The prosody of finiteness and non-finiteness: the accent of Estonian finite and non-finite verbs

1
The prosody of finiteness and non-finitenessthe
accent of Estonian finite and non-finite verbs

• Anne Tamm
• anne.tamm_at_unifi.it
• RIL HAS Budapest
• University of Florence

2
Prosody and finiteness

• This contribution is about the prosodic
  characteristics of finiteness and non-finiteness.
  
• Can we regard prosody as another language module
  where finiteness operates in some languages?

3
Modules and finiteness

• I assume that certain verb forms, constructions
  and clauses can be either finite or non-finite.
• In terms of language modules, previous research
  has established that finiteness and
  non-finiteness phenomena are found in morphology,
  syntax, semantics and pragmatics. How about
  prosody?

4
Morphological, semantic and pragmatic finiteness
overlapping

• (1) Ta on siin.
• henom be.3s here
• He is here.

5
Morphologically, non-morphologically non-finite

• (2) Ta ei olnud siin sugugi
• henom neg be-nud_ptcp here at all
• mõtlemata kõigile tagajärgedele.
• without thinking of all the possible
  consequences
• He was not here without thinking
• of all the possible consequences.

6
Morphologically nonfinite, non-morphologically
finite

• (3) Ta olevat kodus.
• henom be-quot home-ine
• Allegedly, he is at home.

7
What is finite in Estonian negative clauses?

• in Example (4) below containing negation, the
  verb is morphologically non-finite, but
  non-morphologically, it may be regarded finite
• what is finite and non-finite in this sentence?
• is ei a particle,
• a verb,
• and if it is a verb, is it finite?

8
Negation and finiteness

• (4) Ta ei olnud siin.
• henom neg be-nud_ptcp here
• He was not here.
• (I set this apart in the study of finiteness
• and accent)

9
Negation, finiteness, accent, stress

• This talk concentrates on the contrast between
  morphological and non-morphological finiteness
  and non-finiteness phenomena, adding negation.
• Negation is a crucial an object of study, since
• it allows to test for several interdependent
  factors operating on the different levels of
  linguistic description just as (non)-finiteness
• its accentuation and syntactic properties can be
  teased apart from those properties of finiteness
  and non-finiteness that apply because of the
  properties of the affirmative
• that there are many types that are present in
  Estonian allow us to say something more general
  about negation, prosody and (non)-finiteness

10
The main questions

• Can we tease apart the morphological and
  non-morphological finiteness and non-finiteness
  phenomena in terms of prosody, and accent?
• Does negation form any patterns with finiteness?
• Is morphological/non-finite negation special in
  terms of stress and accent?
• What is the module on which we can build the
  prosodic structure?
• Can we predict stress and accent?

11
Negation, finiteness, syntax

• negated verbs assume a position in a sentence
  that can be different from the affirmative
• the arguments may also change their positions and
  accentuation due to the changed pragmatic and
  semantic factors
• in what follows, I establish the stress and
  accent pattern in some simple sentences and
  negation expressions

12
-mata attracts stress

• in production and perception tests, the -mata
  morpheme is distinguished from all other
  non-finite verb forms in stress it is
  frequently stressed, while all others are
  unstressed (tud, nud, tav, v, mas, mast, ma, da)
• it is also bisyllabic, but since there are other
  bisyllabic combinations of nonfinite morphemes
  and case that are not stressed
• (-nuna, -tuna, -tava, -vale), the stress on
  -mata can be considered remarkable

13
-mata attracts accent

• Also, while other non-finite forms tend to be
  unaccented in an identical position, the mata
  forms are accented (lexical exceptions exist).
• (5) Ta on ()ujumas.
• henom be.3sg swim-m_inf_ine
• He is off swimming.
• (6) Ta on ujumata.
• henom be.3sg swim-m_inf_abe
• He has not swum.

14
ei is unaccented

• The accentuation of the negation verb/particle
  ei is perceived as acceptable, but strange in
  perception tests, triggering unusual contexts.
• (7) Ta ei olnud siin.
• henom neg be-nud_ptcp here
• He was not here.
• (accent on olnud or siin)

15
mitte is partly unpredictable

• From among the negation items, the accent of
  mitte is the most unpredictable and depends most
  on its position and the information structure,
  and its lexical surroundings.

16
mitte is partly predictable

• teps mitte not at all (always accented)
• mitte mina vaid sina (never accented)
• ()mitte midagi nothing at all (variably
  accented)

17
eba- depends on the lexicalization

• The stress pattern in the combinations with the
  prefixal eba- seem to depend on the
  lexicalization of the combination (the adjective
  part has always stress if not lexicalized, and
  nouns tend to be stressed).
• ebameeldiv (-meel- has no stress)
• ebauuenduslik (-uuen- has stress)
• ebaküdoonia quince (-küdoonia has stress)

18
Finite verbs unaccented

• The finite verb in a clause is mostly
  unaccented, unless it is not required by
  information structural or other constraints, and
  its accenting is perceived as strange.
• (8) Ta on siin.
• henom be.3sg here
• He is here.

19
Clause-final finite forms accented

• Clause-final finite forms are accented.
• (9)
• Seda ta mulle eile ütles.
• This.part s/he I-all yesterday say.3s.pst
• Yesterday s/he did say it to me.

20
Negative verbs are accented

• However, some listable finite verbs are rather
  accented, and their accentuation can be
  attributed to their role in discourse.
• Semantically negative verbs tend to attract
  accent.
• eemaldama remove vs lisama add
• loobuma decline vs nõustuma agree
• eitama negate

21
Non-finite forms clause-final

• Non-finite forms appear mostly clause-finally,
  and are accented.
• (10)
• Seda sai mulle eile öeldud.
• This.part s/he I-all yesterday say.tud-ptcp
• Yesterday it was told to me.

22
Non-finite forms as main predicates in finite
clauses

• Non-finite forms that are the main predicates in
  finite clauses are not accented (exactly as the
  finite verbs).
• (11) Ta olevat kodus.
• henom be-quot home-ine
• Allegedly, he is at home.

23
Negated verb, variable accent

• Whether the verb form combining with ei is
  accented or not, depends on its position and its
  discourse properties (accented if clause-final,
  and/or on the list of preferably accented
  predicates)
• (12) Ei tea öelda (no accent on tea know)
• Neg know say-da_inf
• One cannot say.

24
Verbs in negative sentences vary in accent

• The accentuation of the negation verb/particle
  ei is perceived as acceptable, but strange in
  perception tests, triggering unusual contexts.
• (13) Ta ei ()olnud siin.
• henom neg be-nud_ptcp here
• He was not here.
• (accent on olnud or siin)

25
Non-finite negated forms as main predicates in
finite clauses

• Non-finite forms that are the main predicates in
  finite clauses and that are negated are accented.
• (11) Ta ei olevat kodus.
• henom neg be-quot home-ine
• Allegedly, he is at home.
• Accenting is a significant difference between
  morphologically finite and non-finite negated
  main predicates.

26
Clause-final verbs are accented

• Clause-final verbs are accented.
• (14)
• Seda ta mulle eile ei öelnud.
• This.part s/he I-all yesterday neg
• 
  say-nud_ptcp
• Yesterday s/he did not say this to me.

27
Non-morphological finiteness as the basis for
accenting

• One can conclude that the type of finiteness that
  is relevant for accenting is non-morphological in
  Estonian.

28
Prosody patterns little with morphological
finiteness

• Prosodic patterns are generally more regular in
  terms of their correlation with the
• (non-)finitness phenomena in syntax, semantics,
  pragmatics, and extra-linguistic or
  semi-linguistic properties (accentuation at the
  end of an utterance/sentence) than in terms of
  morphological finiteness.

29
An interesting correlation

• the correlation of accent and morphological
  finiteness is weak with morphological finiteness
  but some cases stand out as exceptions
• the negative (abessive) morphologically
  non-finite predicates (mõtlemata) and negated
  morphologically non-finite main predicates (ei
  olevat) are regularly accented
• the negative (abessive) morphologically
  non-finite forms can be predicted to carry
  accent, and the stress on the morpheme

30
Final conclusions

• the examples above showed that it is rather the
  negative content that attracts stress and accent
  in Estonian
• non-finiteness and negation together predict
  accent in some well defined environments in
  Estonian

31
Prosody not a module where finiteness operates

• Can we regard prosody as another language module
  where finiteness operates?
• The answer is no for Estonian, but there are
  regularities that might help to tease apart the
  morphological, syntactic, semantic and pragmatic
  finiteness.

32
Bibliography to copy and paste to see better

• Asu, E.-L. 2004. The phonetics and phonology of
  Estonian intonation. Unpublished doctoral
  dissertation. University of Cambridge, UK.
  Dissertation. Abari K, Tamm A., Gábor K., Olaszy
  G. 2007. Számítógépes összehasonlító
  szövegelemzés ügyfélszolgálati tájékoztatók
  legfontosabb prozódiai elemeinek a
  meghatározására. Magyar Számítógépes Nyelvészeti
  Konferencia. Szerk. Alexin Zoltán és Csendes
  Dóra. Szegedi Tudományegyetem Informatikai
  Tanszékcsoport. Szeged. 24-33. Cinque, G. 1993.
  "A null theory of phrase and compound stress",
  /Linguistic Inquiry/ 24. pp.239-97. Ehala, M.
  2001. Eesti keele baassõnajärjest. Keele kannul
  pühendusteos Mati Erelti 60. sünnipäevaks 12.
  märtsil 2001. Publications of the Department of
  Estonian of the University of Tartu 17 24-41.
  Erelt, M., Metslang, H. (2006). Estonian clause
  patterns - from Finno-Ugric to Standard Average
  European. Linguistica Uralica, XLII(4), 254 -
  266. Grosz, B., Hirschberg, J. 1992. Some
  intonational characteristics of discourse
  structure, Proceedings of ICSLP-92, Banff, pp.
  429-432. Help, T. 1996 "Free" word order
  Finnish vs. Estonian and Hungarian. Linguistica
  Uralica 1 321-30. Hayes, B., 1995. Metrical
  stress theory principles and case studies
  Chicago, Ill. University of Chicago Press.
  Huumo, T. 1993."Suomen ja viron kontrastiivista
  sanajärjestysvertailua". In Yli-Vakkuri, Valma
  (ed.) Tutkielmia syntaksin ja pragmasyntaksin
  alalta. Studia comparativa linguarum orbis Maris
  Baltici 1. Publications of the Department of
  Finnish and General Linguistics, University of
  Turku 43. Turku, p. 97-158. Huumo, T. 1994.
  "Näkökulmia suomen ja viron sanajärjestyseroihin".
  Sulkala, Helena Laanekask, Heli (eds.)
  Lähivertailuja 8, Suomalais-virolainen
  kontrastiivinen seminaari Hailuodossa
  7.-9.5.1994. Research reports from the
  Department of Finnish and Saami, Univeristy of
  Oulu 40. p. 21 39. Huumo, T. 1995. "On the
  Position of Subject in Finnish and in Estonian"
  XXI Kielitieteen päivät Oulussa 6. 7.5.1994. Acta
  Universitatis Ouluensis Humaniora B 19. Oulu
  1995. p. 79 87.Huumo, T. 1996. "Remarks on the
  Functions of Word Order in Finnish and in
  Estonian". Congressus Octavus Internationalis
  Fenno-Ugristarum. Jyväskylä 10.-15.8.1995. Pars
  IV Sessiones sectionum Syntaxis et Semantica
  Contactus linguistici et status hodiernus
  linguarum Cetera linguistica. Eds. Heikki
  Leskinen, Sándor Maticsák Tõnu Seilenthal.
  Jyväskylä. S. 47 51. É. Kiss, K., 1995. Discourse
  Configurational Languages. Oxford University
  Press, 1995. É. Kiss, K., Kiefer, F., Siptár, P.
  Új magyar nyelvtan. Budapest Osiris, 1998.
  Komlósy, A. (1989) Fókuszban az igék, In
  Általános Nyelvészeti Tanulmányok, 17. Kornai A.,
  Kalman, L. 1988. Hungarian Sentence Intonation.
  Harry van der Hulst, Norval Smith Linguistic
  Models 11. Autosegmental studies on pitch accent.
  Foris.183-195. Ladd, D., R., Intonational
  phonology. Cambridge Cambridge University Press,
  1996. Lehiste, I. 2007. The Finno-Ugric Prosody
  Project. Linguistica Uralica XLIII 1. Lindström,
  L. 2006 Infostruktuuri osast eesti keele
  sõnajärje muutumisel Keel ja Kirjandus 2006, 11.
  Lindström, L. 2005. Finiitverbi asend lauses.
  Sõnajärg ja seda mõjutavad tegurid suulises eesti
  keeles. Tartu Tartu Ülikooli Kirjastus. Mihkla,
  M. Meister, Eek. 2002. Tekst-kõne süntees. Keel
  ja Kirjandus. Mihkla, M. Kerge, K. Pajupuu, H.,
  Statistical modelling of intonation and breaks
  for Estonian text-to-speech synthesizer. Robert
  Vich (Ed.). Electronic Speech Signal Processing.
  Proceedings of the 16th Conference Joined with
  the 15th Czech-German Workshop "Speech
  Processing". Prague, September 26-28. 2005 (91 -
  98). Studientexte zur Sprachkommunikation. Band
  36. Dresden TUDpress, 2005. Müürisep, K., Eesti
  keele arvutigrammatika süntaks. Dissertationes
  Mathematicae Universitatis Tartuensis 22. Tartu
  2000. Pajupuu, H. Tael, K. 1989. The acoustic
  parameters and the perception of certain elements
  of pragmatic structure. The 28th Acoustic
  Conference on Physiological Acoustics,
  Psychoacoustics, Acoustics of Music and Speech.
  October 3.-6.1989 Strbske Pleso-High Tatras
  Czechoslovakia,1989. Strbske Pleso-High Tatras
  Czechoslovakia, 1989, 298 - 301. Rizzi, L., The
  fine structure of the left periphery. In Liliane
  Haegeman (ed.) Elements of Grammar. Kluwer,
  Dordrecht, pp. 281-337, 1997. Sahkai, H.1999.
  Eesti verbifraasi sõnajärg. Keel ja Kirjandus1
  24-32. Tael, K. 1988. Infostruktuur ja
  lauseliigendus.3. 133-143. Tael, K. 1988.
  Sõnajärjemallid eesti keeles (võrrelduna soome
  keelega) läänemeresoome keeleteaduse sümpoosion
  (Turku, 30.08-2.09.1988) / Eesti NSV TA
  Ühiskonnateaduste osakond, Keele ja Kirjanduse
  Instituut. Preprint KKI 56. Tamm, A. K. Abari,
  G. Olaszy. 2007. Accent Assignment Algorithm in
  Hungarian Based on Syntactic Analysis, in
  Proceedings of the 8th Annual Conference of the
  International Speech Communication Association
  (Interspeech 2007), Bonn, pp. 466-469. Tamm, A.
  K. Abari, G. Olaszy. 2005. Kísérlet automatizált
  szövegelemzési módszerek kialakítására a
  szóhangsúlyok meghatározásához An attempt to
  create text analysis methods for determining word
  stress in D. Csendes (ed.), III. Magyar
  Számítógépes Nyelvészeti Konferencia NSzNy 2005.
  2005. december 8-9. Tamm, A. 2008. Problems of
  automatic accent assignment a comparison of
  Hungarian and Estonian verbs. Traat, M. 2006.
  Information Structure in Discourse
  Representation. Computational analysis of theme,
  rheme and focus with Unification-based
  Combinatory Categorial Grammar. PhD thesis,
  University of Edinburgh. Varga, L. Intonation
  and Stress Evidence from Hungarian. New York
  Palgrave Macmillan, 2002. Varga, L., Rhythmic
  stress alternation in Hungarian. 1994 233254.
  Varga, L., Hungarian intonation contours. 2
  (1996)111144. Varga, L., The Intonational
  Phrase and Secondary Intonational Phrase
  Formation in Hungarian. 3 (1998)251263

33
Acknowledgements

• This research is supported by the grant The
  bidirectional relationships between the speech
  rhythm and Estonian grammatical and lexical
  structures, grant number ETF 2009, ETF7998 of the
  Estonian Science Foundation (Eestikeelse kõne
  rütmilisuse peegeldused grammatilistes ja
  leksikaalsetes struktuurides (ja vice versa)) or
  Modelling intermodular phenomena in Estonian,
  grant number SF 2009, SF0050023s09 of the
  Estonian Science Foundation (Eesti keele
  alusuuringud keeletehnoloogiliste rakenduste
  teenistuses)