In 2011 Quentin Atkinson, a psychology professor at the University of Auckland, sparked debate in the linguistics community with a study published in Science called “Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa.” The basis of this theory is that all human languages are descended from a single African language or a small collection of closely related African languages. His primary evidence for this is the degree of phonemic diversity of languages around the world. After analyzing 504 languages, Atkinson found that languages farther from Africa have fewer phonemes. This theory caught the attention of many popular news sources such as the New York Times and the Global Post, and was often presented as factual. However, many linguists published responses Atkinson’s hypothesis, calling out discrepancies in his argument.
A founder effect in genetics occurs because a smaller group of people naturally has a smaller range of genetic diversity. Around the world it is seen that there is less genetic diversity farther from Africa. This was used as support for the out-of-Africa model of human expansion around the world, with Oceania and South America among the last places to be populated by modern humans. Atkinson applied the concept of a founder effect to linguistics, with phonemic diversity analogue to genetic diversity. Many linguists think that this comparison is unfounded. Ian Maddieson, a linguist at UC Berkeley, wrote in a response to Atkinson that “a subgroup of speakers of a given language does not use a subset of the phonemes, but all of them.” To lose a phoneme completely due to segmentation of a population is unlikely, and would require all the words containing that phoneme to be lost as well. There are processes by which phonemes are lost and gained, but a founder effect alone does not account for these changes. Claire Bowern, a linguist at Yale, also responded to Atkinson’s study, saying that “founder effects in genetics are robust because of the time it takes for genetic diversity to recover after a bottleneck. Linguistic change is much more rapid.” She further invalidates this comparison by explaining that genetic diversity is affected by factors other than founder effects. Any significant population loss, such as plague or famine, has caused loss in genetic diversity, but no observed linguistic changes.
Atkinson defines phonemes as the “perceptually distinct units of sound that differentiate words." While a phoneme inventory doesn’t represent how many different sounds are actually used in a language (allophones), it shows which sounds speakers think of as distinct. Atkinson used the data in the World atlas of linguistic structures (WALS) to compare number of phonemes in 504 world languages, finding that African languages have the highest phonemic diversity. In his article, Maddieson shed some light on how phonemes were counted in WALS and the process was not totally accurate. For example, only basic vowel qualities were considered, and factors like length and nasalization were not accounted for, even though these create distinguishable phonemes in many languages. This means that the data Atkinson used in his study is not representative of how some languages actually function, and underestimated how many phonemes some languages have.
Furthermore, the number of phonemes doesn’t tell us as much about a language’s complexity than phonotactic rules and possible number of syllables. In his analysis, Atkinson uses some aspects of phoneme inventory size and some aspects of phonological possibilities to determine what he labels as “phonemic diversity” — and it’s not entirely clear what this measure is. Maddieson argues that the content of a language’s phonemic inventory should be taken into account as well as its actual size. Some sounds are rarer than others; a larger inventory isn’t necessarily more “diverse” than a smaller inventory. Bowern also questions what exactly Atkinson means by “phonemic diversity.” She believes that founder effects could lead to “sub-phonemic variation” — that is, at the level of the allophone — but not the overall size of phoneme inventories. A small splintered-off group of speakers could conceivably change the ways a phoneme is actually articulated, but this would be classified as phonetic change linguistically, and would not affect the number of phonemes in a language.
The basis of Atkinson’s hypothesis lies in the correlation between group size and phoneme inventory: “small populations have fewer phonemes.” He expands this to conclude that phonemes “are more likely to be lost in small founder populations.” If this is true, it makes sense that African languages would have the most phonemes, since human life is accepted to have started there. Then as humans spread out phonemes were lost, leading to fewer phonemes at the end of the migration track in places like Oceania and South America. While phonemes are always being lost and gained in languages, there doesn’t seem to be any reason for a founder effect to reduce the phoneme inventory. Atkinson also claimed that “contact and borrowing between groups of speakers” was a primary source of phoneme gain. Don Ringe, a linguistics professor at the University of Pennsylvania, analyzed phoneme loss and gain in Indo-European languages in response to Atkinson’s theory. He had several findings that call Atkinson’s work into question. For one he found that “phoneme loss does not greatly outpace phoneme creation.” With this in mind, it seems unlikely that so many phonemes would be lost as humans migrated out of Africa. Additionally, Ringe found that “most of the new phonemes” in the Indo-European family “arose by language-internal processes; only 8% were acquired by borrowing.” Ringe concludes that “the Oceanic reduction of phoneme inventories is a result of internal factors,” specifically the phonological rules of such languages, “not the lack of language contact.” This directly contradicts Atkinson’s theory and its underlying assumption that more phonemes will be lost than gained over time if small populations are isolated. It seems there may be another factor at work here to explain the correlation between small population and fewer phonemes, but it’s clear that the first doesn’t lead to the second.
Ringe also points out that population size is almost irrelevant at the time in history when humans began leaving Africa: “no language could have had more than about 100,000 speakers at most." Even an “origin” language would not have had enough speakers to make a significant difference in size between the main population and fragmented founder groups. Östen Dahl, a linguist at Stockholm University, notes in his response to Atkinson that only size of phoneme inventory correlates with population; there is no decrease in complexity of the language with regard to morphology or syntax for instance. In fact, languages with fewer speakers often “have greater complexity of at least certain aspects of grammar." This suggests that languages do not “simplify” in smaller populations, and equating size of phoneme inventory with complexity is not a fair judgment.
Finally, a lot of problems with Atkinson’s study arise from the data he worked with and how he analyzed it. The 504 languages examined constitute only 5% of languages still spoken today. Furthermore, these are all modern languages; there is no reason to think that our current language distribution is comparable to what the earth was like during the first human expansions. Bowern raises the point that “all the largest language families in the present world are the result of expansions in the Holocene period,” which spans from about 10,000 years ago to today. Humans are thought to have left Africa long before the Holocene period, with current estimates ranging from 40,000 to 70,000 years ago.
In addition to Atkinson’s dataset being unrepresentative of the environment being studied, it is small and incomplete. Vladimir Pericliev, a Bulgarian linguist, points out that different patterns could have easily emerged from different data, more complete data, or different analysis techniques. Atkinson may have interpreted the data from the point of view of someone with the pre-existing knowledge that human life began in Africa. Pericliev demonstrates an alternative interpretation in his response: If we instead look at phonemic complexity from the perspective of child language acquisition, from generation to generation, we would arrive at the opposite conclusion — that language originated in Oceania and South America and grew more complex as humans migrated toward Africa. Of course this is ridiculous as we know scientifically that humans migrated in the opposite direction, but Pericliev was able to expose the ambiguity in the data through this example.
The origin of language has been a topic of speculation for centuries. Although Atkinson’s theory fits with our current understanding of early human migration patterns, there are numerous problems with his data and analysis. Several linguists have responded to his study, pointing out: the unwarranted analogy of founder effects in genetics to linguistics; the inconsistent definition of phonemes and phonemic diversity; the missing link between size of population and phoneme inventory; and other issues. Despite this, there is undoubtedly a lot to be learned by comparing language evolution to early human migration. We may not have enough data on early languages to come to a definitive conclusion, but this is a subject worth exploring.
Sources:
Atkinson, Quentin D. “Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa.” Science 15 (2011): 346-9. ScienceMag. Web. 8 Dec. 2014.
Bowern, Claire. "Out of Africa? the Logic of Phoneme Inventories and Founder Effects." Linguistic Typology 15.2 (2011): 207-16. ProQuest. Web. 8 Dec. 2014.
Dahl, Östen. "Are Small Languages More Or Less Complex than Big Ones?" Linguistic Typology 15.2 (2011): 171-5. ProQuest. Web. 8 Dec. 2014.
Maddieson, Ian, et al. "Geographical Distribution of Phonological Complexity." Linguistic Typology 15.2 (2011): 267-79. ProQuest. Web. 8 Dec. 2014.
Pericliev, Vladimir. "On Phonemic Diversity and the Origin of Language in Africa." Linguistic Typology 15.2 (2011): 217-21. ProQuest. Web. 8 Dec. 2014.
Ringe, Don. "A Pilot Study for an Investigation into Atkinson's Hypothesis." Linguistic Typology 15.2 (2011): 223-31. ProQuest. Web. 8 Dec. 2014.
No comments:
Post a Comment