Against Conflation: 1 Stratification, Abstraction, and Theoretical Architecture in SFL

Why This Matters

Systemic Functional Linguistics (SFL) is not merely a descriptive framework for analysing language. It is a theory of symbolic abstraction: a carefully articulated account of how different kinds of semiotic organisation relate to one another.

Much contemporary confusion in SFL debates arises not from empirical disagreement, but from a failure to respect this architecture — in particular, the distinction between levels of abstraction and types of theoretical relation. When these distinctions collapse, theoretical terms may remain familiar while their explanatory force quietly evaporates.

This post examines one such collapse. We name the alternative model early, but the argument is architectural rather than personal: the issue is not who proposed the model, but what happens when stratification, instantiation, and semogenesis are treated as if they were the same kind of thing.

Stratification Is Not Interaction

In Halliday’s canonical model, stratification is a relation between levels of symbolic abstraction.

• Semantics construes meaning potential

• Lexicogrammar operationalises that potential as structured choices

• Phonology/graphology materialises those choices for transmission and stabilisation

These strata do not interact as modules. Nor do they all “make meaning” in the same sense. Rather:

A lower stratum realises a higher stratum; a higher stratum is realised by a lower stratum.

This asymmetry matters. It is what prevents SFL from collapsing into a flat architecture of mutually influencing components.

By contrast, in Martin’s reinterpretation (most explicitly in English Text, 1992), strata are effectively treated as parallel modules whose relation is interactional rather than abstractive. The claim that “all strata make meaning” conflates semogenesis (the ongoing creation of meaning in use) with stratification (the organisation of symbolic abstraction).

The consequence is subtle but severe: if all strata are equally strata of meaning, then no principled distinction remains between construing meaning, enacting meaning, and materialising meaning.

When Phonology Becomes a Stratum of Meaning

The clearest symptom of this collapse is the treatment of phonology as a stratum of meaning.

In Halliday’s model, phonology does not construe meaning; it stabilises and transmits distinctions already construed at higher strata. Its semiotic contribution is real, but it is not semantic.

Martin’s decision to locate the information unit within phonology (1992: 384, 401) exemplifies the problem. Information is a grammatical system — not a phonological system that independently “makes meaning.”

This is not a terminological quibble. It reflects a deeper misunderstanding of what different levels of abstraction are.

Stratification vs. Instantiation

A second, compounding confusion concerns instantiation.

Instantiation is not something that happens within each stratum independently. It is a perspectival relation between:

• Potential (the system)

• Instance (the text/event)

To say that “all strata instantiate” treats instantiation as a process moving down a ladder, rather than as a shift in perspective on the same semiotic architecture.

This confusion becomes explicit in Martin’s reconstrual of context:

• Context of culture (potential) is redescribed as a stratum of genre

• Context of situation (instance) is redescribed as a stratum of register

Here, a relation of instantiation is replaced by a relation of interstratal realisation. Context — which in Halliday’s model is not language at all, but a higher-order semiotic environment realised by language — is absorbed back into the linguistic architecture as if it were simply another layer of meaning-making.

The result is a theoretical category error: a perspectival relation is mistaken for a structural one.

One Level of Abstraction Disguised as Many

Taken together, these moves reveal a consistent method:

• Distinct theoretical relations (realisation, instantiation, semogenesis) are treated as interchangeable

• All domains are assumed to exist at the same level of symbolic abstraction

• Architectural asymmetry is replaced by modular interaction

The vocabulary of SFL remains, but its theoretical force is redistributed. What looks like innovation is often the product of relabelling across collapsed distinctions.

What This Post Has (and Has Not) Argued

This post has not argued that Martin’s work lacks descriptive value, nor that it has had no productive influence. It has argued something narrower and more consequential:

That Martin’s model systematically misunderstands the architectural commitments of Hallidayan SFL by collapsing distinctions between levels of abstraction and types of theoretical relation.

In the next post, we turn to a specific case study of this method: discourse semantics as a stratum, and the systematic relabelling of existing systems that made this move appear both novel and necessary.