7
Lectur e I,  1 We must begi n ou r account wi th the questi on, of ten posed in recent ti mes,  What is syntax?  If we go simply by its etymology, the answer is easy: syntax is the ancient Greek word  ÆØ, abstract derivative of  ıø  (‘put in order together, organize, arrange’).  The word is attested already in the fth century in several senses derived from the verb, which do not concern us here. We shall conne ourselves to the use of the wor d amo ng tho se who condu cte d research on langua ge. They use ÆØ  in the rst place of the joining together of the letters of the alphabet. Herodian (2nd c.  ad), one of the most eminent of the grammarians, who had perhaps more inuence than any on later centuries, wrote a monograph entitled æd ıø H Øåø  (‘Concerning the Conjunction of the Letters’). 1 In it he asks what sort of letter combinations are possible and, in particular, how syllables are to be divided. Secondly, and analogously, the word  ÆØ  was applied to the combining of words in a sentence. In the Roman Imperial period this is standard, the reference being sometimes simply to the purely formal act of stringing words together, sometimes to the syntactic structure. The word may have been used in this sense already in the linguistic theory of the Stoics. 2  At all events, the term ‘syntax’ achieved signicance and acknowledged cur- rency—and, I would add, renown—through a frequently cited work written by the father of Herodian,  Apollonius Dyscolus, who taught in Alexandria. His nickname refers to the difculty that he caused his pupils in his writings and le ct ur es , and for us, too, hi s expl anat ions ar e fa r fr om easy to unde rstand. 3 Beside a great many other works, he wrote four books  æd ıø. This work has 1  Aelianus Herodianus of Alexandria was active at Rome under Marcus Aurelius, to whom he dedicated his magnum opus, the  ˚ÆŁºØ Œc æ ø Æ, on the accentuation of Greek in  20  books. The monograph to  which W. here refers (GG III.2,  390406) was probably part of a larger work on orthography; for an introduction to research on Herodian since  1890, see Dyck (1993). 2 It occurs in the title of two works ascribed (by Diog. Laert.  7.  1923) to Chrysippus, head of the Stoa from  232  bc and the most inuential of the Stoics after their founder Zeno of Citium (335263  bc), especially in Stoic logic, which included what we would call grammar. On Stoic linguistic theory, see I, 1415 and nn.  67, pp.  245 below. 3  The view that  ˜Œº means ‘difcult to understand’, of his writings is only one of three competing explanations of Apollonius’ cognomen, all of them in the ancient  Life  ( GG II. 3, xi–xii). The suggested alternatives are (a) ‘difcult, peevish’, of Apollonius’ ill temper, and (b) ‘difcult to answer’, of Apollonius’ posing of impossible questions in grammatical contests in the gymnasia.

David Langslow Jacob Wackernagel, Lectures on Syntax With Special Reference to Greek, Latin, And Germanic 2009

Embed Size (px)

Citation preview

  • Lecture I, 1

    Wemust begin our account with the question, often posed in recent times,What

    is syntax? If we go simply by its etymology, the answer is easy: syntax is the

    ancient Greek word , abstract derivative of (put in order

    together, organize, arrange).

    The word is attested already in the fth century in several senses derived from

    the verb, which do not concern us here. We shall conne ourselves to the use

    of the word among those who conducted research on language. They use

    in the rst place of the joining together of the letters of the alphabet.

    Herodian (2nd c. ad), one of the most eminent of the grammarians, who hadperhaps more inuence than any on later centuries, wrote a monograph entitled

    d H (Concerning the Conjunction of the Letters).1 In

    it he asks what sort of letter combinations are possible and, in particular, how

    syllables are to be divided. Secondly, and analogously, the word was

    applied to the combining of words in a sentence. In the Roman Imperial period

    this is standard, the reference being sometimes simply to the purely formal act of

    stringing words together, sometimes to the syntactic structure. The word may

    have been used in this sense already in the linguistic theory of the Stoics.2

    At all events, the term syntax achieved signicance and acknowledged cur-

    rencyand, I would add, renownthrough a frequently cited work written by

    the father of Herodian, Apollonius Dyscolus, who taught in Alexandria. His

    nickname refers to the difculty that he caused his pupils in his writings and

    lectures, and for us, too, his explanations are far from easy to understand.3 Beside

    a great many other works, he wrote four books d . This work has

    1 Aelianus Herodianus of Alexandria was active at Rome under Marcus Aurelius, to whom he dedicatedhis magnum opus, the c , on the accentuation of Greek in 20 books. The monograph towhich W. here refers (GG III.2, 390406) was probably part of a larger work on orthography; for anintroduction to research on Herodian since 1890, see Dyck (1993).

    2 It occurs in the title of two works ascribed (by Diog. Laert. 7. 1923) to Chrysippus, head of the Stoafrom 232 bc and the most inuential of the Stoics after their founder Zeno of Citium (335263 bc),especially in Stoic logic, which included what we would call grammar. On Stoic linguistic theory, see I,1415 and nn. 67, pp. 245 below.

    3 The view that means difcult to understand, of his writings is only one of three competingexplanations of Apollonius cognomen, all of them in the ancient Life (GG II.3, xixii). The suggestedalternatives are (a) difcult, peevish, of Apollonius ill temper, and (b) difcult to answer, of Apolloniusposing of impossible questions in grammatical contests in the gymnasia.

  • been in print since the sixteenth century, but the only edition worth considering

    today is that of Uhlig (1910), which has a reliable text and a jI, 2 brief runningsummary of the contents, which makes it possible to follow the difcult train of

    thought. Apollonius sets himself the task of explaining how independent sen-

    tences can arise through the joining together of words, and this involves him in

    going through the functions of the parts of speech one by one. One indication of

    the great regard in which this work was held is the fact that Priscian (for us the

    most important of the Roman grammarians, who lived, taught, and wrote in

    Constantinople in the rst decades of the sixth century), after presenting the

    phonology and inectional morphology of Latin in sixteen books, followed these

    in books 17 and 18with a Latin version of Apollonius Dyscolus, applying to Latin,of course, what Apollonius had taught for Greek. Priscian translated as

    constructio (putting together, arrangement; 17. 1 GL III, 108. 12).4Thanks to Apollonius, the term syntax became established roughly in the

    sense of sentence structure. I cannot say to what extent it remained alive over the

    centuries, but at all events nineteenth-century grammarians operated with it.

    More important is the question how we are to use the term and what we should

    treat under the heading syntax. This has been much debated in the last few

    decades. Since the question What is syntax? had already led many scholars to

    better-focused reection on matters of principle, particularly great attention was

    paid to the book bearing the question as its title, published in 1894 (2nd edn1927) by John Ries (18571933), a Germanist engaged in independent research inthis area. In fact, nobody before Ries was ever quite clear what was supposed tobe included under what was called syntax.5 Here are some of the ways in which

    syntax has been dened. I conne myself to the most important.

    A serviceable systemin particular because of its perfect consistencyis that of

    the Syntax of the Slavic Languages (186875, 2nd edn 1883) by Franz Miklosich(181391), a leading Slavist, who has done great service also in his work on theGypsy languages. He begins this book, the fourth volume of his Comparative

    Grammar of the Slavic Languages, with the sentence, Syntax is the study of the

    meaning of word-classes and word-forms. He accordingly sets out to determine

    the meaning of the grammatical forms, and carries this through with perfect

    consistency. He has produced an extraordinarily instructive set of material,

    although it must be said that in this approach much goes by the board about

    which we should like to hear and about which we should justiably expect to be

    4 In fact, the rst edition appeared in 1495 (see Uhlig 1910: xiiixxiii); Uhligs remains the standardedition, although there is an excellent edition with French translation and notes by Lallot (1997). For anEnglish translation of the Syntax, see Householder (1981) (with Blanks review, 1993b), and for discussionof its content and its place in the history of philosophy and linguistics, Blank (1982) and (1993a), Sluiter(1990: ch. 2), Matthews (1994: esp. 7896). On Priscian, see Law (2003: 8691) with bibliography.

    5 W. cannot be wholly serious here: on 19th-c. syntax in general, see Morpurgo Davies (1998: 30411)and esp. the excellent monograph by Graf (2001).

    8 general introduction

  • informed: we learn nothing about sentence types as such, nothing about the

    accentuation of the sentence, nothing about word order, nothing about connec-

    tions between sentences.jI, 3A second system is presented in the work of the classicist Christian Carl Reisig

    (17921829), whose lectures on the Latin language were edited (1839) by his pupilFriedrich Haase in Breslau.6 Reisig divided his whole grammar into three parts.In the rst, he presented the morphology, which he called etymology; in the

    second, semasiology, the theory of meaning, where he deals with the meaning of

    the grammatical forms; and in the third part comes the syntax, the theory of

    the combination of words. The substance of his second part is roughly what

    Miklosich presents in his syntax, while in Reisigs third part the means ofcombining and joining words are explicitly addressed.

    A third, related, form of presentation is that which Krger chooses in hisGreek grammar for schools. He divides his syntax into two parts (187391: 1. 2. 1).The Analysis treats in turn (a) the noungender, number, caseand the article,

    and (b) voices, tenses, and moods of the verb, and innitive and participle. In the

    second main part, the Synthesis, Krger discusses the combining of nouns andadjectives; sentence structure; relations of agreement; the construction of simple

    and complex sentences. So Krgers Analysis corresponds more or less toReisigs Semasiology, his Synthesis to Reisigs Syntax.7

    In contrast with these lucid systems, most accounts are characterized by

    complete confusion, comments on sentence structure being made both before

    and after those on the functions of the individual forms. Obviously, such a lack of

    system cannot possibly be of any benet to the description itself. In no scholarly

    discipline is untidiness more out of place than in grammar. But what Miklosichoffers is not satisfactory either, because he says nothing about combinations. And

    now Ries has championed the view that the only proper business of Syntax iswhat Krger calls Synthesis, and that the meaning of case-forms, etc. must betreated as part and parcel of morphology. Riess vote has had great inuence. Inparticular, the leading contemporary exponent of systematic comparative gram-

    mar, Karl Brugmann, has sided with Ries and treated the meaning of theinectional forms under morphology. Strictly speaking, there is of course some-

    thing nonsensical in saying rst of all that this form is the genitive of, say, lucrum,

    and then in another place that the genitive has such and such a set of functions.

    Consideration of function cannot be separated from that of form because not all

    forms which bear the same name have the same value; this very form, for

    example, the so-called genitive lucri, occurs in usages unattested for third-declen-

    sion genitives (cf. II, 217 below). Or, again, it can be shown that the plural of

    6 Note also the later edition, in three volumes, Reisig (188190).7 On Krugers Greek grammar, see further I, 303 below.

    what is syntax? 9

  • neuter nouns jI,4 is not identical in function with plurals of masculine and femininegender (I, 89 below). In Greek the same sort of thing is true of the aorist passive(I, 1379 below).In principle, the procedure advocated by Ries and followed by Brugmann is

    absolutely right: also the boundary that Ries has proposed between grammar andstylistics deserves, on the whole, our approval. But of course we do have to take

    account of the actual state of affairs: in nearly all existing grammar books, and in

    lectures on grammar, themeaning of the inectional forms is not dealt with under

    morphology proper, and consequently syntax has to ll in the gaps. I shall

    therefore treat not only Krgers Analysis and Reisigs Semasiology butalso what is understood as syntax in the narrower sensethis is the approach

    taken also by the most eminent syntactician of the present age, Berthold

    Delbrck (18931900: I, 7386). In the interests of clear presentation, however,we shall do well, as Delbrck does, to keep these two subjects separate.8

    With this in mind, I intend to present a swift overview of the syntax ofGreek,

    Latin andGerman. I am presupposing no specialist knowledge of these subjects,

    merely the knowledge you bring with you from school. I have thought it

    important to deal with German alongside the classical languages because we

    shall thereby enjoy the benet of being able to refer to our own native-speaker

    intuitions, and because one of the tasks of the grammarian is to illuminate living

    languages and not to have his eye only on the past.9 This does not mean that I am

    8 In fact, Berthold Delbruck (18421922) had died a few years before the publication of the 2nd edn ofthe Vorlesungen. Note the biography of Delbruck by Eduard Hermann (1923a), and especially on hisIntroduction (1880), see Morpurgo Davies (1998: 235, 2456); on Delbruck and Karl Brugmann(18491919), note also Sebeok (1966: I, 48996 and 57580). Delbruck confronts John Ries explicitly inthe opening of the third volume of his comparative syntax (18931900: III, 15). There is illuminatingdiscussion of the individuals mentioned in the last few paragraphs in Morpurgo Davies (1998) and Graf(2001), both with further references: of the classicists Reisig and Haase, now regarded as the founders ofGerman Semasiologie, the systematic study of meanings and their interrelations and developments; ofMiklosich and his (de facto) identication of syntax with morphosyntax; and especially of John Ries, hisreaction againstMiklosich, and his inuence. MorpurgoDavies characterizes (1998: 311) Riess nal aim asan integrated theory of form and meaning in which syntax implicitly acquires a primary role since themeaning and function of words can only be established starting from the utterance. She sees (335 n. 47)Riess inuence also in e.g. Meyer-Lubkes introduction to vol. 3, on syntax, of his grammar of theRomance languages (18901902), and she quotes an essentially Riesian denition of syntax from a recentstandard work on English syntax (McCawley 1998). Brugmann sides with Ries in his Kurze vergleichendeGrammatik (19024), with its successive parts on the word-forms and their use (281 ff.) and the formationof phrases and sentences (623 ff.), and in the second edition of the Grundriss (18971916), part 2 of which isdevoted again to the word-forms and their use. In Brugmann & Thumb (1913: 41416), the importantpractical point is made that, while Riess strict partition of the study of (a) the form and (b) the meaning of(1) single word-forms (Wortlehre) and (2) combinations of words (Syntax) is in theory all well and good,it is not always easy or even possible to keep 1b and 2b neatly apart. For a contemporary British reactionagainst the separation of syntax from stylistics, note e.g. Lawton (1900): W. would have sympathized withLawtons main point that nothing can be understood or enjoyed aright when torn out of its properplace. . . . Linguistics is biology, not anatomy.

    9 Linguists recognized already in the 19th c. that this is one of the grammarians tasks partly because thebenets are in fact mutual, that the study of living languagesand the development of synchronic and

    10 general introduction

  • undertaking always to do justice in equal measure to all three languages. Some-

    times one language, sometimes another may tend to hold centre stage. I also

    reserve the right to discuss some features in more detail than others, and occa-

    sionally even to seize on other languages, if it serves our purpose.

    It is a very valuable thing in a linguistic account such as this not to be restricted

    to a single language but to take several together. The solving of a problem

    benets immeasurably from a comparative approach. The comparative method

    has shown itself to be fruitful and indispensable in other disciplines, too. Making

    comparisons between languages presupposes a certain degree of similarity and

    relatedness, but even for the languages which are our present concern, three types

    of relatedness come into question, j I, 5just as in other elds, such as law, there canbe three types of relatedness between different peoples.

    (1) Relatedness based on human nature, on general laws of the human psyche,fundamental relatedness. There is no doubt that there are phenomena of a

    syntactic nature (syntactic in the broadest sense) which may be traced across

    many languages of the world, without there being any connections between the

    peoples who speak these languages. These features are best described precisely in

    terms of their universality, and they show that some expressions that we nd in

    Greek, Latin, and German are deeply rooted in the nature of human speech.10

    Let me pick out a couple of examples. When you study Latin, you learnand

    may be surprised bythe rule that after a comparative adjective the noun denot-

    ing the thing compared goes in the ablative. And if you later on come to learn

    Greek, you are told that Greek possesses an analogous short form of the com-

    parative construction, except that it then puts the noun in the genitive. In the old

    days, when languages were studied in isolation, that Latin ablative was under-

    stood as being in instrumental function and, e.g., maior fratre (greater than his

    brother) was interpreted as literally greater bymeans of, or through, his brother.

    The genitive in the Greek construction, on the other hand, was identied with

    the genitive that is found with the superlative and viewed as a partitive genitive.11

    diachronic description and theoryis essential for assuring and enhancing the quality, the reality, ofstatements about dead and reconstructed languages. See Morpurgo Davies (1998: 2379, 267) with furtherreferences. Conversely, the study of language change provides vital clues to synchronic structure; verygood on this last point is Kiparsky (1970: esp. 31415).

    10 Today, this formof linguistic comparison is called (language) typology andmakes prominent reference,including in the titles of textbooks, e.g.Comrie (1989), Croft (2003), to (language) universals. Croft (2003: 1)distinguishes typological classication (the classication of structural types across languages . . . having to dowith cross-linguistic comparison of some sort) from typological generalisation (the study of patterns thatoccur systematically across languages, generally involving language universals). On linguistic typology andclassication in the 19th c. and before, see Robins (1973) and Morpurgo Davies (1998: 21219).

    11 This account was perhaps encouraged by the apparent (strictly illogical) use of the superlative for thecomparative with a genitive plural, as e.g. in the parodos of Sophocles Antigone 100 H lit. light loveliest of previous (lights), i.e. lovelier than any previous. This is memorablyimitated byMilton, Paradise Lost 4. 323 Adam the goodliest man of men since born jHis sons, the fairest ofher daughters Eve.

    comparing syntax across languages 11

  • In fact, however, the relationship between the Greek and Latin expressions can be

    explained only by the fact that Greek genitive forms generally have ablatival

    function. The comparison makes clear that we must see an ablatival force in

    this form of expression, and this is made certain by Sanskrit, where, in the

    corresponding construction, a pure ablative is usual, a case formally distinct

    from the Sanskrit instrumental. Indeed, the use of an ablatival expression

    with a comparative adjective is grounded in human nature, a fact that can be

    demonstrated from later forms of the classical languages. First, an example from

    modern Greek, where we commonly nd expressions of the type

    Ipe e I, lit. richer from his brother, exactly like Latin ditior fratre (abl.).

    The comparative relation is felt as a distance, as a departure. By being richer, the

    rich man distances himself from the poor man. It is a form of expression that

    cannot be regarded as a sort of imitation of earlier Greek but which springs

    directly from the natural language instinct. Something similar occurs in later

    Latin in the writings of the gromatici (or agrimensores land-surveyors), where

    we nd expressions such as plus a quattuor lapidibus, more than jI, 6 four stones,literally more from four stones (p. 250, 8 Campbell p. 344, 8 Lachmann).12Among cognate languages, Slavic, for example, has an identical usage, e.g. in

    Polish mi mniej od nich, I have less than they; od normally means from, away

    from (see now Ernst Fraenkel 1921: 205);13 so, too, does modern Persianwith az from, out of (cf. I, 303 below). And if we turn to an unrelated languagelike Hebrew, we nd an analogous use of the preposition min away from

    with the comparative; similarly in Turkish, where than is expressed with the

    ablative sufx.14

    You may know from reading Homer that, while the direct object is usually in

    the accusative, when it is just a part of something, the genitive is used, so e.g. at

    Iliad 9. 214 a kr he sprinkled some salt (gen.); similarly in Thucydides at4. 80. 2 ei ktym K to send out some of the helots (gen.). In exactly thesame way in Finnish it is usual to mark an object in its entirety with the

    accusative, a part of it with the genitive.15 In Homer, even the subject can

    12 On the Roman land-surveyors, see nowGuillaumin (1996) and Campbell (2000). For the addition ofprepositions, note already Ovid,Heroides 16. 98, 17. 69. It has been suggested that the retention of the bareablative of comparison in Latin was due to Greek inuence; see Coleman (1977: 143), with references.

    13 In Polish, the construction with od is less bookish than that with niz than (nom.), but is usedonly when (a) the terms of comparison are noun phrases, and (b) the rst term is either the grammaticalsubject of the sentence or the logical subject of the comparative (see Rothstein 1993: 7067).

    14 On the reconstruction of the syntax of comparison in Proto-Indo-European, see Andersen (1980).The separative (i.e. ablative) construction emerges as the most prominent in Stassens recent study (1985)of the comparative construction in a sample of 110 languages; cf. the surveys of comparative constructionsin Andersen (1983) and Crookston (1999).

    15 Or rather, with the partitive, another of the many cases of Finnish, distinct from the genitive; as inW.s next example, the Finnish partitive alternates also with the nominative in subject function. See furtherSulkala & Karjalainen (1992: 21012), Abondolo (1998a: 1568), and on the Finnish cases, I, 301 below andn. 2, p. 502.

    12 general introduction

  • stand in the genitive if only part of it is involved, as in the negative clause at Iliad

    13. 191 h wqoer Y nowhere was (any of ) his skin (gen.) visible.16

    Comparison is fruitful not only in that it makes us recognize similarities but

    also in that it helps us to see variations, and thus saves us from thinking that what

    is familiar to us must necessarily be as it is. There is a difference between our three

    languages, on the one hand, and the Semitic languages, on the other, with regard

    to tense.When we use a verb form, we think above all of the time of the action, or

    tense (Zeitstufe). Anyone familiar with the Semitic languages, however, knows

    that this distinction between tenses is foreign to them. A Semitic speaker cannot

    directly represent distinctions of time. He distinguishes only between complete

    and incomplete actions.17Or again, one of the most important rules of word

    order in the Indo-European languages is that the determiner precedes the thing

    determined. We say, for instance, Gottes Haus (Gods house), Greek says

    e (Zeuss city), while the Semitic expression is fundamentally reversed:

    beth-el houseof God.18

    16 Cf. Janko (1994: ad loc.). With a negative, this use of the gen. is apparently inherited from IE;without, it is not found in Greek before the 5th c. (Herodotus 3. 102); cf. Schwyzer & Debrunner 102.Chantraine (1953: 501) compares Il. 22. 319, 325; Schwyzer adds Arist. Wasps 352.

    17 In Biblical Hebrew, for example, the perfect (of punctual aspect) is used of complete actions, e.g.katabt I wrote, as opposed to the imperfect (of durative aspect), which is used of ongoing or futureactions, e.g. ektob I write; will write; am writing; on Hebrew, see further McCarter (2004: 3478). Thesecontrasting verbal conjugations are traditionally called tenses, at least with reference to the West Semiticlanguages, although aspect would be more appropriate; for surveys of this immensely complicatedand controversial area of Semitic linguistics, see Moscati et al. (1964: 16.2831), Lipinski (1997: 332,334), and Stempel (1999: 12430).

    18 On this Indo-European pattern, which some believe goes hand in hand with the placement of theverb after its object (I, 7 below), see Lehmann (1974: esp. ch. 3) on nominal modiers, Adams (1976) withspecial reference to Latin, and more generally Lehmann (1992: ch. 12). In Indo-European the rule is morecontroversial than one would guess fromW., placement depending vitally on the nature of the determiner(see e.g. Pinkster 1990a: 1848, with further references); in Semitic, things are more straightforward,though again not without principled exceptions; cf. Lipinski (1997: 494504) on attributive, appositionaland genitival or subjoining relations in the nominal phrase.

    comparing syntax across languages 13