Prenasalized stops, that is, homorganic nasal+stop elements that behave as single phonological segments, raise a number of interesting questions concerning the relationship between phonological units and timing in speech. Do complex phonetic elements of this kind occupy the same duration as simpler elements, such as plain stops or nasals? Do prenasalized stops have the same timing pattern as a phonological sequence of nasal plus stop? How do prenasalized stops act with respect to rules which adjust the duration of neighboring segments? For example, would a vowel before a prenasalized stop be shortened by the widespread rule which shortens a vowel in a closed syllable (Maddieson 1985)? It has been argued that the status of prenasalized stops as single segments is directly related to their duration. They have been defined as nasal+stop sequences with the duration typical of other single segments (Herbert 1986). He, and Sagey (1986), in her dissertation on complex segments, both indicate that they would expect phonological consonant sequences to have longer durations than single segments regardless of whether the single segments are phonetically simple or complex. On the other hand, Ladefoged and Maddieson (1986) suggest that there is no demonstrated phonetic difference in timing between nasal+stop sequences and prenasalized stops. Purported language-internal contrasts between these elements actually involve a difference between geminate and single nasals before stops (as in Sinhalese), or between syllabic and non-syllabic nasals before stops. They suggest that deciding if a nasal+stop element is a prenasalized stop is not a phonetic issue but one which concerns solely the phonology of the language in question.

From 18 to 21 August 1989 approximately 120 members of the Association (and a few guests) met in Kiel, West Germany, to work on revisions of the International Phonetic Alphabet (IPA). There had been a great deal of preparation for this convention, involving the soliciting of opinion by mail on a wide range of topics, and considerable research on the present status of the IPA. A previous issue of this journal, JIPA 18(2), collected a number of these preparatory opinions. At the convention itself there were five major working groups which met almost continuously for the first two days. Three groups considered consonants, vowels and suprasegmentals; the other two considered computational aspects of the IPA, and the needs of speech pathologists and others for extensions of the IPA. In addition there were groups concerned with the principles on which the IPA should be based, the form of presentation of the IPA, past successes and failures, and methods of illustration of the IPA. The groups reported back to the whole at intervals, and on the last day the convention met in a series of plenary sessions to consider and vote on the final working reports. The following is a compilation of six of the group reports reflecting the results of the discussions and votes of the plenary sessions. The Association is very grateful to the co-ordinators and co-chairs of the groups. Most of the wording below is taken directly from their reports; in accordance with the traditions of the Association, this report is presented without attribution to particular authors.

SAM, an international project funded by the European ESPRIT initiative, is concerned with the methodology of assessment of the speech interface with computers. The applications at which the SAM Project aims require that recorded-speech corpora be both defined and created on a European scale. Within the present stage of development (called the “extension phase”) some three different aspects of this work have to be specified and developed:

The Association must define and communicate its officially accepted symbols to other organizations and professionals, including publishers, dictionary makers, computer-systems specialists, and programmers. Accordingly, it was decided that each accepted symbol or diacritic should be assigned a unique numerical equivalent, independent of computer-coding conventions, and a unique name which provides a mnemonic description of the character shape. Symbols that have been used in earlier versions of the IPA, but deleted in later revisions, should retain a number, and name, for reference purposes. The numerical equivalent (IPA Number) is to be regarded as a communication-interchange standard, to serve as a basis for creating computer-code translation tables from various phonetic-character-set software to the common IPA Number. This IPA Number is not implemented directly in computer format (for example, ASCII), but is expressed as a simple numerical directory of digit triples to serve as a unique reference. The IPA Number can also serve as a typesetter's guide to the Phonetic Symbol Chart. The systematic numerical listing represents the IPA symbols as presented in the Chart. Therefore, the first digit of the triple indicates the symbol category; inn for accepted IPA consonant symbols, 2nn for former IPA consonant symbols and non-IPA consonant symbols, 3nn for vowels, 4nn for segmental diacritics, 5nn for suprasegmental symbols; 6nn-8nn are reserved for future specification (e.g., for symbols for voice quality settings or for pathological speech). The digit triple 9nn has the function of an escape sequence from IPA-symbol mode into procedures definable for special applications (e.g., Roman, Greek, Cyrillic orthographies, “Comment” mode, etc.).

We have designed special IPA characters for IPA phonemic transcriptions of the following languages: E[nglish—RP and GA], F[rench], G[erman], I[talian], P[olish], R[ussian], Sp[anish] and Sw[edish], using two different programs which can be run under MS-DOS (BIOS) on an IBM PC or compatible. One, the ChiWriter—henceforth CH—is a complete word processor whilst the other, the LETTRIX—henceforth L—has to be combined with a word processor, such as WordStar, which we have used. In either case, we have designed our characters for a 9-pin dot-matrix printer, though CH also allows font design for a 24-pin or a laser printer. A Quality option has to be chosen, which reduces the printing speed by a factor of two with L and four with CH.

The usual description of vowels in respect to their “phonetic quality” requires the linguist to locate them within a so-called “vowel space,” apparently articulatory in nature, and having three dimensions labeled high-low (or close-open), front-back, and unrounded-rounded. The first two are coordinates of tongue with associated jaw position, while the third specifies the posture of the lips. It is recognized that vowels can vary qualitatively in ways that this three-dimensional space does not account for. So, for example, vowels may differ in degree of nasalization, and they may be rhotacized or r-colored. Moreover, it is recognized that while this vowel space serves important functions within the community of linguists, both the two measures of tongue position and the one for the lips inadequately identify those aspects of vocal tract shapes that are primarily responsible for the distinctive phonetic qualities of vowels (Ladefoged 1971). With all this said, it remains true enough that almost any vowel pair of different qualities can be described as occupying different positions with the space. Someone hearing two vowels in sequence and detecting a quality difference will presumably also be able to diagnose the nature of the articulatory shift executed in going from one vowel to the other.

This paper is intended to provide a general orientation regarding the articulatory phonetics of the languages involved in the SAM project, as described in Wells (1987): it comprises a brief statement of the phonemic system of each language (the consonants and vowels, and how they are transcribed), together with notes on the most salient realizational and suprasegmental characteristics. As in Wells (1987), the objective is to assign a computer-coded phonemic notation, referred to as SAM-PA, using only the printable symbols from the 7-bit ASCII series.

The following errata appeared in the article by Jassem and Lobacz in the Journal 19.1:

Page 17, line 30: for F [e] read: F [e].

Page 18, line 24: for Sw /ɦ/ read: Sw /ɧ/.

Page 21, line 20: delete: ones