Abstract:
We experimentally investigate the emergence and persistence of monolingualism in a sender-receiver game with no conflict of interest, where the sender sends a message (without prior meaning) regarding a privately observed state to two other receivers who need to choose actions to match the state.  Subjects simultaneously play both roles of sender and receiver in a three-member society.  Through repeated play, they learn how to ``speak'' and ``listen'' to their group members.  We also employ the strategy method for senders to elicit the message for each state and incentivize the use of a same language.  This design allows us to observe ``multilingualism'' and ``monolingualism'' in the lab. 

To induce convergence to structured languages, we introduce half of the states first for 10 rounds before introducing the rest, which are 90-degree rotations of the initial states for the treatment group. The treatment groups indeed converge to structured languages more often. The receiver matches the true state twice as often when the language is structured, compared to non-structured. 27 out of 32 three-member societies converge to monolingualism in 40 rounds. After the same language emerges, we observe the learning process of ``immigrants'' who move to another society. Results show that immigrants learn faster in societies with more structured languages.