Affective ‘this’

Christopher Potts, Florian SchwarzLinguistic 2010 · DOI

TL;DR

This paper provides quantitative corpus evidence for the long-standing linguistic theory that the word "this" serves as an "affective" marker of emotional closeness and solidarity. By analyzing millions of online reviews, the authors demonstrate that "this" behaves like an exclamative (e.g., "wow!") and is used more frequently when speakers are in high-agreement "solidarity" states with their audience.

What Problem Does This Paper Address?

In linguistics, demonstratives like this and that are usually studied for their "deictic" functions—pointing to things in physical space ("this chair") or in a conversation ("this idea"). However, since Robin Lakoff’s work in 1974, linguists have suspected these words also have an "affective" or emotional dimension. For instance, saying "There was this traveling salesman..." feels more vivid and engaging than "There was a traveling salesman..."

The challenge was that these claims were largely based on researcher intuition and small sets of examples. This paper seeks to bridge the gap between qualitative theory and quantitative data. The authors ask: Can we prove that "this" is a marker of emotion and solidarity using large-scale statistical evidence from the real world?

Methods

The authors employ a "sentiment-informed" corpus analysis. They use the star ratings (1–5) of online reviews as a proxy for the speaker's emotional state.

  1. Log-Odds Distribution: They calculated the frequency of "this" across different star ratings. To normalize for the varying sizes of rating categories, they used a log-odds ratio.
  2. Quadratic Regression: To identify "U-shaped" patterns (where a word is common at the extremes—1 and 5 stars—but rare in the middle), they fitted logistic regression models. A positive quadratic coefficient ($\beta_2$) indicates an "exclamative" profile.
  3. Solidarity Modeling: They tested if "this" usage correlates with how much a reviewer agrees with the "room." They calculated the probability of a specific rating for a product and correlated it with the frequency of "this" in reviews for that same product.

Datasets

The primary dataset is an extension of the UMass Amherst Sentiment Collection, containing:

  • 118,395 reviews from Amazon and TripAdvisor.
  • 18.6 million words total.
  • Metadata including star ratings (1–5), dates, and usernames.

They also briefly utilized a corpus from OpposingViews.com, where commenters align with specific "sides" of a debate, to further test the solidarity hypothesis in a non-scalar (non-1-to-5) environment.

Key Insights and Ideas

  • The "Exclamative" Profile: "This" (both as a determiner like "this book" and a pronoun like "I like this") shows a distinct U-shaped distribution across star ratings. This mirrors the behavior of clear exclamative markers like wow, absolutely, and the exclamation mark itself.
  • Anti-Exclamatives: In contrast, "colorless" functional words like the, but, and quite show an inverted U-shape. They are most common in 2-4 star reviews where authors are being nuanced and "weighing the good against the bad."
  • Solidarity in the Majority: The authors found a significant positive correlation between "this" usage and the level of agreement among reviewers. If you are writing a 5-star review for a product that everyone else loves, you are statistically more likely to use "this," supporting the idea that it functions as a marker of shared sentiment.
  • Proximal vs. Distal: While proximal this (near the speaker) signals closeness, distal that (far from the speaker) showed an "anti-exclamative" or even negative bias in the data, often associated with "censure, dislike, or scorn."

Key Results

  • Statistical Significance: For both determiner and pronominal this, the quadratic coefficient ($\beta_2$) was positive and highly significant ($p < 0.001$), confirming the link to heightened emotional states.
  • German Correspondence: The researchers found nearly identical patterns for the German proximal demonstrative dies- ("this"), suggesting that the affective use of demonstratives may be a cross-linguistic trait rooted in the basic meaning of "proximity to the speaker."
  • Effect Size: While the effect is statistically robust, the authors note it is "mild." The quadratic coefficients for this (around 0.025 to 0.078) are smaller than those for wow (0.243), which makes sense given that this is primarily a functional word, not a dedicated emotional outburst.

Limitations and Future Work

The authors acknowledge that their method is "necessarily approximative." Because "this" is so common in non-emotive contexts (simple pointing), the affective signal is often buried in noise.

Furthermore, the "solidarity" measured in online reviews is somewhat artificial, as reviewers aren't in a direct live dialogue. The authors suggest that future work should apply these quantitative methods to more interactive datasets, such as spoken transcripts or social media threads, where the "fostering of closeness" between participants is more immediate and dynamic.

Summary updated

Abstract

Lakoff (1974) argues that affective demonstratives in English are markers of solidarity, with exclamative overtones deriving from their close association with evaluative predication. Focusing on this, we seek to inform these claims using quantitative corpus evidence. Our experiments suggest that affectivity is not limited to specific uses of this, but rather that it arises in a wide range of linguistic and discourse contexts. We also briefly extend our methodology to demonstrative that and to German diese- (‘this’).
BibTeX
@inproceedings{potts2010affective,
  author = {Christopher Potts and Florian Schwarz},
  title = {{Affective ‘this’}},
  booktitle = {Linguistic},
  year = {2010},
  doi = {10.33011/lilt.v3i.1219},
}
Figure