syntax & semanticspwd/sands01.pdf · presentation of syntax and semantics will depend...

23
Chapter 1 INTRODUCTION 1. Introduction Although the title of this book is Syntax and Semantics , the focus will be more consistently on the semantics of syntax. Before beginning the discussion, it will help to be as explicit as possible about the attitudes which shape the exposition in the following chapters. It should be emphasized that these are orientations and not absolutes. Their value and the value of this presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the inquiry is the belief that every distinctive meaning which is present in speech will have an audibly distinct expression. 1 The second assumption is that if there are distinct forms, then there will be some difference in meaning. That is, meaning sustains form. Language means; and because language means, form exists only in so far as it serves meaning. Form isolated from meaning is not instructive (and non- existent). In the same way that form does not exist in isolation, utterances do not occur in isolation. It is literally impossible. These assumptions combine to imply that there is no autonomous syntax. That is, ‘syntax’, ‘grammar’, or ‘form’ exists only because it encodes some meaning. Because of this, it makes no sense to try understanding grammatical patterns except by asking what they mean. The assumptions which guide the presentation in this book is then a kind of ‘functionalism’. But the word ‘functionalism’ designates a collection of attitudes, which are not necessarily internally consistent. Among the variety which Givón (1995) discusses, the functionalism represented here is ‘naive functionalism’. One distinguishing aspect of such a functionalism is the belief that all differences in form convey some difference in meaning. There are no meaningless forms, yet it is not infrequent that belief in meaningless forms is expressed. Schauber (1979) comes to this conclusion 1 This does not preclude distinct meanings sharing an expression, i.e., ambiguity. It only insists that if a meaning is present, it will be signalled somehow.

Upload: others

Post on 11-Oct-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

Chapter 1

INTRODUCTION

1. IntroductionAlthough the title of this book is Syntax and Semantics, the focus will be

more consistently on the semantics of syntax. Before beginning thediscussion, it will help to be as explicit as possible about the attitudes whichshape the exposition in the following chapters. It should be emphasized thatthese are orientations and not absolutes. Their value and the value of thispresentation of syntax and semantics will depend ultimately on the quality ofthe understanding they yield.

The first assumption that guides the inquiry is the belief that everydistinctive meaning which is present in speech will have an audibly distinctexpression.1 The second assumption is that if there are distinct forms, thenthere will be some difference in meaning. That is, meaning sustains form.Language means; and because language means, form exists only in so far as itserves meaning. Form isolated from meaning is not instructive (and non-existent). In the same way that form does not exist in isolation, utterances donot occur in isolation. It is literally impossible.

These assumptions combine to imply that there is no autonomous syntax.That is, ‘syntax’, ‘grammar’, or ‘form’ exists only because it encodes somemeaning. Because of this, it makes no sense to try understanding grammaticalpatterns except by asking what they mean. The assumptions which guide thepresentation in this book is then a kind of ‘functionalism’. But the word‘functionalism’ designates a collection of attitudes, which are not necessarilyinternally consistent. Among the variety which Givón (1995) discusses, thefunctionalism represented here is ‘naive functionalism’. One distinguishingaspect of such a functionalism is the belief that all differences in form conveysome difference in meaning.

There are no meaningless forms, yet it is not infrequent that belief inmeaningless forms is expressed. Schauber (1979) comes to this conclusion

1 This does not preclude distinct meanings sharing an expression, i.e., ambiguity. It onlyinsists that if a meaning is present, it will be signalled somehow.

Page 2: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

2 SYNTAX & SEMANTICS

concerning Navajo -go. For example, “Navajo does not always distinguishtemporal, conditional, and resultative subordinate clauses” (Schauber1979.224). Consider these (Schauber 1979. 224, 236, 238, 241, 251 & 252):

(1)[horse 2.Perf.rope- 1.father 3.3.Fut.buy]‘If you rope a horse, my father will buy it’‘When you rope a horse, my father will buy it’

(2)[1.father 3.Perf.come- Pl.1.Fut.eat]‘If my father comes, we'll eat’‘When my father comes, we'll eat’

(3) (a)[Mary 1.3.Perf.come-John 3.with 3.be.well]‘Because Mary came to see me, John is happy’

(b)[Mary 1.3.Perf.come-John 3.with 3.Perf.be.well]‘When Mary came to see me, John was in a good mood’

Because of the contrast (Perfective ... not Perfective) in tense-aspect in (3a),there can be no temporal intepretation *‘When Mary came to see me, John ishappy’. Hence, the sensible relation between must be some other one, e.g. acausal one ‘Because Mary came to see me ...’ But in (3b), where the conflictis absent, a temporal sense is possible (Schauber 1979.225). In response tothis array, Schauber (1979.226) states:

It is clear, however, that /-go/ does not mark any specific relationship since it isused to replace when, if , and because. Therefore, if /-go/ has any semantic content,it must be some abstract concept consistent with all three meanings, a concept Iwill refer to as adverbial since it occurs in adverbial clauses and define simply asthe semantic common denominator of such words as if , when, and because.Adverbial is being used here as a heuristic device, one needing no furtherclarification since I will argue that /-go/ in fact has no semantic content at all[My emphasis, PWD].

In Lisu, a similar diversity in the occurrence of ma prompts Hope (1974.89) toremark that

Page 3: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 3

The definition of ma in semantic terms is impossible, as it is open-ended and all-inclusive, referring to as many different objects as there are nouns in the language. In fact it has no meaning [My emphasis, PWD], rather than a lot of meanings.

The assumption in this book is that if there is indeed a formal contrast, thenthere must be a contrast of meaning. The existence of a ‘formal’ difference isnot a given, nor is the recognition of some semantic contrast. There is anegotiation on the part of the linguist which ends with the statement, “Yes,there is a difference and there is a meaning to be described”, or “No, there isno difference”.This assumption is then a directive to search in a certain way,to be as sensitive and as open to possibilities as one can be. It does notpreclude one outcome or the other.2 The relevant question is always “Whatdoes it mean?” (and how do we understand that query and any response to it).3

It is almost never straightforward, and it is never easy to grasp the content ofgrammar, to understand its intricacies, and to give them some reasonableexpression.4 And answers to our inquiry must refer always to usage and itscontext. This view prompts us to search to understand the meanings of piecesof language, where we suspect that they may be different, and not to accept‘meaningless’ forms nor synonymy (“They mean the same thing”) as answersto morphosyntactic puzzles. Examination of shapes is interesting because theshapes provide us an entrance into meaning. That, of course, does not implythat the meanings will be easy to identify and describe. Some will resistprecise understanding for many years, but that does not say that meaning isabsent, only that we are not yet able to articulate it.

2. The Form of SyntaxThe second important effect of naive functionalism concerns the converse

of the belief that if there is a formal contrast, then there must be a contrast ofmeaning. Namely, a contrast in meaning must likewise be supported by aformal difference, as a formal difference is maintained by a semanticdifference.5 The position here is that the converse is not accepted. That is, asemantic difference can exist in the absence of two audibly distinct

2 The semantic problems of Navajo -go and Lisu ma are discussed in Davis Ms.c.

3 The practice that lies behind this description will not de discussed here. Some of theseissues have been addressed in Davis (1995b). We say here, though, that any response to thequestion “What does it mean?” is not accepted as a direct answer of the sort that we want.

4 Harris (1993:6), in a discussion of recent linguistic behavior, calls meaning “elusive”.

5 This is not a denial of the existence of homonymy.

Page 4: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

4 SYNTAX & SEMANTICS

expressions. This is important because it opposes the present concept ofsyntax and semantics even more strongly to the more commonly acceptedview. The issue here is the use of ‘hierarchy’.

An example of this point can be found in Yogad (Austronesian,Philippine):

(4) Nap-pa-pinta kan tu binalay ku[NAG-PA-paint I house my]

Knowing nothing but the utterance itself, a Yogad speaker may hear ‘I hadsomething in my house painted’ or ‘I had my house painted’. In otherPhilippine languages, the two senses are formally distinct, e.g. Kinaray-a (oneof the Visayan languages in the Philippines),

(5) (a) Nag-pa-pinta ako kang binalay ko[NAG-PA-paint I house my]‘I had my house painted’

(b) Nag-pa-pinta ako sa binalay ko[NAG-PA-paint I house my]‘I had something in my house painted’

A common reaction to (4) by linguists would be to recognize the differentsenses as significant and to say that (4) is ‘ambiguous’. That observation isone about the meaning/usage of (4), but the ambiguity requires that thesemantic difference be somehow incorporated as a grammatical difference.Because there is in (4) no audible contrast of order, choice of form, orintonation (or pause), the grammatical difference which maintains thedifferent senses of (4) must be some other formal difference.

That additional formal device is hierarchy. Hierarchy is usuallyrepresented in ‘trees’ or in ‘nested bracketings’. This formal mechanism hashad a role in arguments concerning whether all languages languages have agrammatical contrast between ‘noun’ and ‘verb’, and we will examine aspecific instance of the use of hierarchy in this context. Sapir (1924:84) writesof the language Nootka:

Properly speaking, all ‘nouns’ are indeterminately such, being formally identical

with durative intransitives (e.g. ‘person, to be a person’) until nominalized by

or an equivalent element.

Page 5: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 5

In contradiction to earlier claims about Nootka, Jacobsen (1979) claims thatthe language contains a distinction between lexical nouns and lexical verbs;and Hopper and Thompson (1984:705) conclude that Jacobsen “shows, webelieve convincingly, that the distinction must be maintained, ultimately ongrammatical grounds [emph. mine, PWD]”.6 This suggests that we shouldlook more closely at Jacobsen’s evidence for such a conclusion.

Nootkan (composed of Makah, Nitinat, and the language Nootka) is thesouthern branch of the Wakashan family of languages.7 The examples whichJacobsen cites from Nootkan are taken from Makah.8 Sentences in Makah areconstrained to overt expression of only one PARTICIPANT (Jacobsen 1979:119):

... the language has a strong reluctance to admit more than one nominal argument per

clause ... Thus, although one might say that the word-order norm of the language is

verb-subject-object (VSO), the seemingly more significant fact is that a clause will

usually be just verb-subject (VS) or verb-object (VO), and nothing longer.

Utterances such as the following are then typical of Makah (Jacobsen1979:110 & 119):9

(6)[eat-caus-indic-3sg dogs-art]‘He’s feeding the dogs’

6 It is interesting that the acceptance of hierarchy in some form is so ingrained in linguisticsthat even such functional linguists as Paul Hopper, Sandra Thompson, and T. Givón wouldnot disown it. I think these three would not disagree with my characterizing them asfunctionalists, although they might now have an opinion of hierarchy different from theirearlier expressed ones.

7 The Wakashan family has two subgroups: Kwakiutlan (or Northern) Wakashan andNootkan (or Southern) Wakashan. The Kwakiutlan languages are Haisla, Heiltsuk, andKwakiutl. The Nootkan languages are Makah, Nitinat, and Nootka. Makah had “200 speakersreported in 1977 from a population of 600, on the Olympic Peninsula of Washington,opposite Vancouver Island. Most speakers are middle-aged or older” (InternationalEncyclopedia of Linguistics, Vol. 4, p. 235). Haisla has 250 speakers, Heiltsuk, 450,Kwakiutl, around 1,000 speakers, and Nootka, 500 speakers.

8 “Certainly the Nootkan languages are closely enough related that the general resultsobtained for any one of them would apply to all” (Jacobsen 1979:108).

9 In these examples, the morpheme-by-morpheme glosses are Jacobsen’s. Such glosses implya matching segmentation in the Makah, but it is not provided. Nevertheless, becauseJacobsen notes ‘eat-caus-indic-3sg’ of the first word in (1), we will assume that thosemorphemes/meanings are somehow present in the Makah.

Page 6: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

6 SYNTAX & SEMANTICS

NOOTKAN

NITINAT

NOOTKA

STRAITS SALISH

CowichanMusqueam

SQU

AMIS

HCOAST SALISH

COMOX

PENTLATCH

CHILCOTIN

Anahim

StoneChilcotin

Talio

Kimsquit

Bella CoolaBellaBella

BELLA

COOLA

HEILTSUK

Nahwiti

Koskimo

KWAKIUTLAN

KWAKIUTLKwakiutl

Yucult

Owikeno

MAKAH

LILLOOETFountain

Mount Currie

AlkaliLake

SECHELT

Figure 1: Map of the languages of the Wakashan family.

(7)[eat-indic-3sg dog-art]‘The dog is eating’

Page 7: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 7

(8)[man-indic-3sg work-art]‘It’s a man that’s working’‘The one working is a man’

(9)[ref-give-past-indic-1sg dog-art]‘I gave it to the dog’

Although these sentences are the most frequent expressions, “one can elicitsentences like these” (Jacobsen 1979:119-20):

(10)[point-indic-3sg man-art house-art]‘The man is pointing at the house’

(11)[give-past-indic-1sg boy-art dog]‘I gave the boy a dog’

Sentences such as (11) are avoided in part by the use of incorporation(Jacobsen 1979:120):

(12)[bone-give-past-indic-1sg dog-art]‘I gave the dog a bone’

A second method of avoiding citing more than one PARTICIPANT in a clause isto employ an absolutive form (Jacobsen 1979:115):

It is quite common for normal words to occur without any inflectional ending that

might indicate a function as predicate or argument. Such words are termed absolutive

(or absolute) by Sapir (1924; also Sapir & Swadesh 1939), and the question of

whether different functions can be discerned for them, especially in their occurrence

immediately following an inflected verb form, is central to our problem.

Examples of this construction are (Jacobsen 1979:115):

Page 8: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

8 SYNTAX & SEMANTICS

(13)[many-indic-3 bird fly]‘A lot of birds are flying’

(14)[many-indic-3 baggage get.lost-mom]‘A lot of baggage got lost’

Jacobsen (1979:115) remarks about these sentences:

In each of these, only the first word has the third person indicative ending. I take the

second word (‘bird’, ‘baggage’) to be a noun [sic] acting as the subject of this verb,

and the third word to be a verb [sic] forming a separate clause but understood as

referring to the same third person subject.

But the second word can also be a ‘verb’ as well as a ‘noun’ (Jacobsen 1979:116):

(15)[many-indic-3 fly]‘There’s a lot of flying’

(16)[many-indic-3 get.lost-mom]‘A lot got lost’

(17)[many-indic-3 bird]‘There’s a lot of birds’

(18)[many-indic-3 baggage]‘There’s a lot of baggage’

Now Jacobsen (1979:121) claims that

probably the clearest defining characteristic of nouns as opposed to verbs is that they

may directly occur as subjects (and other nominal arguments), whereas verbs may so

occur only by being nominalized with the article

Page 9: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 9

Sentence (8) provides an example of a ‘verb’ with occurring as anargument. The sentences of (19) illustrate the occurrence with a ‘noun’ withand without (Jacobsen 1979:137):10

(19) (a)[big-indic-3 man-art]‘The man is big’

(b)[big-indic-3 man]‘It’s a big man’

Jacobsen (1979:137) says of (19b) that “the existence of the entity named bythe whole phrase [i.e., ‘man’, PWD] is being asserted, i. e., ... the predicativeending is in immediate constituency with the whole phrase (even thoughoccurring on the first word)”, and of (19a), that “the attribute is predicated ofthe second word, which is then the subject of the predication”.

The difficulty is that the distinction between noun and verb requires thatthe ‘verb’ and the ‘verb’ -Ø be distinct as argument and absolutive,respectively, while the ‘noun’ and ‘noun’ -Ø are distinct as twomanifestations of ‘argument’. Thus, in (17) - (18), ‘bird’ and ‘baggage’ arenot ‘absolutives’ as ‘fly’ and ‘get lost’ are in (15) - (16). Although formallyparallel, there exists a covert, formal distinction between (15) - (16) and (17) -(18). The repeated examples in (20) and (21) juxtapose the ‘contrasting’expressions:

(20) (a)[man-indic-3sg work-art]‘It’s a man that’s working’‘The one working is a man’

(b)[many-indic-3 fly]‘There’s a lot of flying’

10 The of is a morphophoneme which may appear as [] after vowels, or as somealteration in the preceding consonant (Jacobsen 1979:109).

Page 10: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

10 SYNTAX & SEMANTICS

(21) (a)[big-indic-3 man-art]‘The man is big’

(b)[big-indic-3 man]‘It’s a big man’

The difference between (20b) and (21b) is that there is a ‘clause boundary’between the first and second words in (20b) (as well as in [16]), but in (21b)(as well as in [17] and [18]) such a boundary is absent (Jacobsen 1979:116):

I consider the difference between these last two cases to reduce largely to a

difference in clause boundaries: in ... [(17), (18), and (21b)] we have nominal

arguments acting as subjects of verbs within the same clause (or more accurately as

heads of underlying modifying constructions ... [cf. (21b), PWD], whereas in ...

[(15)/(20b) (16)] we have two clauses, the second consisting of a predicate in the

absolutive (third person) form, understood to have the same subject as the preceding

predication ... thus it will be seen that the discriminating of predicates and nominal

arguments involving absolutives hinges crucially on the determination of clauseboundaries [emph. mine, PWD]. One disagreeing with the distinction might well be

led to a Whorf-like conclusion, projecting the latter case onto the former, so that the

nouns are viewed as separate predications, giving ‘there’s a lot; they are birds’ and

‘there’s a lot; it’s baggage’.

Jacobsen’s reason for imposing distinctions where none appear relies on suchbeliefs as these (Jacobsen 1979:118):

... unmarked ‘absolutive’ forms, in their occurrence connected syntactically topreceding predicative words (including those in absolutive form), play several

different syntactic roles, partly correlating with different parts of speech. If they are

nouns, they may act as subjects ... or objects ... of verbs, and as objects of

prepositions ... If they are adverbs, they may act as modifiers of verbs ... If they are

verbs ..., they may be in a construction with the preceding predication to indicate a

complex action.

The following illustrate the differences referred to (Jacobsen 1979:121, 125,131 & 133):

Page 11: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 11

(22)[come-now-indic-3 fog]‘The fog is coming’

(23)[ref-eat-past-indic-1sg hair.seal]‘I ate a hair seal’

(24)[eat-indic-1sg still]‘I’m still eating’11

(25)[know.how-indic-1sg sing]‘I know how to sing’

The differences which Jacobsen appeals to (i.e., ‘subject’, ‘object’, ‘modifiersof verbs’, and ‘complex action’) seem to arise more from viewing Makahfrom the perspective of English, or the European languages more generally.There appears to be nothing in Makah itself which requires that we see (22) -(25) as grammatically distinct. It is the English glosses which prompt theimposition of the distinctions; but if we see these sentences more from theperspective of Makah and suggest more Makah-like glosses:

‘It is coming, namely fog’‘I ate it, namely hair seal’‘I’m eating, namely still [happening]’‘I know how, namely [to] sing’

the opposition between the parts of speech becomes more elusive.Ultimately in Makah, the issue of the grammatical distinction between

noun and verb depends upon the presence of an unobservable clause

11 It should be noted that the Makah root for ‘still’ may also appear as the V in a sentence(Jacobsen 1979:125):

(i)[still-indic-1sg eat]‘I’m still eating’

Any difference between the gloss of (i) and that of (19) is not commented upon.

Page 12: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

12 SYNTAX & SEMANTICS

boundary, which varies tautologously with the equally invisible distinctionbetween noun and verb, ... all of which ignores the observable formalparallelism between ‘nouns’ and ‘verbs’ in the ‘absolutive’ forms, whichbelies their distinction.

Jacobsen (1979:116) goes on to claim that ‘noun’ and ‘verb’ can bedistinguished in (15) - (18) on the basis “that words in the momentaneousaspect are necessarily verbs (acting as predicates rather than arguments)”. But‘fly’ in (15) lacks that verbal stigma as does ‘wait’ and ‘basket-make’ in these(Jacobsen 1979:117):

(26)[tired-now-past-indic-1sg wait]‘I got tired of waiting’

(27)[careless-rep-now-indic-1sg basket-make]‘I’m making baskets carelessly’

And Jacobsen (1979:114) notes that “nouns also occur in Makah in themomentaneous aspect” and provides these examples:

(28)[man-mom-now-indic-3]‘He’s gotten to be a man’

(29)[baggage-mom-indic-3]‘He’s packing’

based upon ‘man’ and ‘baggage’. The distinction betweennouns and verbs once again escapes us. Consistent with the assumptionsannounced above, we will not attribute formal hierarchy to language as one ofits grammatical resources.

2.1 Taxemes and the Origin of HierarchyJacobsen’s use of ‘clause boundaries’ or hierarchy has become common

practice since the first part of the 20th century, and because recourse tohierarchy is so accepted now in American linguistics (both by formalists andby [some] functionalists), it may be instructive first to know that it was

Page 13: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 13

apparently not always so, and second, to trace something of the history of thedevelopment of the notion.

Hierarchy is absent from Bloomfield’s taxemes. The reliance uponhierarchy emerged in the 1940’s as the result of a developing school ofAmerican Structuralism. Historically, its motivation was the practical one oflanguage analysis. Bloomfield’s (1933) conception of grammatical patterningwas based upon a set of formal possibilities in language which he termed‘taxemes’. There were four of them:

(i) ORDER. “the succession in which the constituents of a com-plex form are spoken” (Bloomfield 1933:163).

(ii) MODULATION . Modulation consists of the use of “secondaryphonemes ... of pitch ... of stress” (Bloomfield 1933:163).

(iii) PHONETIC MODIFICATION. It is recognized as “a change in theprimary phonemes of a form” (Bloomfield 1933:163).

(iv) SELECTION. “The meaning of a complex form depends in partupon the selection of the constituent forms ... the features ofselection are usually quite complicated with form-classesdivided into sub-classes” (Bloomfield 1933:165).

The manipulations of (i) - (iv) are the only things to study in syntax. Anexample of a contrast by phonetic modification might be:

(30) (a) Who do you want to drive?(b) Who do you wanna drive?

where it appears that the same sequence of morphemes {want to}, but withdiffering morphophonemics, is the means of marking contrasting meanings. In(6b), with want to phonetically modified by a ‘reduced’ pronunciation, the youis known to be the one who is the passenger. But in (6a), with the ‘non-reduced’ pronunciation of want to, it may be the case that the you will be thedriver. The assumption here is that if a difference in meaning is signalled, thenthe expression is also different. Want to and wanna then represent differentforms, and (6a) and (6b) are examples of (iv).

Within American Structural (Post-Bloomfieldian) linguistics, two types ofpattern — grammar and phonology — exhaustively account for the patterns oflanguage, e.g. Hockett (1954.216):

Morphophonemic and tactical pattern taken together constitute GRAMMATICAL

Page 14: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

14 SYNTAX & SEMANTICS

PATTERN. This, paired with PHONOLOGICAL PATTERN, completes the synchronicpattern of language.

Tactical pattern is where we find syntax, and its terms are:

(i) HIERARCHY. Exemplified by such ambiguities as old men andwomen.“... hierarchical structure [is] ... an integral part of the structureof utterances” (Hockett 1954:218).

(ii) FORMS. I.e., the morphemes can mark tactic patterns. “Thehierarchical structure ... is partly marked by the nature andlocation of these morpheme-occurrences” (Hockett 1954:221).E.g. the agreement morphemes of the Latin:Pater bonus filium amat. ‘The good father loves the son’Pater bonum filium amat. ‘The father loves the good son’

(iii) CONSTRUCTION. “... a construction, in the linguistic sense(within I[tem and ]A[rrangement]), is a relation ... Theconstituents are elements [classes or forms, PWD]; they standin a particular arrangement (linear order). A construction, thenis a class of ordered n-ads of constituents, whereupon aconstruction is a relation” (Hockett 1954:225). E.g. Mandarin(Hockett 1954:222):(a) chau3 fan4 cr1 fan4

‘Fry rice!’ ‘Eat rice!’(b) chau3 fan4 hau3 fan4

‘fried rice’ ‘good rice’(a) differs from (b) only in the constructional relation — not inhierarchy or form, or order. The first in (a) is ‘verb-object’; thesecond in (b) is ‘attribute head’.

(iv) ORDER. May act as “independent of all other factors” (Hockett1954:222) in (a) John and Bill(b) Bill and JohnHierarchy, construction, and form are constant in (a) and (b),leaving order as the only variable.

Hockett (1954:227) likens construction in linguistics to the position ofrelation in mathematics, e.g. greater than. The motivation for the presence ofHockett’s constructions is that the same order of forms in chau3 fan4 has the

Page 15: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 15

same hierarchical organization, yet there is a systematic (thoroughgoingrecurrent) difference in meaning. Thus, chau3 fan4 is either ‘Fry rice!’ or‘fried rice’, and that difference must be attributed to something in addition tohierarchy, form, or order. Examination of other examples such as cr1 fan4 ‘Eatrice!’ and hau3 fan4 ‘good rice’ lack the ambiguity of chau3 fan4 and point toa difference in form class, perhaps ‘adjective’ versus ‘verb’. That is, theconstructions represented require the recognition of parts of speech.

There is a partial overlap between Hockett’s elements of tactical patternand Bloomfield’s taxemes. Bloomfield’s selection is the closest thing toHockett’s construction, and Bloomfield’s construction is an entirely differentthing from Hockett’s (Bloomfield 1933:184):12

The free forms (words and phrases) of a language appear in larger free forms

(phrases), arranged by taxemes of modulation, phonetic modification, selection, and

order. Any meaningful recurrent of such taxemes is a syntactic construction.

The major difference here is the presence of hierarchy on Hockett’s list. Itwas not on Bloomfield’s list of formal resources.13 Notice also that hierarchyis the only member of the list which requires the acknowledgement thatmeaning is involved in syntax.14

The intrusion of hierarchy has its origin ... as do most other modificationsto Bloomfield ... in the reliance upon methodology to generate bothdescription and theory. Wells (1947) provides a fairly challenging formulationof construction (Wells 1947:94):

... a construction is a class C of occurrences, subject to the following conditions: (1)

12 Hockett’s ‘forms’ are closer to what he later calls ‘markers’ (Hockett 1958). E.g. the formand in English marks the construction of coordination in English without participating in it.In his example here, the agreement morphemes of Latin ‘mark’ which elements go togetherin the construction of modification. In the way hierarchy is taken as knitting elementstogether, so do forms (or markers) associate terms as combining into some larger unit.

13 Modulation is recognized as one of the major syntactic constituents of an utterance inImmediate Constituent analysis on a par with Hockett’s form or with Bloomfield’s selection.Phonetic modification is generally discarded in syntax and recognized as an epiphenomenalpresence. Later, contrasts in phonetic modulation such as want to and wanna are taken upagain as evidence for ‘empty categories’ and ‘traces’ (Lasnik & Uriagereka 1988.22).

14 The distinction between constructions, e.g. verb-object and attribute-head, can berecognized by the different privileges of occurrence between the two: chau3 fan4 and hau3

fan4 versus chau3 fan4 and cr1 fan4 ... a formal property, not a semantic one. And of course,choice of forms or choice of order requires no acknowledgement of meaning.

Page 16: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

16 SYNTAX & SEMANTICS

there is at least one focus-class which includes all the sequences of which themembers of C are occurrences; (2) all these occurrences have a certain meaning in

common; and optionally (3) all these occurrences occur in a certain total

environment or in all of a certain class of total environments.

The difference between sequence and occurrence is roughly the same as thatbetween type and token.15 A construction is not an occurrence, an instance,but an abstraction/generalization away from instances. For example, theattribute-head construction of chau3 fan4 ‘fried rice’ and hau3 fan4 ‘good rice’is the class to which they belong. The focus class to which they belong can berecognized, for example, by their common occurrence in the context

___ ‘I like ...’16 Notice that Wells requires that the occurrences have “acertain meaning in common” in the same way that Harris and Hockett requirethat variants of a morphemic unit have a “same meaning”.17 Wells (1947:93)comments that

a theory of I[mmediate]C[constituent]s ... [can] be developed up to a certain point

without a consideration of meaning; that point has now been reached.

The same can be said of morpheme variants ... up to a point, we do not have toconsider that they mean. The point which Wells has in mind is the encounterwith old men and women, the king of England’s people, she made him a goodhusband, chau3 fan4 ‘fried rice’, and the like. They are ambiguous.18 And ifwe sort out the two senses of the king of England’s people, we need to makecertain that the appropriate occurrences of it are associated with my peopleand the Prince of Wales. If, by the king of England’s people, we mean the king... of England’s people, then the occurrence goes with the Prince of Wales(who may be the King of England’s people and substitute for the entire phrase

15 “Therefore, the statement ‘pages and pages contains three morphemes but fivemorpheme-occurrences’ is only a way of saying ‘pages and pages contains three morphemesand every occurrence of pages and pages contains five morpheme-occurrences” (Wells1947:94).

16 Wells’s and Hockett’s constructions are similar in that they are both essentially classes ofoccurrences. Hockett orders form classes to create constructions, and his classes embodyWells’s notion of sequence (type) as opposed to occurrence (token).

17 There is a proportion here. Constructions are to occurrences as morphemes are tomorphemic variants.

18 Wells (1947:93) calls it “equivocation”.

Page 17: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 17

the king of England’s people); but if we intend the king of England’s ...people, then the occurrence goes with my people with my substituting for theking of England’s). And in order to know which goes where, we ... as Wellssays ... have to consider meaning. Cf. his criterion-(2) for definingconstructions.

But none of this yet requires hierarchy. Wells (1947:96) explains itspresence via the use of Immediate Constituent analysis:

... this is the relevance of constructions in the theory of ICs — the IC-analysisshould reflect the construction. Consequently, when the same sequence has, indifferent occurrences, different meanings and therefore (provided that the meaning-

difference cannot be ascribed to the morphemes taken separately) different con-struction [sic], it may have different IC-analyses. The IC-analysis of a sequence

often reflects the semantic analysis of what the sequence means, but the meaning

needs to be considered in making the analysis only [Emphases mine, PWD] when

two occurrences of the same sequence (or of two sequences belonging to all the same

sequence-types) have meanings incompatible with each other.

IC analysis is the successive factoring of larger forms (utterances) into smallerones and finally their ultimate constituents, “to break it up into parts of whichsome or all are themselves expansions” (Wells 1947:83). Like Harris’s andHockett’s prescriptions for morphology, this is technique, one which is likethe others for morphology in invoking “the famous concept of patterning,applied repeatedly and in divers special forms” (Wells 1947:81).19 Byidentifying a focus and an environment (Wells 1947:86), we get theprocedural equivalent of a morph in some grammatical environment and of aphone in some phonetic environment. It is a dichotomous procedure and theresult is unavoidably binary . One compares two focuses which share acommon environment and if they have “a certain meaning in common”, and ifthe combination of the focus and environment creates a sequence(construction) which yields maximally general statements about privileges of

19 Wells (1947:84 & 93):

Our procedure aims only to tell, given two or more mechanically possible dichotomies..., how to decide in favor of one of them ... we do not propose our account as amechanical procedure by which the linguist, starting with no other data than the corpusof all utterances of the language and a knowledge of the morphemes contained in eachone, may discover the correct IC system. For any language, the number of possible IC-systems is very large; but in practice it is easy to see that most of the possibilities arenegligible. Just as when working out the phonemics, the practicing linguist willdiscover many shortcuts.

Page 18: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

18 SYNTAX & SEMANTICS

occurrence, then that IC cut is the preferred one. Once this procedure is begunit continues until we reach the ultimate constituents.

In Wells’s formulation, there remains a space between the hierarchy pro-duced by the method of IC analysis and the concept of construction. They arenot yet the same thing. The former should reflect the latter in order to beuseful, and it is only in the instances of “equivocation”/ambiguity that it issignificant. From Wells’ comments cited above, “when the same sequencehas, in different occurrences, different meanings” such that the “differencecannot be ascribed to the morphemes taken separately ... Differentconstruction[s] ... may have different IC-analyses”. Seven years later,Hockett’s (1954:220) position is different, and more extreme:

Our conclusion, then, is that hierarchical structure is a ‘primitive’ just as are formsand order. The demonstration turns necessarily on two considerations: (1) theremust be many utterances in which the hierarchical structure is unambiguous, toafford a frame of reference; (2) there must be also at least a few in which thehierarchical structure is ambiguous, since otherwise the hierarchical structurewould in every case be determined by forms and order, and hence not a‘primitive’.

With this, hierarchy is extended beyond the instances of ambiguity such thatevery order has hierarchy, whether ambiguous or not. Once hierarchicalized,always hierarchicalized.20 Now the notion of construction is wedded tohierarchy, but the motivation for the presence of hierarchy originates intechnique or procedure. Notice that without introducing hierarchy, Bloomfieldcould account for any of the patterns that this interpretation can. Only at thispoint in American linguistics, do box diagrams, labelled bracketing, trees,etc. become a required part of language.21

Post-Bloomfieldian theory projects two patterns upon language data,relabeling Bloomfield’s Semantics as Grammatical Pattern (according toHockett 1954:216). And because each morpheme has an atomic ‘same’ of

20 Recall “Once a phoneme, always a phoneme”. Hocket (1954:223) recognizes thisproblem:

... machinery which has to be in our workshop for use in certain marginal cases tends toobtrude itself where it isn’t wanted.

21 Harris’s (1946) complementary technique of working from morpheme ‘upwards’ toutterances yields essentially the same picture of syntax; but IC-analysis was the dominantconception. The notion is quickly extended to phonology. Cf. Hockett’s (1955) phonologicalhierarchy of utterances –> macrosegments –> microsegments -> syllables –> phonemes –>components.

Page 19: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 19

meaning that will recur with each instance of that morpheme and because eachgrammatical meaning (Bloomfield’s [1926:157] “constructional meaning”)corresponds to some one combination of hierarchy, form, construction ororder — there is an isomorphy between grammar and any putative semantics.Bloomfield (1933:138) provides us with a slight different alignment ofpatterns:

When the phonology of a language has been established, there remains the task of

telling what meanings are attached to the several phonetic [sic] forms. This phase of

the description is semantic [Emphasis mine, PWD]. It is ordinarily divided into twoparts, grammar and lexicon.

The lexicon is “the total stock of morphemes in a language” (Bloomfield1933:162). And the grammar is “the meaningful arrangements of forms in alanguage” (Bloomfield 1933:163). Figure 1 summarizes the relations and

Bloomfield American Structuralism (Hockett)

SEMANTICS

PHONOLOGY

Grammar + Lexicon

Morphophonemic Pattern + Tactical Pattern

Phonological Pattern Phonological Pattern

Figure 1: A comparison of two theories of language.

contrasts them with their altered interpretations in American Structuralism. Inaddition to the terminological differences, the two views have substantivedifferences in the “grammar”/“tactical pattern”. Now, Phonology and Seman-tics are structured in (semi-)analogous ways. Each has its minimum, mean-ingless element that combines into meaningful combinations (Bloomfield1933: 166):

A taxeme is in the grammar what a phoneme is in the lexicon [sic] –– namely the

smallest unit of form ... In the case of lexical forms, we have defined the smallest

meaningful units as morphemes, and their meanings as sememes; in the same way,

the smallest meaningful units of grammatical form may be spoken of as tagmemes,and their meanings as episememes.

Page 20: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

20 SYNTAX & SEMANTICS

Lexical forms are combined with tactic forms to create grammatical forms.But in no case is a hierarchy required, nor suggested by Bloomfield. And(Bloomfield 1933:169):

Whenever two (or, rarely, more) forms are spoken together, as constituents or a

complex form, the grammatical features by which they are combined, make up a

construction.22

One similarity with Bloomfield remains. Since IC analysis is intended tomirror closely the semantics of syntax, we seem to have a means of describingit; and we can achieve the description in an objective way with the prescribedmethodology. There was little methodology with Bloomfield’s taxemes, butthe expression of the form of grammar was also simultaneously an expressionof the relevant semantics. The particular/peculiar shape of grammar hereresults from a failure to recognize semantics as a distinct pattern of language.As the assumption of an autonomous phonology forced the ‘unnatural’recognition of phonological junctures — a clear intrusion of something else(grammar) into phonology — so the assumption of an autonomous grammarindependent of semantics forces the ‘unnatural’ intrusion of semantic patternsinto grammar as hierarchy. Since there is no distinct semantics to serve aslocus for the description of meaning, semantic pattern is transformed intogrammar as hierarchy. Hence, the wrongful assumption that

hierarchical constructions

are portions of grammar; they are not. They are badly understood semanticpatterns. In the same way that Hockett rejects morphs within phonology, soshould hierarchical constructions be labeled as artifacts of the analysis. Theyare false.

Because linguistics (linguists) cannot analyze these meanings, they arelike Saussure’s signifieds ... unpatterned. There can be no organization ofsememes/constructional meanings except that which is reflected in grammar.There is no separate semantics.23 This interpretation of syntax in the mid-

22 “Any meaningful recurrent set of such taxemes is a syntactic construction” (Bloomfield1933:184).

23 One of the few contradictions to that conclusion is Harris’s (1951) simultaneousmorphological components (Harris 1951:272):

categories [which] are not identifiable as consisting of any particular phonemes in the

Page 21: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 21

1950’s is the beginning point of Transformational Generative Grammar,which initially does not reject the view presented here, but extends it.

2.2 An alternative directionThe acceptance of hierarchy is also present among functionalists. Givón,

for example, offers (1995:177) a quartet of grammatical resources onlyslightly different from Bloomfield and Hockett:24

“observable components of grammatical structure: a. Linear order b. Nestedhierarchic structure c. Grammatical morphology d. Rhythmics: intonation andpauses”.

Givón’s ‘linear order’ repeats Bloomfield’s and Hockett’s ‘order’. Givón’s‘Grammatical morphology’ repeats Bloomfield’s ‘selection’ and Hockett’s‘form’. Givón’s ‘rhythmics’ repeats Bloomfield’s ‘modulation’. Givón’s‘nested hierarchic structure’ repeats Hockett’s, but not Bloomfield’s‘hierarchy’. It is on Givón’s list of observable components even though it is“more abstract and cannot be observed directly”.

In the discussion which follows, there will be no recourse to hierarchy. Iftwo distinct meanings are encoded, there are then only three ways to keepthem apart: order, modulation, and selection. That is, the speaker can vary theorder of the same group of elements; or the speaker can keep the same orderbut vary the choice of form at some point in the order. This is now what keeps(30a) and (30b) apart. Lastly, the speaker can use the same forms in the sameorder but employ contrasting intonations.25 Not all languages systematicallyexploit this last device, but all will depend heavily upon the first two.

This is all there is to syntax. And for that reason, syntax, or grammaticalform, considered separately from meaning is not very interesting. Any

utterances.

Although presented as a grammatical description, this is in effect a covert semantic analysisof sememes (not morphemes) into ‘components’. And again semantics is conflated intogrammar. The recognition of semantic complexity of morphemes is not really recognizeduntil the 1960’s, and then not by linguists, but by anthropologists. Cf. Tyler (1969).

24 Bloomfield’s ‘construction’ is not the same as Hockett’s, yet they share the property thatneither is an elemental component of grammar. As noted above, Bloomfield’s ‘construction’is built from his taxemes. Hockett’s ‘construction’ is complex as is Bloomfield’s in that bothrequire components. Hockett’s ‘relation’ component in ‘construction’ is actually theinvocation of a meaningful contrast cast in formal terms, e.g. ‘verb-subject’ vs. ‘attributehead’.

25 I include here, as well, pauses and all phenomena outside the ‘segmental’ phonology.

Page 22: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

22 SYNTAX & SEMANTICS

apparent complexity of grammatical forms is but a reflection of how alanguage has configured its meanings. The validity of the beliefs outlinedabove will depend upon whether they produce a useful and coherentunderstanding of languages.

3. The Substance of SyntaxHaving described something of our assumptions concerning the form of

grammar, we turn now to anticipate briefly what kinds of semantics willemerge as we examine languages from this perspective. Our goal will be thediscovery of constant principles of semantic organization of the simplesentence. We will find that certain ‘tasks’ are accomplished in the semanticsof all languages, and these may be named as follows:

FOCUS

TOPIC

DETERMINACY, i.e., the relation of PARTICIPANTS to our experienceindependently of the first six above.

EVENT and PARTICIPANTS

VOICE and ROLE

The grammatical expression of these contents provides the characteristiccontour to the simple sentence in all languages. The above seven meaningswill be discovered in every language. There are complex interactions amongthem, and none exists independently of the others. Some are so closelyrelated, that they exist almost as opposed pairs. This is especially true ofEVENT and PARTICIPANTS and of VOICE and ROLE. Part of our task will thenbe to discover what those relations are, as well as how to understand themeanings. And in doing that, we will find that there is at least one additionalsemantic pair:

NUCLEUS and PERIPHERY

In place of beginning with a definition associated with the terms, we shallbegin by looking at a number of languages. The important part of the activitywill be our ability to see that what is happening in one language is somehowsemantically connected to something in another. The plausibility of theseidentities and the order they bring to our understanding of language are theonly arguments for the correctness of what we have done. We shall work fromlanguages outward towards an understanding of them.

Page 23: SYNTAX & SEMANTICSpwd/sands01.pdf · presentation of syntax and semantics will depend ultimately on the quality of the understanding they yield. The first assumption that guides the

INTRODUCTION 23

Nature knows no abstractions, and there is nothing abstract in language.Languages do not have definitions. This implies that I will never try to defineFOCUS nor any of the other semantics. I cannot do it, nor do I want to.Languages are natural objects, and because they are, they are not abstract.They do not have definitions, categories, etc., and the imposition of such uponlanguages will distort them and almost certainly guarantee that wemisunderstand them. Our strategy will be to explore the grammar of as manylanguages as we can, guided by the search for the semantic contrasts. Ourunderstanding will not come from a definition of the semantics, but from howwe each make sense of our encounter with their presence in the languageswhich we examine. Our grasp of FOCUS, for example, will derive from thelanguages in which we study it; and the next language we study maydramatically alter our conception of what FOCUS can be, how it can interactwith other semantics, and finally what is a possible human language. Theterms themselves are not interesting. Having examined some language, andhaving discovered some pattern, we will have the need to label it for easyreferral. That is the only motivation for these terms, which have no statusprior to our having discovered the patterns which they name.

The ultimate goal will be to integrate those seven so that each is not justan item on a list; each will bear some sensible and organic relation to theothers. It is the interrelatedness of the semantics that helps shape thePROPOSITION, which is the content communicated by them. In this book, I willfocus on ‘simple’ PROPOSITIONS. The semantics of ‘complex’ expressions isdealt with elsewhere (Davis Ms.c.).

Because it is not possible to discuss the semantics of syntax in a linearfashion, completing the exposition of one topic before taking up the next, weshall have encountered several examples of, say, VOICE, before we finally turnour attention directly to that topic. In addition to the discussion of thesemantic landmarks, several languages will be discussed at greater length. Itmakes little difference where the investigation is begun. The general plan tobe followed here will be this. First, we shall look at a single language fromour announced perspective just to see all the terms at work and to havesomething concrete to refer back to. Then we shall turn our attention to eachkind of semantics in turn. I shall follow the order in which the semantics werelisted above. As we take up each new semantics, it will be related to what hasgone before, creating ultimately an image of the PROPOSITION.

[Version: May 11, 2005]