Cognitive scientists have good news for linguistic purists terrified about the corruption of their mother tongue.
Using an artificial language in a carefully controlled laboratory experiment, a team from the University of Rochester and Georgetown University has found that many changes to language are simply the brain’s way of ensuring that communication is as precise and concise as possible.
“Our research shows that humans choose to reshape language when the structure is either overly redundant or confusing,” says T. Florian Jaeger, the Wilmot Assistant Professor of the Sciences at Rochester and co-author of a study published in the Proceedings of the National Academy of Sciences Oct. 15. “This study suggests that we prefer languages that on average convey information efficiently, striking a balance between effort and clarity.”
The brain’s tendency toward efficient communication may also be an underlying reason that many human languages are structurally similar, says lead author Maryia Fedzechkina, a doctoral candidate at Rochester. Over and over, linguists have identified nearly identical grammatical conventions in seemingly unrelated languages scattered throughout the globe. For decades, linguists have debated the meaning of such similarities: are recurrent structures artifacts of distant common origins, are they simply random accidents, or do they reflect fundamental aspects of human cognition?
This study supports the latter, says co-author Elissa L. Newport, professor of neurology and director of the Center for Brain Plasticity and Recovery at Georgetown, and the former George Eastman Professor of Brain and Cognitive Sciences at Rochester. “The bias language learners have toward efficiency and clarity acts as a filter as languages are transmitted from one generation of learners to another,” she says. Alterations to language are introduced through many avenues, including the influence of other languages and changes in accents or pronunciation. “But this research finds that learners shift the language in ways that make it better – easier to use and more suitable for communication,” says Newport. That process also leads to the recurrent patterns across languages.
To observe the language acquisition process, the team created two miniature artificial languages that use suffixes on nouns to indicate subject or object. These “case markers” are common to Spanish, Russian, and other languages, but not English. In two experiments, 40 undergraduates, whose only language was English, learned the eight verbs, 15 nouns, and grammatical structure of the artificial languages. The training was spaced over four 45-minute sessions and consisted of computer images, short animated clips, and audio recordings. Then participants were asked to describe a novel action clip using their newly learned language.
When faced with sentence constructions that could be confusing or ambiguous, the language learners in both experiments chose to alter the rules of the language they were taught in order to make their meaning clearer. They used case markers more often when the meaning of the subject and object might otherwise have caused unintended interpretations. So for example, a sentence like “Man hits wall,” is typical because the subject is a person and the object is a thing. But the sentence “Wall hits man,” as when a wall falls on top of a man, is atypical and confusing since the subject is a thing and the object is a person.
The results, write the authors, provide evidence that humans seek a balance between clarity and ease. Participants could have chosen to be maximally clear by always providing the case markers. Alternatively, they could have chosen to be maximally succinct by never providing the case markers. They did neither. Instead, they provided case-markers more often for those sentences that would otherwise have been more likely to be confused.
The findings also support the idea that language learners introduce common patterns, also known as linguistic universals, conclude the authors. The optional case marking that participants introduced in this experiment closely mirrors naturally occurring patterns in Japanese and Korean—when animate objects and inanimate subjects are more likely to receive case markings.
The history of English itself might reflect these deep principles of how we learn language, says Jaeger. Old English had cases and relatively free word order, as is still true for German. But at some point pronunciation changes began to obscure the case endings, creating ambiguity. In contemporary English, word order has become the primary signal by which speakers could decode the meaning, he says.
“Language acquisition can repair changes in languages to insure they don’t undermine communication,” says Fedzechkina. In light of these findings, new generations can perhaps be seen as renewing language, rather than corrupting it, she adds.
By the same token, says Jaeger, many elements of informal speech can be interpreted as rising from the brain’s bias toward efficiency. “When people turn ‘automobile’ into ‘auto,’ use informal contractions, swallow syllables, or take other linguistic shortcuts, the same principles are at work,” he says. Recent research has shown that these types of shortcuts appear only when their meaning is easily inferable from the context, he adds.
Hi Abel,
good question about the monolinguals, but we actually had a very good reason to limit ourselves to the monolinguals. English does only have case on pronouns, i.e. no productive case-system (kinda like Spanish, aehem). We wanted to be as certain as possible that our learners do not transfer whatever bias we see simply by mapping surface properties of their native language to our artificial language. That’s why pretty thoroughly screened for expertise in other languages.
English, like any other language I am aware of for which this tendency has been studied, does have a statistical contingency between animacy and the probability of being a grammatical object (or rather — a contingency between animacy and thematic role). We wanted to know whether this tendency –which we assume to be universal across languages, as it strikes us to be largely a property of the world around us, rather than languages– transfers to the artificial language AND whether this then caused learners to deviate from the input towards a language code (grammar) that make more efficient use of case-marking.
That being said, our lab is actively involved in collaborations that seek to re-run such experiments on speakers with different language background. Hal Tily (formerly a graduate student at Stanford, now with Nuance Technologies) has developed a tool for artificial language learning over the web, which we hope will allow us to pursue this line of research. Let me know if you’re interested and I’ll post some references here.
HTH.
Florian
Hi Chris,
thanks for pointing that out — what a bad mistake. Of course, we know Spanish has no case (except on pronouns). This must have somehow escaped our attention during restructuring of the manuscript.
Florian (one of the authors of the paper)
Actually, on second thought: we wrote that Spanish has differential case-marking. That’s arguably true. At least that’s one common interpretation of the differential use of the preposition before human and non-human object referents. These aren’t suffixes, but neither did we write that. Haha, you confused me!
The same experiment should be conducted using test subjects with various linguistic backgrounds before drawing any definite conclusions. Bilingualism is more common, why use only monolingual English speakers? Certainly, a start, but not enough.
There is a mistake in this article. Spanish does not use “case markers” as there is no grammatical case in Spanish. The endings of Spanish do change according to gender, but that’s not the same as case.
I meant to write: “The endings of Spanish nouns do change according to gender, …”
We wrote that Spanish has differential case-marking. That’s arguably true. At least that’s one common interpretation of the differential use of the preposition before human and non-human object referents. These aren’t suffixes, but neither did we write that. Haha, you confused me!
Hi Florian,
Thanks for taking the time to reply.
I should preface my remarks by saying that I don’t have access to the original paper, so I’m only commenting on the information presented above.
I quote:
“To observe the language acquisition process, the team created two miniature artificial languages that use suffixes on nouns to indicate subject or object. These “case markers” are common to Spanish, Russian, and other languages, but not English.”
This clearly suggests that Spanish uses suffixes on nouns to indicate grammatical case. Perhaps the original paper worded it differently, but as presented above it’s undoubtedly misleading.
Furthermore, it’s hard to argue that the differential use of prepositions in Spanish for human and non-human objects is a case marking. It’s only a case marking to the same extent as any preposition is a case marking, in the sense that a preposition indicates the function of the word which follows.
The function of nouns in both Spanish and English is indicated primarily by their word order, supported by prepositions. There are many other languages which have well-established case systems for nouns such as German, Finnish, etc, so I would have thought one of those would have made a better example to go with Russian.
Best,
Chris
http://www.speaklanguages.co.uk
Hi Chris,
thanks for clarifying. I see now how the first report about our paper is misleading. Here’s what we say in the paper:
“As a test case, we investigate the acquisition of differential case
marking systems (43–45) found in a large number of natural languages (e.g., Sinhalese, Spanish, Russian, and Hindi). Case
marking is the addition of markers on nouns, typically prefixes or
suffixes added to the noun stem, for example, to indicate which
noun is the subject and which is the direct object of the verb.”
I can see how this was confusing for the reporter and we should have caught the mistake in the report. In the paper though, we are talking about differential case-marking systems and Spanish is often considered to be an example of such a system (see, e.g., the references on wikipedia). In the final sentence of our quote we say “typically”, which probably mislead folks into reading that we were saying that Spanish is a language with case-marking affixes, which it isn’t.
In any case, thanks for pointing out this confusing statement. Btw, if you contact us, we’re happy to send you the PNAS paper (I can’t post it online for copyright reasons).
Florian