Professional Documents
Culture Documents
Roar Høstaker
The aim of this article is to study the relationship between Bruno Latour’s theories
and semiotics. In particular the article compares Latour’s concepts to those of the
linguist A.J. Greimas. From Latour’s earliest texts in science studies onwards, semiot-
ics has been a basic theoretical tool. As the article will show, Latour privileges the
autonomy of language in order to avoid the ascription of substance to human and
non-human actors. It is within this autonomous field that his general associology
based on trials of strength can come into play. Furthermore, the article analyses
Latour’s theories concerning the gradual emergence of actors, circulation of refer-
ences and technical mediation. Finally, the article tries to show how Latour’s approach
reaches a limit when it comes to the study of the settings of social action. A way out
of this problem is sketched while at the same time remaining within a semiotic uni-
verse.
It is well known that Bruno Latour’s an- Latour’s concepts: A.J. Greimas and his
thropology of science is to some extent Paris school of semiotics. From his ear-
influenced by semiotic theory. The use liest texts on science studies (Latour and
of concepts like actant, inscription, Fabbri, 1977) to his most recent ethno-
translation, modalities, shifting in, shift- graphic study of a court of law (Latour,
ing out and regime of enunciation attest 2002a), Latour always maintains semiot-
to this fact. The aim of this article is to ics as a basic theoretical tool. Greimas’s
show how most of Latour’s theoretical works usually constitute a majority of his
views can be connected to semiotics, al- semiotic references, hence my interest
though this should not exclude other in this article to study this connection.
readings of his work. As I will show be- In a way, we might say that I am trying
low, many of his analyses give actor-ori- to read Latour’s texts in the light of
ented or political explanations, as well Greimasian theory in order to analyse
as semiotic ones. On the other hand, some of the presuppositions of Latour’s
agency has a central place in the semi- theories and to assess their strength and
otic theory with which I will compare weaknesses.
6
Roar Høstaker
teurization of France (Latour, 1988a: 153- oriented explanations in his texts, while
236). At the outset, Irreductions presents at other times his explanations are semi-
a principle of irreducibility: “nothing is, otics-oriented. Many of Latour’s most
by itself, either reducible or irreducible read texts from the 1980s emphasised
to anything else” which is meant as a actor-oriented explanations (cf. Latour,
forewarning that there are no inherent 1983; 1987) and his associology based on
truths. The second major principle is trials of strength has been much criticised.
“whatever resists trials is real” (Latour, It was claimed that Latour portrayed sci-
1988a: 158). Everything can increase or ence as though it were composed of ex-
decrease in reality through the force pressions of political processes: a sort of
gained or lost by trials of strength. generalised Machiavellianism or even
‘Trials of strength’ was a central theme worse – for Latour everything is war!
of the laboratory studies within the field (Haraway, 1997: 33-34). In the 1990s
of science studies in the 1970s and 1980s, Latour tried to accommodate his views
and the ambition of Irreductions was to to some of this critique by introducing
generalise this principle to an ontologi- concepts inspired by Whitehead (Latour,
cal level. Basic entities (often called 1996a; 1999b), and some of his dissatis-
actants) gain strength by associating faction with Irreductions in the early
with other entities. In this way they 1990s came from the dangers of estab-
might resist some entities and form joint lishing a new meta-language (Crawford,
vectors of force with others (Latour, 1993: 265). Scientific concepts, either
1988a: 160). This is the basis of Latour’s from the social or the natural sciences,
constructivism. But this ‘associology’ usually form an observation language
comes from a ‘crossing’ not only with that in many ways ‘replaces’ the entities
semiotics but also with actor theory; it studies. Latour wanted to form an in-
the actor-based social theory Latour fra-language, whose role is to indicate
most frequently refers to is ethno- only the relationship between entities,
methodology. Although it may sound without ‘touching’ the objects them-
otherwise, it is not far from Greimasian selves. Nonetheless, he seems to proceed
semiotics because Greimas’ theory of in a way that is faithful to the essence of
narratives contains its own theory of the principles of Irreductions (Crawford,
agency (cf. below). Furthermore, Latour 1993:266). During the 1990s and on-
points to a common research stance be- wards, semiotics has continued to hold
tween semiotics and ethnomethodol- a central place in his theoretical formu-
ogy: “Semiotics is the ethnomethodol- lations.
ogy of texts. Like ethnomethodology, it What form of semiotics does Greimas
helps replace the analysts prejudiced represent? Greimas’s school was one of
and limited vocabulary by the actor’s several different attempts in the 1950s
activity at world making” (Latour, 1993a: and 1960s to develop a scientific ap-
131). proach to the analysis of language and
Both semiotics and actor theory seem texts. This usually included a rigorous
to fuse into Latour’s associology, and theoretical apparatus in order to unveil
something that may be bewildering is the underlying structure of texts, or even
the fact that he sometimes gives actor- of the wider discourse. This could be
7
Science Studies 2/2005
8
Roar Høstaker
one hand, informed by notions of inter- pression (signifier) and that of content
ests, alliances, enrolment, spokesper- (signified). These planes presuppose
sons, and strength.1 On the other hand, each other reciprocally. In addition,
this observation language is formed by within these two planes he distinguished
narrative theories. between form and substance (Hjelmslev,
However, what makes the notions of 1993: §13). Substance in this context is
translation become an observational usually understood as meaning or pur-
language? When Latour sets out to study port in as much as they are taken on by
how scientists and engineers construct the semiotic form (Greimas and Courtés,
their objects he has to describe them 1982: 322). If we, for example, take the
somehow, and he has set limits on the phrase “the GRF hormone can cure
way this should be done. Anyone em- dwarfism”2, this phrase can be expressed
barking on such a study should not “use phonetically in different ways even by
culture, the content of science, or dis- native English speakers – it varies in pho-
course as the cause of the phenomenon” netic substance – but it has to conform
(Crawford, 1993: 263). We should not use to some sort of sound pattern and syn-
nature to judge culture or discourse, and tax in order to be understandable (form
similarly, we should not use culture or of expression). Similarly, a fundamental
discourse to judge nature (Latour, meaning of the phrase can be laid bare
1993a). In short, he advocates an agnos- (form of content), and this meaning is
tic stance in relation to the evaluation of given a substance within particular dis-
the substances of the different disci- courses. The phrase can be treated as a
plines constructing objects, whether truth claim within biochemistry, an ethi-
these concern science, discourse or cul- cal or political question. These substan-
ture. In his case studies, Latour can leave tial matters cannot be decided by lin-
judgements of substance to the actors guists, but have to be left to the actors
themselves. They have to be: “left to their themselves. The form (of expression and
own devices. It’s a laissez-faire sociol- content) is necessary for the substance
ogy” (Latour, 1996b: 170). This is the ba- to be part of the signification, but the
sis for Latour’s affirmative view of sci- substance is variable and taken by itself
ence, culture and discourse. His solution it is amorphous.
is to describe the forms or types of ac- For Greimas and his colleagues this
tivities in which the actors are engaged division serves as a way to determine the
in the translation process, whether these object of linguistics to include only the
actors are people, objects or literary de- form of the expression or of the content
vices. (see Figure 1). The substance is part of
This is a solution quite parallel or an indeterminate domain for the semi-
analogous to the one chosen by Greimas. otician. The reason why the semiotician
One of Greimas’ starting points was can say something about signification is
Louis Hjelmslev’s (1993) phonological that the form organises itself into sys-
model of language in which the Saus- tems of relations (cf. Hénault, 1979: 28-
surean model of the sign is transformed. 29). It is not my intention to claim that
Hjelmslev distinguished between two Latour adopts this model, but by limit-
parallel planes of language – that of ex- ing himself to the forms of activities he
9
Science Studies 2/2005
Substance of CONTENT
(amorphous intellectual zone)
Substance of EXPRESSION
(amorphous phonic zone)
10
Roar Høstaker
actant defined as “that which accom- two basic forms - either an utterance of
plishes or undergoes an act, independ- state (être) or an utterance of doing
ently of all other determinations” (Grei- (faire). The justification for this theoreti-
mas and Courtés, 1982: 5). Only gradu- cal articulation was that it allowed for a
ally does it become an actor by virtue of unified conceptualisation of descrip-
being invested with one or more roles. tions and transformations (Hénault,
Greimas based much of this work on 1983: ch.2). In syntagmatic chains utter-
Vladimir Propp’s study of 100 Russian ances will take other utterances as their
folktales (Propp, 1968 [1928]). Propp for- object and modalise them. When an ut-
mulated 31 different functions to de- terance (of state or doing) governs an-
scribe situations in these folk-tales. other utterance (of state or doing), the
Greimas condensed and generalised first utterance is said to be a modal utter-
Propp’s schema during the 1960s and ance while the second one is a descrip-
1970s and gave different versions of a tive utterance (Hénault, 1983: 55-61).
narrative schema in which a subject lacks When an utterance of doing modalises an
a certain object (understood in wide utterance of state we have narrative per-
terms). In the folktale the object might formance. In the phrase “Eve bought a
be a person or thing missing after a mis- red dress” the action (buying) changes
deed by a villain. A sender enters a con- the state of Eve from a state of non-pos-
tract with the subject (receiver) in order session to a state of possession (of the
to liquidate what is missing. In the dress). Performance amounts to the re-
folktale the sender is often a person of alised action in narratives. This realisa-
authority from the social hierarchy tion, however, presupposes the exist-
(queen, king, mother, father) conveying ence of some virtuality or potentiality for
an obligation upon the subject (Greimas, action. In other words, some state must
1983; Hénault, 1979). The sender del- lead to this or that form of doing. The
egates a task to the receiver-subject and subject in the phrase (Eve) must have a
(often) sanctions the results of the ac- certain competence to do what she does.
tion. The sender possesses some knowl- Greimas and colleagues distinguish be-
edge that he transmits to the receiver tween four modal values of competence.
and he also frames the action. What The subject can be seen to have know-
makes a text a narrative text is, accord- ing (savoir) about what to do, to be want-
ing to this theory, a state of disposses- ing (vouloir) to do something, to be
sion or possession of some valued ob- obliged to or have to (devoir) do some-
ject leading to some action to produce thing, and to be able to (pouvoir) do
the opposite state of dispossession or something (Hénault, 1983: 55-61).
possession (Hénault, 1979: 145). Typically a hero acquires different
During the 1970s Greimas trans- forms of competence during the course
formed the narrative schema further by of a story3, but this competence is ac-
concentrating on the relationship be- quired before the hero’s actual perform-
tween the subject and the object as the ance. According to Latour, however, this
principal actants of transformations. relationship is quite the opposite in sci-
Utterances about the relation between entific texts: performance comes first,
a subject and an object were seen to take then competence. In an analysis of Louis
11
Science Studies 2/2005
12
Roar Høstaker
area. The pedologists thereby obtained tion of the forest and the savannah
a synoptic representation of the soil in shows the process by which a science
the area by this organisation of the produces its internal referent. Each step
lumps of earth. This device made it pos- of the re-presentation involves the cog-
sible for the pedologists to assess the nitive activities of the enunciator (the
qualities of the soil at a glance and to researchers) through successive steps of
enable the production of a graphical rep- shifting out by which they frame and
resentation of the soil in the area. Later stage the object they want to say some-
these samples were moved to Paris for thing about (Latour, 1999b: ch.2; cf.
further analysis and the results entered Greimas and Courtés, 1982: 87-91, 259-
the scientific literature in the form of re- 261). In fiction texts, shifting out is a
ports and papers. In this way the pedo- technique used by the author (the enun-
logists turned themselves into (scien- ciator) to move the reader’s attention
tific) spokespersons for the rainforest in away from himself/herself and out to
Boa Vista (Latour, 1999b: ch.2). some action going on at another place,
An important point in this context is another time and concerning someone
that the world can nowhere be seen else (Latour, 1988b: 5). In this way an
apart from words, although the world is, impression of realism is produced by the
of course, much more than words. If we constraints that this shift has placed
limit ourselves to the non-scientific de- upon the actors. In scientific texts, this
scription, we can sense the forest, but we ‘reality’ or ‘truth effect’ is produced quite
cannot produce signification about it systematically. If we return to our exam-
(even in our thoughts) without adapting ple, we find, first, the shifting out from
linguistic form to substance. In the re- the enunciator to the frame of the geo-
search enterprise this is even clearer. metrical pattern, then from these num-
When the researchers divide the area bered squares to the samples of earth,
into numbered squares this is carried from the samples to the pedo-compara-
out by means of the forms found in ge- tor and from this device to the graphical
ometry and arithmetic. When the re- drawing and from the graphical drawing
searchers collect samples of the soil, the to a written summary. The enunciator
samples are not just earth, but, within can at each step, without problems,
this endeavour, they have taken on the move back from one or more of the
form given to them by the researchers. frames – shift in – and say something
They have taken on a significance be- about the condition of the forest and the
yond being just pieces of earth and have savannah in Boa Vista (Latour, 1999b:
become representatives of some part of ch.2). Shifting in is the opposite opera-
the area. They re-present the forest, and tion of shifting out and moves the atten-
these lumps of earth can again be re-pre- tion back to the enunciator or some
sented in the graphic table and the other ‘I’ in the text (Latour, 1988b: 6;
graphic table can be re-presented by Greimas and Courtés, 1982: 100-102).
descriptions in texts. (Latour, 1999b: For each step in the chain of outward
ch.2). shiftings, something is delegated the task
Interpreted in semiotic terms this of re-presenting the object the enuncia-
process of collection and re-presenta- tor really wants to say something about.
13
Science Studies 2/2005
However, this process of delegation de- beyond somehow confirms or denies the
pends at each stage on the relevance and efforts of human beings in their study of
the accuracy of the ‘framing’ made by the nature (cf. Latour, 1987; 1988a; 1999a;
enunciator. It is this ‘framing’ that allows Latour and Woolgar, 1986).
the sciences to claim that they speak
about an external referent and not just From Local Events to the Collective
an internal one. For Latour, however,
every notion of an external referent is We have so far studied how Latour de-
meaningless4 . Reference can only mean scribes quite limited and local events.
the chain of translations of internal ref- However, the ‘translation model’ used by
erents. To what degree scientists speak Latour is also usually understood to pro-
truthfully about nature depends upon vide a more general understanding of
the quality of this chain (Latour, 1999a: the composition of the world. This is
310). usually described in the idiom of politi-
A consequence of this view is that cal activism: mobilisation, enrolment,
there is no known or knowable reality interests, alliances, and spokespersons.
that is a non-linguistic reality. That does (Latour, 1999b: 194; Callon, 1986). In this
not mean that everything is language, way new technical objects gain or lose
but rather that science has become im- existence (Latour, 1996b), and through
manent to language. There is no place the enrolment of technical objects and
where reality can hide outside language; other humans we build a collective and
this view constitutes a rebuttal of realist even a Hobbesian Leviathan (Callon and
epistemology. The rupture between Latour, 1981). However, in Pandora’s
word and world, assumed by realist epis- Hope Latour again describes the basic
temology, involves presuppositions of an relationships between humans and
undifferentiated background existing technical objects in the idiom of semi-
independent of human knowledge. otics, although this is not always obvi-
When scientists propose some state of ous in the text (Latour, 1999b: 176-193).
fact about the world, a common argu- Here he is concerned with how humans
ment is that this factual entity has always and objects fold into each other. How is
been present, even before the fact. The it possible for us to treat technical ob-
fact, lying in a slumbering state, is po- jects around us simply as silent interme-
tentially knowable by us, and when it is diaries? This is the core question in what
discovered it simply manifests its po- Latour calls the four meanings of tech-
tency. The real hero is the entity itself and nical mediation. These meanings and
not the humans who describe it and how they relate to Greimas’s semiotics
make it possible for this knowledge to be are examined below.
formulated. In no other situation is this Latour’s meanings of technical me-
clearer than in scientific controversies. diation can be compared to Greimas’s
From a realist standpoint the production narratology. We have already discussed
of an artefact is usually described in so- how narrative performance is formed in
cial terms while the acceptance of a fact the example “Eve bought a red dress”.
is described as being the thing itself. Eve- This transformation of Eve from the
rything happens as if the reality of the non-possession of an object to the pos-
14
Roar Høstaker
15
Science Studies 2/2005
Figure 2. The modal structure of Having-to-do (devoir) with Latour and Akrich’s
concepts in brackets.
(Opposition)
Having-to-do Having-to-do
(prescribed) (proscribed)
(Complementarity)
(Contradiction)
Not-having- Not-having-
not-to-do (Opposition) to-do
(allowed) (afforded)
16
Roar Høstaker
17
Science Studies 2/2005
18
Roar Høstaker
19
Science Studies 2/2005
1999c; 2002c) while to speak politically gest different subject-positions and differ-
is to create the united representation of ent pre-existing competencies in the per-
a multiplicity (Latour, 2002b). former in order to speak truthfully within
The regime of political enunciation each regime. An important observation,
does not describe a line, but rather, a cir- however, is that a regime of enunciation is
cle. A person trying to speak on behalf not based upon inter-definitions or the
of a multiplicity forms it into a unit. In actors’ own contextualisations, because
order to make the multiple into one, the this regime is a structure that exists a
representative has to claim autonomy priori, and the actors often uncon-
while at the same time being obedient sciously presuppose them. Here, Latour
to the demands of the multiplicity. To has suddenly left his own method of fol-
speak politically is thus concerned with lowing the actors. However, a regime of
both how to produce unity (autonomy) enunciation avoids the trap of context-
and how to be faithful to the represented oriented sociology by not starting from
(heteronomy). This involves work lead- a point outside the context, but from
ing to a circular movement between a within it. Nonetheless, this concept
situation of univocality and the dissolu- amounts to a change of perspective in
tion of this unity in a situation of multi- Latour’s theories since the focus is not
vocality. If the work of mediation and how actors make their own world, but
representation is successful it will form how a world not of their own making
a ‘good curve’ and a political group will over-determines how they express them-
be formed. However, it is an impossible selves (if they are competent performers).
task to make this combination of faith- Latour (1993a: ch.2) provides similar pre-
ful representation and unity for action existing structures in his analyses of mo-
without distorting the immediate de- dernity as a constitution governing how
mands of the relevant multiplicity. This we distinguish between nature and cul-
is why political enunciation is so full of ture, and similarly, the iconoclastic ges-
circular and vague expressions, repeti- ture by which we denounce fetishes of
tive phrases and catchy words. From the different kinds (Latour, 1996c; 1999b:
viewpoint of ‘double-click information’ ch.9).
this is betrayal, but this ‘betrayal’ is an The great difference between such
integral part of the political way of Latourian structural formulations and
speaking. The representative should be- context-oriented sociology, is that in
tray the multiplicity in order to give it Latour’s formulations there are no gen-
one voice. What is important for Latour eral causes intervening from the outside
is to liberate political enunciation from and deciding what the actors say or do.
being valued according to non-political The structures are positions offered us,
criteria (Latour, 2002b). and we can, at least to some extent, avoid
Latour’s use of enunciation seems them. However, a regime of enunciation
clearly to conform to the linguistic defi- or a similar structure is a historically
nition, and to the fact that a regime of constituted entity, but the extra-contex-
enunciation may be seen as a virtual tual relationships keeping it in existence
structure pre-existing the utterance and do not seem to interest Latour. That
coordinating it6. The different regimes sug- would perhaps be to ascribe a given
20
Roar Høstaker
21
Science Studies 2/2005
teur had to align his interests with the tempts to be political Latour seems to
hygiene movement, he had to be a have demonstrated another limit to his
French patriot or nothing. The field of approach.
science in France was, in this way,
overcoded by nationalism. A major point Conclusion
of Deleuze and Guattari is that the molar
is of a different nature than the molecu- Latour’s anthropology of science is origi-
lar and hence have to be studied in it own nal and interesting insofar as he has in-
way. Latour’s emphasis given to the mo- troduced a certain line of reflection into
lecular relations limits his approach. science studies and social theory in gen-
Another avenue of critique can be eral. A part of this line of reflection is to
opened in relation to Latour’s agnostic make semiotic concepts operational in
stance in relation to the substance of sci- social science research. His analyses of
entific research and its social and envi- circulating reference, the gradual emer-
ronmental consequences (cf. above). gence of scientific objects and of tech-
The question is for how long the anthro- nical mediation is important in this con-
pology of science can isolate itself from nection. The limits of Latour’s approach
the matter of science. By limiting his ap- are not due to semiotics, but rather to
proach to what actors do, without pass- his application of it. He has, to a high
ing judgements on their actions and degree, limited his approach by giving
their objects, a Latourian researcher may preference to the observable and actual
easily slip into a mode of acquiescence. over the principles governing the setting.
As a research strategy, this agnosticism Or, in linguistic terms, he seems to pre-
has been very fruitful, but Latour has fer analyses along the syntagmatic axis
recently tried to export it to the field of to those along the paradigmatic axis.
politics in order to “bring science into However, the concept of regime of enun-
democracy” (Latour, 1999d). In the same ciation seems to bridge some of this ten-
way as his studies of scientific practices, sion between the actual and the virtual.
his critique of political institutions con- This also holds true for his analysis of the
cerns their form and not their substance. competencies of techno-social assem-
If he had engaged himself with the lat- blages.
ter, he would have been forced to take One consequence of Latour’s prefer-
into consideration the existing assem- ence for the actors’ actual construction
blages of political enunciation. Instead of the world, is the tendency of Latour’s
he limits himself to sketching new forms case studies to encapsulate themselves
of representation and decision-making within a limited context. This can be
that include science and technology (cf. avoided in other ways and still fall within
Latour, 1999d). However, this politics of a semiotic universe. I have tried to show
nature has no direction. It is without this by introducing Deleuze’s and
substance and hence without any poli- Guattari’s distinction between the mo-
tics at all (cf. Caillé, 2001). In order to lecular and the molar as two different
solve our ecological problems, the only levels and how the latter may overcode
thing he can offer is a rewriting of the the former. Latour concentrates his ef-
rulebook of liberal democracy. In his at- forts on the molecular level and he only
22
Roar Høstaker
23
Science Studies 2/2005
24
Roar Høstaker
25