ii
A b stra ct
This dissertation com pares two Pacific Rim languages, the Asian language Ko rean and the native N orth American language Lushootseed Salish. The former is spoken by approxim ately 70 million people on the Korean peninsula and in the neighbouring P eople’s Republic of China. Lushootseed Salish is now moribund but h ad an estim ated speakership of about 11,000 at the tim e of first European contact.
T he theoretical framework used in this comparison is the Functional G ram m ar of Simon G. Dik. T he K orean d ata are from both native and non-native gram m arians as well as m y own judgem ents as a native speaker. The Lushoot seed d a ta are from the fieldnotes and two gram m ars of Professor Hess.
Following th e theoretical and gram m atical discussion, a traditional story from both languages is provided w ith detailed analysis.
iv
C o n ten ts
T itle p a g e i A b s tr a c t ii C o n te n ts iv A c k n o w le d g e m e n ts vii L ist o f A b b r e v ia tio n s ix D e d ic a tio n xP art I
B ackgrou n d
l
C h a p te r 1 I n tr o d u c tio n 2 C h a p te r 2 G ra m m a r 6 2.1 Functional G r a m m a r ... 6 2.1.1 Predicate-fram e and t e r m s ... 6 2.1.2 O n predication c o n s t r u c tio n ... 10 2.1.3 Assignment of syntactic f u n c ti o n s ... 142.1.4 Assignment of pragm atic f u n c t i o n s ... 20
Them e: ... 22 T o p i c : ... 24 F o c u s : ...i... 26 T a i l : ... 28
P art II
C a u sa tiv es
32Contents ._________________________________ X
C h a p ter 3
T h e ca u sa tiv e co n str u c tio n s 33
3.1 Definition of c a u s a tiv e s ... 33
3.1.1 Comrie: T he num ber of argum ents in c a u s a tiv e s ... 34
3.1.2 Givon: A gentivity to c a u s a t iv e s ... 36
3.1.3 Dik: Predicate f o r m a tio n ... 38
3.2 Syntactic causatives and lexical c a u s a t i v e s ... 40
3.2.1 Preamble ... 40
3.2.2 T he periphrastic and affixal causatives in K orean . . . . 40
3.2.3 T he morpheme -tx “ / -tu- in L u s h o o ts e e d ... 43
3.3 D isc u ssio n ... 44
3.3.1 Doubling of gram m atical r e la tio n s ... 44
3.3.2 Extended d e m o tio n ... 48
3.4 Remedies for the sh o rtc o m in g s... 51
3.4.1 Morphological approach . 51 3.4.2 Semantic properties of th e jif.ru v e r b ... 55
3.4.3 Functional approach to c a u s u c iv iz a tio n ... 57
3.4.4 Subj and Obj functions revisited ... 61
3.4.5 Expression rules from K orean c a u s a t i v e s ... 63
3.4.6 Expression rules in Lushootseed c a u s a tiv e s ... 68
3.5 C onclu sio n ... 73
P art III
T ex ts
75 C h a p ter 4 A K o rea n sto ry: T h e R ig h te o u s B r o th e r s 76 4.1 The righteous b r o t h e r s ... 764.2 A parsing schema for the righteous b r o t h e r s ... 78
4.3 The righteous b r o t h e r s ... 80
4.4 Glossary for Korean folk t a l e ... 82
C h a p te r 5 A L u sh o o tse e d story: T h e Y o u n g M in k an d H is L ittle Y o u n g er B r o th e r , T u ty eek a 89 5.1 The Young Mink and his little younger brother, Tutyeeka . . . . 89
5.2 T he Young Mink and his little younger brother, Tutyeeka . . . . 92
5.3 A parsing schema of the tex t a b o v e ... 94
Contents vi
A p p e n d ix A
v ii
A ck n ow led gem en ts
I wish to thank my supervisor T hom Hess who has showed me unfailing support and many pieces of advice throughout this work. My deepest g ra titu d e goes to him. Professor Geoff O ’G rady serves as a com m ittee m em ber even after his retirem ent. His encouragement was more th an once the source of much of my pleasure while writing this work. Professor B arbara Harris serves as a com m ittee mem ber. Professor Gordon Fulton from English serves as an outside m em ber. Professor Jan van Eijk serves as an external exam iner from the Indian Federated College of Saskatchewan, Univeristy of Regina, Saskatchewan. I th an k th em all.
U ntil this point in my program , m any professors and colleagues have showed encouragements to me. To nam e a few, I would like to m ention Dr. B arry Garlson who gave me directions and suggestions during my coursework. Dr. Dawn Bates (ASU, Tem pe, AZ) and a Lushootseed elder, Vi H ilbert, shared with me their enthusiasm for the Lushootseed morphology. Dr. D onna G erdts (SFU, Burnaby, BC) introduced this p re tty campus to me while I was stru g gling w ith RG framework. Dr. Joseph Kess, Dr. Leslie Saxon and Dr. Ewa Gzaykowska-Higgins often encouraged .ne to keep on writing when I was at a loss w hat to do. I am thankful to them . Iiaura P roctor, Chris W hitney and Evan W illiam s who, saying th a t V ictoria is no longer quiet, left for M cBride, have been great to me during my acquainting w ith this town. It was my pleasure to m ake friends w ith Susan F itzgerald, P au l Hopkins, Erica Hofm ann, S andra K irkham , Levan Kverenckiladze, Tadao M iyam oto, Xue Ping, C hiharu Uda, Wei Yang, and Xia Zhang inside and outside the classroom. Susan has read my work more than once and given me suggestions for b e tte r stylistics. Kevin Cat- tell from C om puter Science helped me find a proper typesetting and solve m inor com puting problem s. Susan and K evin’s friendship is discovered in alm ost every page of this work.
W hile staying in Vancouver Island, the In stitu te 1991 was an excellent opportunity for studying w ith other young scholars and students all over the world. I would like to th an k the Linguistic Society of America which allowed me to be eligible for the in stitu te fellowship during th e session held at U G -Santa Cruz, CA. I still rem em ber the exciting debates at the Prosodic Morphology sem inar directed by Professors J. M cC arthy and A. Prince. I learned a lot from the m eeting. Furtherm ore, I benefited from the friendship shared by th e Nez Perce trib e members (Lapwai, ID; Aug. 1991) and the M akah elders (M akah,
Acknowledgements v iii
WA; Aug. 1993). I am grateful to them . My special thanks go to Larry Bird (Iliakcuk) from Dene nation who kept inform ing me of the Amerindian wisdom accum ulated from the time immemorial.
My international excursion for linguistics was supported by many friends and fam ily m em bers. I would like to thank Dr. Byung-Ho Chung (U rbana, IL), M r. Jinki Chung (U B C ), Mr. Hee M in Lee (Ja k arta), Mr. Sang Nak Lee, Dr. Chang Nyung Lim, Mr. Sang Ryeol Song, Professor Yubun Suzuki (Fukuoka), Mr. and Mrs. Yoon Chee, Dr. and Mrs. Jon ath an Kim, and Mr. and Mrs. Terrence H. Kwon. I also like to than k many friends who have gathered together at M argot’s cosy living room at Oliphant street. I am grateful to my grandparents and parents who gave me continuous support. I wish to thank aunt Sook He Kim and Hae-Joong who have encouraged me while they are staying in N orth A m erica as well as in Seoul. At the final stage of this work, I was lucky enough to get much support from Meekyung who had become my life-time p artner. I would like to take this chance to show my love for her.
List o f A b b rev ia tio n s
i 1st person Nom Nom inative
2 2nd person Nomn Nominalizer
3 3rd person Obj O bject
Acc Accusative Obi Oblique
Ag Agent OC Out-of-control
Aor Aorist Pass Passive
AO Agent oriented P at P atient
Art Article PI Plural
Asp Aspect PO P atient oriented
Ben Benefactive pref prefix
Caus Causative
Q
QuestionComp Complementizer Rec Recipient
Dat D ative S Sentence
Dec Declarative Sg Singular
Det Determ iner Subj Subject
Dem D em onstrative Suf Suffix
DO Direct object T /A T ense/A spect
EDO Em bedded direct object Temp Tem poral EIO Em bedded indirect object Top Topic
ES Em bedded subject Tr Transitive
fut future V Verb
Gen Genitive
Go Goal
Instr Instrum ental
10 Indirect object Irr Irregular Kr Korean Loc Locative Lu Lushootseed Mod M odal MS M atrix subject
X
D ed ica tio n
To the memory of my grandfather
Chi Soo Kim (1906 - 1991)
1
P a r t I
________
2
C h ap ter 1
In tro d u ctio n
This dissertation is a contrastive study of K orean and th e A m erindian language called Lushootseed using th e functional gram m ar m odel set forth by Simon G. Dik. It concentrates on causative constructions.
K orean is spoken by some seventy million people on the K orean peninsula and approxim ately two m illion elsewhere throughout th e world. Lushootseed is a m oribund Salish language form erly spoken around the shores of P uget Sound in W ashington S tate. Anthropological dem ographers estim ate th a t in precontact tim es it h ad about eleven thousand speakers, which m ade it one of the largest of th e tw enty-three known Salish languages.1
T h e d a ta of this study are based on Korean folk stories, m y own knowledge as a n ativ e speaker and th e writings of Korean gram m arians, both traditional and m odern. Sources for Lushootseed are th e extensive tex t collection of P ro fessor T hom Hess and the two pedagogical gram m ars com piled by him.
K orean and Lushootseed differ on most usual points of typological com par ison. K orean is verb final while Lushootseed is verb in itial w ith th e usual related phenom ena: K orean has postpositions, Lushootseed prepositions; and relative clauses precede th e head in K orean b u t typically follow th e head in Lushootseed. W hereas Lushootseed puts possessor after th e possessed item , K orean uses the reverse order (Greenberg, 1964). H ere follow constrasting exam ples of each of these four points:
1See Suttles, W . and B. Lane (1990) “Southern Coast Salish". In: Wayne Suttles (ed.) Handbook o f N orth A m erican In dians, vol. 7. Smithsonian Institute. Washington, D.C. 485-502.
Introduction
? ? ( 1) Kr: sonyon-i ka-ss-ta. Lu: ?u?ux“ ti cacas.
boy-Nom go-T /A -B ec Asp-go D et boy T he boy went.
■W
( 2) Kr: ki kaij-esa Lu: ?al to stutak' Det river-Postp(Loc) P rep Det river In th e river.
( 3) Kr: yano-ka koyagi-ka m ak-in kas-ita. salmon-Nom cat-N om eat-C om p thing-to be Lu: s?uladx“ ti?a? su?aiad ?a ti?a? pispis.
salm on D et N om -Asp-eat-Tr P rt D et cat T he salm on is w hat th e cat ate.
( 4) Kr: i salam-ii kas Lu: sq^abay? ?a ti?a? stubs.
this m an-Poss dog dog Poss Det m an
This m a n ’s dog, or dog of this m an
Lushootseed strives tow ard only one argum ent per clause (although oth ers can be brought in by m eans of oblique constructions) whereas Korean, like English, perm its fairly long sequences of argum ents w ith a p articular verb. How ever, b o th languages are argum ent dropping. Lushootseed com pensates for th e lack of argum ents by m eans of verbal suffixation showing voice and, in subor dinate clauses, person and nu m b er.2 W hereas Lushootseed perm its omission of all argum ents, K orean, on th e other hand, only allows th e omission of subjects and objects. Listeners determ ine these by context.
Morphologically, th e two languages are less disparate. N either language distinguishes a class of adjectives from verbs. B oth have a very large num ber of affixes, p articularly on th e verb, and the num ber of affixes per word can be quite high. On average th e index of synthesis in K orean3 is 1.99 an d 2.275 in
2In m atrix clauses listeners identify third person referents by context.
3A s for the index o f synthesis, I checked three sources in Korean. First, I counted the number o f morphemes in a one hundred word text from theoretical linguistics material. The index was 2.17. Secondly, I counted it from a novel. The index was 1.83. Finally, I counted it from a journalistic passage. The index was 1.98. On average, the index is 1.99.
Introduction____________________________________________________________ 4
Lushootseed. On th e other hand, Korean is m ore agglutinating th a n Lushoot seed. On average th e index of agglutination is 1.53 in Korean4 and 0.832 in Lushootseed.5
Typical of O riental languages, K orean has a large set of num eral classifiers. Superficially, Lushootseed has th e same kind of m orphem e class (except th a t the Lushootseed m orphem es entail wider gram m atical uses th an does the Korean class). On th e o ther hand, K orean compounds extensively while com pounding is rare in Lushootseed. In word building, Lushootseed tends to ‘n e st’ its roots w ithin m any layers of derivation and inflection as in this diagram:
E.g.
-ali
T h e Korean affixes, on th e other han d , ten d tow ard a m ore concatenated structure.
T h e dissertation is divided into th ree parts: Background, which includes this introductory chapter and an overview of Functional G ram m ar as described by Dik; a second p a rt devoted to a discussion of the causative; and a th ird p a rt com prising a te x t and glossary from b o th languages. Several appendices and th e bibliography conclude th e work.
I see two areas in which th is dissertation is a contribution to knowledge. It is th e first a tte m p t a t describing a Salish language using the F unctional G ram m ar m odel of Sim on C. Dik. Since this theory is relatively new, its descriptive
4As for the index o f agglutination, I checked the above one hundred word text, First, from the theoretical essay, the index o f agglutionation was 1.48. Secondly, from the novel, the index was 1.52. Finally, from the journalistic passage, the index was 1.60. On average, the index of agglutination is 1.53.
5The Lushootseed index of synthesis is provided by Professor Thom Hess, and the index of agglutination is calculated by Professor Hess and me from a story in the pedagogical grammar.
Gramm ar 5
power is tested here. Secondly, a contrastive gram m ar of two languages from different parts of the world th a t have never been so studied before enhances linguistic knowledge generally.
6
C h a p ter 2
G ram m ar
2.1 F u n c tio n a l G ram m ar
T he theoretical fram ework which I shall adopt as a tool for com paring some constructions in K orean and Lushootseed m akes use of a relatively sim ple form of functional gram m ar (Dik, 1980a, 1981). T h e theoretical ten et of this gram m ar is based on a functional view of n atu ra l language. According to Dik (1980a:l), a n a tu ra l language is first and foremost regarded as an instru m en t of social in teraction by m eans of which hum an beings com m unicate w ith one another and thus influence one another’s m ental and practical activities.1
2 .1 .1 P r e d ic a te -fr a m e a n d te r m s
T h e form ation of linguistic expressions in a typical functional gram m ar (henceforth FG ) consists of two steps. F irst, a predication is form ed out of the predicates and term s contained in the Fund. The Fund consists of a lexicon containing th e properties of lexical item s th a t m ust necessarily be learned and m em orized, and two rule com ponents generating forms derivable by synchroni- cally productive rules. They are predicate an d term form ation rules. Secondly, a predication is m apped onto its actu al form by m eans of expression rules. Inside th e m odule of expression rules, th e forms and the order of constituents are de
1In this framework the role o f communicative competence (cf. Hymes, 1972) and the actual im plementation o f this competence receives more attention than the argumentation of linguistic expressions via a series o f abstractions.
The origin of functional grammar can be traced back to the Prague school of linguistics and the pragmatic school o f thought in philosophy from Oxford (Austin 1962, Searle 1969).
Gramm ar
term ined. A form of FG is thus a system for constructing linguistic expressions out of th e predicates and term s contained in the Fund, via in term ediate levels of underlying predications according to D ik (ibid:5).
FG specifies functional relations at th ree different levels: sem antic, syn tactic and pragm atic. Inform ation relevant to the sem antic functions of the argum ents is specified in predicate-fram es where are built th e basic stru ctu re of th e nuclear predications, th e sub categorization of predicates, th e num ber of argum ents surfaced, etc. Sem antic functions specify th e roles which th e refer ents of th e term s involved play w ithin th e “state of affairs” designated by the predication in which they occur. Before reaching th e m apping procedure of expression rules, two levels of function assignm ent are necessary for m aking a predication fully specified. These are assignm ent of syntactic functions and the assignm ent of pragm atic functions.
T h e basic elem ent of a predication is a predicate. Some predicates are applied to an appropriate num ber of term s, resulting in th e construction of predications. T h a t is, we can construct predications by com bining predicates and term s. In FG predicates are stored in th e lexicon as th e generic forms of predicate-fram es.2 These predicate-fram es are known as th e basic building blocks out of which underlying predications are constructed. Inform ation re garding th e num ber of argum ents a predicate requires can be retrieved from the predicate-fram es. T h e basic predicate-fram es and th e basic term s comprise the lexicon.
Each predicate is trea ted as p a rt of a predicate-fram e, which defines its sem antic and syntactic properties. T he predicate-fram es them selves serve as the basic m aterials to construct predications. Each predicate-frarne specifies the
2These basic predicate-frames and terms are not formed according to synchronically pro ductive rules; hence, it is assumed that they are provided. Dik (1980a:5) suggests that they must be listed and cannot be generated by means o f rules, and that they m ust be known and cannot be creatively constructed by the language user when he or she needs them.
G ram m ar
detailed inform ation needed for a predication, such as th e form of the predicate, the syntactic category, th e num ber of argum ents which th e predicate takes, the sem antic functions of these argum ents, and th e selectional restrictions with regard to argum ents. If we use th e Lushootseed verbal predicate %/?a6yi ‘give’ in (1), th e following is th e predicate-fram e stored in the lexicon:
( 1) ?abyiy (Xx: hum an(X i)).4s (X2)Go (X3: anim ate(X 3))fiec
The predicate V f a b y i ‘give’ requires th ree argum ents which are an actor, a ‘ta r g et’ of th e act and a receiver of th a t action. However, the predicate fram e in (1) has one constraint, which prevents it from having three argum ents sim ultane ously, when all th ree argum ents are regular nom inals rath er th a n pronom inals. It is a language-specific constraint of Lushootseed in which th ree regular nominals are not usually allowed to surface w ithin a predication. T he predicate-fram e still states th a t ?abyi- is a verbal (V) w ith aq, a;2 and x3 plus their sem antic functions Agent (Ag), Goal (Go), and R ecipient (Rec), where the Agent m ust be hum an and th e Recipient m ust be anim ate.
According to Dik (1981:57), t e r m s are form ed by m eans of term form ation rules following th e general schem a shown in (2).
( 2) (wx;: </>i(x;): ... : <fn (x,-)) (n > 1)
where to indicates one or m ore term operators and each <f>j (x,) indicates an open predication having x,- as a free variable. Since the open predications can include term s of th e sam e stru ctu re as (2), th e procedures for term form ation are recursive. In F G it is presum ed th a t languages possess a productive system of p r e d ic a te f o r m a t i o n rules by m eans of which th e set of basic predicate-fram es can be extended in tan d em w ith a set of derived predicate-fram es. For instance, all derivational m orphology and com pounding, causativization, voice change, and superlative an d com parative form ation are dealt w ith inside the com ponent
Gramm ar
of a set of derived predicate-fram es.
Like predicate form ation, a term cf>j m ay serve as th e head or as a modifier in th e stru ctu re of the term . T h at is, th e te rm form ation rule operates like those in predication form ation. In the case of a modifier, th e open predication can be noun, adjective, participal, adpositional phrase, or a relative clause. Hence, term form ation produces complex term s such as complex noun phrases, endless strings of conjoined phrases, and so on. Of course, th e derived term s m ay feed the predicate-fram es via term insertion, w hich results in an infinite num ber of term s in a m odel world.
In addition to th e basic predicate-fram es and term s given in a n atu ra l language, predicate form ation rules and te rm form ation rules together provide a set of predicate-fram es and a set of term s, which can be com bined into nuclear predications. In order to expand th e basic predicate-fram e tow ard an extended predicate-fram e, for instance, we can add a locative phrase to (1) as follows:
( 3) ?u-?ab-yi-t-ab ?a ti luX ti sq^obay? ?al ti?o? ?al?al. Asp-give-yi Role3-Tr-PO P rt Det m a n P rt D et dog prep Det house ‘T he old m an gave [someone] a dog in th e house.’
In order to get to the predication of (3), we need th e basic predicate fram e in (1), extended w ith a location satellite position.4 Observe the following:
( 4) ?abyiy(xj: h u m an (x i))j4s ( x 2 ) g 0 ( x 3 : an im ate(x 3))Rec (x 4 ) l o c
Here we need four appropriate term s to fill th re e argum ent positions and a satel
3According to Hess (1993:49), an argument which is associated with the -yi- in the pred icate can convey benefactive, dative, indirect object, recipient or second object. Hence, the cover term for these semantic functions is simply the -yi- role. In Southern Lushootseed the equivalent term is based on the slightly different form -si-.
4 Augments are single words within the clause which express locative or temporal notions and are not part o f any other constituent in the clause. They are positionally free relative to the other constituents. However, here Pal UPaP PalPal appears as an adjunct to the predication rather than the augment, Readers are recommeded to refer to Hess (ibid:110)
G ram m ar____________ 10
lite position of th e extended predicate-fram e (4). A fter inserting the appropriate term s b o th in th e argum ent and satellite positions, the following is obtained:
( 5) a. (dlx,-: stubs,v(x,-): luX^x,-)) the old [man] b. (dlx,-: sq“ abay?iv(xj)) th e dog c. (elxfc: stubs^(xfc)) [someone] d. (dlx;: ?al?aljv(x;)) th e house
The te rm form ation rule is fully specified in (5), indicating th a t it has three definite and one existential nom inal categories of which th e argum ent c is not overtly shown after expression rules have applied. T he right column represents the English glosses for each term . T h e predicate-fram e w ith sem antic functions for (4) is illu strated as in (6):
( 6) givev a. (dlx,-: stubsjv(x,-): lu X ^ x ,-))^ b. (d lx ,: sqt"abay?jV(x j))Go c. (e lx fc: stu b sw(xfc))fiec d. (dlxj: ?al?aby(xi))z,oc
T he w in term form ation ru le (2) can be one or m ore operators in functional notation. (6c) has an E xistential Q uantifier as an operator, while in other argu m ents d stands for th e term operator ‘definite/ 1 - rther, it has gender distinction too. Therefore, th e om itted stubs denotes a m ale ra th e r th a n a female. ‘I1 repre sents th e term operator ‘singular.’ T h e colon indicates th a t the inform ation to th e right gives a specification of or a restriction on the possible values of x,-, Xj, Xfc, and x; to th e left.
2 .1 .2 O n p r e d ic a tio n c o n s tr u c tio n
T h e outp u t of term insertion is a predication. According to Dik (1981: 70ff), predications of this ty p e are hierarchically organized structures, b u t they do n o t have a linear order defined over th eir constituents. All perm utations of th e four term s in (6), for instance, contain th e same inform ation. In FG predications are linearized only at th e very end of the descriptive procedure.
Gramm ar_____________________________________________________________ 11
The stru ctu re of a predication is based on individual predicate-fram es which have th e same m eaning across languages. Consequently, a typological universal5
is obtained through term insertion rules at th e stage of predication construction, before linear order has been achieved by expression rules. T he actu al num ber of argum ents which surface is adjustable, since FG perm its a predicate-fram e to be only partially filled. T he predicate-fram e (4) in Lushootseed is only in p art filled by term s, resulting in an open predication w ith a free variable. In short, only one argum ent is sufficient for th e nuclear predication denoting ‘give’ in Lushootseed, because Lushootseed gram m ar allows only one direct com plem ent per nuclear predication.
T he general schema for predications induced from (4) can be form ulated as follows:
(7a) N uclear predication
[ ^ { s i X s Q - f o n ) . ] p red ic a te a r g u m e n ts V ---v— ;--- ' n u c le a r p r e d ic a tio n (7b) E xtended predication [[ [ ( s i X ^ - . f o n ) ] t( y i ) ( y a ) - ( i / n ) . ]] p re d ic a te a r g u m e n ts s a te llite s i — - — > te r m s -r- mmm ^ e x te n d e d p re d ic a tio n where 1. <j> is a predicate,
5According to Dik (1980a:13), this is one o f the features in FG which contributes to typo logical adequacy in that it does not force us to assume a certain fixed order among the terms in a predicate cross-linguistically, which never occurs in actual linguistic expressions.
G ramm ar_____________________________________________________________ 12
2. x i ... x n represent nuclear argum ents of th a t predicate, 3. yi ... yn represent satellites to th e nuclear predication, 4. each $,• and yi is m arked for some sem antic function, and
5. th e nuclear predication is m arked for some type of sta te of affairs, (e.g. some verbal notion).
Among th e infinite num ber of sem antic functions how does one describe the relations of one to another? According to F G , from th e unordered term s in the basic schem a (7a,b) we can ex tra ct a hierarchical notion rather th a n a linear one in th a t th e intrinsic relationship am ong sem antic functions characterizes the term s in a given predicate. It is also presum ed th a t some of these sem antic functions are m ore central to the predication than others. If we extend the notion ‘central’ to (7b), all th e sem antic functions in th e nuclear predication are obviously m ore central th a n those assigned in the satellite term s.
For instance, an agent is a m ore necessary constituent th an a goal which is an optional elem ent in a given action. T he agent m ust be more central to the predication th an th e goal is.6 Along these lines, there can be no recipient w ithout a goal, b u t th ere can be a goal w ithout a recipient. Therefore, it is reasonable to s ta te th a t th e goal is m ore central th an th e recipient in this respect, even though the goal in Lushootseed can be o m itted, as in (8).
/ 5 i
( 8) ?u-?ab-yi-t-ab ti luX ti cacas Asp-give-yi Role-Tr-PO P rep Det m an Det boy ‘T he old m an gave [something] to th e boy.’
A language-specific rule in Lushootseed suppresses constructions w ith three a r gum ents which surface after th e expression rules.
6Goal in FG corresponds to Patient in Bloomfield (1934) according to Dik (1989:104). Note that Patient is consistently used in Hess’s description of Lushootseed.
G ramm ar 13
A lthough we com pare only two sem antic functions at one tim e w ith respect to th e notion ‘central,’ it is clear th a t there is a partial ordering of th e term s of a predication. This p artial ordering is called th e sem antic function hierarchy (SFH).
( 9) Sem antic function hierarchy (SFH) Ag Go Rec Ben In str Loc Temp
Ag stands for Agent; Go stands for Goal; Rec stands for Recipient; B en stands for Beneficiary; In str stands for Instrum ent; Loc stands for Location; and Temp stands for Tem poral. O ther sem antic functions such as Experiencer, M anner, Source, etc., are not included here.
A part from the intrinsic ordering im plied in the SFH, it is also presum ed th a t predications do no t define any further orientation or perspective on the sta te of affairs which they designate. In predications a relationship between the num ber of entities w ith th eir own roles in th e sta te of affairs is encoded at an interm ediate stage before the full-fledged phrases. T h a t is, th e linear ordering am ong constituents in a sentence does not take p a rt in th e process of predica tion construction. A t this stage predications sim ply sta te certain relationships betw een a num ber of entities, once th e sem antic roles are established.
According to Dik (1981:71), in n atu ral discourse people are apparently m ore interested in th e entities which play a role in some sta te of affairs, th a n in the relations obtaining am ong them . If the entity, chosen as a point of depar tu re for describing some sta te of affairs, were always th e en tity w ith th e highest ranking sem antic function in th e SFH, this perspective could be tak en as fixed regardless of gram m atical relations. O n the contrary, it is noted th a t th e actual linguistic expression and th e predicative stru c tu re of a sentence do not often coincide. T here is no one-to-one correlation betw een sem antic functions and gram m atical relations among argum ents. G ram m atical relations m erely repre
G ram m ar 14
sent a certain salience in th e logico-semantic relations am ong th e constituents of the sentence.
P redications can be form ulated for linguistic expressions of quite different structures. These differences result from th e assignm ent of two further types of functions besides th e sem antic functions already given in the predication con struction. T he logico-sem antic relations can be m apped as a set of syntactically encoded sem antic functions am ong the sentential constituents. And additional inform ation can be added to th e actual linguistic expressions after the syntactic inform ation has been m apped. This additional inform ation includes situational and contextual inform ation which is realized by adding intonation, stress, or even utilizing dislocation of constituents in FG term s.
2 .1 .3 A s s ig n m e n t o f s y n ta c ti c f u n c tio n s
As th e point of d eparture for describing th e sta te of affairs, the syntactic function is assigned. In turns, su b j(ect)7 assignm ent determ ines th e perspective from w hich th e situation of th e predication is described.
In Korean, th e expressions for th e predicate cuta ‘give’ are as follows: (10a) noin-i sonyan-eke kae-lil cu-ass-ta.
old-m an (Ag) child (Rec) dog (Go) give-T/A -D ec ‘T he old m an gave a dog to th e boy.’
(10b) kae-ka noin-e iihas sonyan-eke cua-ci-ass-ta. dog (Go) old m an (Ag) child (Rec) gave-Pass-T/A -D ec ‘T he dog was given to th e boy by th e old m a n .’
(10c) sonyan-i noin-ekesa kae-lil pat-ass-ta. boy (Rec) old m an-from (So) dog (Go) receive-T/A -D ec ‘T he boy received th e dog from th e old m a n .’
7From the viewpoint of derived syntax, one argument (usually a noun phrase) from among the arguments of a predicate is singled out for preferential treatment — preferential in that, for instance, it alone undergoes a number of syntactic rules such as movement rules, agreement rules, etc. in generative terms. This argument is called the S U B JE C T of the sentence, according to Comrie (1981b:68).
G ram m ar 15
In Lushootseed, th e actual expressions describing states of affairs desig n ated by a predicate such as VTabyi ‘give’ in (4) take as their point of departure either [someone] (11a), [something] ( l i b ) or the dog (11c) (following Hess’s 1993 illustration):8
(11a) ?u-?ab-yi-tab ?a ti lul( ?o ti sqwabay?. T he old m an gave agent p atien t [someone] a dog. ( l i b ) ?u-?ab-yi-tab ?a ti luX ti cacas. T he old m an gave
agent dative [something] to th e boy. (11c) ?u-?ab-yi-tab ti cacas ?a ti sq^abay?. [Someone] gave th e dog
d ative p atien t to the boy.
The exam ples in each of (10) and (11) are synonymous in describing th e state of affairs, although each exam ple has its own perspective. In other words, in neither (10) nor (11) are examples a, b and c interchangeable w ith one another, although the truth-conditions of th e sentences in (10) and (11) are identical. This im plies th a t there is m ore to th e sem antics of n atu ra l language sentences than w hether th e question of th e truth-conditions bestow synonym ity on the sam e set of sentences.
In FG (Dik 1981:72ff) th e following are th e sim ilarities in handling of Subj assignment:
(12) a. th e types of constituents accessible to Subj assignment; b. the conditions under which such constituents are accessible
to Subj assignm ent;
c. th e form al consequences of Subj assignm ent for th e stru ctu re of th e linguistic expression; and
d. th e subsequent rules of gram m ar which require crucial reference to th e Subj function.
8Cf. Hess (1993:49fT). D ative in Hess (ibid) corresponds to Recipient in FG. It is also noted th at patient a la Bloomfield (1933) in Hess (ibid) corresponds to Goal in FG. Here we use patient in Hess’s description for Lushootseed, while goal is used in D ik’s description for patient.
G ram m ar 16
M ore often th an not, th e notion of Subj assignm ent prom pts a discussion betw een an active and passive construction; however, in FG its association with voice-change is sim ply abandoned. T h e active-passive relation is considered to be only a special case of a m uch m ore general process. In principle, the Subj function can b e assigned to any function in th e SFH (9). From a broad range of d a ta across languages, th e following show how the Subj is employed:
(13) a. Subj assignm ent is not intrinsically restricted to Ag and Go argum ents;
b. Subj assignm ent does not necessarily lead to any m ovem ent of argum ents.
T h e next syntactic assignm ent m ust be th e syntactic function Object. T he possibilities of O bj(ect) assignm ent are m ore restricted th an those of Subj as signm ent in (12). T he occurrence of Obj assignm ent is dependent upon the occurrence of Subj assignm ent. T h a t is, Obj assignm ent m ust follow Subj as signm ent. Obj assignm ent does not norm ally apply to Ag, since Ag argum ents are th e prim ary candidates for Subj assignm ent. In FG only two syntactic func tions - Subj and Obj - are required. Formally, Subj an d Obj assignm ent are expressed as th e sim ple addition of th e Subj and Obj function to argum ents with specified sem antic roles. Consider th e following predication to which syntactic functions have been applied.
(14) ?abyiv (dlx<: m a n ^ x ; ) : o l d ^ x , - ) ) ^ ^ | (d lx j : dogN (xj))Go 1 \ (d lxk : boyN (x k))Rec j 0b.
This predication, which is based on exam ple (11), shows th a t O bj can be as signed to either Go or Rec, e.g. (11a) ?u?abyiteb ?a ti luk ?s ti sqwabay?. ‘T he old m a n gave [someone] a dog,’ and ( l i b ) ?u?abyiteb ?a ti luk ti cacas. ‘T he old m a n gave [something] to th e boy.’ In o ther words, it is possible to specify a perspective for either th e goal or th e recipient via th e Obj function assign m ent. Since linear order is determ ined at the final stage of the F G ’s descriptive
G ram m ar ______ 17
procedure, the predication in (14) can be used for any language. T he predi cation stru ctu re in (14) is independent of th e constituent ordering p attern s of any one specific language. D ik’s FG obtains typological adequacy by sim ply not specifying Obj function to a p articular term a t this stage of the gram m ar.
If we apply predication (14) to Korean, we have th e constructions (15) and (16).
(15) noin-i sonyan-eke kae-lil cu-ass-ta.
old-m an (A g )-S u b j child (Rec) dog (G o )-O b j give-T/A -D ec ‘T he old m an gave a dog to th e boy.’
(16)9 noin-i sonyan-il kae-lil cu-ass-ta.
old-m an (A g )-S u b j child (R ec)-O b j dog (Go) give-T/A -D ec ‘T h e old m an gave th e boy a dog.’
In (15), O bj is added to Go, while in (16) it is added to Rec. T he assignm ent of Obj is selected by m eans of placing th e viewpoint on eith er Go or Rec in (14). E ither way, the stru ctu re of (14) rem ains intact.
T he full inventory of Subj and Obj assignm ent for predication (14) is sum m arized in tab le (17).
(17) Agent Goal Recipient
a. Subject O bject
b. Subject O bject
c. Subject
d. Subject
T he two assignm ents (17a) and (17b) cover exam ples (15) and (16) respectively. (17c) and (17d) are found in examples (18) and (19).
9Example (16) is the well-known double object (accusative) construction in Korean to be explored in chapter four.
G ram m ar 18
(18) kae-ka noin-e jihae sonyan-eke cua-ci-ass-ta. dog (Go) old m an (Ag) child (Rec) gave-Pass-T/A -D ec ‘T he dog was given to th e boy by the old m an .’
(19) sonyan-i noin-ekesa kae-lil pat-ass-ta. boy (Rec) old m an (So) dog (Go) receive-T/A -D ec ‘T he boy received th e dog from the old m a n .’
After th e syntactic functions are assigned to (18) and (19), Subj appears on either th e Go or Rec argum ent, giving (18’) and (19’):
(18’) kae-ka noin-e iihas sonyan-eke cua-ci-ass-ta. dog (G o )-S u b j old m an (Ag) child (Rec) give-Pass-T/A -D ec ‘T he dog was given to the boy by the old m an .’
(19’) sonyan-i noin-ekesa kae-lil pat-ass-ta. boy (R ec )-S u b j old m an (So) dog (Go) receive-T/A -D ec ‘T he boy received th e dog from th e old m a n .’
P u t simply, Subj assignm ent in FG accounts for th e changes in perspective w ith out recourse to an active-passive shift. In fact, (19) sim ply employs a different predicate, nam ely th e word for ‘receive.’
T hus, Subj and Obj assignm ents effect no change in th e input predication, except for those two syntactic functions to the relevant argum ents. T he following points are noteworthy.
(20)
a. T h e underlying sem antic function of the argum ent su b m itted to Subj or O bj assignm ent is retained. T his means th a t even in later stages of the description, Subj and Obj can be differentiated in term s of th eir underlying sem antic functions.
b . Any argum ent not provided w ith a Subj or O bj function sim ply rem ains as it is, characterized only by its sem antic function.
c. All of the form al effects of Subj an d Obj assignm ent are taken care of by la te r rules which are applied only after th e assignm ent of th e pragm atic functions Them e and Tail, and Topic and Focus.
Gramm ar 1 9
T he u ltim ate order of constituents is determ inded by th e interaction of th e syntactic and pragm atic functions. F irst, th e Subj or O bj function determ ines the positions th e constituent has. T hen pragm atic functions determ ine th e ul tim a te linguistic expression. T he Subj or O bj function is also responsible for the way the sentence constituents are m arked by case or adpositions. If th ere is agreem ent betw een th e verb and any other constituent in th e clause, th en th e Subj is the first candidate and the O bj is th e second candidate for fulfilling this agreem ent.
Languages differ w ith respect to which constituents can and which con stitu e n ts cannot be assigned Subj and Obj function. In FG th e variation across languages in th is respect is controlled according to th e SFH as illu strated in (9). T his hypothesis in FG can be presented as follows (Dik, 1981:74):
(21) A g > G o > R e c> B en > In str> L o c> T em p
Subj X X X X X X X
Obj X X X X X X
A pparently, Subj can be added to any sem antic function. Obj can be added to any sem antic function except Ag, when th e Ag is occupied by Subj. T he schem a in (21) shows th a t th e assignm ent of Subj and O bj functions becomes m ore “unusual” or “m arked” as we pass through th e SFH from left to right, both w ithin an d across languages.10
10According to Dik (ibid:78), languages differ in the ‘cut-off point’ for the syntactic function assignments. For example, English has its cut-off point for Obj assignment after Ben(efactive) as shown below:
John bought MaryflenO&j a drink.
As for Subj assignment, English has its cut-off point either before or after Ben. Compare the following with the above:
G ram m ar 20
2 .1 .4 A s s ig n m e n t o f p r a g m a tic fu n c tio n s
W ith in linguistic theory in general, th e principle th a t pragm atic functions of some sort m ust play a role a t some level in the description of n atu ral lan guages is recognized w ithout dispute. For instance, pragm atic functions have been organized into oppositions such as ‘to p ic’ versus ‘com m ent;’ ‘th em e’ versus ‘rhem e;’ ‘given’ versus ‘new ;’ and ‘focus’ versus ‘presupposition.’ These opposi tions are pragmatic functions insom uch as they can be predicated of constituents only w ith respect to som e wider, e x tra sentential setting in which they occur. It is also known th a t th ey are all crucially dependent on pragm atic inform ation in some way or another. In FG th ere are four pragm atic functions. They are them e, topic, focus and tail.
In FG , pragm atic functions, em ploying the so-called left-dislocated, and right-dislocated constituents, are n o t to be regarded as p art of th e predication proper, b u t as constituents preceding and following th e predication, and con nected w ith it by way of some p ragm atic characteristic. To express this char acteristic, Dik (1981) assigns th e pragm atic function Them e to left-dislocated constituents, and th e pragm atic function Tail to right-dislocated constituents. Up to th e point of th e assignm ent of syntactic functions, principles and schem ata have been developed inside th e predication. A gram m ar m ust also provide the m eans for introducing predications preceded by a T hem e and followed by a Tail (Dik, 1981:130ff).
(22) ( x , ) T h e m e , P redication, ( x j)Tau
shows th a t b o th th e T hem e and th e Tail are optional additions to a predication. T h a t is, T hem e and Tail functions are only loosely connected to a predication. According to Dik (ibid), th e prag m atic statu s of these two functions are in ter preted as in (23):
Gram m ar
(23) a. A constituent w ith T hem e function presents a dom ain or universe of discourse w ith respect to which it is relevant to pronounce th e predication th a t follows it.
b. A constituent w ith Tail function presents, as an ‘afte rth o u g h t’ to the predication, inform ation m eant to clarify or m odify (some constituent contained in) th e predication.
Besides them e and tail, Dik (ibid) assumes th a t th e pragm atic functions Topic and Focus are relevant to the predication proper.
(24) a. A constituent w ith Topic function presents th e entity ‘ab o u t’ which th e predication predicates som ething in the given setting. b. A constituent w ith Focus function presents th e relatively m ost
im portant or salient inform ation w ith respect to th e pragm atic inform ation of th e Speaker and th e Addressee.
Dik (1981) em phasizes th a t FG does not adopt th e view th a t each prag m atic function should effect a binary division of the linguistic expression such th a t, if we assign some pragm atic function P; to some constituent of a linguistic expression, the rest of th e linguistic expression should necessarily have some pragm atic function P j. Interestingly, although th e notion of ‘presupposition’ in FG is regarded as a very im p o rtan t pragm atic notion, it is n o t considered to be a pragm atic function. Instead it is im plicitly associated w ith linguistic expressions as in th e following schema:
T he assum ption th a t p is tru e is a presupposition and L is som e linguistic ex pression, then
(25) a. W hen a Speaker assumes th a t p is true, he m ay express L. b. W hen a Speaker expresses L, he m ust assum e th a t p is true. c. W hen an Addressee hears a Speaker express L, he m ay infer
th a t th e Speaker assumes p to be true.
If (25) provides an adequate account for th e notion of presupposition, th e n th e presuppositions m ay not be p art of th e linguistic expressions in which th ey are located. T h e correlation betw een th e presupposition and th e nuclear predication
G ram m ar_____________________________________________________________ 22
is linked by m eans of discourse ra th e r th a n a rigid syntactic rule.
Let us exam ine several constructions in both Korean and Lushootseed employing pragm atic functions.
T h e m e :
B y definition, T hem e is a constituent which presents a dom ain or universe of discourse w ith respect to which it is relevant to pronounce a following pred ication. D espite its relevance to predications, th e T hem e is not to be regarded as a p a rt of th e predication. F irst, we shall review th e T hem e in Korean and Lushootseed, and exam ples in English in which th e T hem e is connected w ith the predication only via th e pragm atic relation of relevance as shown here:
(26) a. tsehapilo-m alham yon kikos-i k,lin cokae(i)ta. gaper-as for that-N om big clam -to be-Dec ‘As for th e gaper, th a t is a big clam .’
b. ca k^in cokas-ka tashap-ita. th a t big clam-Nom gaper-to be-Dec ‘T h a t big clam is a gaper.’
T he following are th e Lushootseed counterparts to (26):
(27)11 a. ha?ac g“ al la?b hik™ s?axwu?. horse clam and-C onn very big clam ‘As for th e horse clam , it is a big clam .’ b. la?b hik™ s?ax“ u? ti? ii ha?ac.
very big clam th a t horse clam
‘(A) really big clam (is) th a t horse clam .’
T he T h em e term s in (26) and (27) are directly related to th e predication in which the T h em e behaves as an elem ent of th e predication set. T he Them es above can also ap p e ar as p a rt of th e predication by repeating th e noun, nom inalization, pronom inalization, clitic, etc. T h e examples above are straightforw ard in th a t
G ram m ar 23
the sentence-initial Them e(s) should be filled by some term (s), which m ay force us to be confined to the ‘ex tra ctio n ’ account for the origin of the T hem e position.
Of course, there are certain constructions in which th e T hem e constituent does not have any n atural source w ithin th e predication th a t follows it. In other words, the T hem e constituent and th e predication th a t follows it do not have a direct relationship which can be interp reted as p art of th e predication. Observe the following (Dik, ibid: 133).12
(28) As for the students, adolescents alm ost never have any sense. (29) As for Paris, the Eiffel Tower is really spectacular.
(30) As for the Centre, well, th e less said about so-called ‘think tan k s,’ the b etter.
T he them e constituents the students, Paris, and the Centre do no t originate from th e predications. T h a t is, a kind of m acro-level universe in discourse is set up for the T hem e constituent and th e predication th a t follows it.
To cope w ith the im plicit13 and explicit relations between a T hem e con s titu e n t and th e predication following it, Dik suggests th a t th e T hem e con s titu e n t is an elem ent outside the dom ain of th e predication. Its location is to the left of the predication as represented in (31).
(31) ( ^ i ) r / i e m e j ( ••• ( x , ) . . . ) P r e d ic a tio n
In th is schema, th e optional statu s of the te rm (ay) inside th e predication allows the T hem e function to be incorporated into th e general schem a of (22) regard less of th e overt relation betw een th e T hem e constituent and th e variable (xy).
12Originally, (28) and (29) are from Ross (1970); and (30) is from Schwartz (1976). 13Quoting Grice f 1:>57), FG adapts the pragmatic relation o f relevance to relational com munication as rougidy shown below.
For any pair of Theme T and predication P to make sense, it must be relevant to pronounce P with respect to T.
Gram m ar_____________________________________________________________ 24
Once again, th e relationship betw een the T hem e and the predication can be characterized by articulating th a t, for a given T hem e x,-, th e predication may be an open predication in x,\ This implies th a t the T hem e constituent m ay be tre a te d as a m em ber outside th e perform ative m odality of th e subsequent predi cation. T h a t is, th e predication in schema (31) has th e full range of perfom ative m odalities such as declarative, imperative, and interrogative. F urther, examples from o th er languages also favor th e schem a (31) above. Observe the T hem e constituent w ith interrogative modality.
(32) M y hom etown? I haven’t been there for ages.
A ssum ing th a t it is correct to assign T hem e function to m y hometown in (32), one m ay have a question from where th e m odality w ithin m y hometown comes. If it is ex tra cted from th e predication, th ere should be a rule which m ay introduce the interrogative m ood to th e T hem e position. Otherw ise, the interrogative m odality m ust be introduced by some previous predication. However, if (31) applies to this situation, th en th ere is no conflict betw een the declarative and the interrogative m ood in (32).
T opic:
In FG T hem e is in principle presented as independent of th e predication, and th e predication is adjusted to th e T hem e rath er th a n th e other way around. However, Topic (Dik, ibid: 141) is considered to be p a rt of a predication. T he Topic function m arks th a t constituent of a predication ‘a b o u t’ which th e predica tion can be taken to predicate som ething. W hile th e T hem e constituent retains its independency of th e predication involved, the Topic constituent is necessary for com pletion of a predication. F urtherm ore, while T hem e is presented in ‘ab solute’ form w ithout any m arking of sem antic and syntactic functions, Topic
Gram m ar 25
often carries th e m arkings signalling its sem antic and syntactic functions w ithin th e predication. However, these m arkings m ay be partially or wholly n eu tral ized in languages which have a special Topic m arker such as K orean -(n)in14 and Lushootseed g^al15 constructions:
(33) cunc^i-nin ssakkoto cunc/lii-(i)ta. herring Top become stale-E m p herring to-be
‘As for herring, [it] is th e best even though it becomes sta le .’ ‘(As for fish,) th e herring is th e b est even though it becomes stale.’ (34) ti?a? s?ax:’"u? g“ ol Pu-dag^-abac-bi-d ?a ti?a? sali?.
D em b u tte r clam Top Asp-inside-LS-2Stem-Tr P rt Dem two ‘As for the b u tte r clam , it is betw een the [other] tw o.’
T h e above examples show th a t th e topic elem ent in the sentence can be replaced by th e Subj function; hence, th e question arises as to w hether or no t there is a distinctively assigned function for th e Topic constituent independent of th a t for th e Subj function. In F G , as expected, th e two functions m ay coincide, b u t th ere is no requirem ent linking th e two constituents w ith different syntactic and pragm atic functions in th e sentence. Therefore, th e general schem a for th e T hem e function in (31) can be extended s follows:
(35) (xi'jThemei (f> ... (Xj^Xopic
As m entioned above, th e value of i m ay be identical to th a t of j. Up to this point, th e reasoning behind th e distinction of T hem e and Topic in FG is said to be th a t th e T hem e is excluded from th e dom ain of predication th a t follows it, while th e pragm atic function Topic is included in th e dom ain of predication.
14Some readers may have more exposure to the Japanese topic marker -w a whose function in Japanese is very similar to -(n)in in Korean.
E.g. Sakana-wa tai-ga ii fish Top red-snapper Subj good ‘As for fish, red-snapper is the best.’
15According to Hess (1993:154), topicalization is achieved by placing gwsl after the topic, in affect treating the topic as an entire clause and using gwol to introduce the following clause. Often, the first constituent, a demonstrative, m ay be omitted.
G ra m m a r_____________________________________________________________26
Note th a t Topic is not specified in term s of linear order.
T h e other two pragm atic functions recognized in FG m ay be thought of as being sym m etrical to th em e and topic. Let us briefly consider the third pragm atic function, Focus.
Focus:
T h e m ost im p o rtan t or salient inform ation with respect to th e pragm atic inform ation th e Speaker has for the Addressee is Focus. C onstituents p e rta in ing to pragm atic inform ation shared by Speaker and Addressee are not Focus. R ather, th e Speaker assigns Focus to those constituents in a discourse situation which contain new inform ation in term s of pragm atic stru ctu re .16 Aside from realization by prosodic elem ents, th e pragm atic function Focus m ay be shown w ith w h-question words. T h e typical operation of Focus assignm ent employing w h-question words is perform ed in th e following order (Dik, ibid: 149).
Suppose th a t a t a given m om ent f,- in a com m unicative exchange betw een Speaker and Addressee th e following conditions hold:
(36)
1. For a given open predication in x;, </> ... (x*) ..., the speaker assumes th a t there is a term 8 such th a t if we su b stitu te 8 for x,- in the open predication, th e resulting predication <j) ... (6) ... will be true; 2. th e speaker does not know the identity of 8\
3. th e speaker assumes th a t th e Addressee knows th e identity of <5; 4. an d the speaker wants th e Addressee to tell him the identity of 8.
Under th e above conditions th e Speaker can form a predication of the form ft
... (Qx,-)i?oc ..., in which a question word (Q) is assigned th e Focus function. T he question o perator can be separated regardless of th e variable x. Hence, it
16According to Givon (1990:699), even contrastive focus overlaps witn other topics o f gram mar, since functional domains are not fully exclusive of each other. In addition, he quotes Bolinger (1961) in which the following is suggested concerning focus of assertion: every semantic peak is contrastive ..."
G ramm ar 27
is presum ed th a t all question term s receive Foc(us) function.
In detail, a wh-question thus counts as a request from the Speaker to the Addressee to give him a tru e value 8 for th e questioned term x,-. Now if we consider the m om ent t,q i, the Addressee will realize th e above pragm atic inform ation launched by th e Speaker. From th e perspective of th e Addressee, if th e following two item s hold a t tf+ i,
(37) 1. the Addresses does indeed know a tru e value 8 for th e questioned term x,-, and
2. the Addressee is prepared to tell th e speaker the id en tity of 8, th en th e Addressee can form a predication of th e form <j> ... {8)foc ••• as an answer to th e question. T he te rm 8 is also assigned Foe function, since th e term 8 is th e only inform ation which is projected as new from the Addressee. Con sequently, th e constituent which receives Focus function provides the requested identity from th e wh-question. Let us consider the following wh-questions in Lushootseed:17
(38) ?os-?axid k“ i dax^a-s-tag^-ox*" ?a ti?a? qawqs Asp-why A rt pref-N om n-hungry P rt this Raven ‘W hy is Raven hungry?’
(39) ?as-calaxtu k“ i iu-s-huy-s.
Asp-how A rt M od(Irr)-Nom n-m ake-3SgAg ‘How will he m anage?’
T he corresponding constructions in K orean are given in (40) and (41); however, notice th a t th e w h-question words in Korean are only optionally fronted after th e expression rules have been applied.
(40) kalkamakwi-ka wae psekop^-ayo?18 raven-Nom why hungry-Q ‘W hy is Raven hungry?’
17From Hess (1993:136).
G ram m ar_____________________________________________________________ $8
(41) ki-ka ottohke haenaka-l-kkayo. he-Nom how m anage-Q ‘How will he m anage?’
T he w h-question words natu rally receive th e pragm atic function Foe, since they are new inform ation in a given discourse between th e Speaker and th e Addressee. We shall re tu rn to a discussion of Focus in appendix B. R ath er, v/a here discuss briefly the fourth pragm atic function in FG , viz. Tail function.
Tail:
By definition, Tail, as an ‘afterthought’ to th e predication, characterizes constituents which present inform ation m eant to clarify or m odify the predica tion. T h e Tail function can be com pared to the T hem e function in term s of the application dom ain of predication. In FG th e Tail function is assigned to a con stitu e n t outside th e open predication, applied to th e explication of the Theme. Tail comes after th e predication in question, as opposed to th e initial position in th e case of Them e.
Let us investigate th e Tail function and see w hether it m atches well with th e T hem e discussed earlier.
As m entioned in th e section on Them e, the left- and right-dislocands are illu strated in th e following passages (from Siewierska 1991:150).
(42) So what I confess to is a certain ... feeling of dissatisfaction ... in some of the habits ’n ’ expressions of students that come to me particularly English expression, spelling, punctuation. Once again, very basic skills, I ’m often very dissatisfied with them.
(43) Both of my brothers came back, uh but one went up to New Guinea and was uh shot through the throat ’n’ the bullet came out in his out his back ... Yes it was there a long time, the bullet, in fact and he had to have it removed, the bullet.
G ram m ar 29
In (42), the initial constituent, very basic skills is resum ed in th e m ain predica tio n by a co-denotational item , a pronoun here. In (43), th e final constituent, the bullet is co-denotational w ith an. entity in the predication, a pronoun again. T h e Tail is typically located after th e predication. Like th e Them e, the Tail will also be set off from th e preceding predication by m eans of an intonational contour. Observe the following w ith italicized Tail functions (Dik, ibid.'153). (44) H e’s a nice chap, your brother.
(45) I didn’t like it very m uch, that book o f yours. (46) I like John very m uch, your brother I mean. (47) John gave th a t book to a girl, in the library.
As w ith every logical possibility in th e distribution of T hem e, th ere is a ques tion of the extent to which the Tail constituents should be regarded as being ‘e x tra c te d ’ from the predication. In m any cases, it is presum ed th a t some kind of extraction m ust be involved. In (44) th e Speaker produces the predication ‘h e’s a nice chap’ w ithout Addressee knowing whom he is talking abo u t, and then only later realizes th a t th e referent of he should be specified as ‘your b ro th er.’ In term s of th e Speaker’s behaviour we would ra th e r assum e som ething like the following to be the case here; in detail, the Speaker has th e full predication cor responding to “your b rother is a nice chap” a t his disposal, b u t a t first believes th a t it is sufficient for th e Speaker to use he for referring to the argum ent; only later does he add th e fuller specification as a Tail to th e predication.
In (46) it is not necessary to assum e th a t at the m om ent th e Speaker pro duces th e word John, the underlying term m ust be m ore fully specified as your brother John. T he Tail in (46) m ay add inform ation which was no t contained in th e original predication. In (47) th e satellite of th e predication, denoting th e Spatial dim ension, is another exam ple of Tail in which only th e second thought is
G ram m ar ____ 30
carried out to th e preceding predication. If this pragm atic inform ation is added, the predication will be schem atically illustrated as follows:
(48) P redication, (x j)Tau
where th e general condition on possible Tails is th a t they should b e interpretable as a fu rth e r specification or m odification of th e predication. N ote th a t the T hem e can show th e m odal difference from th a t in the original predication, e.g. interrogatives, and so on. T he Tail, as th e extra-clausal pragm atic function, has undergone a sim ilar m odification of the predication, b u t not so radical as the changes in m ood as in th e case of Them e. Observe th e following (from Dik, ibid:154).
(49) Jo h n won’t be invited, eh... Bill I m ean.
Tire above shows th a t the Tail need not be a further ‘spelling o u t’ of inform ation already contained in th e predication. T h a t is, th e Tail constituent m ust in some way or o ther specify or m odify th e preceding predication. T h a t partial independence of th e Tail, however, does not com plete predication as is th e case w ith th e Them e. However, it is a sufficisnt condition for spelling ou t of pragm atic inform ation in a given discourse. Let us consider th e following examples in French (modified and adopted from Dik, ibid: 155).
(50) a. Nous lui avons donne le livre hier, a ta soeur. b. *Nous lui avons donne le livre hier, ta soeur. In Korean, th e sam e construction is as follows:
(51) a. uli-ka ace ki-eke ki c,lask-il cu-ass-ta, ki yaca-eke. we yesterday (s)he-to D et book gave Det wom an-Dat b. *uli-ka ace ki-eke ki chaek-il cu-ass-ta, k i y a c a .
we yesterday (s)he-to D et book gave D et woman ‘We gave th e book to her yesterday, to /* 0 th a t w om an’
G ram m ar 31
These examples show th a t in French and K orean Tail m ust have a m arking corresponding to the constituent of th e predication in which fu rth er specification of inform ation is contained. This need im plies th a t th e Tail carries not only sem antic b u t also syntactic function.
