<![CDATA[
<META NAME="description"
CONTENT="Maartje Schreuder">
<META NAME="keywords" CONTENT="maartje schreuder, language, linguistics, phonology, phonetics,
prosody, rhythm, intonation, phrasing, music"><html>
<head>
<TITLE>Research of Maartje Schreuder</TITLE>
</head>
<BODY bgcolor=ffffff
TEXT=black LINK=#147CB5 ALINK=#147CB5 VLINK=#147CB5 BACKGROUND="musicbg18.jpg" bgproperties="fixed" >
<style fprolloverstyle>A:hover {color: #C6004B; text-decoration: underline}
</style>
<TABLE BORDER cellspacing="2" cellpadding="10" width="100%">
<tr>
<td><center><H1>Research of Maartje Schreuder</H1><center>
<P>
<a href="affil.htm" target="_blank"> Affiliation</a> |
<a href="papers.htm"target="_blank"> Papers</a> |
<a href="pres.htm"target="_blank"> Presentations</a> |
<a href="sounds.htm"target="_blank"> Sounds</a> |
<a href="lit.htm"target="_blank"> Literature lists</a> |
<a href="links.htm" target="_blank"> Links</a>
<BR>
</td>
</TR>
<TABLE cellspacing="2" cellpadding="10" width="100%">
<tr>
<td colspan=2>
<BR>
<center><H2>Prosodic variability in language and music</H2></center>
<center><P><I><B>Supervisors:</B></I>
<I>dr. <A HREF="http://odur.let.rug.nl/~gilbers/"target="_blank">Dicky Gilbers</A> (co-promotor)</I> &
<I>prof. dr. <A HREF="http://odur.let.rug.nl/~koster/"target="_blank">Jan Koster</A> (promotor)</I></center>
<BR>
<P ALIGN="JUSTIFY">For the present research I will compare variabilitly in language with variability in music.
Optimality Theory owes a lot to the constraint-based music theory of Lerdahl & Jackendoff (1983).
Based on the great similarities between language and music I claim that musical knowledge can help in
solving linguistic issues.
<P><center><A HREF = "papers.htm#l&m">Full paper</A> | <A HREF = "sounds.htm#l&m">Sounds</A></center>
<BR>
<HR>
</td>
</tr>
<tr>
<td cellspacing="2" cellpadding="10" width="50%">
<BR><P><CENTER><B>Restructuring Rhythm Patterns</B></center>
<p><center><A HREF = "papers.htm#vienna">Full paper</A> | <A HREF = "sounds.htm#rhythm">Sounds</A></center>
<P ALIGN="JUSTIFY">The topic of this paper is how rhythmic variability can be accounted for both phonologically and phonetically. The main question is whether the influence of a higher speech rate leads to adjustment of the phonological structure, or just to 'phonetic compression', i.e. shortening and merging of vowels and consonants, with preservation of the phonological structure. We claim that the phonological structure is adjusted, which implies that every speech rate has its own register, in terms of Optimality Theory (Prince & Smolensky, 1993) (hence: OT) its own ranking of constraints.<BR>
The allegro realisation of e.g. Dutch <I>zuidafrikaans</I> ‘South African’ shows a preference for a triplet pattern <I>(s w w s), </I>whereas the andante realisation<I> </I>prefers <I>(s s w s</I>) (English data: <I>blueberry hill, strawberry fields</I>). Other linguistic data include ‘stress shifts to the left’ as in <I>perfèctioníst</I> <I>(w s w s)</I> in andante tempo, which can be realized as <I>pèrfectioníst (s w w s)</I> in allegro tempo in Dutch, and ‘stress shifts to the right’ as in <I>stúdietòelage</I> <I>(s w s w w)</I> ‘study grant’, which can be realised as <I>stúdietoelàge (s w w s w).
</I><BR>
We claim that the allegro patterns in all the different types of data are caused by clash avoidance; the distance between beats is enlarged. The different structures are described phonologically as a conflict between markedness constraints, such as ‘foot repulsion’ (Kager, 1994), and ‘output-output correspondence’ constraints (Burzio, 1998) within the framework of OT. Whereas the normal patterns in andante speech satisfy correspondence, the preference for triplet patterns in fast speech is accounted for by means of dominance of the markedness constraints.<BR>
In our experiment, the allegro data of six subjects were obtained by means of a multiple-choice quiz in which subjects competed each other in answering twenty simple questions as quickly as possible. As a second task the subjects were asked to read out the answers in a normal tempo. The results show that the quiz design indeed triggers faster speech by the subjects. The mean acceleration factor is 1.4. Moreover, an auditive analysis of the data indeed shows that most subjects prefer the restructured patterns in allegro speech, as described in an OT-ranking with dominant markedness constraints, whereas the pattern reflecting a dominance of the correspondence constraints is observed more often in the andante tempo. This is not an absolute preference, however, sometimes restructuring does not take place in allegro speech, but on the other hand restructured patterns also show up in andante speech for a small minority of the speakers. Therefore, we suggest a random ranking of weighed correspondence and markedness constraints. Each style and tempo has its own preferred ranking, but allows for reranking.<BR>
We attempted to confirm our phonological account with a phonetic analysis in PRAAT (Boersma & Weenink, 1992). We compared andante and allegro data by measuring the main correlates of primary stress: duration and spectral balance (Sluijter, 1995), next to intensity and pitch. In our analysis, however, neither differences in duration nor differences in spectral balance could identify secondary stress. Therefore, our analysis supports earlier work by Shattuck Hufnagel et al (1994), Cooper and Eady (1986), Huss (1978) and Grabe and Warren (1995), who all claim that acoustic evidence for secondary stress cannot be found unambiguously. Although we did find some differences in duration, spectral balance or pitch, these differences were not systematically found in all pairs in which we perceived rhythmic variability. At this point, we have to conclude that there is no phonetic correlate of the perceived phonological stress shifts in the acoustic signal. Are we fooled by our brains or has the real phonetic correlate of secondary stress yet to be found?
</td>
<td cellspacing="2" cellpadding="10" width="50%">
<center><B>Recursive Patterns in Phonological Phrases</B></center>
<p><center><A HREF = "papers.htm#sp04">Full paper</A> | <A HREF = "sounds.htm#sp04">Sounds</A></center>
<P ALIGN="JUSTIFY">In this paper we investigate an instance of phonological recursion, more specifically we investigate iterative rule application in phonological phrases. The question is whether or not edge-marking processes, such as early pitch accent placement, can be applied recursively to phonological phrases that are embedded in larger phonological phrases.<BR>
Recursion is a very common phenomenon in syntax. In phonology, things seem to be different. Although iterative rule application is proposed for e.g. foot assignment (Selkirk, 1980, 1984) or affixation of prosodic words (Itô and Mester, 1992; Zec and Inkelas, 1991; a.o.), prosodic building rules seem to be limited in that sense. The limitations to the prosodic hierarchy are reflected in the Strict Layer Hypothesis (Selkirk, 1984), of which one of the fundamental assumptions is that prosodic structure is not recursive. Recently, several phonologists have shown that in many cases the Strict Layer Hypothesis is violable (Itô and Mester, 1992; papers in Hall and Kleinhenz, 1999); the assumption of non-recursivity has been challenged by a number of authors, in particular involving prosodic words with affixal clitics (Itô and Mester, 1992; Zec and Inkelas, 1991; Selkirk, 1995; a.o.), but also in larger prosodic domains like the intonational phrase (Ladd, 1986, and references
therein).<BR>
Since recursion is very common in syntax, the primary source of evidence for instances of recursion in phonology is probably provided by phonological rules that operate over syntactically defined (recursive) domains. The phonological phrase is one instance of a domain of which the phrase breaks typically coincide with the edges of morphosyntactic phrases (Selkirk, 1984; Nespor and Vogel, 1986). To in vestigate whether recursion can also be found at this higher domain in the prosodic hierarchy, we examined various Dutch phonological phrases, consisting of either a noun modified by one adjective, of the type <I>aardrijkskùndig genóotschap</I> 'geographical society', or by two adjectives, of the type <I>Amsterdàms aardrijkskùndig genóotschap</I>, i.e., a recursive noun phrase (the accents on the adjectives indicate the main stress position in citation form.). The first type of phonological phrases is known to display early pitch accent placement (Shattuck-Hufnagel, 2000 a.o.)
as a means of signalling a phrasal boundary to give the listener a cue to the prosodic structure of the spoken utterance. The question now is what kind of prosodic structure has to be assumed for the second type, the syntactically recursive noun phrases. We hypothesize that an early pitch accent can be realized on the second adjective, indicating a recursive phrase structure with two left boundaries within the same phonological phrase: [[Àmsterdams [àardrijkskundig genóotschap]].<BR>
We conducted a map task experiment to investigate whether these kind of phonological phrases can indeed be produced with a recursive prosodic structure. The phrases of interest were landmarks on the maps, and afterwards the subjects were asked to read the isolated adjectives within a sentence. For the analysis three trained listeners judged the data auditorily and indicated on which syllables in the adjectives they perceived word accent. It turned out that, although there was a strong preference for the subjects only to accent the main stress syllables of the adjectives, early accent placement is also a strong tendency. For the recursive phrases we found an early pitch accent on the second adjective in 30% of the cases. Chi-Square tests showed that the results were highly significant. This confirms our hypothesis that these syntactically recursive phrases can also be recursive prosodically.
</td>
</tr>
</TABLE>
</TABLE>
<HR>
<P>Last update: April 2004, <A HREF = "welcome.htm"target="_top">Maartje Schreuder</A>
</body>
]]>