Models of Language Change

by Assif Am-David, Frankfurt a.M.

For the last 150 years the ultimate representation of language change has been the family tree model. I shall start by explaining what the family tree model is. I shall pursue in presenting some major schools raising contention against this model. Finally, I shall discuss these critical voices in light of the progress in historical linguistics.

The family tree model is closely linked to the comparative method, which was developed by the Neogrammarians in 1860-1880 (Crowley and Bowern, 2010). The comparative method allows us to establish relatedness between languages that evolved from a common proto-language. This is done based on regular sound correspondences between the languages that are examined. These languages are then presented in a family tree which shows how each of the languages belonging to the language family descended from the proto-language, thereby splitting into branches. This model has served as the foundation not only of language classification, but of language change in general.

Basic assumptions of the family tree model

The family tree model and the corresponding comparative method rely on several assumptions which I shall now review based on Campbell (2004):

A) Sound change is regular: This is called the Neogrammarian Hypothesis and was formulated by Karl Brugmann and Hermann Osthoff. This means that whenever sound change occurs it occurs everywhere in the language and admits no exceptions.
B) Language change occurs by the diversification of language alone: A single language splits into several dialects that later become distinct languages. Once a split of a language into dialects occurs there is no further interaction between the branches. Therefore, each of the new dialects or languages develops further autonomously.
C) Any proto-language had a single form only: Whatever is reconstructed for the proto-language was invariably used by its speech community.

Challenges for the family tree model

Despite the great achievements, which the family tree model has led us to, recent decades showed an accumulation of critical views opposing its assumptions and calling for a change. These can be broadly grouped into three schools. I shall now present each of them separately: 1) Creolistics, 2) dialectology and 3) sociolinguistics.

1) Creolistics:

Several language contact situations in history resulted in the emerging of pidgins, creoles and mixed languages. Well-known examples occurred during the colonisation of Africa, South East Asia, the Pacific and the Americas. These account for most pidgins and creoles known to us. However, previous examples are attested. European examples are Basque-Icelandic pidgin used in the 17th century in Westfjords in Iceland and Russenorsk (Russian-Norwegian) that was used until the 19th century on the Kola peninsula in northern European Russia. Non-European examples include Babalia Creole Arabic of Chad (Arabic-Berakou), Gadal of Niger (Songhai-Tuareg), Tangwang of Gansu, China (Mandarin-Santa), Nagamese of Nagaland, India (Asamese-Naga), Vedda of Sri Lanka (Sinhalese-Aboriginal Vedda), Cauque Maya of Guatemala (Kaqchikel-K’iche’) and Kallawaya of Bolivia (Quechua-Puquina).
The extent of these languages clearly demonstrates they are no mere historical peculiarities. Such languages develop in certain intensive contact situations and clearly present an alternative to the traditional family tree model (Crowley and Bowern, 2010). Unlike most languages, pidgins, creoles and mixed languages do not have one parent language, but at least two: the lexifier language and the substratum language. Therefore, the diversification model in (B) above is clearly refuted.
Sign languages too present a special case. Unlike vocal languages, they do not have a constant speech community as they are mostly restricted to deaf speakers with only very few exceptions. This does not mean that sign languages emerge by contact between two languages. As deaf people do not normally live in isolation, surrounded only be other deaf people, their linguistic environment typically consists mostly of vocal languages. However, since vocal languages and sign languages use a different medium of communication, they differ too much to allow sign languages to evolve from vocal languages. Nonetheless, sign languages usually develop autonomously. However, they do not typically evolve by the diversification of a proto-sign language as the break-up from the original speech community is abrupt, owing to physical barriers.
Pidgins, creoles, mixed languages and sign languages are all examples of alternatives to the development of languages through changes in the proto-language. No family tree can accurately model the development of such languages and for this reason they defy traditional language classification.

2) Dialectology:

Parallel to the development of the comparative method by the Neogrammarians, dialectologists Johannes Schmidt and Hugo Schuchardt already laid the foundation of the alternative wave model (Crowley and Bowern, 2010). Through the study of dialectal variation it became evident that dialects cannot be divided into clear-cut genetic groups. Some dialects certainly display a higher degree of similarity than others, but when observed as a whole they can only be accounted for by a continuum. Thereby some features are shared by a group of dialects and not others, whereby another set of features are shared by an intersecting group of dialects. This results in an intricate web of isoglosses that show that features spread among dialects independently of each other. This spread is best depicted by the wave model as circles showing the extent to which innovations spread. These can indefinitely intersect with other circles to form the web called dialect continuum.
The wave model, which keeps enjoying popularity in dialectology to the present day, rejects assumptions (A) and (B) of the family tree model. Sound change does not apply everywhere in the language, but spreads selectively only to some variants and to some parts of the grammar. This stands in contrast to the Neogrammarian Hypothesis. The same can be said of language change in other realms of language. Similarly, a split of a language to new dialects does not prevent them from interacting, mutually affecting their development. This results in the diverse isoglosses and clearly refutes (B).

3) Sociolinguistics:

Starting from the 1960ies linguistics came to regard language change not only diachronically, but also synchronically. The pioneering research of Labov clearly showed that language change can be demonstrated in a single point in time. The synchronic aspect of language change can be best represented by two phenomena: indeterminacy and variability (Crowley and Bowern, 2010).

Indeterminacy is what happens when native speakers are requested to carry out grammaticality judgements. While some expressions are clearly grammatical and others clearly ungrammatical, there is a great range of variety in between. This shows that some phenomena are in the process of either entering the core grammar of the language or leaving it. Unlike what has typically been assumed, a grammar of a language does not consist of a closed set of rules that define it, but is rather incremental in character. This allows a dynamic account of language change also in the synchronic study of language.
Similar to indeterminacy, variability shows that language is more varied that was believed. Unlike indeterminacy, which addresses the problem of grammaticality and therefore competence, variability is concerned with the grammatical variation in the production within one speech community and even in a single speaker (idiolect). It can be clearly demonstrated that native speakers change their language in different social settings. Their use of lexicon and grammar is not steady and depends on non-linguistic factors. This affects the way we regard language change. Some properties of the language are perceived as enjoying high prestige in a certain speech community whereby others only enjoy low prestige. Through high prestige forms taking precedence over low prestige forms language change is often set into motion.
Both indeterminacy and variability stand in sharp contrast to (C). Proto-languages, similar to other languages, were always undergoing a process of constant language change. The idea that we can define languages by a closed set of lexemes and grammatical rules is refuted by both indeterminacy and variability, since on the one hand, grammaticality is incremental, and on the other hand, every language includes several intersecting variants that are used according to the social setting as perceived by the speech community.


Having gone through all major opponents of the family tree model, I shall now discuss their potential impact on the future of historical linguistics. Each of the schools attacks one or more of the underlying assumptions of the family tree model and the comparative method. Creolistics refutes (B), dialectology refutes both (A) and (B) and sociolinguistics refutes (C). With this criticism the entire construct of the family tree model seems to collapse. However, a major problem with these approaches is that, apart from dialectology, they propose no alternative model. Neither creolistics nor sociolinguistics propose a comprehensive model of language change. Dialectology does contribute the wave model, however, this model is far less detailed than the family tree model and cannot serve as basis for reconstruction.
While the criticism against the family tree model seems compelling we cannot ignore the achievements of the comparative method. Although it is clear that we cannot continue to follow the family tree model blindly, disregarding the bulk of evidence against it, neither can we put it aside with no established alternative available yet. Future historical linguistics shall have to address the problems in the family tree model and find a way to revise it or replace it in light of our enriched knowledge of language.


  • Campbell, L. (2004). Historical Linguistics. Edinburgh: Edinburgh University Press.
  • Crowley, T. And Bowern, C. (2010). An Introduction to Historical Linguistics. Oxford: Oxford University Press.