Essence of meaning

Below is a guest post by Robert Shackleton:

---

Peter Dodds and coauthors have recently published research that proposes a significant shift in the essence-of-meaning framework, which traces its lineage back to Charles Osgood’s initial efforts to use dimension reduction to quantify human meaning. The paper, “Ousiometrics: The essence of meaning aligns with a power-danger-structure framework instead of valence-arousal-dominance,” appeared in Science Advances. The abstract:

From work emerging through the middle of the 20th century, the essence of meaning has become widely accepted as being described by the three orthogonal dimensions of valence, arousal, and dominance. These essential dimensions have become the cornerstone of sentiment analysis across many fields. By reexamining first types and then tokens for the English language, and through the use of automatically annotated histograms—“ousiograms”—we find here that the essence of meaning conveyed by words is instead best described by a goodness-power-aggression-danger-structure (GPADS) circumplex framework; that large-scale English language corpora reveal a systematic bias toward safe, low-danger words; and that the power-danger-structure framework is the minimal framework that represents essential meaning. We find remarkable congruences between the GPADS framework and other spaces including mental states and fictional archetypes, and we construct and demonstrate a prototype ousiometer.

Extensions of Osgood’s original work led to a shift in the 1970s from his foundational dimensions of evaluation, potency, and activity (EPA) to a VAD framework that substitutes valence (or pleasure) for evaluation, arousal for activity, and dominance for potency. Dodds et al. discuss problems with the data, methods, and results in that previous work and offer significant improvements. Their main conclusion is that essence of meaning categories are more accurately characterizd by a five-dimensional framework involving goodness, power, aggression, danger, and structure (GPADS), but they also provide a three-dimensional PDS “minimal framework” emphasizing power, dominance, and structure.

A power-danger-structure framework for essential meaning seems to fit rather nicely into Lakoff and Narayanan’s model of cognitive schemas and frames, and it might also have intuitive appeal from an evolutionary standpoint. Members of a primate group inevitably have a profound interest both in potential dangers and in patterns of social dominance. But I wonder whether Dodds et al.'s third dimension, structured versus unstructured, might be better framed as animacy-inanimacy. Animacy appears to be at least as consistent with their analogy with thermodynamics as structure does, and early humans may well have interpreted much of the world through a frame of animacy, as evidenced not only by animacy as a fundamental grammatical category in many languages but also in the persistence of belief in non-human agency in most cultures, including ours.

---

Above is a guest post by Robert Shackleton.

Note that early work on learning semantic distances by projecting words into a meaning space based on orthogonalizing a term-by-document matric, e.g. latent semantic anaysis, was inspired by Osgood's "Semantic differential" method as well as by Gerard Salton's vector space model.