¹ This paper is an expanded version of a chapter of my dissertation, Madurese phonology and morphology (Yale University, 1964; unpublished). Research was carried out in Indonesia from November 1960 to March 1962 with the aid of a Ford Foundation Foreign Area Training Fellowship. Madurese is a Malayo-Polynesian language spoken by about six million people in parts of East Java and the adjacent island of Madura; the dialect of East Madura, considered standard, is used here.

The symbols plus, dot, and hyphen indicate various types of morpheme boundary. R and Rf indicate a reduplication of the root initial and root final syllable respectively.

The phonemic system includes three series of stops (voiceless, voiced, aspirated) in five positions (labial, dental, alveolar, palatal, velar)—p, t, ṭ, c, k; b, d, ḍ, z, g; bh, dh, ḍh, zh, gh—but only four nasal positions—m, n, ñ, ŋ. In addition, there are w, j (high front glide), s, r, l, h, q (glottal stop). All but q occur single or geminated. The geminated aspirates are written bbh, ddh, etc. There are four vowels: i, u, ә, a. The vowel qualities vary greatly and depend, in a complicated way, on the preceding consonants. Loanwords have introduced forms which do not obey these rules; such vowels are written with acute and grave accents. N- is a prefix whose form (m, n, ñ, ŋ_i a, an) depends on the following morpheme.

² Most previous work on this subject, variously called language levels, social levels in language, social dialects, language types (Dutch taalsoorten), or status languages (Dutch rangtalen), has dealt with Javanese and Sundanese. The following works contain information on the topic.

S. Coolsma, Soendaneesche Spraakkunst (Leiden, 1904); L. C. Damais, ‘Les formes de politesse en Javanais moderne’, Bulletin de la Société des Études Indo-Chinoises 25.263–80 (1950); C. Geertz, The religion of Java 248–60 (Glencoe, 1960); J. Gonda, ‘The Javanese vocabulary of courtesy’, Lingua 1.333–76 (1948); E. C. Horne, Beginning Javanese 4, 8, 32, 41, 56, 95–6, 236–8, 341 fn., 355, 357 (New Haven, 1961); J. Kats, Spraakkunst en taaleigen van het Javaansch 182–203 (Weltevreden, 1912); id., Spraakkunst en taaleigen van het Soendaasch (Weltevreden, 1921–7); R. A. Kern, “t Lemes in ‘t Soendaasch’, Bijdragen 59.385–562 (1906); J. Kersten, Balische grammatica 7–14 (‘s Gravenhage, 1948); H. Kiliaan, Madoereesche Spraakkunst 3–5 (Batavia, 1897); id., Javaansche Spraakkunst 3–20, 55–62 (‘s Gravenhage, 1919); id., Madoereesch-Nederlandsch woordenboek (Leiden, 1904). R. M. Koentjaraningrat, A preliminary description of the Javanese kinship system 14–6 (New Haven, 1957); id., ‘The Javanese of south central Java’, Social structure in Southeast Asia 88–115 (G. P. Murdock, ed.; Chicago, 1960); E. M. Loeb, ‘Javanese word formation, high and low’, JAOS 64.113–26 (1944); C. Poensen, Grammatica der javaansche taal (Leiden, 1897); E. M. Uhlenbeck, De tegenstelling krama: ngoko: Haar positie in het javaanse taalsysteem (Groningen, Djakarta, 1950); A. H. J. G. Walbeehm, De taalsoorten in het Javaansch (Batavia, 1897).

³ A theory which might be of interest here but which has not been applied to situations of the Madurese type is presented by R. W. Brown and A. Gilman in ‘The pronouns of power and solidarity’, Style in language (T. A. Sebeok, ed.; Cambridge, 1960).

⁴ C. Geertz, The religion of Java 248.

⁵ For such a discussion on Javanese see Geertz 255–8.

⁶ There are of course homophonous forms belonging to different levels, e.g. apa K ‘what’, apa B ‘fruit of the sirih tree’; bazha B ‘steel’, bazha AT ‘tooth’.

Some words are in a category called kasar ghallu or kasar dhillu ‘very kasar’ (see H. Kiliaan, Madoereesche Spraakkunst 1.4); but these fall outside the level system here discussed. They correspond to the class of words in English called vulgar or dirty.

⁷ The equivalent Javanese categories are

E. C. Horne, Beginning Javanese 236, uses the term ‘humble krama inggil’ for krama anḍap.

The terms alus tinghi and alus manḍhap are not used by the Madurese themselves. These levels are simply called alus. Madurese who know Javanese often use the Javanese words in referring to the Madurese levels.

⁸ Called tәŋ-an by Kiliaan.

⁹ But subuŋ in this meaning is considered substandard by many speakers.

¹⁰ For a discussion of the semantic categories covered in Javanese see E. M. Uhlenbeck, De tegenstelling krama: ngoko: Haar positie in het javaanse taalsysteem 9–11.

¹¹ For a discussion of this in Javanese see Uhlenbeck 12–3.

¹² In Javanese and Sundanese, on the other hand, there are several patterns of submorphemic alternation in kasar and alus words. Some of these are still productive. For discussion and examples see Uhlenbeck 14–9 and 27–8 fn. 34, as well as the Javanese and Sundanese grammars mentioned in fn. 2.

¹³ See Horne 341 fn. for the same phenomenon in Javanese.

¹⁴ Similar to Javanese krama madya style.

¹⁵ Further investigation of this point is needed.

¹⁶ Equivalent to Javanese ngoko sae.

¹⁷ Mrs. Horne states (56) that the Javanese equivalents -(n)e and -(n)ipun may be used to express social differences when attached to a ngoko-krama (i.e. Madurese biasa) word. The usage does not seem to be true of Madurese, but needs investigation.

¹⁸ See Brown and Gilman for a history of similar changes in the second-person pronoun in Europe; and L. C. Damais, ‘Les formes de politesse en Javanais moderne’ 271, for a comment on Javanese.

¹⁹ Some writers fail to make the correct distinction between the equivalents of alus and alus tinghi.

²⁰ The Javanese figures are from A. H. J. G. Walbeehm, De taalsoorten in het Javaansch 30 and quoted in other works; the Madurese figures are from my own count of H. Kiliaan, Madoereesch-Nederlandsch woordenboek. In both cases figures represent the number of morphemes rather than the number of words. The number of ngoko or kasar morphemes is approximately equal to the number of krama or alus morphemes. Alus mandhap words are included in the alus tinghi figure. I have omitted facultative alus, variant and dialect forms, and morphemes limited to a single dialect.

Uhlenbeck (8) calculates a total of 2000 morphemes affected in Javanese. For less reliable figures, according to Uhlenbeck (26), see E. M. Loeb, ‘Javanese word formation, high and low’. It may be possible to arrive at figures for Sundanese by counting the entries in R. A. Kern, “t Lemes in ‘t Soendaasch’. J. Gonda, ‘The Javanese vocabulary of courtesy’ 375, claims that Sundanese has fewer words (morphemes?) affected than Javanese.