A Mechanism of the Analyticity Development in a Language. Elements of the Theory and the Computer Experiment

V.n. Beloozerov, A.a. PolikarpovCOLING 1973

TL;DR

This paper proposes a cybernetic model where language complexity is a function of input and output information stream complexity. They introduce a mathematical model to explore how language analyticization, a simplification of grammatical structures, arises from decreased mutual comprehension during language contact, aiming to optimize communication efficiency.

What Problem Does This Paper Address?

The paper tackles the fundamental question of how and why languages evolve towards greater analyticity, a phenomenon observed in linguistic typology where complex grammatical structures are replaced by simpler, more analytical ones (e.g., using prepositions instead of case endings). While the link between language contact and analyticization has been noted, the underlying mechanisms and transitional stages remain unclear. The authors frame this problem within a general cybernetic theory of communication, viewing language as a system that regulates information flow. They hypothesize that changes in language complexity, specifically the trend towards analyticization, are driven by the need to maintain effective communication in the face of "noise" or reduced comprehension, particularly in situations involving language contact.

Methods

The authors propose a theoretical framework based on cybernetic principles, defining language complexity (L) as a function of input information stream complexity (S) and output information stream complexity (k), represented by the formula L = S/k.

To investigate the mechanism of analyticization, they develop a hypothetical model of communication. This model simulates a speaker attempting to convey a "sense" to a listener using a sequence of signs. The core of the model involves:

  • Probabilistic Sign Comprehension: Each sign has a probability P of being understood by the listener, with "interferences" (noise) present.
  • Message Construction: The speaker selects a subset of signs (m) from a larger pool to form an utterance.
  • Comprehension Threshold: The sense is considered understood if the number of understood signs exceeds a threshold (N).
  • Iterative Communication: If comprehension fails, the speaker repeats the process with new sign selections until the sense is conveyed.
  • Optimization Goal: The model aims to minimize the expected duration of communication, which is evaluated by the expected number of signs (K) produced across all utterances.

This model allows for the calculation of optimal redundancy (Aopt) and the exploration of how communication efficiency (measured by expected duration K) changes with varying parameters like sign comprehension probability (P), message length (N), and the number of signs chosen (m).

Datasets

Not discussed in the paper. The study relies on a theoretical model and computer simulations rather than empirical datasets.

Key Insights and Ideas

  • Language as an Information Regulator: The paper frames language as a crucial component in regulating information streams within social systems, directly linking its organizational complexity to the complexity of information it processes and produces.
  • Analyticization as an Efficiency Strategy: The core hypothesis is that analyticization is not an arbitrary linguistic change but a functional adaptation to maintain communication efficiency. When comprehension decreases (due to factors like language contact), languages simplify their structure to compensate.
  • The Role of Redundancy: The model highlights the complex relationship between redundancy and communication. While increased redundancy can aid comprehension, the model suggests an "optimum redundancy" that minimizes communication time. Exceeding this optimum can paradoxically lead to decreased comprehension, driving analyticization.
  • Synthetic vs. Analytic Language Trade-offs: The authors suggest that synthetic language organization (e.g., agglutinative languages) might be preferable for long-term memory efficiency, while analytic languages (e.g., isolating languages) economize on memory by requiring more on-the-fly construction of messages, potentially at the cost of precision.
  • Quantitative Approach to Linguistic Evolution: The paper pioneers a quantitative, cybernetic approach to understanding linguistic evolution, moving beyond purely qualitative descriptions of language change.

Key Results

The computer simulations based on the proposed model yielded the following:

  • Existence of Optimal Redundancy: For any given information volume and comprehension probability, there exists an optimal level of redundancy that minimizes the expected time for communication. This is illustrated in Figure 1, showing U-shaped curves for expected communication time (K) as a function of message length (m) for different comprehension probabilities (P).
  • Analyticization and Comprehension: The study suggests a correlation where higher degrees of analytic features in a language are associated with a lower level of comprehension. This implies that the drive for analyticization, while potentially optimizing certain aspects of communication, might come at the cost of overall clarity or ease of understanding.
  • Figure 2: This figure illustrates the relationship between optimal redundancy (Aopt) and comprehension probability (P), showing that as comprehension probability increases, the optimal redundancy decreases.

Limitations and Future Work

  • Model Simplification: The authors acknowledge that their model is a simplification and that a more complex model could be constructed.
  • Adaptation of Communicants: A key area for future research is incorporating the adaptation of communicants to each other during the communication process. This dynamic aspect of interaction is not fully captured in the current model.
  • Further Elaboration and Checking: Future work should involve further elaboration of the model and computerized checking of its predictions.
  • Connecting to Sociolinguistic Factors: The authors suggest connecting the model to parameters like the degree of divergence between contacting languages, the proportion of non-native speakers, and the intensity of language contacts. This would lead to a more comprehensive sociolinguistic model.
  • Predictive Power: The ultimate goal is to build an explanatory sociolinguistics theory that can predict language evolution and societal development by understanding the correlations between language states and societal states.

Summary updated

BibTeX
@inproceedings{beloozerov1973mechanism,
  author = {V.N. Beloozerov and A.A. Polikarpov},
  title = {{A Mechanism of the Analyticity Development in a Language. Elements of the Theory and the Computer Experiment}},
  booktitle = {International Conference on Computational Linguistics},
  year = {1973},
}
Figure