The historical origin of the word “twang” is thought to be an example of onomatopoeia: a word that sounds like what it represents. A twang is the kind of tinny, nasal sound produced by an instrument such as a banjo. It also refers to a type of speech usually associated with the English-speaking population of regions of the Midwestern and Southern United States, as well as several country music singers. The behavior required to produce a twang is complex: speakers apply a nasal quality and usually a rise in pitch to several vowels. Acquiring a twang requires physiological mechanisms ranging from perception (infants hear the speech of those in their environment) to a feedback mechanism (imitation and self-correction) and all the body parts used to produce vowel sounds: the tongue, nasal cavity, mouth, and more extensive pharyngeal structures.
Complex speech phenotypes may have a molecular basis within cells and tissues. Speaking with a twang likely involves several regions of the brain associated with speech and learning as well as those responsible for the coordinated muscular activity of the tongue and soft palette and other parts of the mouth and nasal cavities. Researchers have proposed various mechanisms to account for twang acquisition and performance among speakers. Since the behavior is acquired and can be lost again through training or relocation to an environment where speakers have a different “accent”, it is feasible that epigenetic alterations of genes must be involved. (An early study proposing a retrovirus has been discounted.) There is also some evidence that lesions can be associated with the gain of a temporary or long-term twang, or to the loss of a preexisting twang, which may help in identifying regions of the brain that are involved in its performance.
In a study in the latest issue of Nature Genetics, Terris et al. have studied epigenetic markers around genes that have been implicated in language perception and production in previous studies. They compare the status of these genes in regions of the brain thought to play a part in speech and pronunciation to regions less likely to be involved in these behaviors.
The list of candidate genes was obtained from a database hosted at the Quantitative Neuroscience Lab of Boston University (http://neurospeech.org/–sldb). Additional candidates were obtained through a computational analysis of the PubMed literature, harvesting articles meta-labeled with tags such as the following: twang, speech, language, pronunciation, and nasality.
Tissue samples were obtained from speakers who had undergone brain surgery and were judged to have a pronounced twang (or not) by a mixed audience of native (US-born) linguists. Results were compared between this group and five sets of controls: speakers who had never had a twang, those who had had a twang earlier in life but had lost it, native speakers of French (whose speech is not estimated to have a “twang” but is highly nasal), and a few individuals who had lost or acquired a twang through a stroke or other type of cerebral damage. Evaluations were performed using a standardized “Twang scale” developed at a school of performing arts in Los Angeles. (This program was developed to remove the twang of young actors.) Speakers were graded on a scale of 0 to 10 (0 = British accent; 10 = Bob Dylan).
The lab carried out a comprehensive analysis of methylation patterns across the genome from brain tissue samples from target and control regions for all five groups. The primary method used was bisulfite sequencing, which is based on the treatment of DNA with bisulfite. This causes a chemical conversion of cytosine residues to uracil, but only if the cytosines are non-methylated. Methylated cytosines are protected from the change. Comparing the sequences of treated vs. non-treated DNA permits a base-by-base readout of loci where Cs have been transformed to Us, and those which have not. The results from each group were combined and averaged and filtered for significance. They were compared to each other and to a mixed population of all groups.
The resulting patterns were compared on a chart, which revealed spikes (upward = higher methylation, downward = lower) at specific genomic locations. Both extremes are interesting because the twang phenotype might be due to either higher levels of methylation at particular loci, lower levels, or some combination.
Interestingly, the study revealed a number of significant differences between these patterns in “plus-twang” and “minus-twang” groups. The most extreme variation was found in cells of the superior temporal gyrus and primary auditory cortex, with somewhat smaller (although still significant) peaks in adjacent tissue of the brain region known as Wernicke’s area. The highest difference was found in a region ca. 1 Mb from the FOXP2 gene on chromosome 7, a gene which is highly implicated in many aspects of language acquisition and performance. A bioinformatics analysis of this region revealed a high statistical likelihood that it plays a regulatory role in FOXP2 activation, and contains putative FOX transcription factor binding sites. Both this region and the FOXP2 gene have closely related orthologs whose sequences and relative positions are well conserved between mice and humans. Follow-up studies in mice revealed that deleting the putative regulatory region inhibited expression of the orthologous gene in several areas of the brain, and resulted in a shift in squeaking pitch.
The authors remain cautious about their findings. In the paper’s discussion they report: “The exact molecular mechanisms underlying differential methylation remain to be understood, as does the quantitative significance of the identified loci in twang acquisition (or loss).” To address the mechanistic interplay between methylated regions, their regulators, and the twang-phenotype, the group has developed transgenic Cre mice in which particular methylated regions, methyltransferases, and methyl binding proteins can be deleted in a neuron-specific manner. Additionally, libraries of small molecules are being screened for specific effects on squeaking pitch as a phenotypic marker for twang in the mouse model.
Ideally, a potential twang modulator might be found among approved drugs or natural substances, which can be used to study the methylation status of the FOXP2-associated region. The next step would be to assemble a cohort of patients (twang-plus and twang-minus) who have already tried the drug or substance, checking to see whether this exposure has altered their speech patterns.
The author would like to thank Robert Zinzen for critical review of this article.