Reported speed and thought in Kavalan

Haowen Jiang
Rice University
*
Abstract
One of the amazing daracteristics of human language is selfreference, that is, referring to
itself by means of itself. Tis languagewithlanguage phenomenon is most evident in reported
discourse, where speed, thought, and perception tend to be interconnected. Hence, this paper
investigates reported speed (RS) and reported thought (RT) in Kavalan, an endangered Aus
tronesian language, by focusing on the quotative marker zin, whid frames either a speed event
or a mental event. Based on our present corpus, it is found that in narratives reported speed
is nearly twice as frequent as reported thought, with instances of reported speed almost ex
clusively framed by third person pronouns. ln conversations, however, the situation is reversed,
namely, reported thought is twice more frequent than reported speed, with instances of reported
thought almost exclusively framed by rst person pronouns. Our study shows that selfreport of
thought is the norm in conversations while otherreport of speed is predominant in narratives.
Keywords reported discourse: quotative maker: Kavalan: Austronesian languages
1 Introduction
ln daracterizing the nature of general semantics, Korzybski (1))) points out that a map is self
reexive, employing a mapterritory metaphor. To put it another way, a map, or any form
of representation, is capable of selfreference, that is, referring to itself by means of itself. Tis
recursive nature of representation, or specically of language, is thought to play a crucial role in
human adaptation and cognition. Te reexive nature of language is especially evident in reported
discourse (used here as a broad term), whid bears duplex structures and is described as language
within language (Janssen and van der Wur 1))o). Te investigation of reported discourse may thus
reveal us a window on how cognition processes two (or more) levels of representation.
Broadly speaking, reported discourse includes reported speed, thought, and perception since
they are onen intertwined together within or/and across languages. lor instance, Shona, a Bantu
language, has a quotative maker -ti, whidmay denote a mental activity (e.g. to think) or a verbal one
(e.g. to say) (lortune 1)!o). A more impressive example comes from Kambera, an Austronesian
language, where the generic verb w can report mental events (e.g. thought and intention), audible
events (e.g. sound and speed), and even visible events (e.g. motion and sight) (Klamer zccz).
*
Tis paper was rst intended as a term paper for Syntax ll at National Taiwan University, taught by Prof. Shuangfan
Huang, to whom l am deeply indebted. Also, l would like to express my heartfelt gratitude to our Kavalan informants,
including abas, ngengi, Raciang, and buya. Without their assistance and patience, this paper could not appear. All errors
of course remain my own.
1z
like in Shona and Kambera, speedand thought are reported by a common morpheme in Kavalan,
an endangered Austronesian language in Taiwan. To illustrate, the quotative marker zin reports a
speed event in (1) whereas it reports a mental event in (z).'
(1) Reported speed
naRin
xic
sazay
sing
zin-ku
ouo1-1sccix
ci
ix
buya
ix
pa-ipil
c~ulisten
timaikuan
1scioc
l told Buya not to sing to me. (lit. Dont sing and make me listen, l said to Buya.)
(z) Reported thought
mautu=ti
comeiiv
sayza
think
zin-ku
ouo1-1sccix
ci
ix
utay
ix
l think Utay has come.
ln this paper, we shall investigate the reported speed and thought in Kavalan, specically by
looking into the instantiations of zin in both spontaneous narratives and conversations. ln addition
to linguistic context, the syntactic structure of reporting and reported clauses is also expected to
provide language users with enough clues for them to distinguish reported speed from reported
thought.
Aside from the introduction in Section 1 and conclusion in Section , the organization of this
paper is as follows Section z reviews some important issues and ndings in reported discourse:
Section ! explains the methodology and data: Section presents the results and discussion.
z Iiterature reyiew
Discussions of reported discourse in the literature can be roughly classied into two groups (a)
the syntactic relationship between reporting and reported clauses: (b) the grammatical distinction
between direct and indirect report.
z.1 Te syntactic relationship between reporting and reported clauses
Te reported uuerance is traditionally assumed to be an object complement of the verb of speed
(lyons (1)ocz!), Rosenbaum (1)o), Givon (1)cc), Noonan (1)c), among others). Te analysis,
however, is not unproblematic since the speed verb exhibits a more intact relationship with its
subject than with the reported uuerance that is claimed to be the object (see McGregor (1))o)).
Moreover, longacre (1)c) notices the singularity of the syntactic structure in reported speed, sud
as discontinuity and exceptional word order of direct report in lnglish, and thus suggests that the
syntactic relation might be something like sentence, rather than clausal syntagms.
'Kavalan examples in this paper come from two sources. One is from my eldwork notes for the purpose of this present
paper, and the other from National Taiwan University Corpus of lormosan languages. Te orthographic conventions
for Kavalan are identical to lPA except e for sdwa, b for voiced bilabial fricative, d for voiceless alveolar fricative, l for
alveolar ap, y for palatal glide, ng for velar nasal, R for uvular fricative, and for gloual stop. linally, abbreviations
of the gloss are as follows 1 rst person: z second person: 3 third person: sc singular: vi plural: ivi inclusive
plural: rvi exclusive plural: rxz nominalizer: vr proper noun: vrx personal name marker: cir.ruux nonhuman
classier: Ar Agent locus: vr Patient locus: rAr: NonAgent locus: rox nominative case: crr genitive case: oni
oblique case: ioc locative case: rrc negation: vrv perfective: irr irrealis: cAu causative: ixv imperative: ouo1
quotative: rx1 existential: ir1 interjection: bx discourse marker.
Rice Working Papers in Linguistics 1! vol. 1, February 2009
A more extensive account comes from Halliday (1)c), who proposes that the relationship be
tween reporting and reported clauses is paratactic in direct quotes and hypotactic in indirect
quotes, with both being tactic relationships. He also suggests both types of quotation are involved
with projection, one of the two logicosemantic relationships singled out by him (with expansion
being the other). Projection refers to the phenomenon whereby some clause is projected as a piece
of wording or meaning, either spoken or thought. (McGregor 1))oo) lor illustration, the paradigm
for projecting speed and thought in both direct and indirect quote is given in Table 1 below.
paratactic hypotactic
(direct quote) (indirect quote)
reported Percy said lm running away. Percy said he was running away.
speed (RS)
reported Percy thought to himself lll run away. Percy thought he would run away.
thought (RT)
Table 1 Hallidays (1)c1)) paradigm for projecting clause complexes (with examples from Mc
Gregor (1))))
ln response to Hallidays analysis, McGregor (1))) puts forward eight points to argue against
the dependent relationship between reporting and reported clauses. A critical argument is that it is
the reported clause, rather than the reporting clause, that should be viewed as the main clause, con
sidering its freedom of occurrence. Also, the reporting clause is dependent, rather than independent,
in the sense that it modies the reported clause by specifying the source and the type of speed act
of the uuerance (e.g. a question, demand, or answer, etc.). As an alternative, McGregor analyzes the
relationship between reporting and reported clauses as framing, taking into consideration the fact
that they are capable of independent occurrence and at the same time structurally related with ead
other.
Te greatest dierence between dependence and framing relationship is that the former is a part
topart relation while the lauer a wholetowhole one, that is, syntagmatic relationships involving
grammatical items whid are themselves whole units (McGregor 1))o). lraming is in fact a
brilliant metaphor that compares the relationship between a reported and reporting clause to that
between a picture and its frame. Te picture (the framed clause) is the demonstration of a referent
world while its frame (the framing clause) serves as a description that demarcates the picture from
the badground seuing (see Clark and Gerrig (1))c)).
z.z Te grammatical distinction between direct and indirect report
Reported discourse is traditionally classied into two types of modes direct and indirect report.` Te
decisive factor that distinguishes one mode from the other has always been coreferential nominals.
Pike and lowe (1)o)) establish a set of explicit rules to deal with this issue, whid are shown in (!)
and () (taken from lowe and Hurlimann (zccz1z)), where OC refers to the quoted clause and
X to a third party who may or may not participate in the conversation between a speaker (S) and a
hearer (O).
ln expansion, one clause expands on the meaning of another by elaborating it (restating it), extending on it (adding to
it) or enhancing it (providing circumstantial modication) (McGregor 1))oo).
`Other types of modes are also identied, sud as lree Direct Speed, lree lndirect Speed, Narrators Report of Voice,
and Narrators Representation of Speed Act (see leed and Short (1)c1)).
Rice Working Papers in Linguistics 1 vol. 1, February 2009
(!) direct speec
a. Subject (S) of say verb (speaker) is coreferential with rst person pronoun in OC
b. Object (O) of say verb (hearer) is coreferential with second person pronoun in OC
c. Tird role (X) of say verb (bystander) is coreferential with third person pronoun in OC
() indirect speec
a. lirst person of say verb is coreferential with rst person pronoun in OC
b. Second person of say verb is coreferential with second person pronoun in OC
c. Tird person of say verb may be coreferential with third person pronoun in OC
lor example, John (S) is coreferential with the rst person pronoun l in (a) below, but with
the third person pronoun he in (b). ln addition, the rst person me (O) is coreferential with the
second person pronoun you in (a) but with the rst person pronoun me in (b).
() a. John said to me, lll call you tonight, when everything is ready.
b. John said to me that he would call me tonight when everything was ready.
Hence, (a) is a direct quotation while (b) is an indirect one. Tese rules may seem superuous
as well as complicated at rst sight, but it is the orthographical convention in lnglish that makes
the distinction in () straightforward. lor the study of languages like Kavalan, where a standardized
writing system is absent, these rules are in fact quite useful.
However, it is not infrequent that coreferentiality is dicult to identify, especially when only
third person nominals are referred to. ln this case, other symptomatic features have to be looked for.
lirst, tense, aspect, or temporal adverbs are helpful in deciding whether the time in question refers
to the speed situation (in the reporting clause) or to the referent speed situation (in the reported
clause). Second, direct speed may be discontinuous while indirect speed may not be (McGregor
1))). lor instance, the direct reported clause in (a) can be rendered as (oa), where the reporting
clause intervenes between two parts of the reported speed, but this is not possible for the indirect
reported clause in (b). Tird, the quote in direct speed may precede or follow the framing clause,
as in (ob) and (a), whereas the quote in indirect speed almost always follows the framing clause,
as in (b): in cases where it does not, it shows up like an anerthought (McGregor 1))).
(o) a. lll call you tonight, John said to me, when everything is ready.
b. lll call you tonight, when everything is ready, John said to me.
A question that follows from the distinction between direct and indirect quote is whether there
is a preference of one over the other. Generally speaking, unlike reported speed, reported thought
tends to be expressed indirectly (McGregor 1))c). On the other hand, in studying the reported
discourse in Mandarin Chinese, lin (1)))) concludes that direct thought, rather than indirect thought,
is favored when the rst person pronoun refers to the thinker. ln the present study, we also aim to
discover whether there are any correlations among the person of the subject of the framing clause,
the mode dosen in the reported discourse, and the alternation between reported speed and thought.
3 Methodology and data
Te Kavalan data employed here come from a database collected in zcc! and owned by Graduate
lnstitute of linguistics & Multimedia Center at National Taiwan University. Tey consist of two
conversations between a male and a female speaker as well as six narratives. Te conversations
Rice Working Papers in Linguistics 1 vol. 1, February 2009
are about going bad to ones hometown and about a Kavalan speed contest, while four of the
narratives are lrog stories (Mayer 1)o)) and the other two are Pear stories (Chafe 1)cc). Te number
of lntonation Units (lUs), the number of tokens of the quotative marker zin, and the total duration
of uuerances are listed in Table z.
Number of lUs Tokens of zin length (min.)
Conversations z! 1 11cc
Narratives !c z) z1
Total o! cc !oz!
Table z Basic information for Kavalan data
As seen in (1) and (z), the speaker or thinker of the quotative marker zin is marked as genitive in
Kavalan, so for the convenience of the readers a complete paradigm of genitive personal pronouns
is given in Table ! below.
lirst Person Second Person Tird Person
Singular -ku -su
-na
Plural -ta (incl.): -niq (excl.) -numi
Table ! A paradigm of genitive personal pronouns in Kavalan
Although eighty tokens of zin exist in the data, only seventyfour are used. Six tokens are ex
cluded either because of the ambiguity between reported speed and thought, or because of their
mysterious functions in the discourse. lor example, the zin in () may well mean to think or to
say, either way being equally plausible since no symptomatic features (see Section z.z) are available
() tita-an-na
seeii-!sccix
muaza
many
sinsuli-na
plum!sccix
nengi=ti
goodiiv
ala-an
takeii
zin-na
ouo1-!sccix
nani
i
He saw many plums (on the tree), and thought/said its time to plud them o. (pearbuya,
lU !, zcc!)
Another example that has been len out is (c) below. Since it appears in a narrative throughout
whid only the narrator is speaking, it is unknown to whom the -su you is referring. lven though
it refers to anyone who is listening to the story, either you think or you say would be appropriate
in this context. Te zin-su here may have other discourse functions, and to clarify this issue more
data are required.
(c) pasi
likely
yau
ix1
tayan
there
biat
frog
nay
that
zin-su
ouo1-!sccix
nani
i
Te frog might be there. (frogbuya, lU !!, zcc!)
Table ! only shows personal pronouns relevant in this study. lor a complete paradigm of the pronominal system in
Kavalan, please refer to Jiang (zcco1) and the references therein.
ln the examples, pearbuya refers to the Pear story told by the informant buya. Similarly, frogbuya means the
lrog story told by the informant buya. Since there are two types of data used in this study, the sources from whid a
particular example is drawn are indicated by the format XY, where X refers to the genre of the data (either Pear/lrog
stories or conversations) and Y to the name of the speaker.
Rice Working Papers in Linguistics 1o vol. 1, February 2009
Results and discussion
ln the seventyfour tokens investigated, the subject of zin is almost exclusively restricted to rst
person singular (zin-ku, o) and third person singular/plural (zin-na, c), with only three tokens
coupled with rst person inclusive plural (zin-ta, ). Te distribution of reported speed (RS) and
reported thought (RT) across dierent subjects is shown in Table , where only conversations are
counted, in Table , where only narratives are counted, and in Table o, where both types of data are
counted.
zin-ku (1sc) zin-na (!sc/ii) zin-ta (1iii) Total
RS ! 1 c 1 (!!)
RT zv 1 1 !1 (o)
Total !z (c) 1! (zc) 1 (z) o (1cc)
Table Distribution of RS and RT across subjects in conversations
zin-ku (1sc) zin-na (!sc/ii) zin-ta (1iii) Total
RS c 18 1 1) (oc)
RT z o 1 ) (!z)
Total z () z (co) z () zc (1cc)
Table Distribution of RS and RT across subjects in narratives
zin-ku (1sc) zin-na (!sc/ii) zin-ta (1iii) Total
RS ! !c 1 ! (o)
RT !1 z c ()
Total ! (o) ! (c) ! () (1cc)
Table o Distribution of RS and RT across subjects in conversations and narratives
According to Table , in conversations c of the tokens of zin are coupled with a rst person
singular subject and predominantly involved in reported thought. Moreover, the total instances
of reported thought are around twice as many as those of reported speed (cf. !1 and 1 tokens).
lnterestingly, the result in narratives is rather dierent, but the skewness in distribution is equally
obvious. According to Table , coof the tokens of zin are coupled with a third person singular/plural
subject and mostly involved in reported speed. ln addition, the total instances of reported speed
are roughly twice the instances of reported thought (cf. 1) and ) tokens). When conversations and
narratives are examined collectively, instances of a rst person thinker nearly equal those of a third
person speaker (cf. !1 and !c tokens), as shown in Table o.
As for the distinction between direct and indirect report, all instances of reported speed appear
in the form of direct report. lor instance, in ()) below the subject -na they in the framing clause is
Since the speaker/thinker coupled with zin is marked as genitive, the term subject is problematic. lor lad of a beuer
doice, here it refers to the speaker/thinker specied by the genitive pronoun.
Rice Working Papers in Linguistics 1 vol. 1, February 2009
coreferential with the rst person plural pronoun -imi we (exclusive) in the framed clause (under
lined). Hence it conforms to the one of the rules set up by Pike and lowe (1)o)), and can be counted
as direct quote.
()) qa-wiya=ti-imi
iiileaveiiv-1iiixo
zin-na
ouo1-!iicix
Tey say, We are leaving! (frogbuya, lU )c, zcc!)
ln spite of the claim just made, there seems to be one possible counterexample. Te tangi today
in (1c) below, according to the situational context, refers to the speed situation time, rather than the
referent speed situation time, whid is expected of direct report
(1c) [qautu]
come
nay=
that
sudal
paper
nay
that
tangi
today
zin-na
ouo1-!sccix
qia
ix1
(Mother) said that the (ocial) papers would arrive today. (conversationbuya.imui, lU o,
zcc!)
Since it is not impossible for tangi to be intended as the time in the referent speed situation, more
data are required to conrm whether (1c) is truly a counterexample.
Te claimthat reported speedin Kavalan is direct report is further supported by its discontinuity.
As shown in (11) below, the framed clauses are intervened by the framing clauses, whid are noted
here in bold.
(11) a. bula-ika
giveiix~i
tu
oni
babui
pig
tu
oni
u-siq
ciixnuone
ci
ix
buya
ix
zin-ku
ouo1-1sccix
ci
ix
imuy
ix
l told Buya to give lmuy a pig. (lit. Give one pig, l said to Buya, to lmuy.)
b. bula-ika
giveiix~i
zin-ku
ouo1-1sccix
ci
ix
abas
ix
tu
oni
u-siq
ciixnuone
babui
pig
ci
ix
siulan
ix
l told Abas to give Siulan a pig. (lit. Give, l said to Abas, one pig to Siulan.)
On the other side, for reported thought it is dicult to distinguish direct report from indirect
report since neither nominal coreferences nor symptomatic features are available. lor example, in
(1z) there is no coreference between qaniyau they and -ku l, and in (1!) -ita we (inclusive) is
partially coreferential with the rst person and the subject of the framing clause at the same time,
whid satises the coreferential rule for both direct and indirect report.
(1z) supaR
can
ma
i
qaniyau
!iixo
zin-ku
ouo1-1sccix
ka
i
l think they are beuer (at speaking than l am). (conversationbuya.syulan, lU 1z, zcc!)
(1!) qa-mai=ti-ita
iii-xiciiv-1iiixo
tu
oni
sikawma-an
speakxz
zin-ku
ouo1-1sccix
Ranaw
so
Anerwards, l think we will lose our language. (conversationbuya.syulan, lU 11, zcc!)
However, we do nd an example of reported thought that may qualify as direct quote. As illus
trated in (1), the framed clause is intervened by the framing clause in bold
(1) qa-ngasan-iku=
iiilate1scxo
qa-ngasan-iku
iiilate1scxo
zin-ku=
ouo1-1sccix
manan
~ireturn
l think l might return late. (conversationbuya.imui, lU 1cz, zcc!)
Rice Working Papers in Linguistics 1c vol. 1, February 2009
linally, a few remarks can be made about the reference time of the predicate zin, although
Kavalan has no tense maker that distinguishes the present from the past. Te reference time for
the reported thought of the rst person is nearly always identical with the uuerance time while the
reference time for the reported speedof the third person is exclusively anterior to the uuerance time.
Tis result is predictable since selfreport of thought is simultaneous while report of other peoples
speed is possible only aner they have spoken.
S Conclusion
ln this preliminary study, we have investigated reported speed and thought in Kavalan, specically
by looking into the instantiations of zin in both spontaneous narratives and conversations. We have
found that in narratives, reported speed is nearly twice as frequent as reported thought and is al
most exclusively framed by third person pronouns, while in conversations the situation is reversed,
with instances of reported thought almost exclusively framed by rst person pronouns. Our analysis
reveals that selfreport of thought is the norm in conversations while otherreport of speed is pre
dominant in narratives. Moreover, instances of reported speed are almost exclusively direct report,
while the reporting mode in reported thought is yet unclear due to the limitations of the data cur
rently available. We speculate that the distinction between direct and indirect report (even between
the reported speed and reported thought instantiated by zin) in Kavalan might not be as straight
forward or signicant as that in most lndoluropean languages. To some extent, these two modes
of report may be fused together in Kavalan, and this state of aairs is in line with Gldemann and
von Roncadors (zccz) generalization that direct and indirect report are not two clearcut categories,
but instead extreme poles on a continuum of various types of reported mode.
Rice Working Papers in Linguistics 1) vol. 1, February 2009
References
Chafe, Wallace l. (ed.) (1)cc). e Pear Stories: Cognitive, cultural, and linguistic aspects of narrative produc-
tion. Norwood, New Jersey Ablex Publishing.
Clark, H. H. and R. J. Gerrig (1))c). Qotations as demonstrations. Language oo() occ.
lortune, G. (1)). An analytical grammar of Shona. london longmans Green.
Givon, Talmy (1)cc). Te binding hierardy and the typology of complements. Studies in Language (!)
!!!!.
Gldemann, Tom and Manfred von Roncador (eds.) (zccz). Reported discourse: A meeting ground for dierent
linguistic domains. Typological Studies in language, Amsterdam John Benjamins.
Halliday, Midael (1)c). An introduction to Functional Grammar. london l. Arnold, 1st edn.
Janssen, T.A.J.M. and Wim van der Wur (1))o). Reported spee: Forms and functions of the verbs. Amster
dam John Benjamins.
Jiang, Haowen (zcco). Spatial conceptualizations in Kavalan. Masters thesis, National Taiwan University.
Klamer, Marian (zccz). Report constructions in Kambera (Austronesian). ln Tom Gldemann and Manfred
von Roncador (eds.), Reported discourse: A meeting ground for dierent linguistic domains, Typolog
ical Studies in language, Amsterdam John Benjamins, pp. !z!!c.
Korzybski, Alfred (1))). Science and sanity: An introduction to non-Aristotelian systems and general seman-
tics. lnglewood, NJ lnstitute of General Semantics.
leed, Georey N. and Midael H. Short (1)c1). Style in ction: a linguistic introduction to English ctional
prose. london longman.
lin, HsuehO (1)))). Reported spee in Mandarin conversational discourses:. Ph.D. thesis, National Taiwan
Normal University.
longacre, Robert l. (1)c). Sentences as combinations of clauses. ln Timothy Shopen (ed.), Language typology
and syntactic description, Cambridge Cambridge University Press, pp. z!zco.
lowe, lvan and Ruth Hurlimann (zccz). Direct and indirect speedin Cerma narrative. ln TomGldemann and
Manfred von Roncador (eds.), Reported discourse: A meeting ground for dierent linguistic domains,
Typological Studies in language, Amsterdam John Benjamins, pp. 1)c.
lyons, John (1)oc). Introduction to theoretical linguistics. Cambridge, UK Cambridge University Press.
Mayer, Mercer (1)o)). Frog, where are you?. New York Dial Press.
McGregor, William (1))). Te grammar of reported speed and thought in Gooniyandi. Australian Journal
of Linguistics 1(1) o!)z.
Noonan, Midael (1)c). Complementation. ln Timothy Shopen (ed.), Language typology and syntactic de-
scription. Vol. III: Complex constructions, Cambridge, UK Cambridge University Press, pp. z1c.
Pike, Kenneth l. and lvan lowe (1)o)). Pronominal reference in lnglish conversation and discourse A group
theoretical treatment. Folia Linguistica ! oc1co.
Rosenbaum, Peter S. (1)o). e grammar of English predicate complement constructions. Cambridge, MA
MlT Press.
Rice Working Papers in Linguistics 1c vol. 1, February 2009