The grammaticalization cycle of the progressive

A game-theoretic analysis

Abstract

A well-attested phenomenon in morpho-semantic change is known as the progressive cycle, which depicts a directed and cyclic pathway of a grammatical progressive marker through its emergence and disappearance inside the imperfective domain. Deo (2015) offers a model within the framework of evolutionary game theory to study the evolutionary dynamics of four preselected types of progressive-imperfective grammars. Based on her basic game-theoretic model, we investigate which types of grammars would emerge from the first principles in a population of agents under reinforcement learning. In our computational model, the actual progressive-imperfective cycle can be reconstructed from such atomic interactions between learner agents after the addition of several simple assumptions to the basic game-theoretic model.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11

Notes

  1. 1.

    The Swahili-specific glossing abbreviations are taken from Londfors (2003): NPX = Noun prefix, NC = Noun class, IND = Indicative.

  2. 2.

    Here a fast shift means that it takes a short time in comparison to the time a system stays in a stable state.

  3. 3.

    Note that these labels of forms and meanings differ from Deo’s Imperfective Game, which, however, does not change the structure of the game.

  4. 4.

    A contextual cue is a variable for any kind of additional information helping to suggest one reading apart from the form itself. This might be additional linguistic material, the type of the verb itself, or the situation of conversation. The model here abstracts from the concrete materialization of the cue.

  5. 5.

    Note that since the communicative success between speaker and hearer is context-independent, the δ-function can easily be applied for the context-related strategies by abstracting from contexts: \(\delta_{m}(\mathcal{S},\mathcal{H}) = 1\), iff \(\mathcal{H}(\mathcal{S}(m,c),c') = m\) for any \(c,c' \in C\). The utility functions \(U_{s}\) and \(U_{h}\) can be defined in the same way for context-related strategies.

  6. 6.

    Deo (2015) calls this system partially context dependent (pcd), due to the fact that the contextual cue \(c_{p}\) is still helpful for disambiguation, whereas the contextual cue \(c_{s}\) is not needed anymore, since here both meanings are disambiguated by both forms.

  7. 7.

    Admittedly, we have no proof that languages with an optional progressive marker actually use the progressive form according to the strategy outlined here, thus we are not aware of any study that analyzes in what contexts a speaker of a language with an optional progressive marker actually uses the progressive form. We believe that this gap in research is due to the fact that it is not easy to judge if a given context is more likely to license structural or phenomenal readings. Note that the contextual cues of our model are theoretical constructs for encompassing a complex mixture of all possible external linguistic and extra-linguistic cues licensing such a reading. But nevertheless, the OP system as defined in our model follows a particular line of thought: let us assume that we have a ZP language that uses solely contextual cues to disambiguate structural and phenomenal reading inside the imperfective domain. And then a new form appears that is used more and more frequently to phenomenal readings, but optionally next to the old form (ZP → OP shift). When would it be most useful to apply this new form? Admittedly, in situations with contextual cues that is more likely not to license a phenomenal reading, since in those situations with contextual cues that are likely to actually license a phenomenal reading, it is not necessary to use the new form: since the contextual cue helps to disambiguate successfully. Furthermore, it is known from several studies that emerging forms of a grammaticalization process are considered as marked forms (a good example is the German ‘am-Progressive’, which appears to be highly marked and barely considered as a grammatical form of Standard German). Such marked forms are generally used to express a non-prototypical meaning (note that this strategy follows a more general principle in pragmatics and language use: ‘Division of Pragmatic Labor’ (cf. Horn 1984)), and the non-prototypical case in a situation with a contextual cue that licenses a structural reading is the phenomenal meaning.

  8. 8.

    Admittedly, this is not the case for all types of games. But particularly for signaling games multiple studies exhibit that replicator dynamics and reinforcement learning approximate in the long run for diverse configurations, such as game parameters or learning parameters (cf. Barrett 2006; Argiento et al. 2009; Skyrms 2010; Huttegger and Zollman 2011; Mühlenbernd 2013).

  9. 9.

    This condition ensures that there was no change for the last 10 generations; this indicates that an evolutionary stable state is reached.

  10. 10.

    Note that strategies that involve pooling – e.g. speaker strategies that assign the same form to multiple meanings, or hearer strategies that assign the same meaning to multiple forms – are accordingly called pooling strategies.

  11. 11.

    The result is in line with Huttegger (2007), who showed that in binary signaling games, where states are not equiprobable, the pooling strategies have a positive basin of attraction. Furthermore, Enke et al. (2016), showed that in a setting where states are equiprobable, cf. with a cue probability \(c_{p} = 0.5\), all agents learn perfect signaling for both contextual cues.

  12. 12.

    That \(S_{2}\) and \(S_{10}\) emerge more often that \(S_{6}\) and \(S_{14}\) is due to the fact that the allocation \(m_{s} \rightarrow f_{old}\) is biased by being the most common one for context \(c_{s}\).

  13. 13.

    Additional tests showed that for any reduction of contextual cues above 7% a categorical system eventually emerged.

  14. 14.

    For example, both William Shakespeare (1564–1616) and the Irish novelist Laurence Sterne (1713–1768) used OP.

  15. 15.

    Note that for full cue access, as in Experiment I, the optional progressive system \(\langle S_{2}, H_{1} \rangle\) emerged at least for a part of the population.

  16. 16.

    The α-parameter is already used by Deo (2015) who references to Jäger (2007), who used a similar model for analyzing case marking systems with evolutionary game theory. Jäger interprets the α-parameter in terms of speakers priorities: how highly the speaker values linguistic clarity over signal costs. In the given model this parameter can then be interpreted as follows: when α is low then disambiguation by two explicit forms is highly valued, because there are no other means for disambiguation in that language, whereas when α is high, disambiguation by two explicit forms is not highly valued, because there is stronger support for disambiguation by other means. It can be assumed that due to language change the support of such ‘other means’ can vary, and so does the α-parameter in Experiment III. Note furthermore that the α-parameter is not defined for single agents, but a global parameter. In this way it represents changes in the linguistic system as a global construct. Admittedly, more realistic models might consider breaking the α-parameter down to an individual feature of agents which might be part of horizontal and vertical transmission. But we chose to abstract from that, especially since in a complete network structure (such as the one we use in our model) we highly expect individual α-parameters to align and therefore eventually behave as a global value.

  17. 17.

    Note that each simulation run eventually reached a population-wide one form system due to two factors: (i) when the alternating cost parameter exceeds a particular threshold, it makes a one form system more attractive than a two form system; and (ii) the input asymmetry – here implemented as childhood asymmetry – increases the total average probability \(P_{C}(c_{p})\) (note that without input asymmetry it was consistently very small: \(P_{C}(c_{p}) = 0.05\)), and therefore mitigates the difference between \(P_{C}(c_{p})\) and \(P_{C}(c_{s})\). As further experiments showed: such a mitigation supports the emergence of a homogeneous population where only one of both one form systems is used, contrasting with the result of Experiment III, where a mixed population emerged where both one form systems are used.

  18. 18.

    Note that the progressive cycle cannot always be reconstructed, cf. Fig. 11.

References

  1. Aarts, B., Close, J., & Wallis, S. (2010). Recent changes in the use of the progressive construction in English. In B. Cappelle & N. Wada (Eds.), Distinctions in English grammar (pp. 148–167). Tokyo: Kaitakusha.

    Google Scholar 

  2. Ahern, C., & Clark, R. (2014). Diachronic processes in language as signaling under conflicting interests. In E. Cartmill (Ed.), The evolution of language: proceedings of the tenth international conference on the evolution of language (pp. 25–32). Singapore: World Scientific.

    Google Scholar 

  3. Argiento, R., Pemantle, R., Skyrms, B., & Volkov, S. (2009). Learning to signal: analysis of a micro-level Reinforcement model. Stochastic Processes and their Applications, 119, 373–390.

    Article  Google Scholar 

  4. Ashton, E. (1949). Swahili grammar. London: Longman.

    Google Scholar 

  5. Barrett, J. A. (2006). Numerical simulations of the Lewis signaling game: learning strategies, pooling equilibria, and the evolution of grammar. Irvine: University of California. Technical Report MBS 06-09, Institute for Mathematical Behavioral Science.

    Google Scholar 

  6. Bonami, O. (2015). Periphrasis as collocation. Morphology, 25(1), 63–110.

    Article  Google Scholar 

  7. Börgers, T., & Sarin, R. (1997). Learning through reinforcement and replicator dynamics. Journal of Economic Theory, 77(1), 1–14.

    Article  Google Scholar 

  8. Brown, D., Chumakina, M., Corbett, G., Popova, G., & Spencer, A. (2012). Defining ‘periphrasis’: key notions. Morphology, 22(2), 233–275.

    Article  Google Scholar 

  9. Bush, R., & Mosteller, F. (1955). Stochastic models for learning. New York, NY: Wiley.

    Google Scholar 

  10. Bybee, J., Perkins, R., & Pagliuca, W. (1994). The evolution of grammar. Tense, aspect, and modality in the languages of the world. Chicago: Chicago University Press.

    Google Scholar 

  11. Cangelosi, A., & Parisi, D. (1998). The emergence of a ‘language’ in an evolving population of neural networks. Connection Science, 10(2), 83–97.

    Article  Google Scholar 

  12. Comrie, B. (1976). Aspect. Cambridge: Cambridge University Press.

    Google Scholar 

  13. Dahl, Ö. (1979). Typology of sentence negation. Linguistics, 17(1–2), 79–106.

    Google Scholar 

  14. Dahl, Ö. (1985). Tense and aspect systems. Oxford: Blackwell.

    Google Scholar 

  15. Deo, A. (2006). Tense and aspect in Indo-Aryan languages: variation and diachrony. Doctoral dissertation, Stanford University.

  16. Deo, A. (2015). The semantic and pragmatic underpinnings of grammaticalization paths: the progressive to imperfective shift. Semantics and Pragmatics, 8, 1–52, article 14.

    Article  Google Scholar 

  17. Ebert, K. (2000). Progressive markers in the Germanic languages. In O. Dahl (Ed.), Tense and aspect in the languages of Europe (pp. 605–653). Berlin: de Gruyter.

    Google Scholar 

  18. Enke, D., Mühlenbernd, R., & Yanovich, I. (2016). The emergence of the progressive to imperfective diachronic cycle in reinforcement-learning agents. In S. Roberts, C. Cuskley, L. McCrohon, L. Barceló-Coblijn, O. Feher, & T. Verhoef (Eds.), Proceedings of the 11th international conference on the evolution of language (Evolang 11).

    Google Scholar 

  19. Fagyal, Z., Swarup, S., Escobar, A. M., Gasser, L., & Lakkaraju, K. (2010). Centers and peripheries: network roles in language change. Lingua, 120, 2061–2079.

    Article  Google Scholar 

  20. Göksel, A., & Kerslake, C. (2015). Turkish: a comprehensive grammar. London: Routledge.

    Google Scholar 

  21. Goldsmith, J., & Woisetschlaeger, E. (1982). The logic of the English progressive. Linguistic Inquiry, 13(1), 79–89.

    Google Scholar 

  22. Hellinger, E. (1909). Neue Begründung der Theorie quadratischer Formen von unendlichvielen Veränderlichen. Journal für die reine und angewandte Mathematik, 136, 210–271.

    Google Scholar 

  23. Horn, L. (1984). Towards a new taxonomy of pragmatic inference: Q-based and R-based implicature. In D. Schiffrin (Ed.), Meaning, form, and use in context: linguistic applications (pp. 11–42). Washington: Georgetown University Press.

    Google Scholar 

  24. Huttegger, S. M. (2007). Evolution and the explanation of meaning. Philosophy of Science, 74(1), 1–27.

    Article  Google Scholar 

  25. Huttegger, S. M., & Zollman, K. J. S. (2011). Signaling games: dynamics of evolution and learning. In A. Benz, C. Ebert, G. Jäger, & R. van Rooij (Eds.), Language, games, and evolution (pp. 160–176). Berlin: Springer.

    Google Scholar 

  26. Jäger, G. (2004). Evolutionary game theory for linguists (Tech. Rep.). Stanford, University and University of Potsdam.

  27. Jäger, G. (2007). Evolutionary game theory and typology: a case study. Language, 83(1), 74–109.

    Article  Google Scholar 

  28. Jäger, G. (2014). What is a universal? On the explanatory potential of evolutionary game theory in linguistics. In E. McCready, K. Yabushita, & K. Yoshimoto (Eds.), Formal approaches to semantics and pragmatics (pp. 85–103). Dordrecht: Springer.

    Google Scholar 

  29. Ke, J., Gong, T., & Wang, W. S.-Y. (2008). Language change and social networks. Communications in Computational Physics, 3(4), 935–949.

    Google Scholar 

  30. Kirby, S. (2005). The evolution of meaning-space structure through iterated learning. In A. Cangelosi & C. Nehaniv (Eds.), Second international symposium on the emergence and evolution of linguistic communication (pp. 56–63).

    Google Scholar 

  31. Kirby, S., & Hurford, J. (1997). Learning, culture and evolution in the origin of linguistic constraints. In P. Husbands & I. Harvey (Eds.), Proceedings of the fourth European conference on artificial life (pp. 493–502). Cambridge: MIT Press.

    Google Scholar 

  32. Lewis, D. (1969). Convention. Cambridge: Harvard University Press.

    Google Scholar 

  33. Li, P., Maher, S., Newmark, E., & Hurley, J. (2001). The role of parental input in the acquisition of tense-aspect morphology. Journal of Cognitive Science, 2(1), 119–143.

    Google Scholar 

  34. Londfors, A.-L. (2003). Tense and aspect in Swahil. Unpublished manuscript.

  35. MacWhinney, B. (2000). The CHILDES project: tools for analyzing talk. Hillsdale: Erlbaum.

    Google Scholar 

  36. Meillet, A. (1909). Sur la disparition des formes simples du prétérit. In Linguistique historique et linguistique générale (pp. 149–158). Geneve/Paris: Slatkine/Champion.

    Google Scholar 

  37. Mühlenbernd, R. (2011). Learning with neighbours. Synthese, 183(S1), 87–109.

    Article  Google Scholar 

  38. Mühlenbernd, R. (2013). Signals and the structure of societies. Doctoral Thesis, University of Tübingen, TOBIAS-Lib Online Publication.

  39. Nettle, D. (1999). Using social impact theory to simulate language change. Lingua, 108(2–3), 95–117.

    Article  Google Scholar 

  40. Page, K. M., & Nowak, M. A. (2002). Unifying evolutionary dynamics. Journal of theoretical biology, 219(1), 93–98.

    Article  Google Scholar 

  41. Polomé, E. (1967). Swahili language handbook. Washington D.C.: Center for Applied Linguistics.

    Google Scholar 

  42. Rosenbach, A. (2008). Language change as cultural evolution: Evolutionary approaches to language change. In R. Eckardt, G. Jäger, & T. Veenstra (Eds.), Variation, selection, development: probing the evolutionary model of language change (pp. 23–74). Berlin: de Gruyter.

    Google Scholar 

  43. Roth, A. E., & Erev, I. (1995). Learning in extensive-form games: experimental data and simple dynamic models in the intermediate term. Games and Economic Behaviour, 8, 164–212.

    Article  Google Scholar 

  44. Schaden, G. (2012). Modelling the ‘aoristic drift of the present perfect’ as inflation: an essay in historical pragmatics. International Review of Pragmatics, 4, 261–292.

    Article  Google Scholar 

  45. Skyrms, B. (1996). Evolution of the social contract. Cambridge: Cambridge University Press.

    Google Scholar 

  46. Skyrms, B. (2010). Signals: evolution, learning, and information. London: Oxford University Press.

    Google Scholar 

  47. Smith, N. (2002). Ever moving on? The progressive in recent British English. In P. Peters, P. Collins, & A. Smith (Eds.), New frontiers in corpus research (pp. 317–330). Amsterdam: Rodopi.

    Google Scholar 

  48. Taylor, P. D., & Jonker, L. B. (1978). Evolutionary stable strategies and game dynamics. Mathematical Biosciences, 40(1), 145–156.

    Article  Google Scholar 

  49. Wärneryd, K. (1993). Cheap talk, coordination, and evolutionary stability. Games and Economic Behavior, 5, 532–546.

    Article  Google Scholar 

  50. Williams, C. (2002). Non-progressive and progressive aspect in English. Fasano: Schena.

    Google Scholar 

  51. Yanovich, I. (unpublished). Analysing imperfective games. Unpublished manuscript. Eberhard Karls Universität Tübingen.

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to Roland Mühlenbernd.

Additional information

We express our thanks to the audiences of the Tübingen, Munich, and Stanford seminars, of the 17th International Morphology Meeting (Vienna), PLC 40 (Philadelphia), Evolang XI (New Orleans), Germanic Linguistics Roundtable (Berkeley), Diskussionsforum Linguistik (Zurich), and SLE 49 (Naples). Special thanks are owed to Christopher Ahern, Phillip Alday, Sara Sanchez Alonso, Christian Bentz, Heather Burnett, Ashwini Deo, Dieter Gunkel, Olav Hackstein, Gerhard Jäger, Guido Seiler, Thilo Weber, and Igor Yanovich. Any errors that remain are our own responsibility. We gratefully acknowledge the support of the ERC under project EVOLAEMP, http://www.evolaemp.uni-tuebingen.de (RM), the Department of Linguistics at Eberhard Karls University, Tübingen (RM) and the Institute of German Philology of LMU Munich (DE).

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Mühlenbernd, R., Enke, D. The grammaticalization cycle of the progressive. Morphology 27, 497–526 (2017). https://doi.org/10.1007/s11525-017-9310-2

Download citation

Keywords

  • Morpho-semantic change
  • Progressive cycle
  • Evolutionary dynamics
  • Game theory
  • Reinforcement learning