Human Languages with Greater Information Density Have Higher Communication Speed but Lower Conversation Breadth

Languages with Higher Information Density Exhibit Faster Communication but Lower Conversational Breadth

Background

Human languages exhibit extensive differences in encoding information. These differences have been studied extensively within certain semantic domains, such as time, space, color, human body parts, and activities. However, there has been little in-depth research on the global structure of semantic information and its relationship with human communication. The authors first propose that, across a sample of about 1,000 languages, there is a significant variation in information encoding density. They then explore how languages with higher information density more densely configure semantic information. Finally, they investigate the relationship between information density and communication patterns, finding that languages with higher information density tend to communicate faster but cover narrower conceptual ranges in conversations.

Paper Source

This article was written by Pedro Aceves and James A. Evans and published in the April 2024 issue of Nature Human Behaviour. Pedro Aceves is affiliated with the Department of Management and Organization at Johns Hopkins University’s Carey Business School, while James A. Evans is a professor in the Department of Sociology and Knowledge Lab at the University of Chicago and a research fellow at the Santa Fe Institute.

Research Process

1. Measurement of Information Encoding Density

The study used 18 diverse parallel translation corpora, encompassing around 998 languages from 101 language families. Using Huffman coding algorithms, each language’s vocabulary in the given translations was converted to the most efficient binary code, and the number of bits for each document was calculated. This generated a standardized measure of information density for each language, enabling comparisons across all corpora.

2. Measurement of Semantic Density

Next, the study calculated each language’s semantic density using neural word embedding models. These models train a high-dimensional vector space based on the co-occurrence frequency of words in texts, where syntactically and semantically similar words are usually close to each other in the space. The authors found that languages with higher information density also tend to have higher semantic density, meaning the word meanings are more polysemous, and the associations between concepts are higher.

3. Measurement of Communication Speed

To verify whether languages with higher information density can indeed transmit information faster, the researchers used the duration of Bible audio files as test objects. These audio files cover 265 languages. The results showed that languages with higher information density require less time to convey the same information, which aligns with information theory expectations.

4. Measurement of Semantic Breadth in Actual Conversations

Researchers analyzed the text of over 6,000 natural conversations in 14 languages, calculating the conversational conceptual breadth using word embedding models—i.e., the range of semantic space covered in the conversation. The results showed that languages with higher information density tend to cover narrower conceptual ranges in actual conversations but explore topics in greater depth, implying more focused discussions on specific themes from multiple angles.

5. Measurement of Semantic Breadth in Collective Knowledge Output

Finally, the researchers analyzed over 95,000 articles written in different languages on Wikipedia to study the conceptual breadth of collective knowledge output. Similarly, they found that articles written in languages with higher information density are more conceptually focused, further confirming that collective communication in these languages tends to deeply explore smaller conceptual spaces.

Research Results

This research, utilizing large-scale computation and artificial intelligence technology, demonstrated significant differences in language information density and revealed important relationships between this density, semantic density, and human communication patterns. The results show that languages with higher information density can transmit information faster and have narrower conceptual ranges for conversations and knowledge output but engage in deeper discussions. These findings highlight the substantial impact of language structure on human interaction and social behavior.

Research Significance

This study not only deepens our understanding of differences in language encoding methods but also reveals how language structure influences communication speed and breadth of communication content. It expands the concept of linguistic relativity, extending it from mere cognitive frameworks to realms of communication, interaction, collaboration, and collective behavior. This provides new directions for future research on how language information density plays a role in broader social interactions and collective performance.

Research Highlights

• Significant Differences in Information Density: The study recorded extensive differences in information density across languages worldwide.
• Frequent Use and Polysemy: Languages with higher information density have vocabulary that is frequently used in different contexts and has polysemous meanings.
• Faster Communication: Languages with higher information density can transmit information more quickly within a fixed bandwidth.
• In-depth Discussion: Languages with higher information density tend to engage in narrower but deeper discussions in conversations and knowledge output.

Through these research steps, the authors provide new perspectives for understanding how languages influence our daily interactions and social structures. This study lays the foundation for future research, inspiring the exploration of language information density and its broader impacts on social interaction and collective performance.