How do we learn about language history?
There are different methods for studying the history and prehistory of languages, as is explained in Weiss 2011, p. 1-8:
- Documentary Evidence: Examining documents written in the same language at different times. Comparison of earlier and later texts can reveal details about the changes a language has gone through over time.
- Identifying loanwords: They can provide information about language contact, and the kind of loanwords can tell us more about the context of this contact.
- Internal Reconstruction: Internal Reconstruction typically deals with varieties within paradigms or across similar paradigms. We reconstruct on the basis of synchronic language-internal evidence (Bauer 2009, p. 18).
- Comparative Reconstruction (The Comparative Method): Studying the development of languages by comparing features step by step and feature by feature.
The Comparative Method
The Comparative Method, as used by historical linguists, is a study of resemblances within different languages, and their sources. Those resemblances can have various sources (Fortson 2010, p. 1):
- Chance: Words, especially shorter ones, can resemble each other by pure coincidence – there are only so many sounds the human vocal tract can produce.
- Borrowing: Languages, who’s speakers are in contact with each other, can borrow words from each other, for example when learning new techniques from each other, the related expressions can be borrowed, too, and probably even adapted to the other language, and over time change the meaning.
- Language universals: For example, onomatopoeia (sound symbolisms, words sound like what they mean: cuckoo) or nursery talk (words for kinship terms formed from simple syllables, like ma, ba, da, ta).
But when there are resemblances, which can’t be explained by any of the examples above, it may be exactly what we are looking for. If there are many systematic similarities found within two or more languages, and they cannot be ascribed to one of the causes listed above, then the only satisfactory explanation for those is that they are descended from the same parent language (Fortson 2010, p.3)
Those languages are then assumed to be genetically related, which does not necessarily say anything about the speakers in terms of race or ancestry, but purely about the languages which have a common ancestor, a proto language. This is very important to keep in mind.
Once a genetic relation is established, we would further study the ancestor language and its origins to get a better idea of the people who were speaking it. This is where comparative reconstruction starts.
Here I cite Krasukhin (in Klein, 2017, p. 53):
“The main thesis of comparative linguistics is the following: a group of languages is united by common origin, i.e. having a common ancestor language (= protolanguage), which has split in historical or prehistorical times. The latter case requires a reconstruction. The exact methods of reconstruction must be elaborated at all linguistic levels.”
Reconstruction is the process of restoring a no longer existing linguistic state of a language. This state, however, has left traces in existing languages. The role of the linguist is to find and interpret those traces. The main goal is the fullest restoration of the object to be reconstructed. The sum of all reconstructions attainable in a language family can then be the foundation for reconstruction of its proto-language, although the reconstructed proto-language is unattainable as a whole. It is subject to probabilistic judgment about the original condition (Klein 2017, p. 52 ff.).
This method of reconstruction is accomplished through systematic comparison of forms found in the descendant languages (Fortson 2010, p. 3). We can make a table with correspondence sets, and the words in that table will be the cognates from which regular sound correspondences between the languages are established. The more correspondence sets we have, the better we can establish regular sound changes. Thus, the best way to apply the Comparative Method is to work with a system of the following steps:
- Find possible cognates and list them, for example in an Excel sheet – and if possible, already now make sure that they are not subject to borrowing, language universals or chance, see above.
- Establish the correspondence sets
- Find the complementary distribution and
- Reconstruct the possible phonemes – “Phonological reconstruction depends upon the undoing of the various sound changes that have accrued over the life histories of the various daughter languages.” (Weiss 2011, p. 7)
- Examine the results carefully and compare them with the already known typological constraints of the proto language.
To understand the mechanics of sound change is important. Human speech is variable, and therefore language is constantly changing, not just whole words but also sounds. When the speaker of a language pronounces a word, and his or her articulatory system produces a variation of one sound, it might not always be filtered out by the listener, and thatlistener takes the variation over in his own vocabulary. Or, the listener might hear a variation, where there is none or a misconception. A sound change has thus occurred. Sound change is always regular (as it is the statement of the Neogrammarians), so if s changes to z between vowels in one word, “…all tokens of the segment s that are between vowels at that time will become z” (Weiss 2011, p. 7). This lets us establish relative chronologies and identify loanwords.
This is used to reconstruct the phonemic system of a proto-language. It is based on the phonetic law: In language A the phoneme a corresponds to the phoneme b of a cognate phoneme in language B. This sound change is influenced by the phonetic environment, for example accent. Accent is the phoneme’s position in a word, along with the presence or absence of stress or pitch. A given phoneme may be preserved, or it may change, taking either of two different forms:
- Split (a > a1, a2)
- Merger (a, b > b)
The verification of this method needs to be ensured by typological criteria, external comparison if possible and by old borrowings in neighboring languages (Klein 2017, p. 55 f.)
Each language has inflectional and derivational paradigms. When comparing related languages, independent words can be transformed into grammatical morphemes such as cf. Lat. mēns ‘mind, disposition’ with the abl. mente > the adverbial suffix –ment[e] in the Romance languages. All reconstructions at higher levels than the phoneme must be based on the exact application of phonetic laws (Klein 2017, p. 56).
Not all sound change in morphemes is conditioned by environment. An example for morphophonological sound alternation is the ablaut (quantitative as a consequence of stress, and qualitative, which cause is more controversial). An entire system of ablaut-accent can be reconstructed for Proto-Indo-European (Klein 2017, p. 56-57).
The reconstruction of Proto-Indo-European roots is based on the application of laws, and roots are subject to phonotactic constraints (Klein 2017, p. 57 f.).:
- The phonetic structure of roots has some basic rules: In Proto-Indo-European, roots are either a combination of two voiced consonants (*DeD) or a voiceless stop with a voiced aspirate (*TeDH / *DeTH).
- The contour of the root was Consonant(s) – Vowel – Consonant(s). The root may occasionally appear extended with an additional consonant
Reconstructing an entire stem means that the rules for the root as well as for the morphology must apply, for example: OCS brĕme, Greek φέρμα (both meaning ’burden, fetus’), and Sanskrit bhárman– ‘burden’ represent the same prototype and can therefore be reconstructed as Proto-Indo-European *bhér-mṇ.
Probably most important regarding the next chapter about interdisciplinary work is the connection of words with material and nonmaterial culture (Klein 2017, p. 58). When we reconstruct a word, we also suggest a thing, leading us to reconstructing features of a culture (more about this in my Blog post about inter-disciplinary work).
Another method of reconstructing is Internal Reconstruction, which means that we analyze variations within the history of a single language. This alteration can occur by sound change or analogy.
When we deconstruct a sound change using the internal method, the reconstructed form is marked as pre-, for example Pre-Germanic (as opposed to proto-, when a language is reconstructed with the Comparative Method).
A good description of the method is given by Bauer, 2009, p. 18:
“The method is based on the principle that phonological changes occur in specific phonological environments and that these do not always correspond to morphological classes or regularities.”
As an example, she uses Greek δίδωμι “I give” vs. τίθημι “I put”. Both are reduplicated verbforms, but in the case of τίθημιthe reduplication is not complete, the aspiration of the first t is missing, we would expect θίθημι. This change has happened because of a sound change, namely Grassmann’s Law (ChVCh > CVCh, see also 5.2.3 or 22.214.171.124.1). Therefore, in this method, we deal with varieties within paradigms or across similar paradigms (Bauer 2009, p. 18).
As straightforward as the Comparative Method and Internal Reconstruction sound, there are also problems with those methods, things we need to be aware of and things we need to avoid. I summarize the most important points:
- We are dependent on the quality of the written records. The older languages are, the fewer written sources there Sometimes texts are written down by someone who was not a native speaker, or after a long time of oral transmission; there may be spelling mistakes (which on the other hand, when they are obvious, can tell us more about how words were pronounced). Some of the languages named above are hard to reconstruct, as we simply don’t have enough sources.
- We often deal with dead languages, which don’t develop anymore, but that doesn’t mean that our research doesn’t need to be constantly controlled or corrected. Finding new sources can change our results; however, the language in question did not change, but our results have been adapted to the newest findings.
- Language is constantly evolving. Especially when we lack the written resources, we cannot say at which point in time a certain state of development in a language happened. Language change happens, and we know what changes, but the change is rarely linear.
- Which brings us to the topic of sound change being without exception, as Clackson writes: “These modern studies have shown that sound changes are not ‘exceptionless’: some changes may not spread to all words in the lexicon, and indeed some sound-changes may remain restricted to certain groups in a speech community” (Clackson 2011, p. 32 f.) However, in my opinion, even though there are some exceptions, the bigger picture is important. In a longer timeframe, the statement that sound change is exceptionless is much closer to the truth. Smaller nonconformities are less visible when looking at the whole development over 2000 years.
- Raum / Zeit – Modell: We can reconstruct the Indo-European language, but we cannot say much about the exact time and place, such as where and when it was spoken. We know it was not a primitive language: “… es war vielmehr eine der damaligen Sprachen Europas. Es wurde von einer damals schriftlosen Sprachgemeinschaft gesprochen.” (Meier-Brügger 2000, p. 57). The language we reconstruct is no more than that, close enough to the concrete living language, but not the same thing.
- When we set up correspondence sets, we must exclude those where a symbolic relationship obtains between the sounds in a word and its meaning (like onomatopoetic words). (Fortson 2010, p. 4).