<?xml version="1.0" standalone="yes"?>
<Paper uid="C69-3001">
  <Title>WAEG UNG</Title>
  <Section position="4" start_page="0" end_page="7" type="intro">
    <SectionTitle>
3. Data
</SectionTitle>
    <Paragraph position="0"> Starting point was a still unpublished dictionary of German words analysed into morphemic segments and punched on cards by KANDLER at the Sprachwissenschaftfiches Seminar der Universit~t Bonn. When I had access to the material~ it consisted of 117.370 uncorrected entries, most of which entailing morphemic segments of words and markers for categories like word-class~ gender, transitivity, dialect, sociolect, foreign origin etc 5) The entries and categories were copied from &amp;quot;Deutsches W6rterbuch&amp;quot; by MACKENSEN, edition 1955. From KANDLER's material a list of 2.111 kernels - stems of simple words from classes noun, verbs with inlexional ending EN, and adjectives, - was prepared, mainly by automatic data processing 6). As to allostatus, only dictionary forms were available at that stage. Therefore various alloforms of the morphemes were added and marked according to their allostatus. The alloforms consist of what are usually called umlaut- or ablautvariations. Included were &amp;quot;potential&amp;quot; umlautforms; they are kernels that do not appear in inflexion but have vowels that permit Umlaut and in derivation often actually do have it, like, for example, BROT (bread) - plural BROTE - derivation BROETCHEN. The resulting list had 3.613 entries~ marked as ~follows 5)Grammatical markers about gender and transitivity were copied but so far neither checked nor actually used. They appear in the artificial'code list but not in the data statistics.</Paragraph>
    <Paragraph position="1"> 6)The entries were sorted down to about 4000 by computer with the help of the grammatical markers; then mistakes, peculiar entries, unmarked dialectical words etc. were eliminated by hand. Reasons are given in BONTING (1969).</Paragraph>
    <Paragraph position="3"> Simple words: word-class homography and derivation by zemoaffix null The list of 3.613 kernelswas by computer sorted to discover word-class homography 7). Results are shown in table i. While being compared, homograph kernels were marked accordingly and deleted except for one grapheme sequence. The resulting list, which contained 2.759 kernels, was taken as basis for the study of derivations. Table i. The table shows, that somewhat less than half of the kernels are word-class homographs. All three covered word-classes have about the same homograph-haplograph ratio. Of course, only lexicalised homography can be stated here. In German sentences, every verb or adjective can syntactically be used as a noun 8).&amp;quot; 7)Homography according to gender or transitivity etc.</Paragraph>
    <Paragraph position="4"> is marked in artificial code, D or K respectively; homography according to semantic meaning like in SCHLOSS (&amp;quot;palace&amp;quot; or &amp;quot;look for key&amp;quot;) is not treated. null 8)Cf. paragr. 6 for discussion of syntactic and semantic aspects of derivation.</Paragraph>
    <Paragraph position="6"> Traditionally, what is here called word-class homography is called derivation or derivation by zero-afflx.</Paragraph>
    <Paragraph position="7"> I would suggest not to speak of derivation in a synchronic description, because no direction of the derivation can be concluded from the data. Only with historical information is it possible to call TAGEN a verbalisation from TAG (Old High German there was only a noun taga but no verb), and on the other hand to call LAUF a nominalisation from LAUFEN (Old High German verb liofan with pretaerite Ablautform louf, which served as basis for derivation) 9).</Paragraph>
    <Paragraph position="8"> So I propose to speak of word-class homography when lexicalised forms a~e concerned. The use of adjectives and verbs as nouns and of participles as adjectives in sentences ought to be treated in syntax as syntactical trans~Drmations and should be excluded from word-formation. null Derivation by affixation The morphological structure of derivated words was given in the formula K + A (+A) (eF1). To gain precision, some of the following questions have to be answered: which kernels appear together with which affixes, and are there reasonable classification2 What is the surface structure of the morphemes in words? What grammatical functions and - perhaps - semantic meanings are represented by the affixes? It was - and still is - our intention to collect as many data as possible, so that the derlvatlon formulas can be rendered more precise through induction from the data evidence* The first step was to consider surface structure and reformulate the formula in terms of actually used prefixes 9)Only where ahlautvariants are involved can a direction of derivation be safely concluded from synchronic data. - 9 and suffixes. WEISGERBER (1958) proposed to use the Term A b 1 e i t u n g s t y p (derivative type or derivation pattern) for a kind of substitution frame where one or more affixes are the frame and kernels are to he inserted. For a beginning, seven derivation patterns were formulated. Reasons for choosing these particular types are not very sophisticated: we wanted to include nominalisation, verhalisation, and adjectivation as well as prefixation and suffixation just to try the usefulness of the computer as a wordproducer. null Into each frame, all 2.759 Kernels were inserted, which amounted to an output of 15.013 artificial &amp;quot;words&amp;quot;. These were then checked a~ainst the KANDLER-MACKENSEN material for lexicalisation and marked as booked or not booked (a &amp;quot;Y&amp;quot; or &amp;quot;-&amp;quot; in artificial code in a certain column). The count of the results and the derivation patterns are shown in tabel 2.</Paragraph>
    <Paragraph position="9"> As an example, one page of output is copied in table 3. A more detailed analysis, which takes account of word-class information is given in table 4 . From the statistics of these tables and from a comparison of lexicalized and non-lexicalized forms, we hope to gain insight about derivation. To demonstrate how we plan to progress , the pattern / ...-UNG/ is discussed.</Paragraph>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
5.2 Lexicalization
</SectionTitle>
      <Paragraph position="0"> The artificial forms printed in table 5 were also found in the dictionary: of these, 36,7 % contain haplograph verbal kernels, 49,5 % are homographs which can be interpreted as verbal, i0,9 % are haplograph noun kernels, and 3,5 % haplograph adjective kernels. Deverbalization seems to be the predominat function of the suffix UNG.</Paragraph>
      <Paragraph position="1"> Of the ii deadjectives, one is archaic (HARTUNG for - i0 (february), one poetic (WIRRUNG from a novel by Fontane called &amp;quot;Irrungen, Wirrungen&amp;quot;), one a misclassified deverbative (HOEHLUNG from HOEHLEN), and the rest sociolects spoken by hunters (ALTUNG, DICKUNG, SCHALUNG) craftsmen (DUENNUNG, HALBUNG, LASCHUNG, RAUHUNG, SCHALUNG with a different meaning), or sailors (STEILUNG).</Paragraph>
      <Paragraph position="2"> The 34 derivations from noun kernels can likewise be explained as histomically old (e.g. ZEITUNG, WALDUNG)) untypical) or misclasslfied (for details cf. BONTING, 1969, 89 f.).</Paragraph>
    </Section>
    <Section position="2" start_page="0" end_page="7" type="sub_section">
      <SectionTitle>
5.3 Productivity
</SectionTitle>
      <Paragraph position="0"> The deverbative function of UNG is confirmed, when pmoductivity is considemed by looking through the non-lexicalized artificial words. To check my own judgement about the acceptability of various forms, I am currently trying to find ways how to get information from informants about these artificial words. So far, one general conclusion can be drawn from the various questionaires which I had filled in by students: there is wide disagreement about i. dictionaries, 2.</Paragraph>
      <Paragraph position="1"> what individuals think is normal and ought to be in a dictionary, and 3. what individuals think is normal, acceptable, peculiar, unacceptable, unrecognized. I0) There are, however, some characteristic grammatical reslxictions for acceptahallity, which parallel those of lexicalizations: accepted derivations are generally deverbative, and the verbs are predominatly transitive.</Paragraph>
      <Paragraph position="2"> That is no new insight; HENZEN has ~aid so in his &amp;quot;Deutsche Wortbildung&amp;quot; (1965, 181) in regard to lexicalized forms. However, if it holds true for the potential forms, as the test suggests, a general rule can be formulated, where a relation between transizivity of verbs and ability lO)cf. CHAPIN (1967) for evaluation categories  - ii to form derivates with UNG is stated.</Paragraph>
      <Paragraph position="3"> 5.4 Functional and semantic derivation Rendering the general formula for derivations more precise in terms of grammatical functions leads to a more general point: the distinction between grammatical and semantic aspects of derivations. MARCHAND (1967), 13-26) and (1966, 138) suggests the following distinctions and terminology: expansion: no change of word-class derivation: change of word-class  functional derivation: no additional semantic content semantic derivation: additional semantic content.</Paragraph>
      <Paragraph position="4"> According to this terminology, GLEICHUNG ~uld be a semantic derivation (mathematical equation) whereas LOCHUNG (punching) would be a functional one; with UNG there is, however, a complication, because it denotes either the act signified by the verb or the result of the action. MARCHAND' s suggestions should be applied to empirical data; and they should be used in dictionary work, where only semantic derivations deserve an entry and functional derivation ought to be treated by rules, perhaps analogue to those covering purely syntactic transformations as suggested (cf. paragraph 4).</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>