Internal Reconstruction in Linguistics Research Paper

Internal reconstruction is an extremely powerful method, and for this reason some of its results can be both speculative and controversial, though a number of the reconstructions arrived at by this method which were earlier deemed too hypothetical have subsequently become accepted, for example the postulation of a set of ‘laryngeal’ consonants for Indo-European (the ‘laryngeal theory’), whose existence was corroborated following the discovery of Hittite (Lindeman 1987). As in the case of the comparative method, it evolved in practice before it was formally recognized. However, since the method rests on a quite sophisticated formal analysis of language, it did not appear until the late nineteenth century, much later than the comparative method. In a youthful work, for instance, Saussure used aspects of what would now be regarded as internal reconstruction in his study of the IndoEuropean vowels. However, the method was not firmly established until the 1940s, principally in the work of American structuralist linguists such as Henry Hoenigswald.

Internal reconstruction is not a single, unitary method; the term embraces a number of related but distinct procedures, which we may designate as follows:

(a) historical morphophonemics; (b) regularization of systems; and

(c) universal and typological reconstruction.

All of these have in common the attempt to work backwards in time by reversing changes that are assumed to have taken place in the language in question, but they differ in scope and in the factors that are taken into account.

1. Historical Morphophonemics

This is the most limited of the methods of internal reconstruction. It applies only to the reconstruction of phonology, and works by identifying phonological alternations between grammatically or lexically related forms resulting from the differential operation of phonological changes in different linguistic contexts. By eliminating these alternations and reducing the set of alternants to a single form, it effectively reverses the changes and establishes the form that existed prior to their taking effect. In practice, it is equivalent to the synchronic process of establishing ‘morphophonemes,’ or the ‘underlying forms’ of classical Generative Phonology—abstract phonological entities from which the occurring forms can be derived—though it interprets these morphophonemes as historical antecedents.

Consider, for example, the Latin forms given in (1). Nominative Genitive

consul consulis (‘consul’)

genus generis (‘kind’) (1)

corpus corporis (‘body’)

As we see from consul, the genitive is formed by adding the affix -is to the stem of the word, but in the case of genus and corpus the final consonant of the stem is -s in the nominative and -r in the genitive. In synchronic phonology we might regard these two sounds as realizations of the same element, a ‘morphophoneme’ which appears as /s/ at the end of the word and as /r/ between vowels. But the alternation can also be interpreted historically, if we assume that this morphophoneme was actually an original phoneme, assumed to be /s/, which was changed to /r/ between vowels. We can therefore postulate earlier forms of the genitive: *genesis and *corposis.

This method works, therefore, by attempting to undo the effects of the splitting of phonemes in different contexts. Here, the original phoneme /s/ split into two separate phonemes, /s/ and /r/ , depending on the context in which it occurred. This split resulted in a synchronic alternation between the phonemes /s/ and /r/ in different forms of the word. By identifying such alternations, and deriving the alternating phonemes from a single source, we reconstruct the earlier form of the words.

2. Regularization Of Systems

Like historical morphophonemics, this method attempts to reverse changes by eliminating anomalies, but it extends this principle to cover whole systems of linguistic elements.

Consider, for example, the so-called ‘strong’ verbs of the Germanic languages. These verbs form their principal parts not by the addition of affixes (walk– walked; Love–Loved, etc.), but by modifying the vowel of the stem (sing–sang–sung; ride–rode–ridden, etc.). Strong verbs fall into a number of classes, according to the pattern of alternation of the vowels; only the first three are relevant here. For these verbs, we can identify three stems, whose vowels can be reconstructed by means of the comparative method, for the Germanic languages as a whole: as in (2).

Class I i: ai i

Class II iu au u (2)

Class III i/e a u

It can be seen that there is a certain regularity here, though also some anomalies. By the simple expedient of regarding the i: of the first stem of Class I (the colon indicates length) as ii, the first two classes become entirely parallel, with the vowel pattern i–a–zero, where the vowel is followed by i in Class I and u in Class II; in the third stem, the i or u is left as the only vowel. In Class III verbs, the vowel is followed by a nasal or liquid consonant (r or l ), and the i e alternation in the first stem depends on the following consonant: i occurs before a nasal and e before other consonants. We can therefore eliminate this alter-nation; the standard reconstruction assumes an e here, which is also generalized to the first stem of Classes I and II (e became i before the i and u of these classes as well as before the nasal of Class III). Thus, all three classes can be reconstructed with the same pattern (e–a) for the first two stems.

There remains the third stem, for which, as we have seen, we assume a ‘zero vowel’ for Classes I and II. If the pattern is the same for all three classes, Class III should have no vowel at all here, though in fact we find u. This u thus alternates with the zero of Classes I and II, and we can reconstruct a single source for both, which, following the first two classes, is taken to be zero. The nasal or liquid consonant of the stem of this class is assumed to have been ‘syllabic’ (i.e., it had the role of a vowel, as with the n or l of such English words as bitten [bitn] or bottle [botl]). It is presumed that these syllabic consonants later developed u before them. The result of these procedures is an entirely regular system, with the pattern e–a–zero, the differences among these three classes being in the sound which followed the vowel: i in Class I, u in Class II, and a nasal or liquid consonant in Class III. Such a regularized system can then be regarded as existing at an earlier stage of the language.

It is legitimate to object, of course, that, carried to its logical conclusion, this procedure results in the elimination of all anomalies and irregularities, and thus assumes that earlier languages had no such irregularities—an unlikely state of affairs. However, it should be borne in mind that all reconstruction, by whatever method, produces an idealization rather than an attested reality.

3. Universal And Typological Reconstruction

This method attempts to work backwards by invoking known or assumed general attributes of language or processes of linguistic change. This is the most powerful, but also the most speculative form of internal reconstruction, since the evidence we have of such general processes of change is often of a theoretical kind.

The criteria for reconstruction here may be either universal or typological. In the former case, changes are assumed to have occurred which reflect universals, i.e., characteristics of all languages, though these may be diachronic (universal processes of change) or synchronic (universal characteristics of language states which constrain the processes of change). If we therefore encounter a particular feature in the language under investigation which is known to be the result of a certain universal process, then we may assume this process to have occurred and reinstate the form from which the feature may have derived.

Universal processes or constraints have been recognized for all components of linguistic structure: phonology, morphology, syntax, and the vocabulary. One such process, which is currently receiving considerable attention, is grammaticalization, the ‘downgrading’ of lexical words to grammatical words and affixes (Hopper and Traugott 1993); a similar and related process applies in the lexicon, by means of which words are progressively ‘bleached’ of semantic content. Recognition of such universals allows us to determine the direction of language change, and thus to reconstruct earlier stages in conformity with this direction. Thus, if we assume that all grammatical affixes are derived from original lexical words, then a reconstruction which posits grammatical morphemes as earlier stages of lexical morphemes rather than the reverse would be ruled out.

Reconstruction based on typological criteria relies on the observation that languages fall into a number of more or less distinct types, characterized by cooccurring or harmonizing features. A standard typological grouping, for example, divides languages into VO and OV types, according to whether the object (O) follows or precedes the verb (V). A number of other features of sentence structure harmonize with this: in VO languages (e.g., French) the adjective typically follows the noun, and prepositions are used; in OV languages (e.g., Japanese) the adjective typically precedes the noun, and postpositions are used.

To apply the method, we identify the language as belonging to a particular type on the basis of some major feature, and, making the further assumption that languages are ideally typologically consistent, bring the remaining harmonizing features into line. If, for example, we find a language (such as English) which is VO but has the order adjective noun, we will assume that at an earlier stage the language will have either been OV (which is assumed for English) or will have had the reverse order of noun and adjective. Various different typological parameters have been put forward, but their use in reconstruction generally follows the above procedure. Again there are potential weaknesses in the assumptions underlying the method, since most languages are typologically inconsistent to some degree, and it is by no means proven that typological consistency represents an ideal or original state for languages.

4. Conclusion

The methods of internal reconstruction are extremely powerful, but they rest on a range of assumptions about both the nature of languages and the nature and direction of language change. The accuracy and reliability of the results inevitably depend on the adequacy of these assumptions. Not surprisingly, the results of the method have often been challenged as the assumptions themselves have been called into question.

Bibliography:

1. Campbell L 1998 Historical Linguistics. An Introduction. Edinburgh University Press, Edinburgh, Chap. 8
2. Fox A 1995 Linguistic Reconstruction. An Introduction to Theory and Method. Oxford University Press, Oxford, Chaps. 7, 8
3. Greenberg J H 1963 Some universals of grammar with particular reference to the order of meaningful elements. In: Greenberg J H (ed.) Universals of Language. MIT, Cambridge, MA
4. Hock H H 1986 Principles of Historical Linguistics. Mouton de Gruyter, Berlin, Chap. 17
5. Hopper P J, Traugott E C 1993 Grammaticalization. Cambridge University Press, Cambridge, UK
6. Lehmann W P 1992 Historical Linguistics, 3rd edn. Routledge, London, Chap. 8
7. Lindeman F O 1987 Introduction to the ‘Laryngeal Theory.’ Norwegian University Press, Oslo, Norway