierzbicka (2013, p. 229ff). Here, I explicate the semantic content of agency[p] using the Natural Semantic Metalanguage or NSM (Wierzbicka, 2015).

The basic idea of the NSM approach is to “force” the analyst to lay out (via “reductive paraphrase”) the basic semantic content of linguistically indexed categories using only words (called “exponents” in each natural language) derived from a list of sixty or so basic concepts (called “semantic primes”) that have explicitly lexicalized analogs in most of the world’s languages. These basic concepts thus come the closest to being “semantic primitives” or the semantic building blocks of more complex concepts, like agency, which, in standard practice in social theory tend to be “defined” in terms of other even more abstract or complex concepts of equally elusive semantic status (for the latest English list of NSM semantic primes, see here). More specifically, I follow the general approach to explicating abstract nouns discussed in Goddard & Wierzbicka (2013, p. 205ff).

The reductive paraphrase of the process model of agency goes as follows:


  1. Something
  2. People can say what this something is with the word agency
  3. Someone can say something about something with this word when someone thinks like this:
    1. Someone can do something if that someone’s body moves in a way for some time in a place
    2. Someone can think: I can do something if I move my body in a way for some time in this place
    3. Because this someone did this something, something happened in this place at this time
    4. Because this someone did this something, this place is not the same as before
    5. Before, in this same place, the same someone can do other things if that someone’s body moves in another way
    6. Before, in this same place, the same someone can think: I can do something if I do not move

This explication carries the basic message of Giddens’s (1979, p. 55-56) definition and commentary, which I quote in full for comparison with the NSM rendering:

Action or agency, as I use it, thus does not refer to a series of discrete acts combined together [sic] but to a continuous flow of conductinvolving a ‘stream of actual or contemplated causal interventions of corporeal beings in the ongoing process of events-in-the world’the notion…has reference to the activities of an agent…The concept of agency as I advocate it here…[involves] ‘intervention’ in a potentially malleable object-worldit is a necessary feature of action that, at any point in time, the agent ‘could have acted otherwise’: either positively in terms of attempted intervention in the process of ‘events in the world’, or negatively in terms forebearance.

Parts 1 and 2 of the explication are standard for abstract nouns, noting that they refer to an unspecified “something” and that some linguistic community has adopted a lexical form (a word) to refer to this. The other parts of the explication are coded to correspond to the parts of Giddens’s discussion marked in the same color. Part 3 of the explication provides the main conceptual content of the “something” the term agency[p] refers to.

In 3A and 3B (pink) the basic definition of agency entailed in agency[p] is provided. The difference is that in 3A we have the case in which people actually do something, and in 3B we have the case in which people “contemplate” doing it (hence the preface, “people can think” in 3B). Here, it is specified that agency necessarily involves someone doing something by moving their bodies in specific ways for some time at particular places. Thus, Giddens’s agency frame of reference action is a six-way place-holder, involving necessary reference to a person (someone), performing an action (doing), by moving their bodies in particular ways, at specified places for some time.  The presence of an actor or “agent” in Giddens’s parlance (someone in NSM semantic prime terminology) makes sense, as the presence of an actor or agent differentiates action or agency theories from other “agentless” approaches (e.g., Luhmannian systems theory). The reference to actions (“doings”) is also de rigueur. One may be surprised to the further specification that doings are performed by people moving their bodies. Although sometimes analysts speak of “action” as if it is done by disembodied agents, bodies enter the picture via Giddens’s reference to “corporeal beings”; that action is embodied, is, of course, a truism in the tradition of practice theory from which both Giddens and E&M depart.

Finally, the reference to a “way” of moving the body serves to individuate the action as a particular time of doing, which is consistent with the way that skilled activities are generally conceptualized (e.g., bike riding as a way to ride a bike; see Stanley (2011)).  Finally, the schema specifies that the doings are happening at specific places during some strip of time. This is in keeping with Giddens’s emphasis on the fact that no discussion of agency makes sense unless human activity is embedded in “time-space intersections…essentially involved in all social existence” (1979, p. 54, italics in the original). Note that, while all the other terms capture particulars, the temporal reference is left purposefully vague since Giddens conceived of the strip of time within which agency unfolds as unbounded (e.g., continuous and not punctual and thus lacking definite starting and ending times like a “discrete act” would).

The explication in 3C and 3D (blue) clarifies Giddens’s idea that agency necessarily results in consequences or “interventions” into the causal flow of events in the world. Note that an advantage of the NSM approach is that we can lay this claim out using relatively simple notions, namely, people doing things and stuff happening in places as a result; Giddens, on the other hand, has to rely on conceptually complex ideas such as causalintervention, and malleable object-world to convey the same thing. Because these terms are themselves semantically complex and elliptical, they introduce a level of obscurity that is shed in the NSM reductive paraphrase. The paraphrase covers two minimal conditions for an action to make a difference in the causal flow in the world. First, action must itself result in some kind of event in the world (3C); a “happening” that wouldn’t have existed if not for the action. Second (3D), this event itself must leave a mark on the world (indexed by “this place” and time); minimally the counterfactual is that the world is now permanently different because this event occurred. So because the agent acted, the world is no longer the same as it was before the action.

Finally, 3E and 3F (red) convey the basic idea that a necessary component of the Giddensian definition of agency[p] contains two counterfactual references to something that could be rendered in more metaphysically loaded terms as “free will.” First, the fact that what people did is not the only thing they could have done (3E). They could have done other things had they moved their bodies differently and thus enacted an entirely different set of causal interventions into events in the world at a previous point in time (“before”). Second, in 3F we see that the person could have also done nothing, which implies the capacity to not move their bodies (if they wanted to) is itself an instance of the general category of agency[p]. Thus, refraining from action is also an exercise of agency on the part of the actor.

So, there you have it. Giddens’s curiously abstract concept of agency is indeed curiously abstract. However, like most abstract nouns, it is not conceptually empty. Instead, it encodes numerous substantive intuitions (and when not subject to explicit consideration, dogmatic assumptions) about the nature of human action. Some are intuitive (people act when they move their bodies at particular times and places). Others involve somewhat strong metaphysical presumptions (people always in all times and places can act otherwise). Others involve elements of the definition that are seldom noted or explicitly considered (e.g., not acting is a type of agency).


Emirbayer, M., & Mische, A. (1998). What is agency? The American Journal of Sociology, 103(4), 962–1023.

Giddens, A. (1979). Central Problems in Social Theory: Action, Structure, and Contradiction in Social Analysis. University of California Press.

Goddard, C., & Wierzbicka, A. (2013). Words and Meanings: Lexical Semantics Across Domains, Languages, and Cultures. OUP Oxford.

Hitlin, S., & Elder, G. H., Jr. (2007). Time, Self, and the Curiously Abstract Concept of Agency. Sociological Theory, 25(2), 170–191.

Stanley, J. (2011). Know How. OUP Oxford.

Wierzbicka, A. (2015). Natural Semantic Metalanguage. In The International Encyclopedia of Language and Social Interaction (pp. 1–17). Wiley.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to Top
%d bloggers like this: