In 1928 the Polish anthropologist Jan Czekanowski published in the ethnographical quarterly Lud1 a study of the Indo-European languages in which he employed the method of differential diagnosis by quantitative correlation determinations which he had long been using with success in physical anthropology and ethnography. This method, whatever its field, rests upon the recognition of isolable and definable features or traits, which we shall hereafter refer to as elements, whose presence or absence can be determined for a number of populational groups or territorial entities, such as races, tribes, cultures, castes, or, in the present study, languages. The distribution of these is tabulated in terms of plus for presence in a particular group, minus for absence, and the question mark for unknown. Then each group is compared with each of the other groups in terms of the four-cell segregation familiar to statisticians. That is to say, four values are determined: a represents the number of elements common to both groups, b the number present in the first but absent in the second, c the number absent in the first but present in the second, and d the number absent in both. In other words, a and d are agreements, positive and negative respectively; b and c are disagreements. These four values are then substituted in a suitable formula, and a coefficient of similarity between the two groups results. When the coefficients for each pair of the groups being considered are assembled, we get a classification of the relative degrees of similarity between the populational groups or territorial entities which, being objective, has genetic and historical significance. For example, to make this clear in linguistic terms, suppose for the four languages Baltic, Slavic, Indic, and Iranian, we get a high coefficient between Baltic and Slavic, and again a high coefficient between Indic and Iranian, but low coefficients for the four other possible pairs, Baltic-Indic, Baltic-Iranian, Slavic-Indic, and Slavic-Iranian. The coefficients thus make the four languages fall into two classes, Baltic-Slavic, and Indic-Iranian, and it is evident that each class has had a certain history common to its members but not shared by members of the other class.

Sturtevant in HG 144, speaking of relatively transparent nouncompounds in Hittite, goes on to say ‘One may be almost equally sure that tayukas (ta-a-i-ú-ga-aš, da-a-i-ú-ga-aš, ta-a-ú-ga-aš) ‘two years old’ beside yukas (i-ú-ga-aš) ‘one year old’ is a compound whose prior member is akin to tan (ta-a-an, da-a-an) ‘iterum, second’; but our ignorance of the etymology of the second member makes its classification impossible'. This article is an attempt to explain the word yukas ‘yearling’.

In SP and LV the Sanskrit gerund endings tvā and ya are used virtually as in Sanskrit. In addition several Prakritic endings are used. Their application is radically different in two ways: (1) No distinction is made between simple and compound bases. (2) The endings are attached not to the ‘root’ as ordinarily in Skt., but to a thematic present stem (often one which is itself Prakritic and unknown to Sanskrit). This thematic present stem, incidentally, shows a tendency to make itself the base for the whole verb conjugation.

Under breaking I include here the change of a front vowel in a stressed syllable to a diphthong by the influence of following velar elements whether these be consonants in the same syllable or vowels in the next. We have to do then with OE consonantal breaking and with ON breaking, and also with that phase of velarization of front vowels in OE which, since the eighties of the last century, has come to be called u-o/a-umlaut. The last is, of course, a velar-vowel breaking. My thesis is that these three phenomena are not only similar in nature but directly related in origin.

In the following I will discuss a few features of the language of the Dakota Indians which seem to have a wider linguistic interest.

The first refers to the phonetic rendering of a close association of ideas. The initial vowel of Dakota words is preceded by a glottal closure. When, therefore, a word with terminal consonant precedes a word with initial vowel there is a decided break following the consonant. The consonant does not become glottalized but the glottal closure follows it.

Die Methoden der Wortforschung können im allgemeinen nicht den Anspruch erheben, daβ sie mit den neuesten Erkenntnissen der Entwicklungsgeschichte Schritt gehalten haben. Es genügt nicht, Wurzelwörterbücher zusammenzustellen und alle Wörter, die die gleiche Wurzel haben, unter eine Rubrik zu bringen. Das ist eine gewiβ sehr notwendige und oft mühevolle Arbeit, aber sie endet meist auf einem toten Geleise. Was nützt uns beispielweise die Erkenntnis, daβ die semitische Wurzel phr ‘sammel’ bedeutet, und daβ behauptet wird, von dieser Wurzel stammen nun alle Wörter mit den Radikalen phr, einschlieβlich pahru Töpfer' (weil dieser den Ton ‘sammelt’)? Was nützt es uns, wenn die indogermanische Sprachwissenschaft feststellt, daβ agr- das gemeinsame Wort für ‘Acker’ ist, was nützt es, wenn wir eine semitische Wurzel rq aufstellen, um dann zu behaupten, das Wort für ‘grün’ sei davon abgeleitet? Ist denn das ‘Sammeln’ früher als der Mensch, der sammelt, ist das ‘Grün’ eher als das ‘grüne Kraut’, kann man sich bei einem Urwort agr- überhaupt etwas denken? Das soll doch letzten Endes das Ziel der Sprachwissenschaft sein, die Entwicklung des menschlichen Denkens, das in der Sprache seinen beredtesten Ausdruck findet, zu erforschen, wenigstens soweit uns das möglich ist. Und es ist bei einigem guten Willen wirklich möglich, weiter zu kommen, wie nur an obigen Beispielen gezeigt werden soll.

The Fourteenth Annual Meeting of the Linguistic Society of America will be held at Chicago, December 27–8, 1937, jointly with the Modern Language Association of America, which meets on December 28–30. Sessions will be held at The Drake, which will be the headquarters of both associations.