200 The Loom of Language
highly regular system of affixes., including the Bantu dialects discussed
below The veteran philologist Jacob Grimm first emphasi?cd the
merits of Magyar and commended it as a model to people interested in
language planning The existence of such regularity m natural lan-
guages has left a strong impress on projects foi a constructed world
auxiliary
At an early stage in the process of agglutination many words will share
similar affixes, because the latter have not yet suffered much modifica-
tion by fusion with different roots Hence mere regularity of affixes has
sometimes been used as a criterion of the agglutinating type, but
regularity may also result from an entirely different process After
amalgamation has gone far, lifeless affixes tack themselves on to new
words by the process of analogical extension, or old ones may be
rcgulan/cd for the same reason In this way a language with an amal-
gamating part, e g Italian, may approach the regularity of a language m
which few words have yet reached the stage of true external flexion So
the fact that Turkish or Japanese have regular affixes docs not mean
that they have evolved m the same way as Hungarian or Finnish Only
the last two, together with E^tonian^ with the language of the Lapps,
and with dialects of a considerable region of northern Siberia constitute
a truly related group within the heterogeneous assemblage once called
the 1 uraman family
In a language of tic amalgamating type, c.g Sanskrit, Greek, or
Latin, modifications of the sense of the word and the place it takes
m the sentence depend on affixes intimately fused with the radical
(roof) element Since fusion between core and affix is intimate, the
build-up of words is by no means transparent Bven the grammarian
can rarely dissect diem, We can always rccogm/c which accretions are
characteristic of number or case in the various forms of the Magyar
noun (p 197), because all the plural case-forms, as of hajo (ship),
contain the suffix ~k immediately after the root, but comparison of
singular and plural case-forms of an Indo~buropean noun does not
necessarily tell you which part of the suffix attached to the root is
characteristic of a particular case or of a particular number There is no
part of the suffix common to all plural in contrast to all singular case-
forms* In a language such as Latin or Sanskrit there is no part of tie
suffix common to the genitive, singular or plural, in contradistinction
to the different wr/Tw&tfHfbrms of all other case-forms
You can see this without difficulty, if you compare the following
case-forms of a Latin word with our Hungarian example.