Words which in southern Greek dialects contain [azis], [úsis], and other sequences of stressed vowel + sibilant + [i] + sibilant, appear widely in the north with [áys], [úys] etc., and the phenomenon constitutes a major crux in modern Greek phonology. Although the problem has been widely discussed, no really satisfactory solution based on attested processes has been offered. It is the purpose of this paper to review the data, criticize existing treatments, and offer a new proposal.

In a somewhat condescending review of A Dictionary of Canadianisms on Historical Principles some years ago, Mavor Moore complained about “a slighting of spoken practice and of distinctive Canadian syntax,” among his objections being the absence of an entry for eh?, for “both the English and the Americans can spot a Canadian from his ‘eh?’ at the end of a sentence: ‘It’s hot, eh?’” Admittedly, the interjection is not in the DC; moreover, there are no slips for eh? in the citation files of the Lexicographical Centre for Canadian English. This situation certainly indicates that the readers for the DC did not consider it a Canadianism: either that or they were unaware that eh? might be relevant to the collection.

The second of these possibilities may be set aside, for eh? is, in fact, entered in the Intermediate Dictionary and the Senior Dictionary, general dictionaries of English published in Canada for Canadians as part of the Dictionary of Canadian English series, to which the DC belongs. It should be added that eh? is also entered in every general dictionary of English, both British and American, on my shelves—a very considerable collection indeed, and one which includes the Oxford English Dictionary and Webster’s New International Dictionary (Third Edition).

There are in principle infinitely many descriptions to fit the data of language, whether synchronic or diachronic. Although in practice one is seldom confronted with more than a few alternatives within any one descriptive system, the difficulty of choosing between these alternatives in a principled way has been well recognized since Chao. Two sorts of criteria have traditionally been and continue to be invoked more or less explicitly, criteria which may be subsumed under the headings of simplicity and naturalness. Historical linguists have long had to cope with certain sets of data for which neither type of criterion could dictate unique solutions. There is now a growing list of similar dilemmas in generative descriptions. The present paper presents a case where non-unique generative and reconstructional accounts

Martin Joos once said: “In matters of language, Chao never goes wrong,” or words to that effect. I certainly appreciate this great compliment, but in many ways which Martin has not noticed, I did go wrong quite a number of times in matters of language.

It goes without saying that in my pre-adult speech, which was of course way back in the pre-Martin Joos days, there were a number of things in which I deviated from the speech of my environment. In the form of Mandarin spoken around me retroflex initials were distinct from dental sibilants; I used dental sibilants for both. There was also final n after front vowels, but I dropped all the final n’s. Thus, when I noticed that a cat had lapped up my bowl of noodles, I exclaimed: Mhau tsy woode mieh! for Mhau chy woode miann! ‘The cat is eating my noodles!’

In recent years the interests of linguists have turned more and more to the problems of generative grammars. A number of models have been proposed, and in debate sharply opposed to one another, but they all share a number of fundamental assumptions. I will frequently use transformationalists as examples of linguists, and transformational treatments as examples of generative grammar, and I will start from a definition worded in a characteristically transformational-generative way. I do this because they are the most generally familiar instances. But what I have to say will apply with only small adjustments to, say, stratificational generative linguists and linguistics as well.

The source of this word in a borrowing from what we see in late Latinized form as paraverēdus is well known. The form is a compound belonging to technical language, para- + the originally Gaulish verēdus ‘a post-horse (perhaps originally a special kind thereof).’

There is a slight subsidiary problem with the exact value and analysis of the Celtic form; we do not propose to try to solve it here, but it will be sufficient to point out its nature so that an important aspect of the Germanic words may be understood. In the British section of the Antonine Itinerary we find the name Voreda.

1. Contrastive linguistics. This term has had some vogue in recent years as a name for the comparative description of language structures with a view to improving the quality of language teaching. A concise definition is offered by one worker in the field: “Contrastive linguistics offers hypotheses concerning identifications a learner will make between elements of his base and target systems, thus providing predictions and explanations concerning his learning behaviour of presumed high value in planning learning and teaching strategy” (Nemser 1970, forthcoming: 2-3).

In 1959 the Center for Applied Linguistics launched a Contrastive Structure Series under the editorship of Charles A. Ferguson. Six volumes have appeared so far, in which German, Spanish, and Italian phonological and grammatical structures have been compared with English for the benefit of teachers, textbook writers, and ultimately students of these languages. The Center has gone on to organize, under William Nemser’s direction, a number of Contrastive Projects in Yugoslavia, Romania, and other countries, in co-operation with local scholars in linguistics and language teaching.

One of the particularly fine passages in Sir Gawain and the Green Knight is the description of the Castle when Gawain first sees it. Yet when one studies the way in which translators have handled it, one finds that many modern versions seem unnecessarily colourless, and others though more colourful, are freer than needful. I have consequently gathered a fairly full sampling of modern versions of the passage, with the purpose of searching for possible principles of translation which might be useful elsewhere as well as in the passage itself. It seems fitting that the results should be presented here, since the chief tool I have used is the semantic law formulated by Martin Joos in 1953. The law, as it applies to a given context, states that the possible meanings are to be judged as to which contributes least to the totality of the context. I have discussed and used the law frequently with students, and also in print in 1959 (Hill 1959). The present essay constitutes a further presentation of the results of using the law.

Words do not seem to matter equally in different cultures. Nor do they matter equally to all people. It is therefore altogether appropriate that an essay honoring Martin Joos should deal with the meaning of a word in a Middle High German courtly epic; for there have been few cultures in which words counted for so much as in medieval German courtly society, and there are few people to whom words matter so much as Martin Joos. I mean this of course not in the trivializing sense that mere words matter to the detriment of substance, but in the ancient, honorable sense that concern for words and their relations reflects profound involvement in and understanding of language and culture. To Martin Joos, then, I dedicate this exploratory study of triuwe in Gottfried’s Tristan.

In reconstructing prior stages of languages the comparative method (CM) has been primarily applied to phonology. Two standard statements on the CM in recent times illustrate this restricted application. In his article of 1950 on the comparative method, Hoenigswald defines “reconstruction by the comparative method (as) essentially a problem in phonemics, in which the place of allophones is taken by sets of sound correspondences that are partially alike (share one component) and in complementary distribution” (Joos 1957: 298). And in his comprehensive statement (1960:132), Hoenigswald states that “the comparative method is based on the principle that sets of recurring phoneme correspondences between two related languages continue blocks of positional allophones from the mother language.” Handbooks dealing with reconstructed languages reflect the limitation to the phonological component of language. For statements made about syntax are hesitant (Meillet 1967: 29). And one notable attempt to reconstruct a syntactic sequence in Proto-Indo-European, Schleicher’s fable of the horses and the sheep, has been generally condemned, even though it was given some recognition by Hirt (Meillet 1967: 29, Hirt 1939: 113-115). In contrast with the earlier hesitation about applying the CM to the syntactic component of language I would like to suggest that recent developments in linguistics have made it possible to apply the CM to at least some syntactic patterns in much the way historical linguists have been dealing with phonological patterns.