Prominence in Indonesian Stress, Phrases, and Boundaries
Total Page：16
File Type：pdf, Size：1020Kb
Wacana, Vol. 11 No. 2 (Oktober 2009): 197—225 Prominence in Indonesian Stress, phrases, and boundaries ELLEN VAN ZANTEN AND ROB GOEDEMANS Abstract Many (Western) languages have word-based stress, which entails that one, predictable syllable per word is more prominent than all the other syllables in that word. Some linguists claim that such stresses also occur in Indonesian. In this article, we set out to investigate that claim using experimental, phonetic methods. The results confirm our hypothesis that Indonesian lacks word-based stress. Yet, we do observe some kind of prominence pattern. In the last part of this article, we search for the phonological phenomenon that generates this pattern, exploring the level of the phrase to see whether phrasal accents or boundary markers are likely candidates. Keywords Indonesian, phonetics, prominence, stress, accent, and boundary tone. 1 Introduction1 In many course books for Dutch learners of Indonesian (see Teeuw 1984; Steinhauer 2001), you will find a short paragraph about word stress roughly stating that “stress is on the penultimate syllable unless the vowel in the penultimate syllable is schwa (pepet), in which case stress is final”. This implies that, for instance, garúda (ú indicating stress) is the correct way to pronounce 1 We would like to thank in particular Herman Poelman and the staff of the Erasmus Huis Taalcentrum in Jakarta for running some of the perception experiments. ELLEN VAN ZANTEN is a guest researcher at the Phonetics Laboratory, Leiden University. Her main interest is in the shape and function of prosody of Indonesian languages in particular. Together with Harry van der Hulst and Rob Goedemans, she is currently preparing "Stress patterns of the world; The data". E-mail: [email protected]. ROB GOEDEMANS is editor-in-chief for web communications as well as ICT-in-education consultant for the Faculty of Humanities at Leiden University. In the recent past, he conducted research in the phonetics and phonology of stress. When regarding the Indonesian language, the research was conducted with Ellen van Zanten. On a small scale, he is still active in typological investigations into the variety and patterning of word stress. E-mail: R.W.N.Goedemans@ hum.leidenuniv.nl. © 2009 Fakultas Ilmu Pengetahuan Budaya, Universitas Indonesia 01-Goedemans.indd 197 3/31/2010 4:48:58 PM 198 Wacana, Vol. 11 No. 2 (Oktober 2009) ELLEN VAN ZANTEN and ROB GOEDEMANS, Prominence in Indonesian 199 the Indonesian word for eagle. Strangely enough, this almost absolute certainty about the location of stress in Indonesian is only reflected in some theoretical articles about the phonological derivation of word stress in Indonesian (for example, Cohn 1989; Cohn and McCarthy 1994). Most primary descriptive sources we know of are much less resolute in their claims regarding the location of word stress (see Halim 1974), make a different claim (see Samsuri 1971: stress is final) or confirm that the rule given above is a tendency, but not an absolute fixed rule (see Laksman 1994). Some authors, like De Hollander (1984: 27-28) and Alieva et al. (1991:64) claim that, in some cases, stress is maintained on the penultimate syllable of the base word when a suffix is attached to it (for example, bántuan vs. bantúan ‘assistance’, pernyátaan vs. pernyatáan ‘declaration, expression’. Moreover, with respect to the stress ability of the schwa, Van Zanten and Van Heuven (2004: 10) found that stress on the first, second or third syllable were all quite acceptable forsebelum /səbəlum/ ‘before’. Stress on the prefinal syllable was preferred only when this syllable was heavy, as in anaknya ‘his child’. Uncertainty concerning stress position in Indonesian seems to be reflected in difficulties that speakers of Indonesian have with word stress when learning a foreign language. For instance, it is by no means uncommon to hear Indonesian speakers, otherwise extremely fluent in Dutch, make surprising (for their level of command) mistakes in the stressing of Dutch words. One could hear them say inpákken ‘to wrap up’ and Goedemáns ‘a family name’ just as easily as the correct forms ínpakken and Góedemans. Given the fact that a basic characteristic of word stress is that it always occurs on one and the same syllable of a given word, we may wonder about the linguistic reality of this phenomenon for speakers of Indonesian, and ask ourselves the fundamental question: Does word stress exist at all in Indonesian? To understand the ins and outs of this question, and the details of the experiment we conducted to answer it, we must first make a little digression into the phonetic and phonological background of the phenomena we know as stress and accent. 1.1 Word stress: background In many languages, the syllables that form words are not all equally “strong”. Often, some of them are heard as longer, louder, clearer or higher in pitch than the others. Usually, one of the syllables in a given word is the most salient (prominent) because it outclasses all the others in one or more of the aforementioned phonetic properties. This single syllable is said to bear the “main stress”. In some, but by no means all, stress languages, a further division can be made with respect to the rest of the syllables in the word. The more salient of these carry “secondary stress” while others are unstressed. In the majority of the languages that feature stress, the patterns of main and secondary stresses we observe in the words are far from random. The location for main stress is usually limited to one (or a few) preferred position(s) in a given language, and secondary stress (also known as rhythm) usually 01-Goedemans.indd 198 3/31/2010 4:48:58 PM 198 Wacana, Vol. 11 No. 2 (Oktober 2009) ELLEN VAN ZANTEN and ROB GOEDEMANS, Prominence in Indonesian 199 alternates (strong – weak or weak – strong) over the rest of the syllables. This hints at organization on a linguistic level. Word stress in these languages is governed by rules that predict the location of the main and the secondary stresses for most (but not necessarily all, see below) of the words in the language. The speakers of the language know these rules subconsciously. In the remainder of this article, we will focus on main stress only. We will not go too deeply into “metrical phonology”, the phonological theory that deals with the set of rules that derives all the possible stress patterns of natural languages. We simply say that such rules exist, and that they prescribe a single location for main stress in (most of) the words in the language. The rules usually refer to the edges of the word and/or the internal make-up of the syllables to determine that location. In Yimas (Foley 1991), for instance, stress is invariably located on the first syllable of the word, while in Murik (Abbot 1985) stress is located on the first syllable of the word that contains a long vowel, or the first syllable of the word when there are no long vowels; see the examples in (1): (1) Yimas Murik ˈawtmayŋi ‘sugarcane’ anənpʰaˈřɛːtʰ 'lightning’ ˈyamparan ‘stand up’ ˈnuŋgunsařagatʰ ‘green’ In some languages, not all the words have predictable, rule-based stress. Speakers of such languages must learn the stress locations of these so-called “lexically stressed” words. In every case, however, whether stress is lexical or predictable by rules, the main stress for any given word is on exactly the same syllable every time a speaker pronounces that word.2 With this remark, we touch upon a universal principle that governs the rules of metrical phonology. Stress rules conspire to select one and only one syllable to act as the head of the word. A second universal is that there can be only one head. The rules thus lead to exactly the same syllable every time they are applied to a given lexical element. Therefore, if we observe languages in which many words show variation in the syllable that is selected for main stress we must doubt whether rules were applied at all. If we are in doubt about the application of rules, we should be in doubt about the phenomenon we are dealing with, because stress and rules are inseparable partners. Yet, we do hear some form of syllable prominence when we listen to Indonesian speech. Suppose we are indeed dealing with a language that has no rules, and therefore, no stress. Then what could be the phenomenon that governs the prominence we hear? To answer this question, we must separate the word- based phenomenon of stress from phrase-based phenomena like accent. 2 Readers who are interested in the exact details concerning these rules are referred to Hayes’ (1995) excellent expose of stress parameters, or to, for instance, Goedemans, Van der Hulst, and Visch (1996). 01-Goedemans.indd 199 3/31/2010 4:48:58 PM 200 Wacana, Vol. 11 No. 2 (Oktober 2009) ELLEN VAN ZANTEN and ROB GOEDEMANS, Prominence in Indonesian 201 1.2 Accent As mentioned in the previous section, the phonetic correlates of stress, as first defined by Fry (1955, 1958, 1965) are pitch, intensity (‘loudness’), duration and vowel quality. However, pitch is also the phonetic cue for intonation, in which accent is one of the key players. The speaker, to signal to the listener that some part of the utterance contains new information, can, for instance, use accent. The accented part is said to be in focus (Ladd 1980). In English, the accent is realized by a pitch movement on the prosodic head of the focus domain (Bolinger 1958), like in (2) where coffee is in focus; an accent-lending pitch movement should be steep, sufficiently large and in a specific position in the (stressed) syllable (Van Heuven 1994: 19).