Some Questions and Answers on the Prosodic Correlates of Information Structure
6 views 0 purchase
Course
Prosodic Correlates of Information Structure
Institution
Prosodic Correlates Of Information Structure
Some Questions and Answers on the Prosodic Correlates
of Information Structure
Hannes Pirker and Friedrich Neubarth
Austrian Research Institute for Artificial Intelligence (OFAI) ¨
Schottengasse 3, A-1150 Vienna, Austria
{hannes,friedrich}@
Abstract
In this paper a study on the effects of v...
Some Questions and Answers on the Prosodic Correlates
of Information Structure
Hannes Pirker and Friedrich Neubarth
Austrian Research Institute for Artificial Intelligence (ÖFAI)
Schottengasse 3, A-1150 Vienna, Austria
{hannes,friedrich}@oefai.at
Abstract models for scenarios which use concept-to-speech gen-
eration.
In this paper a study on the effects of varying infor-
mation structure and syntactic structure on prosody
is presented. For this purpose a corpus of German
question-answer pairs was designed and established. 2 Corpus Design
The structure and encoding of this corpus is described
and some analysis results are presented.
2.1 Textual material
In order to also facilitate the study of the effects of
syntactic variation, and to trigger variation in phras-
1 Introduction ing, the sentences used in our corpus are organised
into pairs like (1A) and (1B). All sentences display the
It is undisputed, that syntactic structure, ”information same uniform structure namely “np1 vfin np2 vinf1
structure” and more specifically the position of ”focus” und/and np3 vinf2”, with the abbreviations standing
strongly influence prosody ([1]). for noun phrases (np), finite (vfin) and infinite verbs
(vinf) respectively, with the numbers specifying their
Unfortunately, though, there is less agreement on how
linear ordering within the sentence. Their assignment
to define ”focus” properly, let alone what its exact ef-
to the sentences should become clear from the example
fects on prosody are. In addition, the huge number
(1A/1B) below.
of interdepending influencing factors complicates the
analysis of duration in particular ([2], [3]). This makes np1 vfin np2 vinf1
it difficult to identify and quantify the effects of a sin- (1A) Peter verspricht dem Freund zu verweilen
gle factor. One of the few widely accepted assumptions Peter (promises the friendDat ) to stayV i
on information structure is, though, that focus can be Peter promises (to) the friend to stay...
controlled by using wh-questions: The constituent cor-
responding to the wh-word in the question, should - per (1B) Peter verspricht den Freund zu entlasten
definitionem - be focused in the answer. Peter promises (the friendAcc to relieveV t )
Peter promises to relieve the friend...
In order to shed more light on the effects of varying in-
formation structure and syntactic structure, a corpus np3 vinf2
of question-answer pairs was constructed. This allows ... und den Dieb zu bewachen
for the manipulation of focus conditions on lexically ... and the thief to guard
identical sentences. Thus both the “focus definition”- ... and to guard the thief
problem, and the “degrees of freedom”-problem of in-
10 pairs with this structure form the textual basis of
fluencing factors is diminished.
the corpus. The text in these sentences is constructed
The current work is motivated by two different scenar- as follows: All words at position np1 (grammatical
ios: Advantage of its rigid structure is taken within a subject) are disyllabic proper nouns (e.g. Peter, Lisa,
project, where machine-learning techniques are applied Heike,...). At positions np2 and np3 common nouns
for producing duration-models for speech synthesis. In with varying numbers of syllables are used ( e.g. Fre-
this context the sub-corpus allows for complementary und, Chefin, Beamten, Gastgeberin, Wanderprediger)
analysis which help to fine-tune the encoding and selec- and the same words are systematically interchanged
tion of factors prior to automatic learning ([4]). The between position np2 and np3. For positions vfin, vinf1
precisely defined information structure, on the other and vinf2, each slot has its distinct non-overlapping set
hand, is particularly valuable for improving prosodic of finite (vfin) and infinite (vinf1, vinf2) verbs.
Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.
Quick and easy check-out
You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.
Focus on what matters
Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!
Frequently asked questions
What do I get when I buy this document?
You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.
Satisfaction guarantee: how does it work?
Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.
Who am I buying these notes from?
Stuvia is a marketplace, so you are not buying this document from us, but from seller Tutorgrades. Stuvia facilitates payment to the seller.
Will I be stuck with a subscription?
No, you only buy these notes for $12.99. You're not tied to anything after your purchase.