In a text-oriented cognitive approach, it has been proposed  that the high incidence of Complex Noun Phrases in (written or oral) textualization, as contrasted with the high incidence of Main and/or Subordinate Clauses, mirrors the so-called underlying processes of 'Entification' and 'Propositionalization', respectively. This phenomenon has been particularly described as being systematically correlated with specialists’ v/s non-specialists’ language use, with reference to any particular field of knowledge.
According to the theoretical assumptions behind this approach, the process of propositionalization makes reference to an analytic textual orientation, according to which certain logico-semantic primary propositions are ratified as ‘analytic propositions’ and, then, syntactically codified in the form of Main or Subordinate Clauses. Alternatively, the process of entification makes reference to a synthetic textual orientation, according to which the mentioned primary propositions are further processed as ‘synthetic propositions’ and, then, syntactically codified in the form of Complex Noun Phrases.
This empirically-based proposal is theoretically motivated by both (a) Givon’s hypothesis that the notio-functional and structural organization of language is intimately connected with human cognition, perception and neuropsychology, and (b) Lyon’s original proposals about the different types of entities in the world: 1st, 2nd, and 3rd order entities.
In a nutshell, and as far as my understanding goes, the story unrolls like this: as soon as a given phenomenon is perceived, it is “cognitively” configured (or internalized) as a ‘primary proposition’. As part of cognitive-semantic processes, such proposition can be represented either as 2nd order entity (or ‘synthetic proposition’, meaning a 3rd order entity that is perceived and “conceptualized” as though it was a 1st order entity), or, alternatively, as a 3rd order entity proper (or ‘analytic proposition’). Additionally, it has been observed that the specialists’ tendency to produce complex noun phrases (and, therefore, making an underlying entification capacity explicit) contrasts non-specialists’ tendency to produce clauses. In fact, the mentioned contrast is said to throw lights on the specialists’ distinctive tendency to produce “holistic discourses”, where the chosen syntactic arrangements nicely allow these language users to synthesize or agglutinate referents in ways that analytic syntactic arrangements do not.
Consider the following cases in 1 and 2:
1) In (i), an instance of propositionalization is codified in the form of a main clause, and in (ii), entification  is codified in the form of a complex noun phrase (i.e. a potential referring expression at a socio-pragmatic level):
i) Adams and Aizawa are reticent to accept that cognitive processes extend into the tools people use. [Main Clause or Sentence à NP(S) + VP(P) + AjP(C)]
ii) Adams and Aizawa’s reticence about the possibility that cognitive processes extend into the tools people use. [Complex Noun Phrase à NP/NP poss (PreM1) + N(H) + NP/PP(PostM1))]
2) The NP in (iii) shows a synthetic (or “encapsulated”) proposition, where at least one of the arguments can be left out in the textual structuring (i.e. a case of “propositional compression”). Added to that, the NP is said to derive from the underlying sentence in (iv):
iii) The invention of the famous theory of recollection (by Plato)
iv) Plato invented the famous theory of recollection
In a radically alternative perspective, the empirically studied incidence of syntactic structures (such as Complex NPs) described above could be considered, at least by way of provisional exploration, to be playing the role of epistemic material mediating structures. Understood as a hybrid material artefact, human language can be conceived of as a physical structure whose use and organization are dependent on (and probably shaped by) the writer/speaker’s “scaffolding-hungry cognitive performance.” So long as certain syntactic structures can help unearth (or make explicit/clear/available) information that is not immediately (or easily) available to any potential reader/listener, these structures can be said to be playing an epistemic function. The same would apply for any chosen structural or functional arrangement at the textual level.
As a (certainly most) simple example, this is what happens when one considers the textual role of the introductory sections in the traditional organization of expository scientific papers. The need of a title, an abstract, and an introduction, for example, can just illustrate the importance of counting on (referent-agglutinating) mediating structures that enable readers to gradually “unearth” (and therefore facilitate her understanding of) the main information “conveyed” in such articles (hopefully, the writer/speaker’s ultimate “intended” message) . Suffice it to consider the case when those textual scaffoldings are intentionally omitted (as in an ad hoc experiment), reducing the cognitive agent’s problem-solving capacities dramatically.
As a provisional thesis presented here for critical discussion, referent-agglutinating syntactic structures (like complex noun phrases in specialists’ textualization cases) may not be the surface manifestation of logico-semantic representational (language-like) cognitive processing, but rather than that, the evolved material organization of a complex mechanism for continuously generating (more or less elaborated) epistemic mediating structures. As a general character, such mediating structures may very well be regulated by processes of continuous “semiotic disembodiment”, so long as this phenomenon is understood as a continuous generation of material niches for epistemic action.
In this alternative perspective, whereas specialists’ tendency to a holistic textual orientation can be viewed as the result of tool users' expertise, holistic texts themselves (and its implications regarding more or less effective communication) may be considered to be a case of tool transparency
Soto, G. y C. Zenteno. (2004). “Los sintagmas nominales en textos científicos escritos en español”. ELUA: 275-292
 The involved nominalization at the lexical level is subsumed here by the construction of noun phrases, which is considered to be more relevant to a text-oriented approach.
 A related case in point may be the known hierarchical sub/topic structure observed in (but not exhausted to) both formal and informal argumentative texts. More on this hierarchical organization in: G. Soto y C. Zenteno (2001-2003). “La subtopicalización en el discurso científico escrito”, Lenguas Modernas, 28-29, pp. 29-52.