Lotmanian referentiality and (cultural) semantics

1 Introduction

Being able to systematize the nebulous concept of culture in as efficient a way as Juri Lotman did is by no means an easy feat. Lotmanian cultural dynamics are, one could argue, part of a holistic program to understand how culture is done, what is done through it, and what its discrete parts are. Lotmanian semiotics is thus not reducible to its main conceptual tools – signs, texts, semiospheres, specific aspects of cultural dynamics and so on – and so our intent in this paper is limited to presenting a rereading of modeling systems theory that tackles the cognitive challenge raised by Thomas Sebeok’s level-shifting proposal.

First, we quickly review Lotman’s modeling systems theory (MST) in the context of his semiotics of culture and then describe Sebeok’s challenge and proposal, before finally reformulating Lotman’s MST into a referential semantic theory. The upshot of such a reading is that it retains consistency within Lotman’s larger system and makes Sebeok’s challenge unwarranted, contra the standard view of MST, particularly as expressed within the more naturalistic varieties of semiotics. We start this brief journey by revisiting Lotman’s formulation of MST.

2 MST in Lotmanian thought

In “The place of art among other modelling systems” (2011),^[1] Lotman’s systematic presentation of his conception of modeling presents a cognitive argument: that a model is a cognitive substitute of a perceived object (1.2.1). Models can be systematic and rule-bound, including natural languages (1.3.1, 1.4). The introduction of a distinction between primary and secondary modeling systems (MS) is made immediately after this: A primary MS is to be understood, per 1.3.1, as a language, which means that Lotman attributes to natural language the property of being analogous to perception or cognition; at the same time, Lotman states that a secondary MS is a MS that is built upon natural language. The immediate example is that of the literary arts as built upon natural language. In this case, natural language is analogous to reality, structured, and rule-bound, and the literary arts build upon the rules of language to make something more – simultaneously similar and dissimilar (2.0). Lotman gives us a furtive, but useful definition of a sign as a “substitution of a denotat” (2.2) – in other words, a sign is a realized reference to a certain referent. This is important for us later on. A model, following the same point, is a perceptually efficient realized reference to a referent. In order to help us understand this difference, Lotman states in 2.3 that works of art are built on iconic signs.^[2] Taking from Lotman (1977),^[3] we see Lotman make this clearer by stating that “iconic signs are constructed on the principle of a causal relationship between expression and content […]. A sign models its content” (21). In the visual arts this may be simpler to see: An artistic depiction of an individual ceases to be solely a depiction of an individual and instead it becomes a representation of itself.

Lotman assigns special roles to different modeling activities, such as art and play, but we won’t concern ourselves too much with their characteristics. What matters to us is simply the distinction between primary and secondary, as these are the generalizations made and seen as relevant for this kind of semiotics. In The structure of the artistic text (1977), Lotman sees natural languages as “dealing with sequences of signs” that allow us “to stratify the natural linguistic structure into individual layers, each with a completely immanent function” (279) – these strata can be phonological, grammatical, lexico-semantic, and syntactic (phrasal and supra-phrasal). Earlier, in Lotman et al. (2013), we see the reference to natural language as primary (in point 6.1.3) simply reaffirming this notion,^[4] echoed throughout the secondary literature (for instance, to name but a few: Birnbaum 1990; Chang 2003; Shukman 1977). Secondary MS, being dependent on the primary MS, build upon it while creating something different. The previously mentioned strata are possible in a secondary MS, but insufficient to explain its functioning and requires us to look at how these levels interact with each other – that is, we may find more relevance in one level or another as opposed to how we would look at a non-artistic text (1977: 279). We can imagine how poetic phrasing of a sentence may work in this sense, where a metaphor or an odd syntactic structure may become more prominent in their reading.

The distinction between primary and secondary MS is what allows us to analyze the construction of culture – human communicative activities that seem to transcend a bare-bones linguistic exchange. As such, an important caveat moving forward is that MST entails only part of a larger vision on how cultural activity works – MST allows us to analyze some cognitive features of culture, but it does not paint much of a picture of cultural dynamics nor of large-scale cultural taxonomy, two points that are better reflected in other aspects of Lotmanian thought.

In what follows, we look at the Sebeokian challenge to Lotmanian MST and his proposal for a solution to this alleged issue.

3 The Sebeokian cognitive challenge

Throughout the last couple of decades, the original dual formulation of MST has seen a massive shift into a tripartite MST, with very little resistance. This is for good reason: The points raised by Thomas Sebeok (and later Marcel Danesi) are coherent with the current paradigm in semiotics – if not a direct cause of it. In this section, we look at the objection to the scale of MST, its rationale, and a proposed solution.

Recognizing the centrality of MST in the Tartu–Moscow School of Semiotics, Sebeok first interprets the concept of a secondary MS as an ideological model of the world in reciprocity to another MS, and a model of the world as a behavioral imperative (Sebeok 1991). Taking a Peircean route, and coupling it with Uexküllian concepts, Sebeok first understands language as a model by appealing to how Soviet linguistics owes a debt to Uexküll’s theory of meaning,^[5] namely, that a model of the world is a perceptual given (333). This he extends to how humans have developed models of the world based on their evolution, where brain size growth led a possible non-articulated linguistic system into a fully-fledged one, and that this may be an evolutionary trait that increased human fitness and was only later exapted for communication (334–335). In other words, language is a MS in human cognition. Invoking the Peircean notion of the sign – and the idea that thought comes in the form of signs – Sebeok presses the point that other organisms also model the world through simple, nonverbal models that “fit” reality well enough for survival (336). It is here then that Sebeok sees the perceptual–morphological pairing as the primary MS, pushing language to the second stratum. The reasoning for the latter is that the syntactic component of language seems to be a unique feature that allows humans to represent infinite possibilities, including, but not limited to, reality. At the same time, this pushes art and other cultural expressions into a tertiary stratum.

In the glossary at the end of Sebeok (2001), we further find “primary model” defined as a “simulative form (icon),” and “primary modelling system” as the “instinctive ability to model the sensible properties of things (i.e. properties that can be sensed)” (155). Later, Sebeok defines “secondary model” as “either an extension of the physical form or meaning of a simulacrum or an indexical form,” with “secondary modelling system” as a “system that allows for indication or the extension of forms” (156). Finally, “tertiary model” is conceived as “a symbolically devised form” and “tertiary modelling system” as a “modelling system that undergirds highly abstract, symbol-based modelling” (157). It is easy to see the Peircean interpretation of MST here, where the sign types are, to some extent, related to the level of modeling proposed by Sebeok’s reading of Lotman. Sebeok’s challenge to Lotman, to be precise, lies in the consideration of natural language as a secondary vehicle of cognition. The assumption that Sebeok makes is that perception is indirect and it models the external world, and that this representation of the world becomes re-represented in natural language. Danesi (2014) makes this relation clearer, by tying primary MS to Peircean iconicity, secondary MS as related to indexicality, and tertiary MS to symbolicity.

Sebeok and Danesi (2012) give us the most complete account of this proposal, building from the same Peircean account. This proposal has, in fact, been profoundly influential, particularly in biosemiotics. Kull (2010) sees modeling as a perceptual radical,^[6] but notes that Lotman could regard any sign system as a language (55). However, the focus of Lotman in human culture did not allow for noticing the perceptual basis of modeling.

The implications are far-reaching: By equating MST to levels of interpretation (or Peircean sign types) and identifying the primary MS as the Uexküllian Umwelt, Lotmanian theory has been reconstructed to be part of the modern biosemiotic synthesis. The Sebeokian cognitive challenge seems to successfully argue for an interpretation of Lotman that requires us to think of a substratum of meaning-making in cognition, something not found in Lotman’s original formulations. This interpretation, I argue, is not, however, a fair interpretation of MST and relies on incompatible assumptions about semantics that are inferential to Lotman’s MST.

In what follows, I introduce a rereading of Lotman’s MST that is arguably closer to the original notion, does not rely on expanded assumptions about sign types, and is still compatible with semiotic thinking about both cognition and culture.

4 Referential semantics in MST

Lotman had accepted that, to some degree, primary modeling was not necessarily limited to a single language (Gramigna 2013: 347) and that primary modeling was not concerned exactly with linguistic capabilities (Kull 1999: 19) – though this may have come to be a consideration for Lotman after Sebeok, as the later Lotman did consider primary MS within biological codes too (Velmezova and Kull 2022: 94). The point of this is that Lotman’s thinking about primary modeling is not as clear-cut as simply stating that it was all about language.

Birnbaum (1990) notes the loose definition of language employed by Lotman but equates primary MS to “ordinary language” (55). Umberto Eco, in the introduction to Lotman (2001), sees Lotmanian semiotics as a “cognitive science” (x), and in the same volume, Lotman asks us to not forget that, per Saussure, language may be considered as a system, not only as an expression in speech (5–6). Among important details to keep in mind is the point that the communicative function of language is but one aspect of language. Though conceptually ambiguous, Lotman’s notions of language separate the aspect of expression from its earlier systematic existence. In this sense then, we can distinguish at least two forms of language as used by Lotman, capital-L Language as the system underlying lowercase language, its functional expression – the Saussurean distinction between langue and parole. From this division we then assume the following: that capital-L Language is the cognitive system of representation, and that lowercase language is but a part of it.

MST is, at its core, linguistic and representational, or rather, these are the fundamental properties of a primary MS. That is, it assumes that capital-L Language is the system of thought, and that it is used to represent the world.^[7] In our rereading of Lotman, we add that because primary MS is linguistic and representational, it is propositional. That is, we assign propositional forms to linguistic expressions.

From this, we can build a simplified propositional account of a two-level MST.

5 Concluding remarks

It is undeniable that the Sebeokian synthesis of Lotman, Peirce, and Uexküll has been both profoundly influential and productive for semiotics. What this paper aims to show is that there are ways to retain the usefulness of MST without relying on the synthesis. And there may be good reasons to doubt the synthesis itself: The appeal to sign types in this capacity makes the case that sign types are unidimensional despite the plethora of different perceptual systems across organisms. If perception is at the base of modeling, it has to be given to the organism in some capacity. Iconicity, when understood as similarity, precludes different bases for how perception is processed and directly relies on a processing mechanism to state that the percept [apple] is somehow similar to the object apple. MST, when freed from this constraint, does not need to stipulate anything of the like. It instead relies on making the perception of an apple able to be expressed propositionally. This leads to a reduction of problems in terms of what counts as one sign type or another, a crucial problem to deal with particularly in biosemiotics when considering that we have no reportability in non-linguistic organisms.

That is not to say that this rendition of MST is incompatible with Peirce.^[10] For instance, Stjernfelt (2014) makes the case that protolinguistic capabilities may include common noun predicates (73), and that the usage of Dicisigns may satisfy the requirement of representing thought in non-linguistic organisms. A thorough examination of this parallel could indeed shed more light on how a more formal approach to MST can be carried out while also admitting forms of interpretation that emerge from the Peircean synthesis that exists in current semiotics.

This presentation of MST has, as a potential upside, its formalizable nature. The abductive nature of secondary MS can also be advantageous in that it allows us to find where potential semantic unknowns can be found, as in the assumption that, say, [[apple]+[pie]] may contain an unknown {x} semantic value. Although post-hoc knowledge of the secondary MS is necessary to bring full accounts of significations in this form of MST, the Sebeok–Danesi interpretation falls prey to the same issue in that tertiary modeling is not derivable from either primary or secondary modeling.

This interpretation comes, of course, with a number of shortcomings. Namely, we still need to answer what the scope of a primary MS is. Can we account for emotions or abstract objects within a primary MS, for instance? We may find answers to these questions through experimental means, one would hope. For example, Piantadosi and Cantlon (2017) show that numerical perception appears in wild baboons, and that numeral cognition is performed similarly between humans and non-human primates. However, this may only allow us to make generalizations about specific features of MST in primates. This does not mean, however, that we cannot generalize with this flavor of MST, but we may be cautious in ascribing the same generality of propositions to all forms of living beings. In any case, this issue is one shared across all interpretations of MST, for what is an icon in perception may be a function of the perceptual system itself.

Another issue – which is a much more general issue in semiotic views of cognition – is the reliance on representations, which makes this flavor of MST incompatible with radical embodied accounts of cognition insofar as it makes claims about thought in a propositional manner. Uexküllian theory is still compatible, though, as the formalization of perception in Umwelts may still be given in propositional form.

Lotman’s MST as expressed originally has been a springboard for thinking about semiotic cognition. Our proposal here has tempered some of its reliance on Peirce while retaining the strength of the original proposal, bound by a reinterpretation of the concept of language as used by Lotman. Although much work would have to be done on formalizing the expressions admitted in a primary MS – and this was beyond the purview of Lotman’s work – our interpretation retains the form of the original proposal, grounding it on cognition within the scope of language. As mentioned earlier, this interpretation does not say anything about the other aspects of Lotman’s thought, but it can be seamlessly integrated within it without recurring to conceptual resignifications or the integration of incompatible philosophical positions. The strength of the interpretation lies then in its simplicity and applicability to both levels of modeling, opening a number of research avenues on the semiotic logic of thought and creating fruitful questions, such as what cultural mechanisms exist to bind a secondary semantic object to a primary reference. In that way, we retain the original structure, simplify the assumptions made on modeling, and propose an expanding, formalizable way to look at MST.