<?xml version="1.0" standalone="yes"?>
<Paper uid="W91-0208">
  <Title>Touche Ross Management Consultants</Title>
  <Section position="9" start_page="83" end_page="85" type="concl">
    <SectionTitle>
7. Formal Properties of LIRs - 3: Semantically-Based Exceptions
</SectionTitle>
    <Paragraph position="0"> Beside these uncanny exceptions to LIRs based on features of the word-forms involved, LIRs are also subject to systematic exceptions which pick on features of the sense. In drawing attention to them, we notice some surprisingly intricate structure in the lexicon.</Paragraph>
    <Paragraph position="1"> Cutter Cut enables nouns representing cutting instruments to be used as verbs. So people can be knifed, bayoneted, speared, harpooned, axed, tomahawked, or scythed down; various tools can scissor, drill or saw through material. Why then do sword, dagger and their hyponyms take no part in this verbal carnage? Why is it impossible to speak of *swording, *daggering, *poniarding, *bodkinning, *rapiering, *scimitaring, *sabring, *stilettoing, *claymoring or whatever else one might do to an adversary with a specific bladed weapon? It begins to look like more than an accident, since there are plenty of non-derived verbs to cover, in less specific ways, the stabbing, slashing, chopping, hacking, pricking and piercing actions that would be connoted.</Paragraph>
    <Paragraph position="2"> Another strangely coherent semantic constraint is on Food Item - Mass. This allows a count noun for food to appear as a mass noun when the units of that food are not evident. Hence although the primary meaning of the nouns is the discrete animal, fruit or vegetable, one can nevertheless refer to some egg, some crab, some salmon, some potato, some carrot, some apple, some cabbage etc., connoting a mass of the relevant food substance typically on a plate. Only one form of food seems to resist: pulses. No matter how finely you grind or mash them, you cannot refer to &amp;quot;*some pea&amp;quot;, &amp;quot;*some bean&amp;quot;, &amp;quot;*some lentil&amp;quot;. Refried beans are still beans, not bean. Neither of the authors is a native user of garbanzo (a North American name for chickpea), but we would predict that you cannot have a side-dish of garban zo. 13 8. Formal Properties of LIRs - 4: Specificity to Language, and to Dialect Further evidence comes from cases where an LIR appears to be present in one language, but not in another.</Paragraph>
    <Paragraph position="3"> The form of the argument is quite simple: consider two languages (for the sake of argument, called A and B) and two sets of words, apparently translation equivalents one for one between the two languages. Suppose that the members of the set in A all have an extra sense which members of the set in B lack. Such cases are clear evidence of that the assignment of senses is language-dependent; and the facts are explained by positing an LIR to be present in A and not in B. The only alternative would be to appeal to pragmatic interpretation of the context. 14 But pragrnatics is a species of commonsense: it cannot explain why the interpretation of vocabulary differs systematically from one speech-community to another.</Paragraph>
    <Paragraph position="4"> 13The historical situation here seems a little-muddled, since pea is a singularized hack-formation from pease, originally a mass-noun meaning peas. This itself seems to have been a reanalysis of a descendant of an originally plural noun in Latin, pisa. In modern Indian English, dal is used quite happily as a mass-noun for a different son of pulses.</Paragraph>
    <Paragraph position="5"> 14As, for example, Aronoff (1980: pp.755-7) attempts to explain away the need for specific details in rules of morphological semantics by appealing to features of the context in interpretation. His argument is significantly weakened by cases where languages with close morphological processes (e.g. agent nouns in English and French) apparently show quite different constraints on their interpretation: e.g. c'est un bon conducteur has no sense synonymous with the simple il conduit bien ; but cf. he's a good driver = he drives well.</Paragraph>
    <Paragraph position="6">  An example may make the point clearer. In Ancient Greek, there is a set of adjectives (clearly marked as such morphologically) which associate their head-noun with a particular time of day. ( E.g. eo:os - at dawn, early; orthrios - in the morning; mese:mbrinos - at noon; deilinos - in the afternoon; hesperios - in the evening; nukteros - at night; mesonuktios - at midnight; pannukhios - all night long). English also has such a set of adjectives but they are few, and confined to a learned register (e.g. matutinal, nocturnal ); the major translation equivalents in English are attributive uses of nouns (e.g. morning papers, noon-day sun, afternoon tea etc.). Ancient Greek goes on to use its temporal adjectives predicatively, with the sense of adverbial qualifiers of the action predicated: e.g.</Paragraph>
    <Paragraph position="7"> orthrios he:ko:n - coming in the morning (Plato, Protagoras 313B) egre:i mese:mbrinos - you wake at noon (Aristophanes, Wasps 774) eudon pannukhioi - they slept all night long (Homer, Iliad 2.2) Evidently, there is no such predicative use of English temporal adjectives or nouns: prepositional phrases (as used above in the glosses) are the closest that can be offered. 15 Hence there is an adverbial sense of Greek temporal adjectives, used predicatively, which has no parallel in English. A neat formulation of the case would thus be to formulate an Ancient Greek LIR, absent in English, which goes: LIR Time Predication (16) ADJ : temporal, C/p(head-NP) -&gt; ADJ predicative: temporal, t~(head-S) 16 To avoid positing an LIR here would take some special pleading. In this case one would need to argue that the distinction of senses made in English is simply absent in Greek, so that temporal adjectives are available to do more promiscuous service. This would amount to saying that Ancient Greek simply made no linguistic distinction between assigning a time-reference to an action and assigning a time-reference to the agent of that action.</Paragraph>
    <Paragraph position="8"> However, our point here is not to argue for a specific analysis of the Greek facts. It is simply to point out that sense-shifts of the type that LIRs represent are sometimes language-specific. Either we can encapsulate the differences in a single LIR, present in language A, absent in B. Or else we can posit a different assignment of semantic primitives, with B being systematically stricter in the senses allowed for a given class of words (-- in this case, English will not allow its temporal adjectives to assign a time-reference to a whole action). Either way, we shall have gained some concrete evidence on linguistic semantics. What we are not free to claim is that pragmatic considerations alone can resolve such systematic differences between languages.</Paragraph>
    <Paragraph position="9"> A similar cross-linguistic argument for the linguistic reality of LIRs can be made from dialectal differences within the same language.</Paragraph>
    <Paragraph position="10"> Consider for example the treatment of &amp;quot;day-nouns&amp;quot; in Standard British English as against their treatment in various other dialects, including Cockney and General American. Whereas the indexical nouns today, tonight, yesterday, tomorrow occur as adverbs in all dialects, the days of the week Sunday through Saturday cannot  occur adverbially in Standard British English: he came *(on) Tuesday. To account for this, one could posit an LIR Day Noun - Adverb, with a more restricted domain in Standard British English than in the other dialects.</Paragraph>
    <Paragraph position="11"> But to posit this is already to accept that the LIR is a feature of the grammar of English. Contrastive analysis, whether applying within a language or between languages, will, we expect, continue to raise a host of problems for any approach which tries to account for the fine detail of semantics as side-effects of pragmafics. The real world, and the human condition, is not so different from one languagecommunity to another (let alone between dialect-areas). But as languages and their grammars differ, so does their lexical semantics.</Paragraph>
  </Section>
class="xml-element"></Paper>