How do emotional words affect our thinking? An In-depth Exploration on Word Semantic Density

Have you ever wondered why some words make you feel instantly happy while others make you pensive? For example, “happiness” directly conjures positive emotions, while “gift” may imply a pleasurable expectation. These two types of words are known in psychology as “emotion-label” (EL) and “emotion-laden”(EN) words.

What is the difference between EL and EN words?

• EL words can directly express emotions, such as “happy,” “sad,” “angry,” “afraid,” etc., with more efficient cognitive processing because their semantic information is more clustered.

For example, when we see the word “happy,” our brain directly activates the emotional memory associated with it without additional thought, like the emotion when we receive a birthday gift in happiness.

• EN words indirectly trigger emotions, such as “gift,” “funeral,” “wedding,” “war,” etc., but they can evoke emotional responses through personal experience or cultural context; cognitive processing requires more associations, so reaction times may be longer.

For example, “gift” does not directly express an emotion. Still, it may make you think of  “happiness.” It is, therefore, an emotionally loaded word, as if we realize the experience of receiving a birthday gift before feeling happiness.

Why are EL words easier to process than EN words?

The Semantic Density Hypothesis can explain this:

• The semantic information of EL words is more concentrated because they are directly related to specific emotional concepts.
• EN words have more diffuse semantic information because they rely on personal experience and contextual associations.

For example, when we see the word “anger,” we can quickly determine that it is an emotional word as we require less time to memorize our negative emotional experiences. In contrast, when we see the word “quarrel,” we may need to think about it to link our non-emotional experience to emotional experiences before determining it as an emotional word.

How can emotionality valence, positive vs. negative, affect us to differentiate emotional words?

The valence of an emotional word, i.e., whether it is positive or negative, also affects our cognitive processing. In general, positive emotion words (e.g., “joy,” “love”) are easier to process than negative emotion words (e.g., “pain,” “fear”) are more likely to be processed.

1) Cognitive Advantages of Positive Emotion Words

• Faster reaction time: people recognize positive emotion words more quickly than negative emotion words.
• More substantial memory effect: positive emotion words are easier to remember, probably because they are associated with the reward system.

2) Unique Effects of Negative Mood Words

Although negative emotion words are often perceived as slower to process, they may trigger stronger sensory experiences in certain situations. For example, “pain” can remind us of specific physical sensations (e.g., tingling, burning sensations). The intensity of this sensory experience may cause negative emotion words to show higher semantic similarity on some measures.

The Tools: Two different semantic similarity measures

To test whether the semantic density hypothesis can figure out the difference in cognitive mechanisms between EL and EN words with distinctive valences, two different computational models were used for the 427 emotion words in this secondary-data analysis:

1) Sensorimotor Distance Calculator

This computation calculates semantic similarity based on the scores of words on sensory and motor dimensions in the cosine algorithm. Specifically, it measures:

6 perceptual modalities: touch, hearing, smell, taste, vision, internal sensations (e.g., pain, hunger) & 5 action effects: mouth/throat, hand/arm, foot/leg, head, torso.

For example, the word “coffee” may score high on the “smell” and “taste” dimensions, while “running” may score higher on the “foot/leg” action dimension.

2) Word Association Model

This model measures semantic similarity through a free association task. Cue words (e.g., “happy”) are associated with the response words (e.g., “smile”, “sunshine”) after free association, and then calculate the degree of semantic overlap of these association words in the “cue-response” pair in a word associative strength algorithm.

What do we find?

EL vs. EN Words

The semantic similarity of EL words is significantly higher than that of EN words, supporting the semantic density hypothesis and suggesting that EL words are processed more directly in the brain; in contrast, EN words require more associative processes.

Positive vs. Negative Emotional Words

• Unexpectedly, semantic similarity was higher for negative emotion words in the Sensorimotor Distance Calculator. Perhaps negative emotions (e.g., pain and fear) tend to be accompanied by more intense bodily sensations and are more similar in the sensory dimension.
• In the Word Association Model, semantic similarity was higher for positive emotion words (possibly related to more consistent association patterns). This may be because positive emotion words (e.g., “love”, “happiness”) are more likely to trigger consistent associations (e.g., “family”, “warmth”). 

Interaction of Word Type and Valence

• Among EL words, semantic similarity was higher for positive words (e.g., “happy” was more focussed than “sad”).
• Among EN words, semantic similarity was higher for negative words (e.g., “funeral” elicited more consistent associations than “gift”).
  

  ---

What can we improve?

1. The sensorimotor distance calculator relies more on sensory experience, whereas the word association model relies more on free association so that the results may differ.
2. Differences in measurement tools: Alternative algorithms (e.g., Euclidean distance) can be incorporated to evaluate the robustness of the outcomes from the Semantic Density Hypothesis.
3. Influence of cultural factors: This is mainly based on English vocabulary, but different cultures may have different understandings of emotion words. → In the future, comparing how emotional words are processed in different languages could be explored.
   

   ---

The next time we read ‘happiness’ or ‘funeral’, we might realize how differently they journey through our brains—one triggers the emotion directly, the other requires an association; one may make you smile, and the other may make you ponder. That’s the magic of language!

References:

1. Allison, P. D. (1999). Multiple Regression: A Primer. Pine Forge Press.

2. Altarriba, J., & Basnight-Brown, D. M. (2011). The representation of emotion vs. emotion-laden words in English and Spanish in the Affective Simon Task. International Journal of Bilingualism, 15(3), 310–328.

3. De Deyne, S., Navarro, D. J., Perfors, A., Brysbaert, M., & Storms, G. (2019). The “Small World of Words” English word association norms for over 12,000 cue words. Behavior Research Methods, 51(3), 987–1006.

4. Lynott, D., Connell, L., Brysbaert, M., Brand, J., & Carney, J. (2020). The Lancaster Sensorimotor Norms: Multidimensional measures of perceptual and action strength for 40,000 English words. Behavior Research Methods, 52(3), 1271–1291.

5. Unkelbach, C., Fiedler, K., Bayer, M., Stegmüller, M., & Danner, D. (2008). Why positive information is processed faster: The density hypothesis. Journal of Personality and Social Psychology, 95(1), 36–49.