Scholarly article on topic 'Irregular Sound Change due to Frequency in Germanic Languages'

Irregular Sound Change due to Frequency in Germanic Languages Academic research paper on "Languages and literature"

0
0
Share paper
Academic journal
Lingua Posnaniensis
OECD Field of science
Keywords
{""}

Academic research paper on topic "Irregular Sound Change due to Frequency in Germanic Languages"

DOI: 10.2478/v10122-010-0004-z

IRREGULAR SOUND CHANGE DUE TO FREQUENCY IN GERMANIC LANGUAGES

WITOLD MANCZAK

Abstract. Witold Manczak. Irregular Sound Change due to Frequency in Germanic Languages. Lingua Posnaniensis, vol. LII (1)/2010. The Poznan Society for the Advancement of the Arts and Sciences. PL ISSN 0079-4740, ISBN 978-83-7654-030-6, pp. 55-65

Until now, irregular sound change due to frequency has been considered as something sporadic, affecting only the vocabulary, whereas, according to the present writer, irregular sound change due to frequency, which concerns also reductions in morphemes, especially in inflectional ones (which are even more frequently used than words), is the third essential factor of linguistic evolution, in addition to regular sound change and analogical development. There is a synchronic law according to which the linguistic elements which are more often used are smaller than those which are less often used . There is a kind of balance between the size of linguistic elements and their frequency. But if a linguistic element (morpheme, word or group of words) becomes too long in relation to its frequency, it must be shortened

Witold Manczak, Zak^tek 13/59, 30-076 Krakow, Poland

1. INTRODUCTION

The notion of irregular sound change due to frequency is not new. It would be difficult for me to say who was the first to use the term, but in any case it was used as early as 1846 by Diez (1846: 12), the founder of the comparative grammar of Romance languages, who considered Fr. sire < Lat. senior as 'durch häufigen Gebrauch verkürzt'. Some years later, the famous etymologist Pott (1852: 315) stated that It . andare, Sp . andar, and Fr. aller derive from Lat . ambulare 'mit zwar ungewöhnlichen, aber durch die Häufigkeit des Gebrauchs von diesem Worte gerechtfertigten Buchstabenwechseln' . Other linguists followed them . There is, however, an essential difference between the opinions of my predecessors and mine on this subject. Until now, irregular sound change due to frequency has been considered as something sporadic, affecting only the vocabulary, whereas, to my mind, irregular sound change due to frequency, which concerns also reductions in morphemes, especially in inflectional ones (which are even more frequently used than words) is the third essential factor of linguistic evolution, in addition to regular sound change and analogical development: in any text of some languages, even one third of the words show an irregular sound change due to frequency.

In brief, the theory of irregular sound change due to frequency can be presented as follows. There is a synchronic law according to which the linguistic elements which are more often used are smaller than those which are less often used . There is a kind of balance between the size of linguistic elements and their frequency. Anyhow, the size of linguistic elements is not stable . As a result of regular sound change, the size of words may change considerably as the comparison of some Old and New High German words shows:

OHG üf (2 phonemes) > NHG auf (3 phonemes) - increase of 50%; ¡era (4) > Lehre (4) - no change; sköni (5) > schön (3) - decrease of 40%.

Since the frequency of words is not stable either, it may happen that the balance between the size of a word or of a morpheme and its frequency is disturbed. If a word or a morpheme becomes too short in relation to its frequency, it is replaced by a longer one. But if a linguistic element (i . e . a morpheme, word, or group of words) becomes too long in relation to its frequency, it must be shortened, and then there are two possibilities: either a mechanical shortening (autobus > bus) or an irregular change due to frequency (master > mister, you are > you're, Lat . deb-ebat > Fr. dev-ait) .

If irregular sound change due to frequency is far advanced, it consists of the decay of one or more phonemes, e . g . God be with you > good-bye. However, if the development due to frequency has just started, it may only consist in a partial reduction of a phoneme, e . g .

(a) the long vowel undergoes a reduction: Goth sunus shows a short vocalism athough the vowel was long in Proto-Indo-European, cf OI sünü-, Lith. sünus, OCS synü;

(b) the degree of the vowel opening is subject to a reduction (a > e > i or a > o > u), e . g . in OHG nemames > nemumes, the vowel a narrowed to u, in OHG stan > sten, to e;

(c) the full vowel is changed into a reduced one: shall may be pronounced with an [a];

(d) the nasal vowel is subject to denasalization: Pol. bqdzie 'will be' may be pronounced with e instead of q;

(e) the palatal consonant undergoes a depalatalization: the Russian reflexive pronoun may be pronounced as [sa];

(f) the voiceless consonant becomes voiced, the pronunciation of voiced consonant seeming to be easier: an irregular voicing occurs in the ending -s (plural, genitive, and 3rd pers. sing.) and in some very frequent words like as, of, or the.

There are six arguments which can be mentioned in favour of the theory of irregular sound change due to frequency.

2 . FIRST ARGUMENT

If a frequency dictionary for a given language and for a given epoch exists, we may use it, since the majority of words showing an irregular change due to frequency belong to the thousand words most frequently used in the given language. E. g. in French, the distribution of these words is a follows:

First thousand 99

Second thousand 9

Third thousand 4

Fourth thousand 2

Fifth thousand 1

Sixth thousand 0

For more details, see Mañczak 1969: 20.

3 . SECOND ARGUMENT

In addition to irregular sound change due to frequency, there are other irregular sound changes, namely assimilations, dissimilations, metatheses, and expressive and overcorrect forms. These irregular sound changes are characterized by the fact that they occur in different words in different languages, e . g . a dissimilation took place in Ger. Fibel < Bibel, a metathesis in Born < Bronn; however, in another Indo-European language, it would not be easy to find a word meaning 'primer' with a dissimilation or a word meaning 'a well' with a metathesis . In other words, there is no parellelism between the words showing irregular changes such as assimilations, dissimilations, etc in different languages, whereas irregular sound change due to frequency occurs in various languages more or less in parallel, which is explained by the fact that the most frequently used words are nearly the same in all languages . Here are some examples .

Mister < master and sir, which derives from Fr. sire, also irregularly developed from Lat. senior, show irregular reductions . The same is true for Fr. monsieur < monseigneur or Sp . don, being used alongside the more regular dueño < Lat. dominum . Although the opinion on the etymology of Pol. pan or Czechpán is not unanimous, it is unquestionable that these words derive from a longer form, as is the case with Russian barin, which has developed from bojarin .

The fact that a whole mosaic of Old High German forms anti, ande, enti, endi, indi, inti, inde, inte, int, in, unta, unte, unti, un correspond to NHG und shows explicitly that we face here an irregular sound change due to frequency. Similarly in none of the Romance languages, Lat . et has developed normally, cf Fr. et, It . e, or Sp . y. The conjunction a, attested in many Slavic languages, also shows a reduced form.

OHG gangan is irregularly reduced to gan, gen, whence modern gehen. Similarly, reduced forms occur in other Germanic languages, cf E go, Dutch gaan, Dan . ga, etc . Lat. ambulare undergoes a reduction in the Romance languages, cf Fr. aller, It. andare, Prov ana, etc. In the same way, Common Slavic *sídlü (> Pol. szedl, R sel, etc . ) exhibits an irregular reduction, namely the occurrence of í instead of e (alternating with o in *choditi).

4. THIRD ARGUMENT

If in a given language, a morpheme, word, or group of words occurs in a double form (regular and irregular), irregular sound change due to frequency is characterized by the fact that the irregular form is usually used more often than the regular one. In all living Germanic

languages, the suffix -isk- shows a double development: Engl-ish and Wel-sh, Scott-ish and Scot-ch or Scot-s, Ger. französ-isch and deut-sch, Dutch Olymp-isch and Nederland-s, Swed . nord-isk and sven-sk, etc. The reduced forms of the suffix in question are to be explained by irregular sound change due to frequency.

As far as Swedish goes, I excerpted several pages in the newspaper Dagens Nyheter, where I found the following forms in -isk and -sk:

-isk: 13 politisk; 7 ekonomisk, nordisk; 6 brittisk, symmetrisk; 4 faktisk, praktisk, saudiarabisk, 3 demok-ratisk, psykisk, skandinavisk, socialdemokratisk; 2 arabisk, babylonisk, dramatisk, europeisk, käUkritisk, mekanisk, österrikisk, saudisk; 1 aktivistisk, automatisk, belgisk, byrakratisk, diplomatisk, elektronisk, etc .

-sk: 35 svensk; 30 dansk; 11 amerikansk, fransk; 5 afrikansk, tysk; 4 engelsk, norsk; 2 japansk, rysk, västtysk; 1 holländsk, inhemsk, medicinsk, platonsk, polsk, skansk, stockholmsk, utländsk, västerländsk .

From the statistical point of view, these data are as follows:

Number Number Mean frequency

of attestations of words of occurrence

-isk 116 57 2

-sk 120 20 6

The mean frequency of occurrence of the derivatives in -sk (with irregular sound change) is higher than that of the words in -isk (with regular sound change)

As far as German is concerned, I excerpted several pages in the journal Die Zeit, where I found the following forms in -isch and -sch:

-isch: 8 europäisch; 7 klinisch; 4 genetisch, kritisch; 3 englisch, gentherapeutisch, technisch; 3 britisch, französisch, gentechnisch, mitteleuropäisch, römisch; 1 amerikanisch, belgisch, biophysikalisch, editorisch, etc

-sch: 22 deutsch(land)

From the statistical point of view, these data are as follows:

Number Number Mean frequency

of attestations of words of occurrence

-isch 63 33 2

-sch 22 1 22

The mean frequency of the word deutsch (with irregular sound change) is much higher than that of the derivatives in -isch (witch regular sound change) .

As far as Dutch goes, I excerpted several pages in Leidsch Dagblad, where I found the following forms in -isch and -s:

-isch: 4 logisch; 3 economisch, historisch; 2 academisch, botanisch, democratisch; 1 automatisch, calvin-istisch, etc

-s: 66 Leids; 40 Amerikaans; 9 Engels; 6 Nederlands; buitenlands; 3 Duits(land), Hollands; 3 Anglicaans, Japans, westers; 1 aards, Amsterdams, binnenlands, Brits, etc .

From the statistical point of view, these data are as follows:

Number Number Mean frequency

of attestations of words of occurrence

-isch 22 12 2

-s 152 25 6

The state of affairs is similar in other Germanic languages (Danish, etc.). Only English is an exception because English, where -ish is preserved, is more frequently used than derivatives like Welsh, Scotch, Dutch, French, Scots, Norse, or Manx, where the suffix underwent a reduction. This is to be explained by the fact that the consonant cluster preceding -ish in English made the reduction of the suffix impossible.

In French, Latin -ënsem and Germanic -isk have converged in one suffix, which appears under a double form, -ois and -ais. The suffix -ois, originally pronounced [we], either regularly developed into [wa] or was reduced into [e], undergoing irregular sound change due to frequency. I excerpted several pages in the newspaper Le Monde, where I found the following derivatives in -ois and -ais:

-ois: 13 chinois; 10 François; 2 québécois, suédois; 1 berlinois, dauphinois, hongrois, Luxembourgeois, munichois, patois, Rueillois.

-ais: 112 français; 23 polonais; 22 anglais; 11 japonais; 10 libanais, néerlandais; 3 hollandais, lyonnais, portugais; 2 new-yorkais; 1 aveyronnais, bourbonnais, écossais, irlandais, pakistanais.

Number Number Mean frequency

of attestations of words of occurrence

-ois 34 11 3

-ais 200 15 13

All these statistical data demonstrate that there is a connection between high frequency and irregular reductions of the Germanic suffix -isk- both in French and in Germanic languages .

The situation is similar if there are not double suffixes, but double words. It is not necessary to consult frequency dictionaries in order to know that the irregular a, my, Mr.,

Mrs., sir, business or Ger. hübsch, Herr are more often used than, respectively, one, mine, master, mistress, sire, busyness, Ger. höfisch, hehrer. The same observation concerns groups of words, e . g . good-bye, Ger. zwar are more frequently used than God be with you, Ger. zu wahr.

5 . FOURTH ARGUMENT

If irregular sound change due to frequency occurs within a paradigm or within a word family, it may be recognized by the fact that only the more commonly used forms are subject to it, whereas the forms used less frequently remain regular; e. g. the Old High German verb stantan, standan is reduced to stan, sten (> stehen). That this development is due to frequency is proved by the fact that the change concerns the more frequently used present tense and does not apply to the less frequently used preterite (stand). Besides, it should be pointed out that in Old High German texts collected in Braune's chrestomathy, the reduction takes place in the simple forms (stantan > stan, sten, whereas, in the majority of the compound forms such as gistantan, ajstantan, intstantan, ufstantan, ufarstantan, umbistantan, widarstantan, it does not occur (except in farstantan and arstantan). Obviously, the simple forms are generally used more often than the compounds.

The same applies in the case of the Old High German verb gangan, which is reduced to gan, gen in the present tense, which explains today's difference between the irregular gehen and the regular ging. Claiming our attention also is the fact that among the compound forms

of this verb mentioned in Braune, 14 do not show any reduction (ar-, bi-, fer-, ful-, fram-, in-, int-, missi-, ubar-, uf-, untar-, uj-, ujar-, zigangan).

When we consider Ger. haben, it appears that the forms in the singular present indicative, which are more often used, are shortened (hast, hat), whereas the plural forms haben, habt are regular. As far as the relation of this verb to its compound forms is concerned, it is worth comparing the irregular E has, have, had to the regular behaves, behave, behaved. Anyway, a similar phenomenon occurred in Old High German: haben was shortened to han, but the compound forms anthaben, bihaben did not exhibit any reduction.

6 . FIFTH ARGUMENT

If one compares two irregular sound changes due to frequency in a linguistic atlas, the area of the more frequent form is larger that that of the less frequent one . In different languages, the infinitive suffix undergoes an irregular reduction, e. g. in English, where give is shortened whereas Ger. geben is regular (Manczak 1993) . The same applies to Romance, Slavic, and Baltic languages. In a French text, the infinitive in -er occurs 101, that in -ir 36, and that in -oir 18 times . In the Atlas linguistique de France, I found that r is not pronounced in the infinitives in -er in 291 villages, in the infinitives in -ir in 188 villages, and in the infinitives in -oir in eight villages . In other words, the area of dropping r in the frequent verbs of the type aller is larger than that of dropping r in the less frequent verbs of the type dormir, while the area of dropping r in the rare verbs of the type avoir is the smallest .

7 . SIXTH ARGUMENT

If a frequency dictionary and a reverse dictionary are available for a given language, it is useful to examine series of words beginning or finishing in the same letter or letters. Here are some statistical data which I have established on the basis of the frequency dictionary by Thorndike and Lorge (1944) and the reverse dictionary by Lehnert (1971) .

In English, there are 15 words in -ave, e . g . save, among which only one shows a monophthongization of the diphthong, namely have, and it is worth noting that have is the most frequent of the words in -ave, cf 'the Lorge magazine count': have 24456, save 872, wave 478, grave 243, brave 216, slave 123, crave 74, shave 62, rave 53, cave 40,pave 38, suave 27, stave 11, knave 2, lave 2 .

There are 16 monosyllabic weak verbs in -ay, e . g . play, among which only one shows irregular reductions, namely say (says, said), and it is important to call attention to the fact that say is the most frequent of the verbs in -ay, cf 'the Lorge-Thorndike semantic count': say 3168, play 1456, pay 1270, lay 717, stay 463, pray 307, nay 205 (unfortunately, different parts of speech are not distinguished by Lorge and Thorndike), ray 148, hay 86, spray 77, clay 66, tray 26, fray 24, flay 10 .

There are 5 words in -ayer (in the dictionary by Thorndike and Lorge), e. g. layer, among which only one shows a reduced pronunciation, namely prayer, and prayer is the most frequent word in -ayer, cf 'the Thorndike general count': prayer 350, layer 90,player 90, slayer 16, payer 3 .

There are 9 words in -een, e . g . seen, among which only one may have a reduced pronunciation, namely been, and been is the most frequent word in -een, cf 'the Lorge magazine count': been 9870, seen 1402, green 1025, queen 342, screen 306, keen 168, sheen 13,preen 7, spleen 4 (while other words in -een, e . g . ween or peen, are not mentioned in the frequency dictionary).

There are 218 feminines in -ess, e . g . princess, among which only two underwent a reduction, namely mistress > Mrs. and Miss, and Mrs. and Miss are the most frequent feminines in -ess. For lack of space, I mention only the disyllabic feminines: Mrs. 3651, princess 254, actress 166, hostess 157, duchess 109, mistress 97, waitress 39, countess 22, goddess 21, empress 18, heiress 15, laundress 10, tigress 7, abbess 3, huntress 3,priestess 1.

There are 34 monosyllabic words in -f, e g if, among which only one shows an irregular voicing of the final consonant, namely of, and of is the most frequent word in -f, cf 'the Lorge magazine count': of 112601, if 14506, half 1984, chief545, roof 417, brief340, golf 278, poof 194, leaf 155, beef 148, proof 141, grief 137, wolf 121, self 115, shelf 100, loaf 89, scarf 88, thief 65, deaf 53, gulf 52, wharf 46, calf 44, dwarf 43, chef 39, hoof 35, sheaf 40, surf 21, turf 17, waif 10, elf 6, coif 5, woof 5, reef 4, serf 2 .

There are 34 words in -ill, e g still, among which only one may have a reduced pronunciation, viz . will, and will is the most frequent word in -ill, cf 'the Lorge magazine count': will 9573, still 2356, bill 1403, fill 1137, kill 919, till 453, thrill 432, ill 341, hill 335, chill 310, mill 253, skill 104, shrill 83, spill 74, drill 60, grill 54, sill 46, frill 44, jill 42, gill 41, pill 30, windmill 23, downhill 18, refill 17, standstill 15, uphill 15, trill 14, sawmill 14, distill 13, instill 12, quill 8, rill 4, whippoorwill 3, twill 3.

There are 15 words ending in fricative + -in, e . g . coffin, among which only two underwent a reduction, viz cousin and basin, and these words occupy the first and the second positions, cf 'the Thorndike general count': cousin 350, basin 160, coffin 90, raisin 90, dolphin 50, muffin 28, assassin 18, paraffin 18, resin 18, toxin 18, dauphin 14, rosin 14, elfin 12, regamuffin 6, griffin 4.

There are 12 monosyllabic words in -ine, e g line, among which only one underwent a reduction, viz . mine > my, and my is more often used than any word in -ine, cf 'the Lorge magazine count': my 22184, line 1498, mine 1119,fine 1078, nine 468, shine 328, dine 326, pine 172, wine 156, vine 119, whine 75, thine 5, sine 2 .

There are 700 derivatives in -iness, e g business, among. which only one was shortened, viz business, and business is the most frequent word in -iness, cf only several examples from 'the Thorndike general count': business 700, happiness 300, weariness 90, readiness 57, loneliness 28, loveliness 18, greediness 14, friendliness 12, steadiness 12, unhappiness 8, tardiness 7

There are 36 words in -ire, e g fire, among which only one underwent an irregular reduction, viz sire > sir, and, from the point of view of frequency, sir occupies the second position, cf 'the Lorge magazine count': fire 1319, sir 876, tire 865, desire 779, require 594, wire 449, entire 406, inquire 288, admire 257, hire 248, acquire 221, retire 165, inspire 154, empire 89, dire 30, attire 27, sapphire 25,perspire 17, spire 17, conspire 16, aspire 15, mire 14, esquire 14, bonfire 14, expire 13, satire 12, afire 10, squire 9, vampire 9, umpire 8, sire 7, wildfire 6, transpire 5, quire 2, ire 2, grandsire 1.

There are 11 monosyllabic words in -one, e g tone, among which only one underwent

a monophthongization, viz . gone, and gone is the most frequent word in -one, cf 'the Lorge magazine count': gone 1859, tone 536, bone 393, stone 386,phone 272, zone 51, throne 50, drone 42, lone 38, cone 24,prone 19 .

According to Berndt (1960: 132), 'früher als in starktonigen Wörtern gleicher Struktur schwindet ausl. -3 in im Satz schwächer akzentuierten Wortformen... son (< ae. sona "bald")'. In reality, the irregular sound change of soon is due to frequency because soon is more often used than any word in -oon, cf 'the Lorge magazine count': soon 1445, afternoon 1033, moon 285, noon 236, teaspoon 211, tablespoon 154, spoon 115, honeymoon 110, saloon 95, balloon 69, coon 57, cartoon 25, boon 24, maroon 20, macaroon 17, loon 15, croon 14, cocoon 13, forenoon 11, lagoon 11, dragoon 10, baboon 9, platoon 7, swoon 5, harpoon 4, raccoon 3, racoon 3, buffoon 2, monsoon 2

Welna (1978: 40) claims that 'the development of /o/ before /rd/ is not clear. The forms with lengthened /o/, like LOE bord 'board', ford, hord 'hoard' and LNthb * sword, seem to have survived as late as ENE and the existence of the long vowel is confirmed by the spelling oa in MoE board, hoard, which however indicates an earlier /o:/, not /o:/... LOE word was later replaced by its variant with short /u/.' This shortening is due to frequency because word is more often used than the other words, cf 'the Lorge magazine count': word 2845, board 825, sword91, ford 89, hoard 22 .

There are 9 monosyllabic words in -our, among which only one may show a reduction of -our to [a], viz. your, and your is the most frequent word in -our, cf 'the Lorge magazine count': your 9052, our 7599, hour 2485, four 1637,pour 556,flour 396, tour 149, sour 102, scour 20

There are 6 words in -over, e g clover, among which only one may have a reduced pronunciation, viz . over > o'er, and over is the most frequent of these words, cf 'the Lorge magazine count': over 7520, moreover 153, clover 29, drover 6,plover 3, rover 2 .

According to Jespersen (1922: 259), fellow 'in careless everyday pronunciation is often made [fela], in novels, etc., written feller, fella'. This reduction is to be accounted for by frequency because, among disyllabic words in -ow, fellow occupies the second position, cf 'the Lorge magazine count': window 1564, fellow 860, shadow 491, widow 181,pillow 161, elbow 134, sorrow 130, eyebrow 115, meadow 70, willow 47, rainbow 45, arrow 39, burrow 37, harrow 27, barrow 20, billow 19, furrow 17, sparrow 11, hedgerow 10, tallow 6, minnow 2

There are 7 numerals in -teen, e g thirteen, but the numeral ten, which, from the etymological point of view, is identical with -teen, underwent a reduction, and ten is used more often than all numerals in -teen, cf 'the Lorge magazine count': ten 1260,fifteen 410, eighteen 215, sixteen 194, fourteen 143, seventeen 139, nineteen 109, thirteen 92 .

For more examples see Manczak 1987.

8. IRREGULAR SOUND CHANGE DUE TO FREQUENCY IN MORPHEMES

It takes place in formative morphemes, e. g. the German suffix -lich, which occurs in many adjectives, is subject to a reduction only in the two most frequently used derivatives of this type, namely welch and solch. However, irregular sound change due to frequency

occurs in inflectional morphemes considerably more often. This is due to the fact that inflectional morphemes are more frequently used than derivative ones. Here are some examples.

The opinion on the origin of the weak preterite in the Germanic languages is not unanimous: some consider it to be derived from the 2nd person aorist of the middle voice, others, from the verb corresponding to Ger. tat. In the light of the theory of irregular sound change due to frequency, the latter hypothesis should be considered the correct one. In Gothic, the conjugation of the preterite indicative and optative was the following:

Indicative Optative

Singular nasida nasidedjau

nasides nasidedeis

nasida nasidedi

Plural nasidedum nasidedeima

nasidedup nasidedeip

nasidedun nasidedeina

Dual nasidedu nasidedeiwa

nasideduts nasidedeits

The distribution of forms with and without reduplication is the following: (1) the less frequent optative displays reduplication in all its attested forms, whereas the more frequent indicative does not show it in all the attested forms; (2) within the indicative forms, the less frequent plural and dual exhibit reduplication, but the more common singular does not. Thus, everything suggests that, originally, the reduplication existed in the singular preterite indicative, and its disappearance should be accounted for by an irregular sound change due to frequency, which first attacked the forms used most often and only then the forms used less commonly. This resulted in the state known from the Germanic languages attested later than Gothic, where there is no trace of any reduplication in the forms of the weak preterite It is understood that the weak preterite exhibits other unclear points, but this often happens when irregular sound change due to frequency operates. If they had not known Latin, the comparativists would have maintained that the forms of the perfect Fr. chant-a, It. cant-d, and Rum . cint-a derive from three different forms with asterisks . However, it is known that all these forms derive from one form cant-avit, and the variation chant-a, cant-d, cint-a is to be explained by the fact that the reductions due to frequency may manifest themselves in different ways, cf the Old High German counterparts of the modern und, quoted above

Another example: in Middle High German in some dialects, the -n of the infinitive disappears, e . g . nemen > neme. A similar change has taken place in English, where a difference exists between e . g . to give and the participle given. This change is also due to frequency, which is proved by the fact that infinitives without n (typical both of strong and of weak verbs) are more often used than participles with n (only typical of strong verbs). In the same way, frequency accounts for irregular simplification of a geminate in the Old High German gerunds of the type nemanne 'zu nehmen' > nemane.

Further example . In Gothic and in Old High German, the declension of the a-stem nouns was of the following shape:

Gothic

Sing . Nom . *-a giba +

geba + geba +

Old High German

Acc . *-am giba +

Gen. *-as gibös

Dat. *-ai gibái

geba, gebu + geba, gebu +

Plur. Nom . *-as gibös

geba + geba + geböno geböm

Acc . *-ans gibös Gen. *-öm gibö

Dat. *-amis giböm

As is known, the Indo-European *a and *o result in *o in Proto-Germanic. Therefore, the development of the forms not marked by crosses is regular. In historical grammars, this double development is accounted for by the existence of the acute and the circumflex intonation in Proto-Indo-European. This explanation gives an impression of an ad hoc explanation for two reasons: (1) the distribution of regular and irregular endings in Gothic differs considerably from that of regular and irregular endings in Old High German; (2) everything indicates that the Balto-Slavic intonation arose independently of Greek; therefore there is no proof that any intonation existed in proto-Indo-European. For these reasons, the irregular endings of the Gothic and OHG a-stem nouns are to be accounted for on the basis of their frequency, which is proved by the fact that, both in Gothic and in Old High German, the irregular development occurs in the more frequently used endings, since it is known that (a) the singular is used more often than the plural, (b) the nominative and the accusative are used more often than the dative and the genitive. As a parallel, we may cite the fact that, in the frequently used Latin nominative singular of the type tabul-a, the final vowel underwent an irregular shortening, whereas, in the less frequently used ablative singular tabul-a, the old length was preserved

Still another example. It has been asserted that, in Old High German, in the final syllable, Proto-Germanic *a disappeared earlier than Proto-Germanic *i and *u, which means that e g oHG tag is as regular as oHG wini or sunu. In my opinion, this assertion should be doubted for two reasons:

(1) The disappearance of final unstressed vowels occurs in various languages, but I do not know a language where the final unstressed a (which is considerably wider than i, u) disappeared earlier than i, u (whereas the disappearance of i, u prior to the disappearance of a is often found) .

(2) In Old High German, there are some words which are not o-stem nouns, and which end in a, e . g . ana (> an), aba > ab, unta (> und), fona (> von). The majority of these words derive from words ending in *a or *o in Proto-Indo-European: ana < *ana, aba < *apo, unta < *nta. Since it is known that Proto-Indo-European *a and *o merge into *a in Proto-Germanic, the question arises of how to reconcile the fact that the Proto-Indo-European *a, *o were preserved in OHG ana, aba, or unta, and the fact that the ending *-o-s in the nominative singular of the o-stems disappeared in Old High German, which resulted in the nouns of the type tag

In my view, Old High German words of the type ana, aba, unta show a regular change of the final unstressed Proto-Germanic *a (< Proto-Indo-European *a, *o), whereas the disappearance of *-o-s in the nominative singular of the o-stem nouns is to be accounted for

by an irregular sound change. To put it differently, the final vowel disappeared irregularly in the frequently used o-stem nouns (OHG tag), whereas it was regularly preserved in the less often used i- or M-stems (OHG wini, sunu). Such an interpretation of these phenomena in Old High German is supported by parallels drawn from outside the Germanic languages. In Old Church Slavic, the i- and M-stems show a regular development (gost-i, syn-u), but the masculine o-stem nouns show an irregular reduction (grad-u), whereas the less frequently used o-stem neuters exhibit a nearly regular development (let-o). A similar situation occurs in Lithuanian . Although no difference exists between the o-, i-, and M-stem nouns (vilk-as, ant-is, turg-us) in the literary language, there are dialects where the reduction takes place in the o-stems (vilk-s), whereas the development is regular in the i- and M-stems. An analogous case occurred in Old Prussian, where the more frequently used o-stem masculines exhibited both the full and the reduced endings (deiw-as and deiw-s), whereas the ending was always regular in the o-stem neuters (assar-an) .

REFERENCES

Berndt Rolf. 1960 . Einführung in das Studium des Mittelenglischen. Halle (Saale): Niemeyer.

Diez Friedrich. 1846. Altromanische Sprachdenkmale. Bonn .

Jespersen Otto . 1922. A Modern English Grammar. I . Third ed . Heidelberg: Winter.

Lehnert Martin . 1971. Rückläufiges Wörterbuch der englischen Gegenwartssprache. Leipzig: VEB Verlag Enzyklopädie.

Manczak Witold . 1969 . Le développement phonétique des langues romanes et lafréquence. Krakow: Uniwersytet Jagiellonski.

Manczak Witold . 1987 . Frequenzbedingter unregelmäßiger Lautwandel in den germanischen Sprachen. Wroclaw: Ossolineum.

Manczak Witold. 1993. "Loss of the final n in English." KwartalnikNeofilologiczny 40, 21-30.

Pott August F. 1852. "Plattlateinisch und Romanisch." Zeitschrift für vergleichende Sprachforschung auf dem

Gebiete des Deutschen, Griechischen und Lateinischen 1, 309-350. Thorndike Edward L . , Lorge Irving. 1944. The Teacher's Word Book of 30,000 Words. New York: Columbia University.

Welna Jerzy. 1978. A Diachronic Grammar of English. Warszawa: PWN .

Allatum die 4 mensis Novembris Anno 2009