<P> We know little about the timing of language's emergence in our species . Unlike writing, speech leaves no material trace, making it archaeologically invisible . Lacking direct linguistic evidence, specialists in human origins have resorted to the study of anatomical features and genes arguably associated with speech production . While such studies may tell us whether pre-modern Homo species had speech capacities, we still don't know whether they actually spoke . While no one doubts that they communicated vocally, the anatomical and genetic data lack the resolution necessary to differentiate proto - language from speech . </P> <P> Using statistical methods to estimate the time required to achieve the current spread and diversity in modern languages today, Johanna Nichols--a linguist at the University of California, Berkeley--argued in 1998 that vocal languages must have begun diversifying in our species at least 100,000 years ago . </P> <P> More recently--in 2012--anthropologists Charles Perreault and Sarah Mathew used phonemic diversity to suggest a date consistent with this . "Phonemic diversity" denotes the number of perceptually distinct units of sound--consonants, vowels and tones--in a language . The current worldwide pattern of phonemic diversity potentially contains the statistical signal of the expansion of modern Homo sapiens out of Africa, beginning around 60 - 70 thousand years ago . Some scholars argue that phonemic diversity evolves slowly and can be used as a clock to calculate how long the oldest African languages would have to have been around in order to accumulate the number of phonemes they possess today . As human populations left Africa and expanded into the rest of the world, they underwent a series of bottlenecks--points at which only a very small population survived to colonise a new continent or region . Allegedly such population crash led to a corresponding reduction in genetic, phenotypic and phonemic diversity . African languages today have some of the largest phonemic inventories in the world, while the smallest inventories are found in South America and Oceania, some of the last regions of the globe to be colonized . For example, Rotokas, a language of New Guinea, and Pirahã, spoken in South America, both have just 11 phonemes, while! Xun, a language spoken in Southern Africa has 141 phonemes . The authors use a natural experiment--the colonization of mainland Southeast Asia on the one hand, the long - isolated Andaman Islands on the other--to estimate the rate at which phonemic diversity increases through time . Using this rate, they estimate that the world's languages date back to the Middle Stone Age in Africa, sometime between 350 thousand and 150 thousand years ago . This corresponds to the speciation event which gave rise to Homo sapiens . </P> <P> These and similar studies have however been criticized by linguists who argue that they are based on a flawed analogy between genes and phonemes, since phonemes are frequently transferred laterally between languages unlike genes, and on a flawed sampling of the world's languages, since both Oceania and the Americas also contain languages with very high numbers of phonemes, and Africa contains languages with very few . They argue that the actual distribution of phonemic diversity in the world reflects recent language contact and not deep language history - since it is well demonstrated that languages can lose or gain many phonemes over very short periods . In other words, there is no valid linguistic reason to expect genetic founder effects to influence phonemic diversity . </P>

When do most specialists conclude that language as we know it appeared