Unity and disunity in evolutionary sciences: process-based analogies open common research avenues for biology and linguistics
- 1.8k Downloads
For a long time biologists and linguists have been noticing surprising similarities between the evolution of life forms and languages. Most of the proposed analogies have been rejected. Some, however, have persisted, and some even turned out to be fruitful, inspiring the transfer of methods and models between biology and linguistics up to today. Most proposed analogies were based on a comparison of the research objects rather than the processes that shaped their evolution. Focusing on process-based analogies, however, has the advantage of minimizing the risk of overstating similarities, while at the same time reflecting the common strategy to use processes to explain the evolution of complexity in both fields.
We compared important evolutionary processes in biology and linguistics and identified processes specific to only one of the two disciplines as well as processes which seem to be analogous, potentially reflecting core evolutionary processes. These new process-based analogies support novel methodological transfer, expanding the application range of biological methods to the field of historical linguistics. We illustrate this by showing (i) how methods dealing with incomplete lineage sorting offer an introgression-free framework to analyze highly mosaic word distributions across languages; (ii) how sequence similarity networks can be used to identify composite and borrowed words across different languages; (iii) how research on partial homology can inspire new methods and models in both fields; and (iv) how constructive neutral evolution provides an original framework for analyzing convergent evolution in languages resulting from common descent (Sapir’s drift).
Apart from new analogies between evolutionary processes, we also identified processes which are specific to either biology or linguistics. This shows that general evolution cannot be studied from within one discipline alone. In order to get a full picture of evolution, biologists and linguists need to complement their studies, trying to identify cross-disciplinary and discipline-specific evolutionary processes. The fact that we found many process-based analogies favoring transfer from biology to linguistics further shows that certain biological methods and models have a broader scope than previously recognized. This opens fruitful paths for collaboration between the two disciplines.
This article was reviewed by W. Ford Doolittle and Eugene V. Koonin.
KeywordsProcess-based analogies Language evolution Protein assembly Word formation Lateral transfer Constructive neutral evolution Similarity networks Incomplete lineage sorting
Biological objects on Earth have been evolving for billions of years. The origin of language evolution dates back to only about 200 000 years ago. The specific aspects of the evolution of life forms and the evolution of languages are traditionally investigated by the disciplines of evolutionary biology and historical linguistics. The research objects of the two disciplines differ greatly. Biology deals with substantial objects, that is, objects with a concrete physical manifestation. Languages, on the other hand, are ‘products of the human mind’ (, p. 144). They are intellectual objects (, p. 72), that is, objects whose manifestation is based on the interaction between humans. They are realized physically, be it when they are spoken or written down, but their realization is dependent on the existence of individuals who speak and understand them, and in this way, language systems are constantly being reconstructed by new speakers who learn them .
Apart from similar models and methods developed independently, there was and is also a considerable amount of explicit transfers between the two disciplines. An early example is the intimate intellectual exchange on Darwin’s evolutionary theory and its implications for the study of languages between the biologist Ernst Haeckel (1834–1919) and the linguist August Schleicher (1821–1861) . According to this correspondence, it was Haeckel who brought Schleicher’s attention to the work of Darwin. Schleicher was deeply impressed by the similarities of the research objects in such different domains (, p. 6). He emphasized, however, also that these parallels would only hold for the essential features, not for the details (, p. 29). Haeckel, in turn, took inspiration from Schleicher’s language tree diagrams to promote evolutionary tree drawing in biology (, p. 300).
In the 20th century, especially the early work on genetics, not long after the correct modelling of the structure of DNA by Watson and Crick , was characterized by a strong linguistic influence. This is reflected in the multitude of linguistic terms, like ‘alphabet’ and ‘word’  or ‘translation’ , which were used to describe biological phenomena in the biological domain . While, as indicated by Eugene V. Koonin (one of the reviewers of this manuscript), the majority of these terms reflected mere metaphors of which only a minority became later integrated into the standard terminology of biology (see also ), we can also find examples for the explicit transfer of linguistic methods and theories to the biological domain. Thus, up to today, the theory of formal grammar  plays an important role in addressing certain problems in bioinformatics , like RNA folding and protein structure analysis, and it is not uncommon for biological textbooks on sequence comparison to also include a chapter on formal grammars (, pp. 233-259). This influence is not restricted to classical models of grammar . Advanced models, like tree adjoining grammar, have likewise been used for RNA structure prediction , and inherently linguistics methods, like methods for document prediction, have been successfully applied for the task of protein classification . During the last twenty years the direction of interdisciplinary transfer has turned, and many methods originally designed for applications in evolutionary biology have been applied to linguistic data. These include algorithms for phylogenetic reconstruction [46, 47], phylogenetic network approaches [48, 49, 50, 51, 52], multiple sequence alignment [53, 54, 55], and homolog identification [55, 56].
In the following, we will argue that these transfers can be further enhanced. By shifting from the comparison of research objects to the comparison of processes affecting the research objects in the disciplines, wrong analogies due to an exaggeration of similarities and a neglection of differences can be avoided. At the same time, the identification of important processes, common to language and biological evolution, can give rise to new, potentially fruitful analogies. For linguistics, these transfers offer new theoretical and practical ways to explain the mosaic distributions of words across related and unrelated languages, with and without invoking processes of lateral transfer. A new analogy between the process of word formation in linguistics and protein assembly in biology offers a fresh perspective on the idea of a protein grammar  and can inspire new methods and models in both fields. Invoking a system perspective can further help to demystify the phenomenon of convergent evolution in languages resulting from common descent.
The striking similarities between biological and language evolution opt for a systematic investigation of analogies in the two disciplines. Such an investigation may cumulate in a program whose objectives would be (a) to investigate the isomorphy of processes, methods, and models in the two disciplines, (b) to foster the development of models lacking in either of the disciplines, and (c) to reduce the duplication of effort. Such a program, very close to the one proposed by the Society for General Systems Research in 1954 (as reported by (, p. 13)), would further ‘promote the unity of evolutionary science through improving communication among specialists’ (adapted from (, p. 13)). A multitude of analogies between biology and linguistics has been proposed in the past 200 years . Languages have been compared with organisms (, p. 16f), species , microbes [49, 50], mutualist symbionts , and populations . Words have been compared with cells (, p. 23f), amino-acids , codons [65, 66] and genes . Sounds (phonemes) have been compared with nucleic bases [65, 67] and atoms . Only a small amount of these analogies has received broader attention, many have been rejected quickly after they were first proposed, and only recently, an explicit transfer of methods and models has been initiated .
The focus on processes produces potentially fruitful novel analogies. It can also identify processes that seem to be exclusive to one of these two historical sciences (Fig. 2). Among the exclusively linguistic processes, we identify such processes as sound change (Fig. 2:14), semantic change (Fig. 2: 16), or purification (Fig. 2: 10). Neither of these processes seems to have a biological counterpart: It has been proposed to compare sound change in linguistics with concerted evolution in biology , but we think that the analogy between the two processes does not completely hold. In concerted evolution, two traits change in a similar manner. During sound change, the phoneme system of a language changes . An analogous process in biology would be a process in which the canonical amino acids constantly changed during evolution. During semantic change, the associations between words and concepts are restructured (, pp. 24–27). One might think of comparing this with changes in the regulation of genes in a genome which may yield drastic changes in function . However, while biological function is still determined and restricted by the nucleic and proteic forms, no necessary limits are imposed on the association between forms and meanings in natural languages: the association is arbitrary in the sense that a substantial link between form and meaning in languages is not necessary [72, 73]. Purification is a process by which language change is actively triggered with the goal to preserve the pure state of one’s mother tongue. One paradigmatic example for this kind of change is the Romanian language which was heavily influenced by neighboring Slavic varieties, until, around the end of the 18th century, nationalist movements triggered a purification process by which Slavic loanwords were successively replaced with native Romance words .
Exclusively biological processes include, among others, asexual (Fig. 2:6) and sexual reproduction (Fig. 2:12), but most likely also natural selection in a strict sense (Fig. 2:9). Some scholars claim that there is evidence that certain aspects of languages, like their sound systems, correlate with environmental factors , while other aspects, like their morphological complexity or the way they change, correlate with demographic factors [76, 77]. But languages are not independent of the ones who use them. They replicate via acquisition (of one’s first language, Fig. 2:4) and learning (of a further language, Fig. 2:1). Although we cannot exclude, that selection processes in biology and linguistics are similar and that a common theory of fitness could be derived , and that languages, for example, differ regarding the difficulty with which they can be learned, we think it would be premature to draw any process-based analogies here. Linguists tend to avoid the discussion of the fitness of languages due to its political and cultural implications, emphasizing that all natural languages are learnable within the normal time span that children need to acquire a language. There are also no known cases of languages becoming abandoned by their speakers due to their difficulty, since speakers always slightly adjust their languages to fulfil their communicative needs and thus maintain the functionality of their most important communication tool. Even if ease of transmission was a factor potentially influencing language evolution, as suggested by W. Ford Doolittle (the first reviewer of this manuscript), learning difficulty is by no means the sole factor that leads to language spread. The spread of English as a major second and first language, for example, was largely due to political factors, depending on those who carry the language rather than the language itself. It was not the rather simple grammatical structure of English that favored its spread but the fact that large powerful countries in different parts of the world use English as their first and official language. That the speaker size and especially the amount of second language speakers may have an impact on the way languages evolve is most likely [76, 77]. In order to be able to assess the various factors more substantially, however, much more research is required in the future, and we are careful in drawing any analogies with biological processes, as we still do not know enough about all the mechanisms involved in language evolution. For this reason, we are careful in identifying a direct counterpart process of natural selection in the linguistic world. There is ample evidence that some kind of selection occurs during language evolution [79, 80]. This selection is often called cultural selection, and we place it among the exclusively linguistic processes (Fig. 2:7).
The large amount of disciplinary-internal processes for which we could not find any counterpart is a challenge for current research in the evolutionary sciences, and a specific challenge for biologists and linguists. One the one hand, future research may show that some of these processes actually have counterparts in the other discipline, on the other hand, we may make progress in explaining why those processes are unique to a specific domain. In both cases, we will gain deeper insights into both the unity and the disunity of evolutionary processes across disciplines. But at least as important as the differences are the newly identified commonalities, which we will discuss in detail in the following section.
New analogies for biology and linguistics
The PBAs which we identified can be roughly divided into three categories, depending on the type of process which is involved. Tree-like processes represent the classical Darwinian framework of descent with independent modification between lineages, like divergence, and drift. Introgressive processes represent a network model of evolution in which lineages can influence each other after divergence, be it lateral transfer and borrowing (Fig. 2:13), hybridization and creolization (Fig. 2:8), or protein assembly and word formation (Fig. 2:15). Systemic processes represent a systemic model of evolution in which the interdependence between the components of evolving objects has a direct impact on the way they change (Fig. 2:17).
Biological methods can help to automatize the identification of homologous words
While the process of vertical descent is well established in both linguistics and evolutionary biology, it is notoriously difficult to define which words or other linguistic features are historically related across languages. Identifying words of common origin, for example, is of fundamental importance to compare diverging languages. In linguistics, the term cognate is used to address those words which share a common origin in which no lateral transfer occurred. So cognacy is, strictly speaking, not the same as homology in evolutionary biology , although it is often used interchangeably. Just like gene trees can be used to infer species trees in biology, sets of cognate words can be used to infer the relationships between languages [61, 82]. Problematically, the identification of cognates suffers from numerous practical limits. Traditionally, cognates are identified manually in linguistics, without any help of computational methods. But since the classical approaches to cognate identification are notoriously difficult to apply, the number of words used in phylogenetic language comparison is restricted to very small parts of the lexicon which are assumed to be neutral with respect to culture and present in all languages across all times. These basic parts of the lexicon, which are supposed to change slowly, only consist of about 200 words per language .
The overall number of words across languages varies drastically, and it is difficult to come up with a reliable statistics. However, given that near-native abilities of second language learners for the major European languages require the knowledge of about 4,000 to 5,000 words , it is obvious that cognate sets in computational applications cover an extremely restricted set of words. Despite this extreme restriction, only a fragment of the 7,000 languages spoken today have been thoroughly investigated. Given a large and increasing amount of digitally available data, the discipline can no longer be handled by manual inspection alone.
Incomplete lineage sorting as an introgression-free explanans for mosaic cognate patterns
Network approaches shed light on introgressive processes in language evolution
Introgressive processes in language evolution are not restricted to processes like borrowing, in which two or more languages interact, but they can also occur in one and the same language. Words are often created from smaller meaningful units from the same language (morphemes) via processes of word formation . Word formation can be roughly divided into two processes: derivation and compounding . While compounding creates new words by merging existing ones, derivation uses affixes which cannot be used in isolation but only when being attached to other words (compare, e.g., the -ness in English sick-ness). Word derivation and word compounding result in the emergence of word families, that is, groups of words which are cognate within one and the same language. Word families play an important role in lexical organization: by decomposing words into smaller meaningful units (morphemes), speakers can quickly induce the meaning of words, even if they hear them the first time. As a result, speakers can understand between one and three times as many words as they know . The size of word families can vary drastically, be it within one and the same or across several languages. The 60,000 words of the standard lexicon of German, for example, can be assigned to 8,000 word families comprising between 1 and 500 words .
The immediate consequence of word families is that cognate words across different languages are not necessarily completely cognate but may often exhibit different degrees of partial cognacy . In Mandarin Chinese, for example, the regular word for ‘moon’, yuè liàng, consists of two morphemes, the first one originally meaning ‘moon’ in isolation, and the second one meaning ‘shine’ in isolation. In combination, they now mean simply ‘moon’. In Cantonese, the Chinese variety spoken in Hongkong, the regular word for ‘moon’ is jyut 6 gwong 1, with the first morpheme being cognate with Mandarin yuè, but the second element, which means ‘light’ in isolation, being not cognate with the second element in Mandarin. Although methods for automatic cognate detection have been substantially improved over the last years [55, 103], none of the methods proposed so far is able to handle partial cognates across different languages. Word formation, especially word compounding, however, is very productive in many languages, especially in South-East Asian language families like Sino-Tibetan, Austro-Asiatic, Hmong-Mien, and Tai-Kadai (, pp. 62–67) which constitute more than 10 % of the worlds languages . Compounding is not restricted to specific realms of the lexicon but also affects the core vocabulary of languages which is used in phylogenetic approaches. In the Chinese dialects, for example, about 50 % of all nouns and more than 30 % of all words in basic vocabulary are derived from fusion or derivation . In biology, sequence similarity networks have been used to detect composite genes . In a similar manner, word similarity networks could be used to automatically identify compound words, as illustrated in Fig. 5 b. In a recent pilot study, it is further shown how a careful adaptation of similarity networks to linguistic needs allows to identify partial homologies (as the one between the Mandarin and Cantonese words for ‘moon’ shown above) with a high accuracy .
Towards a new linguistics of proteins
Invoking a system-perspective to demystify the mysteries of language drift
Almost 100 years ago, Edward Sapir (1884–1939) made the strange observation that language change may produce strikingly similar phases after the divergence of lineages, independent of areal contact or environmental influence [122, 123]. Sapir called this phenomenon of convergence, seemingly conditioned only by common ancestry, drift. Up to today, a more thorough investigation of the phenomenon is lacking, and many linguists even discard it as a mystical observation . If we look at the evolution of systems, that is, the evolution of interdependencies between components of evolving objects as yet another common process in biology and linguistics (Fig. 2:17), we find a possible explanans for this specific kind of language change. Evolutionary biologists distinguish two classes of interdependencies, depending whether they evolved neutrally (as in presuppression) or as a result of some selection. Typically, the evolution of several complex macromolecular machineries (such as the ribosomes or the splicesomes,  could be explained by a neutral increase of interdependencies between their elemental components, while convergences in regulatory networks (i.e. the fact that some patterns are more frequent than by chance, such as the feed forward loops in transcription networks) can be explained by considerations on the structure of these networks, e.g. the fact that sets of dependencies between elements stabilize or destabilize the function of the collective system that these elements form .
From a linguistic perspective, the use of the systemic perspective as an explanans for linguistic phenomena is by no means new. The structuralist movement, originally initiated by Ferdinand de Saussure (1857–1913) and later popularized by Roman Jakobson (1896–1982) was systemic in its core, assuming that ‘each system necessarily manifests as evolution, while, on the other hand, evolution necessarily bears systemic character’ (, p. 68). In historical linguistics, there is a large amount of literature on system-driven processes of language change. These include work on grammaticalization , direction in language change , and interaction between the varieties of one given language . Likewise, it might be useful to consider ratchet-like (irreversible) processes which would affect linguistic systems in specific states, just as processes of constructive neutral evolution are assumed to affect biological systems . The common change of languages which once diverged from a common ancestor is thus no longer mystical, but simply a consequence of the interdependencies which they inherited from their ancestor. It is more than likely that the many components of languages present interdependencies affecting their stability and rates of changes. For example, a recent use of sequence similarity networks on phoneme diversity across Chinese dialects revealed that phoneme diversity correlates with the grammatical classes to which these words belong . Hence the internal grammatical structure of languages certainly affects their evolution. Unfortunately, the majority of investigations on interdependencies in linguistics is neither formalized nor quantified. investigations on interdependencies in linguistics is neither formalized nor quantified.
We reported unities and disunities between evolutionary processes in historical linguistics and evolutionary biology. Common processes encourage the transfer of methods that had not been proposed earlier. The successful methodological transfer between the disciplines in the past encourages us to systematize the efforts of unification while at the same time being careful to not exaggerate the degree of similarity. Given the strong influence of biological approaches to quantitative research in historical linguistics in the past, the still low degree of quantification in historical linguistic research, and the new analogies which we proposed in this paper, it is clear that biologists may have an important role to play, given that their methods have a wider scope than anticipated earlier. On the other hand (following Schleicher’s idea proposed in 1863 ), given the amount and the subtlety of available historical documentation about the evolutionary processes that triggered linguistic diversity on earth, linguistic data could serve as an additional litmus test for the accuracy of biological methods, and biologists could profit from this advantage in detailed documentation.
In concrete terms, we showed, how biological methods can help to automatize the identification of homologous words in linguistics, how incomplete lineage sorting may serve as an introgression-free explanans for mosaic cognate patterns, and how similarity networks can be used to shed light on introgressive processes in language evolution. Furthermore, by refining the analogy of protein grammar, as a process-based analogy between the processes of protein assembly in biology and word formation in linguistics, both fields could profit from an interdisciplinary exchange and a deeper discussion of similarities and differences between the processes underlying the grammar of proteins and the processes underlying the grammar of words. The increasingly recognized need to account for the systemic dimension of evolution will likely prompt further unification across these fields and further interdisciplinary transfers. In the context of the theory of constructive neutral evolution, it may, furthermore, offer the long missing explanation for the mystical theory of parallel drift in the evolution of diverging languages.
Recalling that – apart from new analogies between evolutionary processes – we also identified processes which are specific to either biology or linguistics, it is important to keep in mind that the use of analogies should always be handled with great care. Not all evolutionary processes accounted for in one discipline necessarily need to have counterparts in other evolutionary disciplines, even if it is possible that future research will add process-based analogies where we failed to identify them. General evolution cannot be studied from within one discipline alone. Although unifying strategies can be fruitful, evolutionary explanations will remain fundamentally pluralistic since there is no reason to assume that all processes are common between biology and linguistics. In order to get a full picture of evolution, biologists and linguists need to complement their studies, trying to identify cross-disciplinary and discipline-specific evolutionary processes. If we want to understand how evolution triggered the diversity of substantial and intellectual objects on earth, we need to consider at least these two sister-disciplines.
We are very grateful to the reviewers for taking all the time to critically read our manuscript and to comment on it in their reviews.
Reviewer’s report 1: W. Ford Doolittle, Dalhousie University, Canada
I confess that I put off reviewing this because I feared that I would not understand it, or else would find it unoriginal: how could there be anything new to say about the similarities between historical linguistics and molecular phylogenetics? But I was wrong: I understand much of the paper and do think it says some important new things.
Basically what the authors propose is that we get even more serious about looking at the cross-applicability of methods and concepts being developed in linguistics and phylogenetics, particularly as these latter focus on evolutionary processes – rather than on the entities that evolve (words and proteins) – and also pay attention to the constraints that give direction to such processes such as syntax and molecular coevolution. Equally useful will be identification of processes that do not appear to be analogous between the domains. The authors suggest sound change, semantic change and purification as purely linguistic processes (the latter involving intent), and asexual/sexual reproduction and natural selection as purely biological.
It would be fun to argue about selection. The authors admit that there might be “cultural selection” (based on “egocentric”? or “content”? bias – see authors’ citation 70 ) that affect acceptance of certain elements within a language. Might it not also be that certain languages as systems are more likely to persist than others, either because of their ease of transmission (surely some languages are easier to learn than others) or affect on their speakers (surely language structure affects cultural “evolvability” somehow and unwritten languages have obvious limitations)? It may also be that in conceptualizing linguistic natural selection we should accept that evolution by natural selection can result from differential persistence as well as differential reproduction. Frédéric Bouchard (with whom the senior author has worked) has extensively developed this concept for biological evolution.
Authors make a number of observations which seem (to me, in my linguistic ignorance) novel, and well worthy of pursuit. For instance, applying models of incomplete lineage sorting (of alleles) to data in rapidly diverging languages seems a good idea, as does analogizing “the process of word formation in linguistics and protein assembly in biology”. It would be good to hear more about this and about using networks to identify composite words, as the senior author has already done for proteins (see their reference 94). It is also amusing that the numbers here are so close. Authors claim that there are about 200 universally conserved “basic parts of the lexicon”, and that second language learners need only master 4,000 – 5,000 words. There are maybe 200 universally conserved genes among all genomes, and the average prokaryotic genome has about 5,000 genes!.
Authors show a curious reticence to go all the way in analogizing language and genome evolution. They consider languages to be special since they are ‘products of the human mind’ and note that “If there was no speaker of the English language, a book containing Shakepeare’s Hamlet would just be a collection of paper with ink blots”. Actually, probably not. Surely clever Mandarin- (or even Martian-) speaking cryptographers could make some sense of the blots. And anyway, it’s analogously true that the sequence of bases in the human genome would only be just a sequence of bases without all the evolved machinery of gene expression and environmentally-affected epigenetic baggage, as opponents of genetic reductionism correctly but so tediously insist.
Authors’ response: We thank the reviewer a lot for the summary. We are glad that despite the initial reservations of the reviewer our manuscript turned out to be comprehensible enough, also for those who are not experts in the field of linguistics. The reviewer mentions that it would ‘be fun to argue about selection’ in the linguistic domain, pointing to the possibility that persistence of languages is linked to the ‘ease of transmission’ or ‘affect on [...] speakers’. Although in preparing the manuscript, we talked a lot about this issue in our interdisciplinary team, we decided to cut it short in the paper, given not only the difficulty to exhaustively grasp the forces at work in language evolution but also due to the heat with which the topic is discussed in linguistics. We refined the relevant passage by adding some further reasons why we are still careful in drawing the analogy, concluding, that in order to be able to assess the various factors triggering “cultural selection” more substantially, much more research is required in the future. Nevertheless, we agree with the reviewer that it would be very interesting to follow up these questions in more detail and we hope that our paper encourages researchers from different disciplines to increase their interdisciplinary work, looking for solutions to this and other problems related to language evolution. We have slightly modified the relevant passage in the main manuscript, trying to take the reviewer’s suggestions more closely into account.
Regarding the proposed process-based analogy between word and protein compounding, the reviewer further mentions that it ‘would be good to hear more about this and about using networks to identify composite words, as the senior author has already done for proteins’ [ 107 ]. As a matter of fact, we have, while waiting for the reviews of this manuscript, managed to carry out some more detailed pilot studies along these lines, and a manuscript with the title ‘Using sequence similarity networks to identify partial cognates in multilingual wordlists’ has been accepted for publication in the “Proceedings of the Association of Computational Linguistics 2016 (Short Papers)”. In this study, which would have gone beyond the scope of the current paper, we show how a careful adaptation of sequence similarity networks to linguistic needs allows us to identify partial homologies in linguistic datasets with a high accuracy [ 106 ]. We have now modified the manuscript in such a way that we directly mention this study along with a brief example, thus showing that similarity networks can indeed successfully be used to detect homologies across compound words in different languages.
As a final point, the reviewer mentions, with a certain regret, that we ‘show a curious reticence to go all the way in analogizing language and genome evolution’, which is definitely correct, but not necessarily since we ‘consider languages to be special’, but more since our experience with parallels proposed between the two fields in the past has led us to be rather cautious. In earlier work on the development of the family tree model in the discipline of linguistics, in which the first author was involved [ 91 ], it could be shown that – in contrast to the conviction of many scholars – it was an independent development in both disciplines, evoked by the emerging paradigm of uniformitarianism that triggered the development of the tree model rather than interdisciplinary transfer. One could thus argue that – if only the processes are strikingly similar – scholars may sooner or later come up with similar ways to handle them, with or without analogies drawn between disciplines. On the other hand, many of the analogies that were proposed so far, be it the one between languages and organisms by August Schleicher that was mentioned earlier in the manuscript, or the recent one between sounds in languages and nuclein bases in biology, turned out to be disappointing, unfruitful, and at times even completely wrong. While holding back ourselves, we hope, nevertheless, that our idea to start from common processes when searching for potentially fruitful analogies will offer us and our colleagues a tool to channel future methodological transfer across different disciplines. Furthermore, the reviewer has convinced us that our statement that Shakespeare’s work would ink blots on paper if there were no speakers of the English language to read it was essentially ill-chosen, not serving the point we wanted to underline, namely, the fact that the medium in which the research objects are realized differs largely in biology and linguistics, and that – in contrast to biology – the aspect of transmission via learning represents a different process of replication and manifestation. We therefore deleted the sentence from the manuscript.
Reviewer’s report 2: Eugene V. Koonin, NCBI, NLM, NIH, USA
The article by List and colleagues draws multiple analogies between evolutionary processes in biology and linguistics. To me, all, rather numerous articles and a few books that I have read on comparisons between biology and linguistics share the same, rather regrettable aspect: they seem very attractive and enticing to begin with but then, disappoint rather sorely. Regrettably, the present article is no exception. Quite frankly, I find that the title of the paper [original title: “Explaining evolution in biology and linguistics using common processes”, note by the authors] is a misnomer: nothing is explained here neither in biological evolution nor in the evolution of languages.
I agree that the ’process-based analogy’ touted by the authors makes more sense than the (apparently, more traditional) object-based analogy. I can also accept that there is substantial ILS in linguistic evolution and that there is some logic in the analogy between protein folding and word formation. The problem is that, as a student of biological evolution, I cannot formulate the new perspectives or ideas that I get from this article. Sadly, I think that I learned nothing truly new and substantial except for some details on the history of evolutionary linguistics and the interactions between linguists and biologists, in particular Schleicher and Haeckel (these historical details are fascinating). I cannot rule out that linguists do get something fresh out of this but the article has been submitted to a biology journal, so one could expect there to be something biologically relevant and perhaps interesting.
Authors’ response: We thank the reviewer very much for his critical review. First, we agree that the title may have been ill-chosen and changed it accordingly in order to reflect more clearly the scope and content of the manuscript. The new title “Unity and disunity in evolutionary sciences: Process-based analogies open research avenues for biologists and linguists” hopefully gives a much clearer emphasis on what we wanted to discuss in the paper, namely that we face common and distinct processes in the evolutionary sciences, and that a focus on common processes rather than similarities in objects might help better in identifying fruitful analogies between disciplines which may eventually open new possibilities for future research.
Second, regarding the reviewer’s disappointment that while showing potentially interesting possibilities of methodological transfer from biology to linguistics, we do not offer ‘something biologically relevant and perhaps interesting’, we think it is important to emphasize that the scope of this paper regards evolution in general. What we want to show is that neither linguistic nor biological evolution are reducible to one another, even at the level of their processes. Therefore, understanding evolution requires (at least) these two complementary fields, which means that the lessons from biological evolution (and from historical linguistics) will never be self-sufficient to account for what an evolutionist ultimately cares for: evolutionary diversity. As biologists, we are compelled to work closer with linguists if we want to learn about aspects of evolution that are simply – and will otherwise remain – foreign to us. That is one lesson: our biological models are incomplete to account for evolution in general, so it would be not only unfortunate but also wrong-headed to forget about linguistic evolution in our accounts of the history of life. Biology Direct could almost have a section for issues related to evolution in general. As for the linguistic perspective, we have shown that in addition to the biological methods for phylogenetic reconstruction which are now regularly applied by historical linguists, there are many more potentially fruitful analogies which could give rise to methodological transfer (such as lessons from incomplete lineage sorting and sequence similarity networks). So linguists should and usually do care for evolutionary biology. But even if it might not yet seem obvious why linguistics might become methodologically relevant for biologists, we should not forget that quite a few methods have already been transferred from linguistics to biology, especially from the disciplines of computational linguistics and natural language processing [ 43 ]. Not only classical models of formal grammar (following the hierarchy of the linguist Noam Chomsky [ 40 ]) are used by biologist, but also advanced models like tree adjoining grammar, which can be used for RNA structure prediction [ 44 ], or inherently linguistic methods for document prediction which can be applied in protein classification [ 45 ], or stochastic analyses of syntax, being applied to study protein domain promiscuity [ 121 ]. In order to substantiate this claim, that – despite the many disappointing examples of failed analogies – there are examples for methodological transfer in both directions which could be labelled success stories, we have added further references and elaborated the details in the text.
To summarize, we hope that readers will get at least two major ideas from this work: (a) it makes sense to embrace a less biology-centered perspective on evolution in evolutionary studies (that is our ignorabimus); (b) introgressive processes are fundamental to make sense of both linguistic and biological change, so a network perspective constitutes, despite the dissimilarity between both fields, the broadest and most fruitful deep commonality to achieve a form of systemic unification. There is a common core of processes between biology and linguistics, which is why evolutionary biologists and linguists should care about each other’s findings. Overall, however, it is true that for all evolutionary sciences such systemic, process-based unifications will remain incomplete. Evolutionary sciences will remain pluralistic in methods and concepts, and another type of unification, i.e. operating in a piecemeal fashion and preserving the singularities of both evolutionary disciplines, will be needed to speak of evolution in general.
Reviewer recommendations to the authors
The authors themselves notice that in the early days of genetics, and molecular genetics in particular, linguistic analogies and metaphors have been quite common. Some of these indeed became integral to the molecular biology lingo (transcription, translation), some are used much more sparingly (word, grammar), others have gone practically out of use (suffix, prefix, flexion). Regardless, though, why do these analogies do not really go beyond metaphors? Somehow it appears to me that this is not for the lack of effort on part of those interested in the linguistics-biology comparison. I feel that there is some deep disparity that precludes any substantial cross-fertilization. And here lies my major dissatisfaction with this paper. The problem is not that List et al. fail to find truly productive analogies between linguistics and biological evolutionary processes: many have tried and (at least, in my opinion) they all failed. The regrettable aspect of the paper is its rather careless but baseless optimism. I think the article would have been much improved if the authors embarked on a true critical discussion of these analogies and the reasons they do not appear to come across as genuinely fruitful.
Authors’ response: We agree with the reviewer that many largely disappointing analogies have been drawn between both disciplines, and it is for this reason that we have showed what reviewer 1 called a ‘curious reticence to go all the way in analogizing language and genome evolution’. There is a deep dissimilarity between evolutionary biology and historical linguistics, even at the level of processes. There is nonetheless a possiblity of substantial cross-fertilization between both fields, especially around introgressive processes and network-like evolution, and as we can see from the application of formal grammars in biology (mentioned above) and the recent popularity of phylogenetic methods in linguistics, fruitful transfer of methods and models has already taken place in the past and in both directions. Currently, the direction of transfer goes especially from biology to linguistics, and this means that linguists import methods and concepts from biology, adapting them to their needs. Given the rapid growth of computational research in the area of natural language processing, however, it is by no means sure that the situation will always remain as this, and it might well be that even in the nearer future our proposed analogy between word compounding and protein assembly offers biologists who study linguistic approaches and patterns new insights into the phenomena in their discipline. Future will tell whether this claim is careless optimism, or whether exploiting common processes between linguistic and biological evolution will not only turn out to be fruitful but potentially also inspire cross-disciplinary research on a larger scale. But even if our optimism turns out to be unjustified, it will essentially contribute to our understanding of evolutionary processes if we can further narrow down the exact ratio of unity and disunity in the evolutionary sciences.
Nevertheless, we understand that we might have been exaggerating our optimism, and we have tried to trim it down to a level which is hopefully acceptable for the reviewer. First, we changed Fig. 2 to reflect more closely that the amount of common processes is presumably much smaller than the general amount of processes (we also try to indicate our own uncertainty by showing a scale with a question mark as value). We also modified the manuscript in several passages to reflect justified scepticism more closely, and we also added references that further substantiate the reviewer’s scepticism.
In what sense did Watson and Crick ’detect’ DNA? They did not even discover it, they built the correct structural model of DNA that allowed them to explain replication.
Authors’ response: We agree and rephrased the sentence accordingly.
ILS, incomplete lineage sorting; PBA, process-based analogies
We thank the two reviewers for their challenging critics and helpful advice. We are grateful to David Morrison for sharing his knowledge about early tree- and network models in biology both in personal communication with JML and in multiple blog posts.
EB and JSP are supported by the European Research Council under the European Community’s Seventh Framework Programme, FP7/2007–2013 Grant Agreement # 615274. JML is supported by the German Research Foundation, Research Fellowship Programme, Grant # 261553824.
Availability of data and materials
The Additional file 1 contains the data and source code needed to reproduce the analyses to retrieve the networks shown in Fig. 5. It can be downloaded at https://zenodo.org/badge/latestdoi/5137/lingpy/process-based-analogies.
EB initialized the study, JML and EB set up the first draft. JSP and PL commented and revised later versions of the draft. All authors substantially contributed to the redaction of the manuscript and have given final approval of the version to be published. All authors read and approved the final manuscript.
JML is post-doctoral research fellow at UPMC (Equipe AIRE) and EHESS (CRLAO) Paris. JSP is a doctoral student in the Equipe AIRE (Adaptation, Integration, Reticulation & Evolution) at UPMC Paris. PL and EB are team leaders of the Equipe AIRE.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
- 1.Popper KR. Three worlds. Tanner Lect Hum Values. 1978;:143–167. http://tannerlectures.utah.edu/_documents/a-to-z/p/popper80.pdf.
- 2.Slingerland E, Collard M. Creating Consilience: Integrating the Sciences and the Humanities. Oxford: Oxford University Press; 2012.Google Scholar
- 3.Kirby S. The role of I-language in diachronic adaptation. Z Sprachwiss. 2000; 18(2):212–25.Google Scholar
- 4.Schleicher A. Die ersten Spaltungen des indogermanischen Urvolkes [The first splits of the Indo-European prehistoric people]. Allg Monatsschr Wiss Lit. 1853; 3:786–7.Google Scholar
- 5.Darwin C. On the Origin of Species by Means of Natural Selection. London: John Murray; 1859.Google Scholar
- 6.Schottel JG. Ausführliche Arbeit Von der Teutschen HaubtSprache [Exhaustive Examination of the German Main Language]. Braunschweig: Christoff Friederich Zilligern; 1663.Google Scholar
- 7.Stiernhielm G. De linguarum origine Præfatio [On the origin of languages] In: Stiernhielm G, editor. D,N Jesu Christi SS. Evangelia Ab Ulfila [The Gospels by Wulfila]. Stockholm: Typis Nicolai Wankif: 1671.Google Scholar
- 8.Gallet F. Arbre Généalogique des langues mortes et vivantes. Illustration; ca. 1800. http://gallica.bnf.fr/ark:/12148/bpt6k8546015.
- 9.Hickes G. Institutiones Grammaticae Anglo-Saxonicae et Moeso-Gothicae [Lectures on Anglo-Saxon and Moeso-Gothic grammar]. Oxoniæ: E Theatro Sheldoniano; 1689.Google Scholar
- 10.Sutrop U. Estonian traces in the tree of life concept and in the language family tree theory. J Estonian Finno-Ugric Linguist. 2012; 3:297–326.Google Scholar
- 12.Leclerc de Buffon GL, Vol. 5. Histoire Naturelle Générale et Particulière [General and Specific Natural history]. Paris: Imprimerie Royale; 1755.Google Scholar
- 13.Rühling JP. Ordines Naturales Plantarum Commentatio Botanica [Botanical Commentary on the Natural Order of Plants]. Goettingae: Abrah. Vandenhoeck; 1774.Google Scholar
- 15.Čelakovský FL. Čtení O Srovnavací Mluvnici Slovanské [Lectures on Comparative Slavic grammar]. Prague: V komisí u F. Řivnáče; 1853.Google Scholar
- 16.Darwin C. Notebook on Transmutation of Species; 1837. http://darwinonline.org.uk/content/frameset?viewtype=side&itemID=CULDAR121.-&pageseq=38.
- 17.Lamarck JB, Vol. 2. Philosophie Zoologique [Philosophy of Zoology]. Paris: Dentu; 1809.Google Scholar
- 18.Morrison DA. Genealogies: Pedigrees and phylogenies are reticulating networks not just divergent trees. Evol Biol. 2016. doi: 10.1007/s11692-016-9376-5.
- 19.Brugmann K, Vol. 1. Einleitung und Lautlehre: Vergleichende Laut-, Stammbildungs- und Flexionslehre der Indogermanischen Sprachen [Introduction and Phonetics. Comparative Studies of Sound Systems, Stem Formations, and Inflexion Systems of Indo-European Languages], GrundriSS der vergleichenden Grammatik der indogermanischen Sprachen [Foundations of the comparative grammar of the Indo-European languages]. Berlin, Leipzig: Walter de Gruyter; 1886.Google Scholar
- 20.Hennig W. Grundzüge Einer Theorie der Phylogenetischen Systematik [Foundations of a Theory of Phylogenetic Systematics]. Berlin: Deutscher Zentralverlag; 1950.Google Scholar
- 21.Sokal RR, Michener CD. A statistical method for evaluating systematic relationships. Univ Kansas Sci Bull. 1958; 28:1409–38.Google Scholar
- 23.Dixon RB, Kroeber AL. Linguistic Families of California. Berkeley: University of California Press; 1919.Google Scholar
- 24.Kay M. The Logic of Cognate Recognition in Historical Linguistics. Santa Monica: The RAND Corporation; 1964.Google Scholar
- 25.Haas MR. The Prehistory of Languages. The Hague and Paris: Mouton; 1969.Google Scholar
- 30.Christy C. Uniformitarianism in nineteenth century linguistics: Implications for a reassessment of the neogrammarian sound-law doctrine In: Koerner EFK, editor. Progress in Linguistic Historiography. Amsterdam: Benjamins: 1980. p. 249–56.Google Scholar
- 31.Wells RS. The life and growth of language: Metaphors in biology and linguistics In: Hoenigswald HM, editor. Biological Metaphor and Cladistic Classification: An Interdisciplinary Perspective. Philadelphia: University of Pennsylvania Press: 1987. p. 39–80.Google Scholar
- 32.Croft W. Typology and Universals. Cambridge: Cambridge University Press; 1990.Google Scholar
- 33.Schleicher A. Die Darwinsche Theorie und die Sprachwissenschaft [The Darwinian Theory and the Science of Languages]. Weimar: Hermann Böhlau; 1863.Google Scholar
- 34.Hoenigswald HM. On the history of the comparative method. Anthropol Linguist. 1963; 5(1):1–11.Google Scholar
- 52.List JM. Network perspectives on Chinese dialect history. Bull Chin Linguist. 2015; 8:42–67.Google Scholar
- 53.Kondrak G. Algorithms for language reconstruction. Toronto: Dissertation, University of Toronto; 2002.Google Scholar
- 54.Prokić J, Wieling M, Nerbonne J. Multiple sequence alignments in linguistics. In: Proceedings of the EACL 2009 Workshop on Language Technology and Resources for Cultural Heritage, Social Sciences, Humanities, and Education. Stroudsberg: Association of Computational Linguistics: 2009. p. 18–25.Google Scholar
- 55.List JM. Sequence Comparison in Historical Linguistics. Düsseldorf: Düsseldorf University Press; 2014.Google Scholar
- 59.Percival K. Biological analogy in the study of languages before the advent of comparative grammar In: Hoenigswald HM, editor. Biological Metaphor and Cladistic Classification: An Interdisciplinary Perspective. Philadelphia: University of Pennsylvania Press: 1987. p. 3–38.Google Scholar
- 60.Schleicher A. Zur Vergleichenden Sprachengeschichte [On Comparative Language History]. Bonn: König; 1848.Google Scholar
- 62.van Driem G. Language as organism: A brief introduction to the Leiden theory of language evolution In: Lin Yc, Hsu Fm, Lee Cc, Sun JTS, Yang Hf, Ho D, editors. Studies on Sino-Tibetan Languages. Taipei: Academia Sinica: 2004. p. 1–9.Google Scholar
- 64.Zwick M. Some analogies of hierarchical order in biology and linguistics In: Klir G, editor. Applied General Systems Research: Recent Developments & Trends. New York: Plenum Press: 1978. p. 521–9.Google Scholar
- 65.Enguix GB, Jiménez-López MD. Natural language and the genetic code: From the semiotic analogy to biolinguistics. In: Proceedings of the 10th World Congress of the International Association for Semiotic Studies (IASS/AIS). La Coruña: Association of Semiotic Studies: 2012. p. 771–80.Google Scholar
- 66.Jakobson R, Vol. 2. Rapports Internes et Externes du Langage [Internal and External Relations of Language]. Paris: Les Éditions de Minuit; 1973.Google Scholar
- 70.Bermúdez-Otero R. Diachronic phonology In: de Lacy P, editor. The Cambridge Handbook of Phonology. New York: Cambridge University Press: 2007. p. 497–517.Google Scholar
- 71.Allen U. Introduction to Systems Biology: Design Principles of Biological Cuircuits. London: Chapman & Hall/CRC; 2007.Google Scholar
- 72.de Saussure F. Cours de Linguistique Générale [Course on General Linguistics]. Lausanne: Payot; 1916.Google Scholar
- 73.Merrell F. The Routledge Companion to Semiotics and Linguistics In: Cobley P, editor. London and New York: Routledge: 2001. p. 28–39.Google Scholar
- 74.Mallinson G. Rumanian In: Harris M, Nigel V, editors. The Romance Languages. London and Sydney: Croom Helm: 1988. p. 391–419.Google Scholar
- 81.List JM. Beyond cognacy: Historical relations between words and their implication for phylogenetic reconstruction. J Lang Evol. 2016; 1. doi: 10.1093/jole/lzw006.
- 82.Atkinson QD, Gray RD. How old is the Indo-European language family? Illumination or more moths to the flame? In: Forster P, Renfrew C, editors. Phylogenetic Methods and the Prehistory of Languages. Cambridge and Oxford and Oakville: McDonald Institute for Archaeological Research: 2006. p. 91–109.Google Scholar
- 83.Swadesh M. Lexico-statistic dating of prehistoric ethnic contacts. Proc Am Philol Soc. 1952; 96(4):452–63.Google Scholar
- 84.Milton J. The development of vocabulary breadth across the CEFR levels. a common basis for the elaboration of language syllabuses, curriculum guidelines, examinations, and textbooks across europe In: Bartning I, Martin M, Vedder I, editors. Communicative Proficiency and Linguistic Development: Intersections Between SLA and Language Testing Research. York: Eurosla: 2010. p. 211–32.Google Scholar
- 88.Scally A, Dutheil JY, Hillier LW, Jordan GE, Goodhead I, Herrero J, Hobolth A, Lappalainen T, Mailund T, Marques-Bonet T, McCarthy S, Montgomery SH, Schwalie PC, Tang YA, Ward MC, Xue Y, Yngvadottir B, Alkan C, Andersen LN, Ayub Q, Ball EV, Beal K, Bradley BJ, Chen Y, Clee CM, Fitzgerald S, Graves TA, Gu Y, Heath P, Heger A, Karakoc E, Kolb-Kokocinski A, Laird GK, Lunter G, Meader S, Mort M, Mullikin JC, Munch K, O’Connor TD, Phillips AD, Prado-Martinez J, Rogers AS, Sajjadian S, Schmidt D, Shaw K, Simpson JT, Stenson PD, Turner DJ, Vigilant L, Vilella AJ, Whitener W, Zhu B, Cooper DN, de Jong P, Dermitzakis ET, Eichler EE, Flicek P, Goldman N, Mundy NI, Ning Z, Odom DT, Ponting CP, Quail MA, Ryder OA, Searle SM, Warren WC, Wilson RK, Schierup MH, Rogers J, Tyler-Smith C, Durbin R. Insights into hominid evolution from the gorilla genome sequence. Nature. 2012; 483(7388):169–75.PubMedPubMedCentralCrossRefGoogle Scholar
- 89.Schmidt J. Die Verwantschaftsverhältnisse der Indogermanischen Sprachen [The Relations of the Indo-European Languages]. Weimar: Hermann Böhlau; 1872.Google Scholar
- 90.Schuchardt H. Über die Klassifikation der Romanischen Mundarten. 1319 Probe-Vorlesung, Gehalten zu Leipzig Am 30. April 1870 [On the 1320 Classification of Romance Dialects. Test Lecture, Held at Leipzig on April 1321 30 1870]. Graz. 1900. https://archive.org/details/berdieklassifik01schugoog.
- 91.Geisler H, List JM. Do languages grow on trees? the tree metaphor in the history of linguistics In: Fangerau H, Geisler H, Halling T, Martin W, editors. Classification and Evolution in Biology, Linguistics and the History of Science. Concepts – Methods – Visualization. Stuttgart: Franz Steiner Verlag: 2013. p. 111–24.Google Scholar
- 94.Haak W, Lazaridis I, Patterson N, Rohland N, Mallick S, Llamas B, Brandt G, Nordenfelt S, Harney E, Stewardson K, Fu Q, Mittnik A, Banffy E, Economou C, Francken M, Friederich S, Pena RG, Hallgren F, Khartanovich V, Khokhlov A, Kunst M, Kuznetsov P, Meller H, Mochalov O, Moiseyev V, Nicklisch N, Pichler SL, Risch R, Rojo Guerra MA, Roth C, Szecsenyi-Nagy A, Wahl J, Meyer M, Krause J, Brown D, Anthony D, Cooper A, Alt KW, Reich D. Massive migration from the steppe was a source for Indo-European languages in Europe. Nature. 2015; 522(7555):207–11.PubMedCrossRefGoogle Scholar
- 95.Tadmor U. Loanwords in the world’s languages In: Haspelmath M, Tadmor U, editors. Loanwords in the World’s Languages. Berlin and New York: de Gruyter: 2009. p. 55–75.Google Scholar
- 99.Booij G. The Grammar of Words. An Introduction to Linguistic Morphology. Cambridge: Cambridge University Press; 2005.Google Scholar
- 101.Wichmann S, Müller A, Wett A, Velupillai V, Bischoffberger J, Brown CH, Holman EW, Sauppe S, Molochieva Z, Brown P, Hammarström H, Belyaev O, List JM, Bakker D, Egorov D, Urban M, Mailhammer R, Carrizo A, Dryer MS, Korovina E, Beck D, Geyer H, Epps P, Grant A, Valenzuela P. The ASJP Database. Version 16. Leipzig: Max Planck Institute for Evolutionary Anthropology; 2013.Google Scholar
- 104.Goddard C. Languages of East and Southeast Asia. An Introduction. Oxford: Oxford University Press; 2005.Google Scholar
- 106.List JM, Lopez P, Bapteste E. Using sequence similarity networks to identify partial cognates in multilingual wordlists. In: Proceedings of the Association of Computational Linguistics 2016. Short Papers. Stroudsberg: Association of Computational Linguistics: 2016. p. 599–605.Google Scholar
- 114.Hartmann S. The diachronic change of German nominalization patterns: An increase in prototypicality. In: Selected Papers from the 4th UK Cognitive Linguistics Conference. Lancaster: Cognitive Linguistics Association; 2014. p. 52–171.Google Scholar
- 115.Heide J, Lorenz A, Meinunger A, Burchert F. The influence of morphological structure on the processing of German prefixed verb In: Onysko A, Michel S, editors. Cognitive Perspectives on Word Formation. Berlin and New York: de Gruyter Mouton: 2010. p. 375–98.Google Scholar
- 117.Hammarström H. A naive theory of affixation and an algorithm for extraction. In: Proceedings of the Eighth Meeting of the ACL Special Interest Group on Computational Phonology and Morphology at HLT-NAACL 2006. Stroudsberg: Association for Computational Linguistics: 2006. p. 79–88.Google Scholar
- 118.Grönroos SA, Virpioja S, Smit P, Kurimo M. Morfessor FlatCat: An HMM-based method for unsupervised and semi-supervised learning of morphology. In: Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers. Dublin and Stroudsberg: Dublin City University and Association for Computational Linguistics: 2014. p. 1177–1185.Google Scholar
- 119.Griffiths S, Purver M, Wiggins G. From phoneme to morpheme: A computational model In: Baayen H, Jäger G, Köllner M, Wahle J, Baayen-Oudshoorn A, editors. Proceedings of the 6th Conference on Quantitative Investigations in Theoretical Linguistics. Stroudsberg: Association of Computational Linguistics: 2015.Google Scholar
- 122.Sapir E. Language. An Introduction to the Study of Speech. New York: Harcourt, Brace; 1921.Google Scholar
- 123.Aikhenvald AY. Semantics and pragmatics of grammatical relations in the vaups linguistic area In: Aikhenvald AY, Dixon RMW, editors. Grammars in Contact: A Cross-linguistic Typology. Explorations in linguistic typology. Oxford: Oxford University Press: 2007. p. 237–66.Google Scholar
- 124.Trask L. Trask’s Historical Linguistics, 3rd ed. London and New York: Routledge; 2015.Google Scholar
- 126.Tynjanow J, Jakobson R. Probleme der literatur- und sprachforschung [Problems of literature and linguistic research] In: Viehoff R, editor. Alternative Traditionen [Alternative Traditions]. Braunschweig: Vieweg: 1928. p. 67–9.Google Scholar
- 128.Haspelmath M. On directionality in language change with particular reference to grammaticalization In: Fischer O, Norde M, Perridon H, editors. Up and down the Cline – The Nature of Grammaticalization. Typological Studies in Language. Amsterdam and New York: John Benjamins Publishing Company: 2004. p. 17–44.Google Scholar
- 129.Oesterreicher W. Historizität, Sprachvariation, Sprachverschiedenheit, Sprachwandel [Historicity, language variation, language difference, language change] In: Haspelmath M, editor. Language Typology and Language Universals. Berlin and New York: Walter de Gruyter: 2001. p. 1554–1595.Google Scholar
- 131.Lopez P, List JM, Bapteste E. A preliminary case for exploratory networks in biology and linguistics: the phonetic network of Chinese words as a case-study In: Fangerau H, Geisler H, Halling T, Martin W, editors. Classification and Evolution in Biology, Linguistics and the History of Science. Concepts – Methods – Visualization. Stuttgart: Franz Steiner Verlag: 2013. p. 181–96.Google Scholar
- 132.In: Hóu J, (ed).Xiàndài Hànyǔ Fāngyán Yīnkù [Phonological Database of Chinese Dialects]. Shanghai: Shànghǎi Jiàoyù; 2004.Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.