WORD

ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: https://www.tandfonline.com/loi/rwrd20

Reviews

Joseph H. Gbeenberg, Robert L. Politzer, André Martinet, André Martinet, G.

Bonfante & F. P. Magoun Jr.

To cite this article: Joseph H. Gbeenberg, Robert L. Politzer, André Martinet, André
Martinet, G. Bonfante & F. P. Magoun Jr. (1949) Reviews, WORD, 5:1, 81-92, DOI:
10.1080/00437956.1949.11659354

To link to this article: https://doi.org/10.1080/00437956.1949.11659354

Published online: 04 Dec 2015.

Submit your article to this journal

Article views: 8005

View related articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rwrd20
 REVIEWS
MALCOLM GUTHRIE, The Classification of the Bantu Languages. Oxford University
Press; London, New York, Toronto. 1948. Pp. 91.
This is the third in the valuable series of linguistic works currently being issued
by the International Mrican Institute. The author, Malcolm Guthrie, is known
for his competent descriptive grammar of Lingala, a Bantu language of the
Congo. Moreover, he has a wide first-hand acquaintanceship with Bantu lan-
guages. In fact, in the introductory section of the present work, he claims intimate
knowledge regarding languages of no less than 14 of the 16 zones into which he
here divides the Bantu family and the work itself gives evidence of a wide variety
of observations regarding many areas in which published material is meager.
Nevertheless, for reasons which I shall state below, one must regretfully conclude
that the work does not answer to the expectations aroused and that it represents
a relatively small contribution to the subject it treats.
The book is divided into four sections: I) Introduction, II) Identifying the
Bantu Languages, III) Methods of Classification and IV) The Bantu languages
Classified. It is accompanied by a separate map in which the Bantu languages
are numbered in accordance with the system described in the fourth section.
In the brief introductory section certain general matters are treated, such as
the establishment of a standard nomenclature for Bantu languages and methods
of transcription. The author seems well aware of the problem of phonemic vs.
phonetic transcription in dialect area work and one could have wished for a more
extended treatment of this interesting topic.
The section on the identification of the Bantu languages which follows is much
the weakest of the work and reflects a curious lack of understanding of the method
of genetic classification of languages. The author here chases the will-o'-the-wisp
of the discovery of 'criteria' for 'defining' the Bantu languages. The author re-
flects that it is 'interesting' that Bleek, the pioneer of Bantu linguistics, did not
try to define the term Bantu. He also finds it 'interesting' that, Meinhof, the
father of Bantu comparative work, did not make such an attempt. That the con-
temporary investigators, Doke and Tucker, have enumerated some character-
istics of the Bantu languages strikes him as encouraging but inadequate. We are
told that "the most that has been achieved is a more or less complete statement
of the characteristic features of Bantu languages, scarcely any one of which is
found to apply to all the languages which everyone has accepted as Bantu."
This is evidently not enough for Guthrie, who is seeking a full scholastic definition
per genus et differentia specifica.
At this point we might have expected the writer to reflect that perhaps Mein-
hof, Doke and the others had good reason for not attempting such a definition
and he might have asked himself whether such definitions exist for Indo-European
Semitic and other well-known language families. But he presses on, We are now
given four sets of criteria, the first two of which are called principal, the latter
two subsidiary-not, we are told, because they are less important but because
81
82 REVIEWS

they are more difficult to apply through "contraction and attrition." One Cri-
terion is the existence of a set of grammatical genders marked by prefixes in-
volving "no correlation with sex references or with any other clearly defined
idea" (a purely gratuitous assumption). Another is the existence of a "balanced
vowel" system, that is, the usual Bantu five or seven vowel system. Presumably
if any Bantu language were to differentiate the a phoneme into separate front
and back phonemes, it would cease to be Bantu. The only criterion enumerated
which is relevant to historical analysis is his second, "a vocabulary, part of which
can be related by fixed rules to a set of hypothetical common roots."
Having at length arrived at an elaborate set of criteria, by their application,
he succeeds in excluding some perfectly good Bantu languages. It is perhaps
excusablethatBamum, Bafut and other languages of the Cameroon 'Semi-Bantu'-
Bantu borderline are not included for this is an ingrained, traditional miscon-
ception, but what shall we say of his refusal to admit the Manenguba group
(Nkosi and others) which even unsophisticated observers like Johnston and
Dorsch saw were Bantu?i Moreover, these languages show special resemblances
to those of the Duala group and Duala has always been accepted as Bantu and
is so classified by Guthrie.
Apparently all four criteria must be present for a language to be Bantu. What,
then, of the languages which satisfy some but not all of the requirements laid
down by the author? Guthrie solves this problem by the establishment of anum-
ber of low caste and outcaste groups (possibly the result of past linguistic mice-
genation?).
We are told of languages such as Bira "which are incompletely Bantu" because
while they have noun prefix classes, adjective agreement is distinctly fragmen-
tary. These languages are included by Guthrie in his enumeration, but their lack
of full status is indicated by the use of italics. Distinct from these, we are told,
is another spawn of half-breeds called Bantoid, which are excluded from con-
sideration altogether. These languages have all the Bantu characteristics except
a common vocabulary. The example given is Bafut of the Cameroons, but of the
six noun root morphemes quoted, three are easily derivable from standard Proto-
Bantu forms, not to mention the prefixes. In fact, Bafut is one of the languages
of the 'Semi-Bantu'- Bantu area whose Bantu affiliations have been traditionally
ignored. What other mythical hybrids may roam the African landscape, we are
not told. But whoever they may be, they will not lack congenial companions,
Bantoid and Semi-Bantu and the centaur-like Nilo-Hamitic.
In his third section, Guthrie discusses methods of classification. We are offered
several, the historical, the empirical and the practical. The historical method is,
according to the writer, the setting up of a genealogical tree of dialects and it is
impossible here because "with practically no historical records, true historical
study, as distinct from comparative study is impossible." The empirical method
is the study of the distribution of isoglosses. The various types phonetic, lexical,
syntactical etc. each comes in for separate discussion. Guthrie considers the em-
pirical method unusable because different isoglosses lead to different classifica-
tions. The method which he advocates, he calls the practical, that is, "the
 REVIEWS 83

presence of some arbitrariness is admitted as an essential modification of the

empirical method." Evidently, all that the writer means by this is that no one
isogloss or restricted set of them can be used as the sole criterion. For we are
told somewhat later that at the meeting of zones B, C and H there is a bunching
of isoglosses of different types and that phenomena of this kind determine his
classification. The author's technique of classification is set forth in the final
section. The Bantu area is divided into sixteen zones, each designated by a letter.
Within each zone, each language is numbered, a new decade being employed for
each subgroup. Thus the designation of the Sumba language as F 23 places it
within the second group of languages in zone F. The problem of the definition of
language and dialect is discussed at this point and a'practical' as opposed to a
'scientific' solution is adopted. By 'practical' is apparently meant non-linguistic.
Thus we are told that Sukuma (F 21) and Nyamwesi (F 22) are hardly more
than dialect variants of the same language, but "for political and demographic
reasons, we have to consider them as separate languages." Since Guthrie never
tells us, except for the few examples cited at this point, where these considera-
tions have entered into his classifications, his division into languages becomes
almost useless. Besides, what is 'practical' in the usual sense of the word about
this procedure is difficult to see. The administrator, for example, wondering
whether the same set of school text-books could be used for Sukuma and Nyam-
wesi, might be misled by Guthrie's classification into assuming they were far
more different than they really are. The opposition of 'scientific' and 'practical'
here is altogether unfortunate. The linguist, no more than the chemist, has two
separate sets of principles, one for his laboratory and another for practical
application.
The final section is concluded by a listing of languages in each zone by sub-
classes. For each zone, Guthrie gives comparable information concerning a num-
ber of important features, e.g., the presence of a gender indicating the diminutive,
the presence and formation of negative tenses, etc. This is much the most valu-
able portion of the work since it enables us to plot the distribution of some im-
portant features over the entire Bantu area. Some mapping of lexical isoglosses
has been done by other writers, but the field remains on the whole neglected.
Yet it is a subject of the highest interest, for here we have a vast dialect area,
unique in the primitive world, in which the conclusions of European dialect geog-
raphy can be tested under very different geographic and cultural conditions. It
is as an advance in this direction that the present work makes a distinct contri-
bution.
Columbia University JosEPH H. GREENBERG

.ANDRE MARTINET, La Prononciation du Fran«;ais contemporain: temoignages

recueillis en 1941 dans un camp d'officiers prisonniers. (Societe de Publications
Romanes et Fran«;aises, No. 23.) Paris, Librarie E. Droz, 1945. 249 pp.
The subtitle of the book is by far more descriptive of the work than the some-
what misleading main title, which might cause some readers to expect a manual
of French pronunciation. Nothing, of course, is further from the intention of
84 REVIEWS

the author. Prof. Martinet's book is purely descriptive and does not attempt to
set any standard as to the pronunciation of French; furthermore it is not pre-
cisely a book on pronunciation at all, but a book on phonemic analysis.
The originality of the work lies chiefly in the fact that the author thought of
a group of French middle-class people thrown together by chance in a prisoner-
of-war camp in Germany in 1941 as a random sample of French speakers to be
analyzed with respect to the phonemic pattern of their speech. He therefore
decided to compose a phonemic questionnaire which he distributed among his
fellow prisoners.
The questionnaire requested first of all the age, profession, birthplace, and
for obvious reasons all other places of prolonged residence. There followed a
series of forty-five questions of which numbers 1 and 2 inquired about the
linguistic background of the individual; numbers 3 to 31 concerned the extent
and nature of vocalic phonemic oppositions, while 32 to 45 concerned consonantal
oppositions in the phonemic pattern of the speaker. The typical question at-
tempting to establish the existence of a phonei?-ic opposition was worded
"Pronon<_;ez-vous de fa<_;on identique: a) la et las? ... b) rat et ras? ... "etc.
(No. 7). The typical question attempting to determine the nature of the phonemic
opposition was worded-somewhat awkwardly, as the author admits (p. 18)-
"Si vous faites une difference entre rat et ras, est-ce une difference: a) de timbre
(voyelle plus ou moins ouverte ou profonde)? ... b) de longueur (duree de la
voyelle)? ... " (No. 8). The awkwardness of th!;:l question lies in the fact that
it can be-and indeed was in some instances-interpreted to mean ''if you do
make a difference in an exceptional case". This is only a small example of the
type of problem which faces anyone who is trying to formulate a questionnaire.
Martinet received 409 answers out of 750 questionnaires distributed. In order
to gain a statistical picture of the phonemic patterns presented by the answers,
he decided (p. 16) not to deallwith the phonological structure of the speech of
each individual, but to consider each phonemic problem seperately. He felt that
this method of approach would be less influenced by errors which might have
been committed by individuals answering the questions.
The 409 individuals were classified with respect to geographic origin as well
as to age. The geographical classification brought out the fact that 59 had trav-
elled in their early youth to such an extent that an a priori classification seemed
impossible in their case. The other 350 were grouped in 11 geographic regions,
which the author established perhaps somewhat arbitrarily, taking into con-
sideration, however, certain factors such as traditional unity of the region or
existence of well-known dialectal divergences. The regions were of unequal
size and representation: thus all of Southern France is grouped into one area
with 101 representatives, including a sub-group Southwest (12 people), while
the comparatively small Paris area, smallest in size of all the regions represented,
follows with the next largest number of speakers, 66. The smallest sample comes
from the region of Burgundy (8 individuals).
The classification as to age divides the sample into three categories: under
31, 31 to 41, and over 41. Geographically, the age classification is applied first
 REVIEWS 85

to all of France except the South, then successively to the Paris area, the East,
the Southeast and the South. In exceptional cases the age breakdown was given
also for other areas.
The method used throughout is to present first in percentage form the oc-
currence of a phonemic opposition in the various geographic divisions, and in
the three age categories. Whenever the author wants to compare the type of
opposition, he utilizes a fraction composed of the number of individuals specify-
ing that they differentiate by timbre over the number of those differentiating
by length.
The organization of the book is extremely clear-cut and simple. After an
introduction dealing with the questionnaire and the method utilized in interpret-
ing it, Prof. Martinet discusses first his analysis of the questions dealing with
the vocalic system under the following headings: L' E caduc; Les voyelles
d'aperture maxima: l'archiphoneme A; Les voyelles d'arriere d'ouverture
moyenne: l'archiphoneme 0; Les voyelles de fermeture maxima: les archi-
phonemes U, tJ et I; Les voyelles anterieures d'ouverture moyenne non labial-
isees: l'archiphoneme E; Les voyelles anterieures labialisees d'ouverture moy-
enne: l'archiphoneme ill; L'harmonisation vocalique; Les voyelles nasales. Then
follows the discussion of the various consonantal phenomena included in the
questionnaire, under the headings: La sonorite; Yod et la mouillure; Un phoneme
etranger: la nasale velaire; L'aspiration; Les consonnes geminees.
After the chapters containing the detailed discussion, there follows an at-
tempt to establish a typical phonemic system for each of the regions under
consideration: Essai de syntheses (pp. 201-222). At this point Martinet turns
to two additional problems which can be solved only after having reached the
main conclusions of the investigation. The first of these is a comparison of the
phonemic patterns of those individuals who were teachers with the phonemic
pattern of the rest of the group studied. The second is an attempt to deter-
mine the age at which phonemic patterns are fixed by comparing the phonemic
patterns of the geographically non-classifiable individuals with the typical
phonemic pattern of each area where each of these individuals had lived.
The first inquiry leads to the not very surprising result that the phonemic
habits of teachers are rather conservative, and more influenced by orthography,
while the second indicates that the basic phonemic oppositions are generally
at least acquired at a rather early age, before the lOth year.
There is no place here to discuss the individual conclusions reached by Prof.
Martinet. The striking contribution made by his book is the approach it takes
to the entire problem of phonemic description. Martinet recognizes that even
in a socially comparatively homogeneous group we find a variety of phonemic
patterns actually limited only by the number of individuals in the group. If
this is true, then we are forced to ask just what the meaning of generic terms
such as "French" or even "Standard French" or "Slow Colloquial French"
really is. Whenever anyone, even a trained linguist, has tried to give a description
of French pronunciation, Prof. Martinet tells us, he has been apt to give his
own pronunciation. Martinet's approach, the attempt to construct the phonemic
 86 REVIEWS

system of a group from the observation of individuals, is in my opinion an im-

portant contribution toward evolving a type of linguistic description which
concerns itself with the group rather than with the individual. Ferdinand de
Saussure stated that "la phonologie ... n'est qu'une discipline auxiliaire et ne
releve que de la parole". 1 But the analysis undertaken by Martinet shows us
more than the parole of 409 individuals. It gives us an idea of the French lan-
guage as a dynamic social force changing in the dimensions of time and space.
Prof. Martinet's study drives home the point that whenever we speak about
"French" or "Colloquial French" we refer to an average speech which may or
may not be identical with that of any individual.
The details of the methodology employed by Martinet present a tremendous
number of problems of which he is quite aware, and of which some can be in-
timated, although not solved, at this point. Of particular importance is the
problem concerning the collection of the data: Is the use of the questionnaire
preferable to the oral interview by a phonetically trained observer? With regard
to determining the existence of the opposition in the speech of the individual,
the carefully worded questionnaire is certainly more reliable, since a trained
observer might detect a difference which would be of no significance if the speaker
himself were not aware of it. However, with regard to the nature of admittedly
existing differences, the trained observer's testimony would be of greater value,
since the speaker could perhaps not always be relied upon to judge accurately
the phonetic nature of the opposition.
Another problem connected with the collection of data is the extent to which
the individual is influenced by orthography or other elements which might
prejudice the speaker's idea of his own pronunciation. This problem exists
regardless of the means of collecting the data, and is handled by Martinet in
the only reasonable way: he acknowledges the problem, but assumes that for
the purpose of comparison on a geographical basis it can have little effect on
his conclusions, since such influences would exist to about the same degree
throughout. Furthermore, by attaching particular significance to percentages
higher than 50, or to percentages higher than the average of the entire area
(this applies only toN orthern France), he safeguards himself against attributing
significance to error.
One set of problems concerns merely the question of how to obtain an ac-
curate description of the phonemic systems presented by the 409 individuals.
Another and perhaps even more important problem, however, concerns the
question of whether the groups of individuals considered are a reliable sample
of the totality of which they are assumed to be representative. Statistics is
divided into two branches: descriptive statistics, which is the type used by
Martinet throughout the book, and inductive statistics, which concerns the
determination of the reliability of a measurement based on a sample as repre-
sentative of the same measurement in the unknown universe from which the
sample is drawn. Prof. Martinet is aware of the existence of sampling problems,
1 Ferdinand de Saussure, Cours de linguistique generale (publie par Charles Bally et

Albert Sechehaye), Lausanne, Paris, 1916, p. 57.

 REVIEWS 87

but not of the more precise scientific means of dealing with them. He is careful
enough to attach significance only to large differences in percentage in his
larger samples, and thus avoids the pitfalls of attributing importance to varia-
tions which can be reasonably accounted for by mere sampling error. 2 The
inclusion of certain tests of significance, as for instance a test for the significance
of the difference of two percentages, a test for the significance of a classification,
or a test for homogeneity would have added to the scientific precision, but very
little to the linguistic conclusions of the work. 3
Certainly no statistical test of significance could have procured for Prof.
Martinet a larger and more representative sample at the time he made his
investigation. However, it is to be hoped that in the preparation of a phonemic
atlas of France-of which this book might be an embryo-the methodologies
evolved in quantitative research in sociology or marketing problems will be
fully utilized and adapted to linguistic inquiry.
Columbia University RoBERT L. PoLITZER

A. RosETTI. Le mot, Esquisse d'une theorie generale. Deuxieme edition revue et

augmentee. pp. 57. Copenhague, Einar Mumksgaard, Bucure1;1ti, Institutue de
linguistica romana, 1947, Dan. Cr. 9.00.
This sketch is one of a. series of monographs published by the Societe roumaine
de linguistique generale. Its title is somewhat misleading, as it suggests an original
synthesis rather than what the book actually is, namely a review of various opin-
ions concerning a number of problems connected with the notion of 'word'. The
author does not refrain from expressing personal preferences. But his intention
was obviously to present a general survey rather than a critical evaluation.
Though the extensive bibliography (pp. 46-57) contains no less than 122 items,
this survey is far from exhaustive, a fact which is not to be wondered at, con-
sidering the number and complexity of the questions touched upon. And yet,
even within the narrow frame of a monograph, we should have expected to find
some names and titles of works which R. has not included. A quotation from
Morris Swadesh's The Phonemic Principle is the only reference to American lin-
guistic thinking. It was, no tloubt, difficult for many European linguists to get
acquainted with American publications during and immediately after the war.
But this can not justify omitting, among others, Sapir and Bloomfield. Even if
the reader does not espouse the mechanistic views expounded by the latter, he
cannot help being struck by the one-sidedness of Rosetti's purely mentalistic
exposition.
2 "Sampling error" may be misleading to anyone not familiar with statistical termin-

ology. It means simply that the difference between the measurement in the sample and the
corresponding measurement in the universe from which the sample is drawn is due merely
to chance.
3 For the purpose of checking the significance of Prof. Martinet's conclusions, I applied

statistical tests of significance to several of the percentage differences to which the author
attributes significance. In each instance the possibility of the difference having been pro-
duced by sampling error (chance) was not greater than about 4 in 100, which offers a reason-
able margin of safety in assuming the findings to be significant.
 88 REVIEWS

A first chapter is devoted to the problem of the general relations between lan-
guage and thought. Whether the word as such should be involved there, may be
doubted. As a matter of fact, in that part of his monograph, Rosetti constantly
refers, not to the word, but to the sign. It is true that when we say 'sign', we
usually think 'word': boeuf and Ochs, traditional examples since de Saussure, are
of course both signs and words. But, in the discussions in which they appear they
are meant as signs, not as words. The basic question, in a linguistic treatment of
the problem of the word, is whether we are justified in positing a linguistic unit
of that name. Now it is a fact that most linguists have so far taken the word for
granted. The definitions of it which have been proposed are, as a rule, not such
as might enable us in doubtful cases to decide whether a given section of an
utterance is a word, several words, or a part of a word. They are, more often than
not, based upon observations limited to one type of linguistic structure. They
frequently leave out of account 'grammatical' words. Among the notable excep-
tions, we should mention Buyssens' discussion and proposals (Les langages et le
discours, Bruxelles 1943, a work Rosetti does not mention in his bibliography).
Bloomfield's definition of the word as 'the minimum of free form,' though widely
and uncritically accepted on this side of the Atlantic, cannot be described as a
valid solution: can we say that the 'word' of is more of a free form than the suffix
hood? It should be clear that 'word' is a term that linguists have taken over from
everyday parlance with its rambling semantic contents. Before it can be used
in a scientific way, it needs to be defined. In the present state of our knoledge,
it would seem safer to redefine it every time we consider a new linguistic pattern
nd by reference to that pattern. Otherwise we run the risk of positing a unit
which, when imposed upon some structures, might tend to confuse tne actual
pattern instead of elucidating it.
Our problem, with 'word', is very much the same as with several other terms
like 'vowel', 'syllable', etc., which are widely used by both laymen and specialists.
Specialists are loath to discard them because they are handier and less likely to
antagonize the beginner than are learned substitutes. On the other hand, they
should be accurately defined if we want to make them serve scientific purposes.
But the difficulty is to define them in such a way as to covery pretty much the
same things as in everyday language, which is the only way whereby the peda-
gogical advantage of retaining known signifiers can be preserved. When a layman
speaks of 'words' in reference to English, he usually means dictionary entries,
and their inflected variants, which, in a text, appear separated by a space from
others of the same kind. Our first concern should be to determine the reasons
historical, psychological, and linguistic, why the spaces in the text are where they
are and not elsewhere. Whether these reasons are the same in two different lan-
guages may well be doubted: the accentual factors which contribute to justify
writing little boy with a space between little and boy do not exist in the case of
petit gar~on; the syntactic pattern, probably combined with lexicographical ex-
pediency, which led the Germans to make one (written?) word of aufgeben, did
not exist in the case of English give up (two words?). Even if we eliminate such
 REVIEWS 89

historical and psychological data as seem irrelevant to our purposes, we emerge

with a number of linguistic factors which, we know, will differ from one language
to another. Yet, were one of them to be found in all languages, we might make it
the criterion for our definition of the word. As a matter of fact, the criterion of
non-separability, supplemented by certain reservations, would seem to yield
fairly satisfactory results in the sense that the 'word' thereby defined would,
in a great majority of cases, coincide with the layman's idea of what is to be
considered a word in the best-known languages.
The nearest Rosetti gets to this is when he deals with the phonetic word, the
phonemic word, and the compound word. But since his conception of the word
seems basically psychological, all these aspects are bound to be treated more or
less as side issues. The problem of the word as a truly linguistic unit is never
clearly presented.
With all its one-sidedness and psychological bias, this little book can render
some service provided the reader is duly aware of its various limitations. It reads
well. The material presentation is excellent.
Columbia University ANDRE MARTINET

Glossaire des patois de la Suisse romande, redige et publie sous Ia direction de

Karl Jaberg; fasc. XXII, boursoufle-branko. Pp. 673-728, in-4°. Victor
Attinger, Neuchatel et Paris, 1948.
Le vingt-deuxieme fascicule de cette oeuvre monumentale suit le precedent
a onze mois d'intervalle. Ceci represente, pour une entreprise de longue haleine,
un rythme satisfaisant si l'on considere toute la matiere contenue dans ces 56
pages de grand format. Pour gagner du temps, il a ete decide de reduire la part
consacree aux etymologies. Les raisons de cette decision nous sont exposees
dans le cinquantieme Rapport annuel de la Redaction, et elles nous paraissent
fort sages. Le veritable but de ce Glossaire est de nous presenter des materiaux
utilisables par divers specialistes et notamment par les linguistes qu'interessent,
au premier chef, l'etymologie. Ceux-ci doivent trouver dans le Glossaire tous
les renseignements relatifs aux patois de la Suisse romande susceptibles d'eclairer
leurs recherches. Mais il n'est pas indispensable que les redacteurs de cet ouvrage
s'efforcent, des maintenant, de tirer de leurs observations des conclusions
etymologiques qu'illeur faudrait etayer par des donnees qu'ils devraient parfois
chercher bien loin du domaine qui est le leur.
De fa~on generale, on peut dire que les necessites pratiques qui obligent la
direction et les redacteurs a condenser un peu, n'auront pas pour effet de diminuer
la valeur de cette oeuvre qui, dans le domaine roman, reste unique par la densite
geographique des sondages, l'etendue et Ia precision des renseignements linguisti-
ques et culturels de tous ordres qu'elle nous offre. On regrette un peu d'apprendre
que Karl Jaberg quitte Ia direction de l'entreprise, mais on peut etre assure
qu'illa laissera dans de bonnes mains.
N ous sommes heureux de trouver exprimee dans le Rapport annuel qui est
de la main de M. Jaberg, !'opinion que le Glossaire ne rend pas inutile des mono-
96 REVIEWS

graphies consacrees a la langue et aux coutumes de communautes particulieres.

Tout en reconnaissant pleinement !'importance et le bien-fonde de l'entreprise
dont nous consider'ons ici les resultats, les linguistes habitues au point de vue
structural sont tentes de deplorer qu'on ait choisi, pour presenter les materiaux
recueillis, la forme d'un glossaire a partir duquel il est fort difficile de reconstituer
la structure particuliere des ensembles linguistiques et culturels qui ont ete
soumis a !'observation. Sans doute existe-t-il, des aujourd'hui, un bon nombre
de monographes consacres a des parlers franco-proven<;aux de la Suisse romande.
Plusieurs de ces travaux sont excellents. Mais on peut regretter et s'etonner
qu'au pays de de Saussure, personne ne se soit encore a vise, a notre connaissance,
de presenter une description fonctionnelle et structurale d'un parler local.
Columbia University ANDRE MARTINET

R. A. HALL, Descriptive Italian Grammar, Cornell University Press and Linguistic

Society of America, Ithaca and New York, 1948,228 pp.
It is difficult to determine what purpose has been served by the writing of
this book. It contains no study of dialects beyond a couple of commonplace
remarks on pp. 2-3; no bibliography; no history of the language; no comparison
with other languages, living or dead; and no attempt at a semantic, stylistic,
literary, or cultural interpretation of any word, form, or class. The phonemic
description is short, and by no means new.
We learn that "standard" or "literary" Italian (as the Italians call it) has
seven vowels, that the plural of delitto is delitti, and that cantero means "I
shall sing;" all of which we suspected a long time ago.
Italian was known and studied several centuries before any other modern
European language, and there has been no lack of good Italian grammars, even
though they did not characterize themselves as descriptive. With that in mind,
one wonders why this one was published. We can only suggest that in the future,
the Linguistic Society of America, the American Council of Learned Societies,
and the Cornell University Press make better use of their money than to publish
any more books of this kind.
There is no need to go into the details of the book. For example, some phone-
ticians might object to the use of the term gutturals for velars (228). I shall only
remark that Mr. Hall shares a common prejudice about the number of speakers
of Italian (although the mistake is rarely as gross as in this case). He writes (2)
that "the approximate number of speakers of Italian is given as 43,700,000."
Now, the population of the Italian Republic in its present boundaries is estimated
at about 47 millions, practically all of them speakers of Italian. Since there are
many more in Corsica, Tessin, Venezia Giula, Malta, Dalmatia, Tunisia, France,
Belgium, Switzerland, Egypt, Turkey, America, Africa, etc., it is obvious that
Hall's number is from fifteen to twenty-five millions short of reality.
I must also remark that there is no such thing as Tyrol in Italy (2). Perhaps
Mr. Hall meant the region called Alto Adige which since the "option," contains
less than 200,000 German speakers, the vast majority of whom are bilingual.
 REVIEWS 91

In Piedmont, the Aosta valley speaks not a Proven~al (same page), but a "Franco-
Proven9al," i.e. essentially, a French dialect of an archaic type, very close to
Old French.
Princeton University G. BONFANTE

JoHANNES HEDBERG, The Syncope of the Old English Present Endings: A Dialect
Criterion (Lund Studies in English, 12). Lund: Gleerup, 1945.
Dr. Hedberg has written a most useful book on a significant point of OE
descriptive grammar and OE dialectology, namely, the relation and proportion
of syncopated vs. unsyncopated forms of the 2nd and 3rd sg. pres. ind. of the
strong verbs and of the weak verbs of class I excepting for short-stemmed jan-
verbs in -r and a few others (pp. 284-5). The material is drawn from all OE
prose and is evidently complete. The full listing of all contract forms is most
valuable from the point of view of purely descriptive grammar, and is thus an
index unlikely to be superseded. The material occupying most of the book (47-
279), is arranged according to the consonantal ending of the verb root; a Verb
Index (306-10) guides the reader to the proper place for his illustrations. There
is a compendious survey of previous theories (28Q-3), none quite satisfactory,
concerning the origin of these contract forms. The result of the study (296 ff.)
is to show that: 'What is evident is this: the unsyncopated form was as typically
Anglian as the syncopated was West Saxon and Kentish.'
But the book has much more general utility and interest than the foregoing
statement of its structure and essence might suggest. First there is the list of
Sources (13-43), a list of all published OE prose manuscripts and a first-rate
bibliography. A few minor suggestions: St. Gregory's book (13) should be
entitled Regulae Pastoralis Liber, Cura Pastoralis and the like being a false
title, for over a century at least the rather exclusive property of specialists on
OE literary matters (see Mediaeval Studies 10, 129 ff., Toronto, 1948). The
'Blickling Homilies' (35), formerly in the possession of the Marquess of Lothian,
is at present owned by Mr. William H. Scheide of Princeton, N. J. 'Anglo-
Saxon Chronicle' (38), better the 'Old-English Annals': the F -version, a bilingual,
is Latin and English, not French and English, and use might well have been
made of the edition of this text in Benjamin Thorpe's six-text edition with cor-
rections according to C-H. Fernquist, Studier i modern Sprakvetenskap 13, 43-52
(Uppsala, 1937) and Magoun, Modern Language Quarterly 6, 371-80 (1945).
The C-text has been competently edited by Harry Rositzke as Vol. 24 of Max
Forster's Beitriige zur englischen Philologie, 1940. The representative of West-
Mercian culture in King Alfred's entourage (286) was Archbishop Plegmund of
Canterbury. Wulfstan's origin (288) is a complicated and controversial matter
on which see Max Forster, Der Flussname Themse, etc. (Sitzungsberichte d.
bayer. Akad., phil.-histor. Kl. 1, 257-64, 1941).
For future writers of text-books for beginners in OE the book will be an
everlasting treasure-trove and will enable them conveniently to supply with
the principal parts of all verbs in question the contracted 3rd sg. pres. ind. This
92 REVIEWS

essential and for a beginner often difficult form should be supplied paranthetic-
ally in a glossary along with the ordinary principal parts in the case of every
verb in que&tion (vs. the mere indication of class numbers and back references
to a grammar) as follows: beodan (biett), bead, budon, boden. Such is the practice
in up-to-date grammars of German, a language which offers pedagogical problems
virtually identical with OE.
Cambridge, Mass. F. P. MAGOUN, JB.