The techniques of historical linguistics
In the course of the 19th century when Indo-European studies evolved as a science in its own right various techniques and methods were developed which help the linguist to arrive at solid facts about previous stages of a language. Of all the methods, the two listed at the beginning below are the main ones, the next two representing additional techniques which can be useful occasionally; the last phenomenon quoted below is important when one is considering the plausibility of change.
1) COMPARATIVE METHOD This refers to the practice of comparing forms in two or more languages with a view to discovering regularities of correspondence. A simple instance from English and German concerns /t/ and /s/. With a series of native words, i.e. not loans, one can see that where English has /t/ German has /s/: water : Wasser, better : besser, foot : Fuss. It is obvious here that English /t/ corresponds to German /s/ in non-initial position. The question which remains is whether the /t/ or the /s/ is original. Here one can quote other Germanic languages, e.g. Swedish has vatten, betra, fot for water, better, foot and this would imply that it is German which has changed the original /t/ to its present /s/. One could also use the arguments under (5) below to show that a fricative is more likely to develop from a stop, through a general process of weakening, rather than vice versa (unless through assimilation to another stop).
A further instance of the comparative method in operation is the reconstruction of morphological endings. Take the nominative singular masculine ending of Germanic. This is postulated to have been /-z/. The evidence is as follows. Whatever sound there was at the end of masculine nominatives, it became /-r/ in North Germanic through rhotacism and is still seen in Icelandic which is particularly archaic, compare English wolf with ulfr; ulf in Modern Swedish shows the loss of the final consonant. Now in Finnish there are a number of well-preserved old Germanic loans such as kuningas ‘king’ which shows a final /-s/. But Finnish does not have /z/ and we know from rhotacism in other languages (such as Latin, compare flōs : flōris ‘flower') that /r/ arises from a voiced sibilant so that we are justified in assuming /-az/ as the ending of the nominative masculine singular in Germanic.
Another major concern of the comparative method is justifying a postulated original form which is not attested. An example of this would be the vowel which was originally present in the words home and Heim (the present-day German pronunciation is /haim/ but as the orthography and southern German dialects show, this developed from an earlier /heim/). This is termed West Germanic /ɑ:/. The assumption is that it was a low back vowel and that the development was as follows.
|Scottish [hem],||OE ham|
|German [heim]||English [houm]|
The reason for assuming /ɑ:/ as the original vowel is that this requires the shortest movement to both a mid front and a mid back vowel. Furthermore, long vowels tends to rise more than to fall so that postulating /ɨ:/ as a mid vowel half way between the front /e, ei/ and the back /ou/ would be quite improbable.
2) INTERNAL RECONSTRUCTION This is the second major technique in reconstructing previous stages of languages. The basic principle is that one uses evidence from within a single language to gain knowledge of an earlier stage. Such evidence is usually available in forms which embody unproductive processes which are remnants of those which were formerly active.
A simple example can be given from Old English where there was an allophony which led to morphophonemic alternations in Modern English as seen in word pairs like roof : rooves, wife : wives, life : lives. The change between singular and plural here is between a voiceless and a voiced fricative. The reason for this is that in Old English [f] and [v] were allophones of each other. The [f] occurred at the beginning and end of words as well as in the environment of other voiceless consonants. The [v] was found intervocalically and in a voiced environment. Unchanging words like five and live show [v] in the singular from the previously intervocalic position of the fricative.
The voiced environment was provided in the plural where the ending /-as/ caused the word-final fricative of the singular to be in an intervocalic position and hence voiced: [ro:f] : [ro:vas]. This alternation has remained although the automatic voicing rule has been lost and both /f/ and /v/ are now phonemes.
3) CONSISTENCY OF ORTHOGRAPHY Latin orthography is known in its entirety and much is known of other systems as well, for instance that þ, ð in Old English were realised as [θ, ð], because the first letter is a known Runic symbol and the second a ‘crossed d’, a fricative voiced alveolar stop used elsewhere as well. Æ is a ligature and symbolised a sound between [a] and [e]. Equally this principle tends to apply to orthographical diphthongs such as eo and ea in Old English.
The orthography is not always reliable, however. Take the practice in Early Modern English of writing ye as a shorthand for the. The y never had any phonetic basis although it has led to a curious spelling pronunciation /ji:/ which is found in names of supposedly traditional pubs and restaurants such as Ye Olde Shippe /ji: əʊld ʃɪp/.
4) RHYME MATERIAL AND REVERSE SPELLING If a word is made to rhyme with another whose pronunciation is known then the same sound value can be assumed for the first word. A reverse spelling is where a writer does not use the usual spelling for a sound A but that for another sound B, as when Middle English writers used wright for write. The conclusion to be drawn here is that the sound indicated by -gh- [x] was already lost by this stage so that the spelling -igh- was interpreted as an adequate representation of /i:/ as in the word write (pre-Great Vowel Shift value).
5) GENERAL KNOWLEDGE OF LINGUISTIC PROCESSES If two languages have [k] and [tʃ] the one can safely assume that [tʃ] is from [k] as [tʃ] to [k] is an non-attested sound change (palatalisation as a process always involves a forward movement from the velum to the palate). Another example of a general process would be rhotacism, the development of /r/ from /z/. This is attested in a wide variety of languages and language groups such as Latin and Germanic (see above). The direction is generally from the fricative to the sonorant, though examples in the opposite direction are not unknown. Yet another example of general reasoning would concern front vowels. If a language has /y/ and /ø/ then one can assume that i-umlaut (the anticipation of a high front vowel in a preceding syllable) has occurred as this is generally the source of front rounded vowels. There is a second possible origin in language contact though this type of source is not accepted by all linguists.
Again an instance of general knowledge helping in an individual case would be with morphology. If a language has fewer inflections than another then it is probably right to assume that the latter is older at least more conservative as inherited inflections tend to be lost by phonetic attrition and to be gained by grammaticalisation of semantically bleached lexical elements.
Applying general knowledge in particular cases assumes that linguists have an accurate conception of what constitutes a typical and what an unusual change. There is not always agreement among scholars on this point and it is difficult to quantify ‘typical’ and ‘unusual’. Despite these difficulties the notions are nonetheless useful. For instance palatalisation is a very common phenomenon. It involves the shifting of an articulation from a velar position to a palatal one, normally with a change in manner from stop to affricate, this later being simplified to a fricative in many instances. This is to be seen clearly in Slavic and Romance languages, e.g. Latin camera became chambre in French, first with /tʃ/ then with /ʃ/ by affricate simplification. Old English also shows this change with /tʃ/ in southern forms and /k/ in northern forms, in Scandinavian and in German, e.g. chin versus Kinn.
A common principle may be seen to apply to a specific process in language change. For instance there is a general principle that words normally maintain their quantity, despite segmental changes within them. Thus on consonant loss, there is frequently compensatory lengthening by a short vowel becoming long. In Middle English the /x/ sound was lost in southern English and the vowels before this segment were lengthened, thus maintaining the entire quantity of the word, e.g. light was originally /lɪxt/ and later became /li:t/ (and /lait/ with the Great Vowel Shift). If one considers single consonants and short vowels as consisting of one unit of quantity then one can interpret the long vowel (a segment with two units of quantity) as arising due to the adoption of the quantity released by the loss of /x/.
Unusual changes are understandably not very widespread. To demonstrate unusualness consider palatalisation again. This is the fronting of the point of articulation. The reverse practically never occurs but Rhenish German (Cologne-Bonn area) is an exception in that it has a general velarisation of alveolars as in Hund [hɔŋk], Seite [zɪk], Pein [pɪŋ].