9.

The Adaptation of Foreign Elements

9.1 Adaptation of French Words to Middle English

During three centuries of bilingualism several thousand word stems were taken into English from socially and culturally dominant French [Baugh 1957: 200–209], but little else was permanently adopted.

The imported word stems were given native inflections.

The phonemes of these word stems were articulated in the English way: thus /p, t, k / before stressed vowels were aspirated; or else they were replaced by phonically similar native phonemes: thus the French rounded mid front vowel / ö / by the unrounded / ē / of ME, as in beef, people. Contrastive vowel length was imposed in accordance with native phonotactic rules: hence ME long vowels in word final and prevocalic positions, as in the ME antecedents of MnE vow, agree, blame, move, robe [Luick 1914-27: 446 ff.].

The stress patterns of the adopted French words were in the course of time also largely adapted to native patterns, as in cíty, énvoy, surface, fore stress replacing end stress [Kurath 1964: 146- 48].

Aside from the lexicon, the effect of French upon the English language is thus remarkably slight. Only one French phoneme, the / ɔi / of joy, joint, was added directly to the English inventory; another, the / ž / of vision, measure, developed via / žj / from a French sequence. Initial /ğ, v, z /, as in joy, veal, zeal, which in ME occurred only medially and finally, is the only other contribution of French to English phonology.

The behavior of English in its confrontation with French, the culturally dominant language of Great Britain for several centuries, may seem to be unusual, but is, in fact, rather typical. The well integrated grammatical and phonological system of a language and its prosodie features strongly resist remodeling under foreign influence, whereas word stems are freely adopted and adapted to the native system phonologically, prosodically, and morphologically.

9.2 Phonological Adaptation In South Bavarian

The persistence of the native system of phonemes and phonotactic rules is strikingly shown in the behavior of a South Bavarian dialect (Carinthian), spoken on the southern slope of the Austrian Alps, in its confrontation with a regional type of Standard German taught in the schools.

This folk dialect has four ingliding diphthongs, as in / liəb / “lieb, dear,” /guət / “gut, good,” /rεəstn / “rösten, roast,” /prɔəsn / “brosamen, bread crumbs.” Of these, /iə/ and /uə / are stable, but / εə / and / ɔə/ were being replaced, word by word, at the turn of the century, the former by / ē / or /ε̄/ and the latter by /o/. Thus rustic /grεəsr /”grösser, greater” was being replaced by refined /grēsr /; / wεə / “weh, ache” by / wé /; / grɔəs / “gross, great” by /grōs / [Kurath 1965: 56–57].

This replacement of the diphthongs is prompted by school-taught German, which lacks ingliding diphthongs and has long monophthongs in such words, i.e., /ē /, /ȫ/, and /ō /. However, the choice of the replacement depends solely upon the resources of the dialect, which has the monophthongs /ē, ε̄,ō/ from other sources, as in /lēbm / “leben, live,” /ε̄ptn / “beten, pray,” /xōfn / “hoffen, hope,” and lacks rounded front vowels. Hence no alien phoneme is adopted in these cases. The “induced” changes conform to the indigenous system of vowels; only their lexical incidence is changed.

The “tyranny” of phonotactic rules is also well exemplified in this South Bavarian dialect. With few exceptions resulting from rather recent changes—the late indigenous shortening of the earlier geminates /pp, kk / and the replacement, under the influence of standard German, of final / -nkx / by / -η /—vowels are long before single consonants, short before consonant clusters [Kurath 1965: 5-6]. When a word is adopted from Standard German or from Viennese— which, like other Central Bavarian dialects, has a different positional regulation of vowel duration [Koekoek 1955: 35-36; Kufner 1957: 175-84]—this rule is regularly imposed. Thus Standard German /knoxən / “bone” becomes /kxnōxn / (the native term is / pān, / corresponding to E bone); and Viennese inflected /nƐttə / “nette, neat” and /fƐšša / “fesche, fashionable, handsome” appear in this dialect as / nε̄tə / and /fε̄šə /.

9.3 The Adaptation of English Words to Norwegian Word Tones

That the tonal patterns of words are imposed by monolinguals upon words adopted from a foreign language is shown in full detail by E. Haugen in his investigation of the Norwegian spoken in some of the Midwestern states [1953: 416–19].

It is a characteristic feature of Norwegian that the stress pattern of polysyllabic words is invariably combined with one of two tonal patterns, one of them simple, the other complex. The simple word tone consists of a rise or a fall in pitch (the former in the eastern and the latter in the western dialects of Norway), and the complex word tone of a rise-fall-rise. The tonal pattern extends from the stressed syllabic to the end of the word or phrase.

When English words are adopted, one of the Norwegian tonal patterns is imposed upon them, the choice depending primarily upon Norwegian morphological types. The onset of the tonal pattern is determined by the location of the main stress of the English word.

Symbolizing (a) the simple tonal pattern by the acute and (b) the complex tone by the circumflex, this process of adaptation can be illustrated by the following examples cited by Haugen: (a) héndel < handle, bísnes < business, píknik < picnic, fottbål < football, lármklokka < alarm clock; (b) fârmar < farmer, hômmstedd < homestead, bêsment < basement, bêbi < baby, hârvista < harvest v., fênsapåst < fence post.

9.4 Adaptation of Latin to Celtic Speech Habits in Northern Italy

Since the latter part of the nineteenth century, possible influences of underlying pre-Latin languages upon Italian regional dialects have been proposed and vigorously debated. In a stimulating paper published in 1930, G. Rohlfs [1952 < 1930: 61–79] reviewed the several problems and came to the following decisions. (1) In southern Italy several syntactic constructions and the survival of the Latin perfect as a preterit can be attributed to the Greek substratum; some Greek words have been adopted, but no phonological features. (2) OscanUmbrian and Etruscan influence cannot be demonstrated and must be rejected; (3) in Northern Italy, the Celtic substratum has produced some striking changes in the Italian dialects.

Rohlfs’ discussion of the effects of the Celtic underlay is summarized below.

The dialects of northern Italy differ markedly from those spoken in other parts of the peninsula. The bundle of heterophones that follows the Apennines from Spezia-Luca in the west to Rimini-Ancona on the Adriatic is the most prominent dialect boundary within Italy. Not only that: it outranks the dialect boundary between Italy and France, and is conventionally recognized as the dividing line between the Western and the Eastern Romanic languages.

Rohlfs cites the following well known groups of heterophones that set the North Italian dialects off from Tuscan:

(1) The loss of final unaccented vowels, except a, as in gros ≠ Tuscan grosso “big, thick,” nas ≠ naso “nose”;

(2) the syncopation of pretonic and posttonic vowels, as in slar ≠ sellaio “saddler,” vedva ≠ vedova “widow”;

(3) the elimination of double consonants, as in boca ≠ bocca “mouth,” spale ≠ spalle “shoulders”;

(4) the voicing of voiceless phonemes between vowels, as in ortiga ≠ ortica “nettle,” fradel ≠ fratello “brother,” cavei ≠ capelli “hair”;

(5) the occurrence of nasal vowels, as in pã ≠ pane “bread.”

Since these features of North Italian have parallels in all of France, he attributes them “without hesitation” to the Celtic substratum shared by northern Italy with France. The fact that the earliest written documents exhibit these innovations is taken to support the inference from the areal congruence of these linguistic subareas of Western Romanic with the originally Celtic domain.

Rohlfs does not comment on the phonemic background of these alleged substratum effects beyond stating that Insular Celtic exhibits the loss of initial and final unstressed vowels, which (following H. Pedersen 1909: 1.260) he attributes to the strong primary stress. This is clearly a weakness in his treatment of the presumed influence of Celtic on the Romanic overlay.

Rohlfs [77-78] rejects the widely held view that the fronting of Latin ū in northwestern Italy resulted from the Celtic substratum on the following all but irrefutable grounds:

9.5 Adaptation of Latin to Celtic Speech Habits in West Romanic

Several phonological changes in the Western Romanic languages have been attributed to the influence of the Celtic substratum and vigorously debated [A. Kuhn 1951: 52–57], among them the “weakening” of word-medial Latin /p, t, k /, as in Latin piper “pepper” > Provençal pebre, French poivre; Latin sēta “hair, bristle” > North Italian seda, Spanish seda, French soie “silk”; Latin lacus “lake” > North Italian lago, Spanish lago, Old French lai. Whether this development resulted from the persistence of Celtic speech habits in the acquired Latin or from phonemic or prosodie “drives” inherent in Latin itself is by no means easy to decide with anything like finality. Even a brief consideration of some of the circumstances should make that clear.

Influence of Celtic speech habits upon the Latin overlay is possible on ethnological grounds, since northern Italy and most of Western Europe, conquered by the Romans between 200 B.C. and A.D. 100, was Celtic territory at the time of the conquest. Thus there is areal congruence between the old Celtic language area and the West Romanic area in which Latin medial /p, t, k / were “weakened”—a fact that suggests possible influence of Celtic upon Romanic.

Evidence for the appearance of “weakened” reflexes of proto-Indo-European /p, t, k / in Celtic and in West Romanic is skimpy and circumstantial. For Gaulish (Continental Celtic), which was submerged by Romanic between 200 B.C. and A.D. 400, there is no spelling evidence for the “lenition” of /p, t, k / in the inscriptions. Insular Celtic “lenition” is fully established when written documents first make their appearance—ca. A.D. 700 in Irish, later in Welsh. It is with reference to the British branch of Insular Celtic that H. Pedersen [1909: 1 242, 436, 462] assumes the “strong” — “weak” positional allophones /p^c ~ p, t^c ~ t, k^ ~ k / for Gaulish. If Pedersen’s assumption stands, these positional variants could have been carried over into Romanic by speakers of Gaulish, the “weaker” allophones later yielding / b, d, g /, whence the further changes illustrated in the examples given above.

It is of considerable importance to point out that “lenition” of /p, t, k /—as also the shortening of all geminate consonants—is confined to word medial position in West Romanic, whereas in Insular Celtic it occurs also in phrase medial position. “Strong” /1, k / thus survive in Celtic only as phrase initials and in certain clusters of obstruents. This distribution of the reflexes of earlier /1, k / results in contextual variation of the initial consonants of words in Celtic for which West Romanic has no parallel.

It should further be noted that “lenited” /t, k / appear as the voiced plosives / d, g / only in the British branch of Insular Celtic (e.g., Welsh). The Goidelic branch (e.g., Irish) has the voiceless fricatives /θ, x / from this source [Pedersen 1909: 1.429].

For “weakened” reflexes of medial Latin /p, t, k / in West Romanic there is no spelling evidence before ca. A.D. 750. Hence the time of their first appearance is unknown.

The tangible evidence that has, or may have, a bearing upon the interpretation of the positional weakening of Latin /p, t, k / in West Romanic can be briefly summarized as follows.

(1) In the Romanic languages this development is largely confined to an area in which Romanic was superimposed upon Celtic.

(2) The positional phonemic split of /p, t, k / is shared by West Romanic with Insular Celtic, as shown by spelling evidence from ca. A.D. 750 onward.

(3) West Romanic differs from Celtic in the distribution of the weakened reflexes of /p, t, k / in that they are confined to word medial position.

(4) Within Insular Celtic, lenited /t, k / yield voiced plosives, /d, g /, in British but voiceless fricatives, /θ, x /, in Goidelic.

(5) Continental Celtic (Gaulish) is assumed to agree with the British branch of Insular Celtic.

(6) A further factor, usually not taken explicitly into account, is the practical identity of the system of simple and geminated plosives in Proto-Celtic and Proto-Italic, which might lead to parallel development.

Under these circumstances it is difficult to decide with finality whether the treatment of word medial Latin /p, t, k / in West Romanic resulted from internal trends of Latin or whether it was introduced by speakers of Gaulish. Most scholars in the Romanic field take it as a reflex of the Celtic substratum, as does Rholfs (see above).

A. Martinet [1952: 216], who might be expected to champion indigenous development in West Romanic, concludes his discussion of this intricate problem as follows:

9.6 An Etruscan Phonological Reflex in Tuscany?

This phonological change, the so-called gorgia toscana, has been widely attributed to the Etruscan substratum of this part of Italy, where the Etrusian language was spoken no later than A.D. 200. This explanation was emphatically rejected by G. Rholfs [1952 < 1930: 71- 75], who pointed out that Etruscan phonological evidence is at best highly ambiguous, that Dante makes no mention of it, and that on the evidence of the Linguistic Atlas of Italy [Jaberg-Jud 1928-40] this phenomenon is now restricted to western Tuscany, including Florence and Siena.

In a short article, R. A. Hall, Jr. [1949] brings the full evidence of the LAI to bear upon this problem, which should dispel the ghost of the survival of Etruscan speech habits for ever. He maps the dissemination of affricates and fricatives derived from intervocalic /p, t, k / in 12 words for each of the three phonemes and makes the following discoveries:

(1) affrication of /k / is fully established in northwestern Tuscany, including the cities of Florence, Siena, and Pisa; (2) affrication of / t / is regular in Florence-Siena and environs, but increasingly sporadic farther west (as in Pisa); and (3) the affrication of /p / is largely confined to central Tuscany, being most frequent in Florence and Siena (see Figure 47).

From the behavior of the isoglossic lines Hall draws these inferences: (1) Florence is the center in which the affrication originated and from which it has been spreading southward via Siena and westward down the valley of the Arno in the direction of Pisa. (2) The affrication (aspiration) of the velar, being most widely disseminated, must be earlier than that of the dental, and that of the labial lags far behind. (3) Diffusion of this phonological feature from Florence- Siena is still in progress.

The unmistakable areal evidence for a rather recent shifting of the intervocalic voiceless plosive /p, t, k / to affricates or fricatives is supported by the fact that there is no evidence for this development before the sixteenth century and that the literary language of Italy, based upon the Tuscan dialect, completely lacks this phenomenon.

FIGURE 47: Tuscany: Affricated and Fricative Reflexes of Latin /p, t, k/

1. [kh ~ x ~ h] in fuoco “fire”

2. [th — θ] in marito “husband”

3. [ph ~ f] in rapa “turnip”

Based on Maps 1, 2, 3 in R. A. Hall, Jr., A Note on “Gorgia Toscana.”

Hall concludes: “We must therefore abandon any idea of correlation between the present-day extent of the ‘gorgia toscana’ and the boundaries of ancient Etruria.”