Select your words carefully.

A dvanced speech and language skills are among the most striking traits that separate humans from all other forms of life. Now two groups, Enard et al (2002) and Zhang et al (2002), have taken a step forward towards identifying recent genetic changes that may provide the underpinnings of these uniquely human characteristics. A major goal of human evolutionary biology is to identify the factors that contributed to the development of complex human traits. Both genetic and environmental factors influence our capacity for speech and language: one of the most striking human-specific traits. These new studies use statistical evolutionary genetics methods to identify one of the genes that may have played a key role in the evolution of the neural circuitry involved in human speech and language (Bishop, 2002; Fisher et al, 2003). Both studies are based on the premise that an advantageous allele of a language-related gene will have fixed in the human lineage around the time of the emergence of three human language and left behind signatures of genetic selection in modern human DNA. The FOXP2 gene is an attractive candidate gene to screen for selection since it is involved in a rare inherited human speech and language disorder (Fisher et al, 2003). For example, carriers of a missense mutation (Arg553His) have impairments in mouth control movements required for speech, written language, and in the comprehension and production of grammar (Lai et al, 2001; Fisher et al, 2003). Although its exact role in language development is unclear, it is known that FOXP2 is a member of the forkhead/ winged-helix (FOX) family of transcription factors that is expressed in fetal and adult human brain. Its expression pattern in mouse also implicates it in the development of lung, intestine, and cardiovascular tissue (Fisher et al, 2003). FOXP2 amino-acid sequences are very conserved in mammals: it is among the 5% most-conserved proteins in a survey of 1880 pairs of human-rodent gene pairs (Enard et al, 2002). This makes it all the more significant that two of the three amino-acid differences between human and mice occurred in the human lineage after our separation from chimpanzees and bonobos, approximately 5–7 million years ago (Figure 1). In both the new studies, statistical tests showed an increased rate of nonsynonymous (amino-acid altering) to synonymous (silent) sequence changes only in the human lineages compared to other mammals. The two amino-acid differences between humans and mouse are closely spaced together (Thr303Asn and Asn325Ser) in exon 7. Zhang et al (2002) surveyed 20 mammalian FOXP2 orthologs, finding that one of these changes (Asn325Ser) also occurs in carnivore lineages. Thus, it is difficult to ascertain whether the Thr303Asn change alone was critical for selection or the presence of both the amino-acid changes were needed. Alternatively, both amino-acid differences may not significantly affect FOXP2 function, and cis-acting elements in the promoter region or motifs in the 30and 50-UTRs may be under selection. One can test for selection in FOXP2 in humans by comparing sequence variation in the gene between and within human and nonhuman primates (eg chimpanzees, bonobos, and gorillas). If an advantageous FOXP2 allele became fixed in early human history, it should have caused a local loss of variation (selective sweep) limited to the selected site and its flanking region (Smith and Haigh, 1974). Selective sweeps in the FOXP2 gene coinciding with the development of speech and language in modern humans in the past few hundred thousand years should be detectable because most existing human polymorphism arose between 0.5 and 2 million years ago and human population structure does not predate 200 000 years (Kreitman, 2000; Przeworski et al, 2000). Both studies tested for selection by comparing levels of polymorphism near the FOXP2 locus to those at other human loci. The introns flanking FOXP2 exon 7 are less variable than most neutral noncoding regions examined in humans to date (Harris and Hey, 2001; Enard et al, 2002; Zhang et al, 2002) as expected if a selective sweep has occurred recently. Furthermore, much of the polymorphism that is present consists of a statistical overabundance of low-frequency alleles. The latter observation would also be expected, given that such alleles should have arisen since the fixation of the advantageous sequence change within or near-