next up previous contents index
Next: Prosodic information Up: Lexical surface information Previous: Orthographic information

Pronunciation information

Pronunciation information is much more application specific (and indeed theory specific) than orthographic information. Standardly, information about phonemic structure  is included in the form of a phonemic transcription   of a standard canonical  or citation form  pronunciation, i.e. the pronunciation of a word in isolation in a standard variety of the language. Often the phonemic transcription  is enhanced by including prosodic information  such as the stress  position (Dutch, English, German), the type of tonal accent  (Swedish), syllable  and word boundaries  in compound  words, and word and phrase boundaries in phrasal idioms . Morphological  information (morph boundaries , as well as the boundaries of words and phrases) is relevant to stress  patterning, and is sometimes also included.

A particularly thorny question is the inclusion of information about pronunciation variants, of which there are two main types, rule-governed allophonic and phonostylistic variants, and idiosyncratic lexical variants. The following rules of thumb can be given:

Although phoneme  is a technical term with somewhat different definitions in different theoretical contexts, and although there are technical arguments due to Generative Phonology [Chomsky & Halle (1968)] which show that the notion of phoneme  leads to inconsistencies, the core of phoneme  theory is relatively standard. In linguistics handbooks, the phoneme  is commonly defined as the minimal distinctive (meaning-distinguishing) unit (temporal segment) of sound. In the following fairly standard definition, the distinctiveness criterion is implicit in the concept of a system; the concept of a sound (= phone, allophone) covers possible variants of a phoneme  (e.g. English aspirated  word-initial /p/ as opposed to unaspirated  /p/ in the context /sp.../ [Crystal (1985), p. 228,]:

PHONEME (PHONEMIC(S)) The minimal unit in the sound SYSTEM of a LANGUAGE ...Sounds are considered to be members of the same phoneme if they are phonetically similar and do not occur in the same ENVIRONMENT.

A fairly complete definition is thus based on distinctiveness, minimality, phonetic similarity and distributional complementarity. Phoneme  definitions are differential or relational definitions, illustrated by the notion of minimal difference between two words in minimal pairs such as the items in the set of English words pin-tin-kin-fin-thin-sin-shin-chin-bin-din-gin-win-Lynne-Min-Nin, (in standard SAMPA  computer readable phonemic transcription:   /ptex2html_wrap_inline45169n - ttex2html_wrap_inline45169n - ktex2html_wrap_inline45169n - ftex2html_wrap_inline45169n - tex2html_wrap_inline45321tex2html_wrap_inline45169n - stex2html_wrap_inline45169n - Stex2html_wrap_inline45169n - ttex2html_wrap_inline45205tex2html_wrap_inline45169n - btex2html_wrap_inline45169n - dtex2html_wrap_inline45169n - tex2html_wrap_inline45337n - wtex2html_wrap_inline45169n - ltex2html_wrap_inline45169n - mtex2html_wrap_inline45169n - ntex2html_wrap_inline45169n/) (the last three are names). Phonemes  defined in this way are further classified as bundles of phonological distinctive features. Operationally, phonemes  are defined by procedures of segmentation  and classification (reflected, for example, in the recognition and classification components of automatic speech recognition systems):  

In contrast to orthographic representations, which for social and cultural reasons, are highly standardised common knowledge, lexical representations of pronunciation are theory and application specific. The most widely used representations in pronouncing dictionaries for human use, such as in foreign language teaching, and in spoken language systems,  are phonemic transcriptions .

Phonemic descriptions are available for several hundred languages, and phonemic transcriptions   based on these are suitable for constructing roman orthographies for languages which have orthographies based on different principles (e.g. syllabic or logographic) or no orthography at all. For a given language, phonemic descriptions differ peripherally (for instance, it is controversial whether diphthongs  and affricates  are to be analysed as one phoneme  or two?). Phonemes  are in general the units of choice for practical phonological transcriptions   in spoken language system lexica. Other, more specialised types of representation such as the feature matrix representations required by all modern phonological descriptions, and autosegmental lattice representations, or metrical tree graph and histogram  representations [Goldsmith (1990)] are increasingly finding application in experimental systems [Kornai (1991), Carson-Berndsen (1993), Kirchhoff (1996), Church (1987b), Church (1987a)] because of their richness and their more direct relation to the acoustic signal, in contrast to phonemic representations. However at the lexical level, they can generally be calculated relatively easily from the more compact, but less general, phonemic representations. Because of the widespread use of phonemes , the concept is discussed in more detail below; for fuller explanations, textbooks on phonology should be consulted.

The central question in phonological lexical representation, in cases where the notion of phoneme  alone is not fully adequate, is that of the level of representation (level of description, level of abstraction). There are three main levels, each of which is an essential part of a full description, and which needs to be evaluated for all but the simplest applications, morphophonemic, phonemic, and phonetic, which are characterised below.

The morphophonemic level  provides a simplification of phonological information with respect to the phonological level; the simplifications utilise knowledge about the morphological  structure of words, and permit the use of morphophonemes , (a near-synonym is archiphoneme ) which stand for classes of morphologically  and phonologically related phonemes .

A standard example of a morphophoneme  is the final obstruent in languages with final obstruent devoicing , including Dutch and German. For example, the phonemic representation German Weg /ve:k/ `way' - Wege /ve:gtex2html_wrap_inline45173/ `ways' corresponds to a morphophonemic  representation {ve:G} - {ve:G+tex2html_wrap_inline45173}, which simplifies the description of the stem  of the word. The morphophoneme  {G} stands for the phoneme  set {/k/, /g/}, and selection of the appropriate member of the set (the appropriate feature specification) is triggered by the morphological  boundary and neighbouring phonological segments. Alternatively the morphophoneme  may be said to consist of the underspecified feature bundle shared by /k/ and /g/, or more technically, the feature bundle which subsumes the feature bundles of /k/ and /g/.

An example from English is the alternation /f/ - /v/ in plural formation in one class of nouns, as in knife /natex2html_wrap_inline45169f/ - knives /natex2html_wrap_inline45169vz/, which can be represented morphophonemically  as {natex2html_wrap_inline45169V} - {natex2html_wrap_inline45169V+z}. The morphophoneme  {V} stands for the phoneme  set {/f/, /v/}. Here, too, selection of the phoneme  (specification of the underspecified subsuming feature bundle) is determined by the morphological boundary  and the phonological properties of neighbouring segments.

A corresponding level is necessary for the description of spelling: cf. variations such as English y-ie in city - cities, or German s-ss-ß as in Bus - Busse, Kuß (Kuss in the new orthography) - Küsse and Fuß - Füße.

Morphophonemic  representations augmented by realisation rules are a useful compression technique for reducing lexicon size :

For requirements such as these, the use of morphophonemic  representations, supplemented by morphological  construction rules and morphophonemic  mapping rules is recommended ([Koskenniemi (1983)], [Karttunen (1983)], [Ritchie et al. (1992)], [Bleiching et al. (1996)] for descriptions of various practical approaches).

There are no standard conventions for the representation of morphophonemes , whether computer readable or not (but see the SAMPA  alphabet for French, Appendix B); capital letters are often used in linguistics publications. Note that this use of capital letters at the morphophonemic  level should not be confused with the use of ASCII upper case codes in the SAMPA  alphabet at the phonemic level.

Citations  of morphophonemic  representations are often delimited with brace brackets {...}.

The phonemic level is a standard intermediate level corresponding to criteria outlined in more detail below. The standard European computer readable phonetic alphabet is SAMPA  (Appendix B): this alphabet is used for the main languages of the European Union, and is recommended for this purpose. The internationally recognised standard alphabet for phonemic representations is the International Phonetic Alphabet (IPA ). The IPA alphabet is used for the most part in the text of this handbook, and is shown in Appendix A. One of the main functions of the International Phonetic Association  since its inception over 100 years ago has been to coordinate and define standards for this alphabet.

Until relatively recently, the special font used for the IPA  has made it difficult to interface it with spoken language systems , and for this reason a number of computer-readable encodings of subsets of the IPA  have been made for various languages [Allen (1988), Esling (1988), Esling (1990), Jassem & tex2html_wrap51367 obacz (1989), Ball (1991)]. The standard computer phonetic alphabet for the main languages of the European Union is the SAMPA  alphabet, developed in the ESPRIT  SAM  and SAM-A projects [Wells (1987), Wells (1989), Wells (1993b), Wells (1993a), Llisterri & Mariño (1993)]; see also Appendix B. SAMPA  is widely used in European projects, both for corpus transcription  and for lexical representations (see also the chapter on Spoken Language Corpora).

However, there is a standard numerical code for IPA  symbols (cf. [Esling (1988), Esling (1990)]; Appendix B), and developments in user interfaces with graphical visualisation in recent years are leading to the increasing use of the IPA  in its original form, particularly in the speech lab software which is used in spoken language system  development.

Citations  of phonemic representations are standardly delimited by slashes /.../.

At the phonetic level further details of pronunciation, beyond the phonemically minimal features, are given. Since the relation between the phonemic and the phonetic level can be described by general rules mapping phonemes  to their detailed realisations (allophones ) in specific contexts [Woods & Zue (1976)], it is strictly speaking redundant  to include these regular variants in a lexicon. However, for reasons of efficiency, detailed phonetic word models  for speech recogniser training or for speech synthesis may be calculated using phonological rules and stored. Essentially this is a software decision: whether to use tables (for efficiency of lookup) or rules (for compactness and generality) for a given purpose.

A specific version of the phonetic level of transcription  is phonotypic transcription , defined as a mapping from the phonemic level using regular phonological rules of assimilation , deletion , epenthesis  [Autesserre et al. (1989)]; this level is frequently used for generating additional word models  to improve speech recognition . Since the amount of phonetic detail which can be processed depends heavily on the vocabulary size   and the number of phonological rules which are considered relevant, no general recommendation on this can be given.

There is no widely used standard ASCII encoding of the entire IPA  for computer readable phonetic representations and therefore no recommendations can be given on this. A proposal by John Wells, the originator of SAMPA,  is under discussion. Currently, individual laboratories use their own enhancements of phonemic representations. However, the fuller encodings mentioned in connection with the phonemic level of transcription   are eminently suitable for interface purpose at the phonetic level, and will no doubt be increasingly used where more detailed phonetic information is required.

Citations  of phonetic forms are standardly delimited by square brackets [ ...].

Chapters 4, 5, 3 should also be consulted in respect of levels and types of corpus representation.

next up previous contents index
Next: Prosodic information Up: Lexical surface information Previous: Orthographic information

EAGLES SWLG SoftEdition, May 1997. Get the book...