Вы находитесь на странице: 1из 8

International Journal of Educational

Science and Research (IJESR)


ISSN(P): 2249-6947; ISSN(E): 2249-8052
Vol. 6, Issue 5, Oct 2016, 11-18
TJPRC Pvt. Ltd.

THE COMPARISON OF PROSODIC BOUNDARY FEATURES WITH


ENCLITICIZED AND PROCLITICIZED FUNCTION WORDS
MIR JEONG1 & YOUNG MEE KIM2
1

Bio and Brain Engineering, Korea Advanced Institute of Science and Technology, Dae-Jeon, S. Korea
2

Humanities and Arts, Korea Science Academy of KAIST, Busan, S.Korea

ABSTRACT
In spontaneous speech, unlike read speech, prosodic phrasing doesnt match with syntactic structure often.
Previous research confirms that this phenomenon usually occurs at intermediate phrase boundaries where function
words are encliticized or procliticized. The research aims at observing prosodic boundary feature change as intermediate
boundary positions were relocated. Ten pairs of the sentences selected from the audio book Hoot
(spontaneous speech) and high school English textbook (read speech) were extracted and the function words were
encliticized to proceeding host words or procliticized to succeeding host words. After measuring the length of the pauses,
final syllables, and the changes of fundamental frequencies using Pratt, differences among prosodic boundary features
were analysed. The results showed that the pause lengths at intermediate phrase boundaries shortened, final syllable

accents occurred at the early phases of boundary positions. In conclusion, prosodic phrasing exposes the overall
sentence structure at the early stages of utterance, and listeners can anticipate the upcoming development of utterances.
Considering that prosodic phrasing is essential for language learning, learners of English as a second language should
practice prosodic phrasing rather than syntactic phrasing.
KEYWORDS : Analytical Thinking, Computational Methodology, Encliticized and Procliticized Function Words,

Original Article

length increased and high phrase accents prevailed over low phrase accents. Final syllable lengthening and phrase

Language Education, Pratt Analysis, Prosodic Boundary Features

Received: Aug 03, 2016; Accepted: Aug 29, 2016; Published: Sep 01, 2016; Paper Id.: IJESROCT20162

INTRODUCTION
In English, spontaneous speech is done by chunking prosodic boundaries into prosodic constituents,
which is called Prosodic phrasing. However, prosodic phrasing in spontaneous speech often does not match with
syntactic structure in read-aloud speech. This is because structuring of prosodic boundaries is essential for
facilitating information processing in listeners (Heldner & Megyesi[1], 2003). Therefore EFLs need to learn
prosodic phrasing based on the understanding of how prosodic boundary cues are realized.
Prosodic Phrasing and Intermediate Boundary
Plan of speech is done based on syntactic and semantic relations but actual speech is done based on
prosodic units. Prosodic boundary features are attached to each prosodic boundary layer so that identification of
prosodic features and layers is possible (Ostendorf [2], 1990). Intermediate boundary identification is especially
important. Intermediate phrase acts as a kind of phrase that is below intonation phrase (Se-Peung Oh [3], 2009).
Prosodic phrasing functions as giving the basic skeleton of speech and intermediate boundary plays a key role
(Frazier et al.[4], 2006). Intermediate boundaries not only gives syntactic structure of English, but also semantic
www.tjprc.org

editor@tjprc.org

12

Mir Jeong & Young Mee Kim

and detail parts.


Prosodic Boundary Features
Boundary location and boundary salience are an important component in prosodic phrasing. Prosodic boundary
features include pause, breathing, fundamental frequency change, syllable lengthening in front of the boundary, speech
speed, intensity change, sound quality, and abbreviation. Relative differences in pause length reveal syntactic hierarchy and
discriminating the difference of meaning is possible (Schafer [5], 1997). In ToBI, pause length is scaled from 0 to 4 by an
analyzer subjectively. Final syllable lengthening occurs at front of the prosodic unit boundaries such as at the end of the
prosodic words, intermediate phrase, and speech unit and the extent of the lengthening differ depending on each hierarchy
(Zvonik[6], 2004). Final syllable lengthening is the most reliable clue that represents prosodic boundary features in
intermediate boundary (Wightman, Shattuck-Hufnagel, Ostendorf & Price [7], 1992). Intermediate phrase consists of at
least one pitch accent and phrase accent. Phrase accent represents the relationship between the current phrase and the
succeeding phrase. Low phrase accent (L-) represents that currently spoken speech needs to be separately interpreted with
the following phrase while high phrase accent (H-) demands that both the current phrase and the following phrase form
upper layer of interpretation unit (Pierrehumbert & Hirshberg [8], 1990). Phrase accent, along with function word,
represents the syntactic and informational relationship of the whole sentence structure.
Syntax Violating Prosodic Boundary and Prosodic Restructuring
Prosodic boundaries of 20% in a Swedish radio interview material occurred in syntactically unmotivated positions
(Strangert [9], 2004 b). Prosodic boundaries of 35% in the subset of the Boston Radio News Corpus were not consistent
with syntactic structures (Fach [10], 1999). Syntactic factors, semantic factors, phonological factors, pragmatic factors,
processing factors, speech speed and style influence the syntax violation of prosodic phrasing in free speech
(Cole et al., 2010; Bachenko & Fitzpatrick, 1990; Fitzpatrick[11-13], 2001).
Function Words and Prosodic Boundary
To clarify the relationship between intermediate phrase boundary and function words, identifying the role of
function words is required. There are few types of function words but occur frequently and consistently in place.
They can restrict words that will come next by appearing at front of the phrase or can inform the type of phrase
(Christophe et al.[14], 2008). Prosodic features change noticeably where function words take place. Function words in
conjunction with prosody provide cues to prosodic phrasing (Gerken & McIntosh [15], 1993). When function words are
encliticized, listeners can predict next content of speech earlier. By encliticizing the function words, they become
positioned near the matrix verb, which determines the whole sentence structure, during the speech so that the whole
syntactic structure is revealed early.
Encliticization and Salience of Function Words
To make modified intermediate phrase around function words salient, it is necessary to adopt prosodic boundary
features distinct from general prosodic boundary features based on syntactic structure. The prosodic features of boundaries
change from strong to weak ones (Heldner & Megyesi, 2003; Strangert & Carlson [1,16], 2006). In cases except where
emphasis is needed for contrast, function words cannot be given accent, so by lengthening final phrase it is possible to
make it salient. High phrase accent (H-) will be used more frequently because it indicates theres more to say, connects the
preceding and anteceding phrase, attracts listeners by increasing fundamental frequency, and plays role in making up the
Impact Factor (JCC):4.5129

NAAS Rating: 2.72

The Comparison of Prosodic Boundary Features with Encliticized and Procliticized Function Words

13

speakers intent earlier

OBJECTIVES
Based on the related works, this research seeks to clarify hypotheses below.

Position and prosodic features of pauses will be changed when the function words are encliticized at intermediate
phrase boundaries in spontaneous speech.

The length of the pauses following the encliticized function words will be shortened compared with the ones when
procliticized.

Duration of encliticized function words will be lengthened.

High-phrasal accents will prevail when function words are enclicitized.

METHODS
Data Selection
The data were selected from the audio book HOOT for spontaneous speech and from the Korean English
textbook of high school for read-aloud speech. 11 pairs of utterance which contain various classes of function words
(in, to, a, have been, and, or, than, if, who, when, that) were analysed at the intermediate phrase boundaries. Monosyllable
function words were selected because accent might arise inside poly-syllable function words. For simplicity in the analysis,
voiced sound function words were selected as much as possible. To and that were selected as they occur frequently, but
were analysed with caution as they contained voiceless rupture sounds. Sentences where pause position is affected due to
pronouns or articles that comes right after the function words were excluded. Complex sentences and long speech were
selected so that speaker does not consciously adjust the speech due to pause or final syllable lengthening.
Procedures of Analysis
Speech speeds were normalized by aligning the number of syllables per second in order to eliminate its effect on
the analysis (Yeona Jeong[17], 2006). There are intra-speaker variation, inter-speaker variation, and listener variation.
Rather than subjective labelling, it was based on contrasting the relative differences in values using Pratt. Pause length,
final syllable length, phrase accent pitch were analysed using Praat program, mostly with spectrogram and waveform.
Phrase accents and pause length index were labelled by the author based on ToBI system. Subjective decisions were
somewhat inevitable during the labelling. However, prosodic research focuses on the relative relationships between the
prosodic properties; the methodology in the research has validity to some extent (Byeong-Seop Ahn[18], 2010).

RESULTS
Examples of Analysis
Procliticized in

www.tjprc.org

editor@tjprc.org

14

Mir Jeong & Young Mee Kim

Figure 1: Analysis of Procliticized in


According to Pratt analysis, there was 347ms pause in front of the function word in. The function word length
was measured to be 70ms. According to To BI labeling, H-H% boundary tone appeared in front of the function word in.
Encliticized in

Figure 2: Analysis of Encliticized in


According to Pratt analysis, there was 79ms pause right after the function word in and is 268ms shorter than the
procliticized case according to ToBI labeling, the pause length is classified as BI2 rather than BI3. The function word
length was 89ms, 19ms shorter than the procliticized case, exemplifying final word lengthening. In front of phrase
boundary, H-H% phrase tone appeared.
Pause length

Figure 3: Length of Pause when Procliticized and Encliticized

Impact Factor (JCC):4.5129

NAAS Rating: 2.72

The Comparison of Prosodic Boundary Features with Encliticized and Procliticized Function Words

15

The pause length in front of the function word for procliticized case is represented in blue bars and the pause
length behind the function word for encliticized case is represented in red bars. The average of procliticized case is
229.9ms and 56.3 for encliticized case with p<0.01. Pause length for procliticized case ranged 62~357ms and can be
grouped into short 100~200ms group and long 300~400ms group. These groups correspond to intermediate boundary and
intonation boundary, respectively (Fant et al[19], 2003). In other words, strong boundary characteristics appear in
procliticized case. For encliticized case, the pause length ranged 18~120ms and according to ToBI system, it is classified as
BI2 rather than BI3.
Function Word Length

Figure 4: Length of Function Words when Procliticized and Encliticized


Blue bars represent the length of the function word for procliticized case and red bars represents for encliticized
case. The average of procliticized case is 93.6ms and 138.3ms for encliticized case with p<0.01. Function word have been
has been excluded when averaging as it can distort the average value. The final syllable lengthening should be significant
enough to be perceptible. According to previous research, the final syllable lengthening should be more than 20% (Park
Mun-Kyu[20], 2003) or 25% (Yoon Young-Sook[21], 2004). Since the final syllable lengthened by 47.8%, it can be
concluded that the final syllable lengthening is perceptible.
Patterns of Phrasal Tones

Figure 5: Patterns of Phrasal Tones when Procliticized and Encliticized


Blue bars represent the frequency of phrasal tones when procliticized. Red bars represent for the encliticized case.
In procliticized case, intonation boundary was prevalent and in encliticized base, phrase boundary was prevalent. H- phrase
accents prevailed over L- phrase accents in encliticized case. This indicates that when function words are encliticized,
phrase accent changes and acts as a means to represent that the preceding and the following phrases have
closely-connected relation.

www.tjprc.org

editor@tjprc.org

16

Mir Jeong & Young Mee Kim

DISCUSSIONS
The research focused on changes in prosodic boundary features when the function words are encliticized and tried
to interpret the phenomena by relating predicting the functions of function words and connection of its constituents in
acoustic language. Unlike preceding research (Wellmann et al., 2012; Seidl & Cristia, 2008; Mannel[22-24], 2009), the
explanation for pause length shortening when function words are encliticized was not just that pause is not relatively
significant in recognizing intermediate boundary, but it is a strategic way to make connections of the constituents
prominent.
There was no analysis about the apportionment of functions of prosodic boundary features relating to function
word encliticization. Caution was needed in the inference as there wasnt complete specification of how each prosodic
feature interact, how they are fused and realized, whether type of union depends on the condition, or which features play
the most important roles. This research only reinforced the inference using previous research and further research needs to
be done in order to secure proof.

CONCLUSIONS
According to the research, hypotheses in 2.6 are verified. It is shown that prosodic phrasing in spontaneous speech
differs from prosodic phrasing in read speech. Caution is needed when generalizing because the data set was small,
however. In recorded high school textbook contents, it was confirmed that prosodic structures coincided with syntactic
structure. In audio book HOOT, it was not a completely free-speech sample, but still prosodic phrasing violating
syntactic structure was observed. If strictly free-speech sample were used, the hypotheses would be much strongly verified.
One of the important aspects of the future work is to verify that encliticization of function words is the result of
evolution and adaptation of language. Perhaps due to the fact that spoken language was introduced much earlier than the
written language, there are discrepancies in prosodic boundary features. Another research method is by experimenting with
infants. Since there are difficulties in experiments with infants, using ERPs generated from infants brain would be helpful.
The research asserted the learning strategy related to function word encliticization, but future research is imperative in
order to prove the effectiveness of this approach. There should be research that contrast between English native and EFL
and between EFLs who used English learning methods in this research and those who didnt.
REFERENCES
1.

Heldner, M, & Megyesi, B, The acoustic and morpho-syntactic context of prosodic boundaries in dialogs, PHONUM 9, 117120., 2003

2.

Ostendorf, M., Price, P., Bear, J., & Wightman, C. W., The use of relative duration in syntactic disambiguation, Proc. Third
DARPA Speech and Natural Language, 1990

3.

Se-Peung Oh, The problems of English intonation labelling and its solution., 2009

4.

Frazier, L., Cahrson, K., & Clifton, C., Prosodic phrasing is central to language comprehension, TRENDS in Cognitive
Sciences, vol. 1.0 no. 6, 244-249., 2006

5.

Schafer, A. J. Prosodic Parsing: The role of prosody in sentence comprehension, unpublished doctoral dissertation, University
of Massachusetts., 1997

6.

Zvonik, E. Parsing and the temporal organization of phrases: An experimental study of read speech, Unpublished Doctoral

Impact Factor (JCC):4.5129

NAAS Rating: 2.72

The Comparison of Prosodic Boundary Features with Encliticized and Procliticized Function Words

17

Thesis of National University of Ireland., 2004


7.

Wightman, C. W., Shattuck-Hufnagel, S., Ostendorf. M., & Price, P. J., Segmental durations in the vicinity of prosodic phrase
boundaries, Journal of the Acoustical Society of America, 91. 1707-1717., 1992

8.

Pierrehumbert, J. & Hirschberg, J., The Meaning of Intonational contours in the Interpretation of Discours, in F. Cohen, J.
Morgan, and M. Pollack, (eds). Intentions in communication, MIT Press: Cambridge, MA. 271-311.,1990

9.

Strangert, E, Speech Chunks in Conversation: Syntactic and Prosodic Aspects, Proc Speech Prosody 2004, Nara, 305-308.,
2004

10. Fach, M.L A Comparison Between Syntactic and Prosodic Phrasing, in Proceeding of the European Conference on Speech
Communication and Technology., 1999
11. Cole, J., Mo, Y., & Baek, S., The role of syntactic structure in guiding prosody perception with ordinary listeners and every
speech, Language & Cognition Processes, 25:7. 1141-1177., 2010
12. Bachenko, J. & Fitzpatrick, E., A Computational Grammar of Discourse Neutral Prosodic Phrasing in English,
Computational Linguistics, 16(3), 155-170., 1990
13. Fitzpatrick, E., The Prosodic Phrasing of Clause-Final Prepositional Phrases, Language, 77(3), 544-561., 2001
14. Christophe, A., Millotte, S., Bernal, S., & Lidz, J., Bootstrapping Lexical and Syntactic Acquisition, Language & Speech
51(1&2), 61-75., 2008
15. Gerken, L, & McIntosh, B.J, Interplay of Function Morphemes and Prosody in Early Language, Development Psychology,
29(3), 448-457., 1993
16. Strangert, E., & Carlson, R., On Modeling and Synthesis of Conversational Speech, Proc. Nordic Prosody IX, Lund, 255-264.,
2006
17. Yeona Jeong The research of ambiguous relative deterministic clause constituting prosody, Korean Kyo-weon University
master thesis, 2006
18. Byeong-Seop Ahn, Korean prosody and phonology., 2010
19. Fant, G., Kruchenberg, A., & Ferreira, J. B., Individual variations in pausing: A study of read speech, PHONUM 9, 193-196.,
2003
20. Park Mun-Kyu Experimental phonetic research in France rhythm structure: Comparing free speech and read speech, Korean
Franceology Dissertation, 42. 51-66, 2003
21. Yoon Yeong-Sook Prosodic phenomena in spoken language in front of boundary syllable, Acoustic Science, 11(2), 137-150,
2004
22. Wellmann, C., Holzgrefe, J., Truchenbrodt, H., Wartenburger, I., & Hohle, B., How each prosodic boundary cue matters:
Evidence from German Infants, Frontier in Psychology, vol 3, 1-12., 2012
23. Seidl, A. & Cristia, A., Developmental changes in the weighting of prosodic cues, Developmental Science, 11(4), 596-606.,
2008
24. Mannel, C. & Friederici, A. D., Intonational phrase structure processing at different stages of syntax acquisition: ERP studies
in 2-, 3-, and 6-year-old children, Developmental Science 14(4), 786-798., 2009

www.tjprc.org

editor@tjprc.org