Phonetic transcription represents the phonetic properties of an actual or potential utterance in a written form. Firstly, it is necessary to have an understanding of what the phonetic properties of speech are. It is the role of phonetic theory to provide that understanding by constructing a set of categories that can account for the phonetic structure of speech at both the segmental and suprasegmental levels; how far it does so is a measure of its adequacy as a theory. Secondly, a set of symbols is needed that stand for these categories. Also required is a set of conventions that tell the reader what the symbols stand for. A phonetic transcription, then, can be said to represent a piece of speech in terms of the categories denoted by the symbols. Machine-readable phonetic and prosodic notation systems can be implemented in electronic speech corpora, where multiple linguistic information tiers, such as text and phonetic transcriptions, are mapped to the speech signal. Such corpora are essential resources for automated speech recognition and speech synthesis.
Claire Brierley and Barry Heselwood
Yu-Ying Chuang and R. Harald Baayen
Naive discriminative learning (NDL) and linear discriminative learning (LDL) are simple computational algorithms for lexical learning and lexical processing. Both NDL and LDL assume that learning is discriminative, driven by prediction error, and that it is this error that calibrates the association strength between input and output representations. Both words’ forms and their meanings are represented by numeric vectors, and mappings between forms and meanings are set up. For comprehension, form vectors predict meaning vectors. For production, meaning vectors map onto form vectors. These mappings can be learned incrementally, approximating how children learn the words of their language. Alternatively, optimal mappings representing the end state of learning can be estimated. The NDL and LDL algorithms are incorporated in a computational theory of the mental lexicon, the ‘discriminative lexicon’. The model shows good performance both with respect to production and comprehension accuracy, and for predicting aspects of lexical processing, including morphological processing, across a wide range of experiments. Since, mathematically, NDL and LDL implement multivariate multiple regression, the ‘discriminative lexicon’ provides a cognitively motivated statistical modeling approach to lexical processing.
David M. Bunis
The Ibero-Romance-speaking Jews of medieval Christian Iberia were linguistically distinct from their non-Jewish neighbors primarily as a result of their language’s unique Hebrew-Aramaic component; preservations from older Jewish Greek, Latin, and Arabic; a tradition of translating sacred Hebrew and Aramaic texts into their language using archaisms and Hebrew-Aramaic rather than Hispanic syntax; and their Hebrew-letter writing system. With the expulsions from Iberia in the late 15th century, most of the Sephardim who continued to maintain their Iberian-origin language resettled in the Ottoman Empire, with smaller numbers in North Africa and Italy. Their forced migration, and perhaps a conscious choice, essentially disconnected the Sephardim from the Spanish language as it developed in Iberia and Latin America, causing their language—which they came to call laðino ‘Romance’, ʤuðezmo or ʤuðjó ‘Jewish, Judezmo’, and more recently (ʤudeo)espaɲol ‘Judeo-Spanish’—to appear archaic when compared with modern Spanish. In their new locales the Sephardim developed the Hispanic component of their language along independent lines, resulting in further differentiation from Spanish. Divergence was intensified through borrowing from contact languages of the Ottoman Empire such as Turkish, Greek, and South Slavic. Especially from the late 18th century, factors such as the colonializing interests of France, Italy, and Austro-Hungary in the region led to considerable influence of their languages on Judezmo. In the 19th century, the dismemberment of the Ottoman and Austro-Hungarian empires and their replacement by highly nationalistic states resulted in a massive language shift to the local languages; that factor, followed by large speech-population losses during World War II and immigration to countries stressing linguistic homogeneity, have in recent years made Judezmo an endangered language.
We discuss here the considerable amount of phonological variation and change in European French in the varieties spoken in France, Belgium, and Switzerland, the major francophone countries of Europe. The data discussed here derive from the perceptual and especially behavioral studies that have sought to extend the Labovian paradigm beyond Anglo-American variable linguistic phenomena to bear upon Romance. Regarding France, what emerges is a surprisingly high degree of uniformity in pronunciation, at least over the non-southern part of the country, and most Southern French varieties are also showing convergence to the Parisian norm. Pockets of resistance to this tendency are nevertheless observable. The Belgian and Swiss situations have in common the looming presence of a supralocal and indeed supranational norm playing a role often attested in other discussions of standard or legitimized languages, that of the variety representing what commonly corresponds to the nonlocal. Indeed, it may be that Belgium and Switzerland typify the local–standard relation most often reported, while the French situation, because of its relatively leveled character, is less easily described as one of standardization.
In the source-filter theory, the mechanism of speech production is described as a two-stage process: (a) The air flow coming from the lungs induces tissue vibrations of the vocal folds (i.e., two small muscular folds located in the larynx) and generates the “source” sound. Turbulent airflows are also created at the glottis or at the vocal tract to generate noisy sound sources. (b) Spectral structures of these source sounds are shaped by the vocal tract “filter.” Through the filtering process, frequency components corresponding to the vocal tract resonances are amplified, while the other frequency components are diminished. The source sound mainly characterizes the vocal pitch (i.e., fundamental frequency), while the filter forms the timbre. The source-filter theory provides a very accurate description of normal speech production and has been applied successfully to speech analysis, synthesis, and processing. Separate control of the source (phonation) and the filter (articulation) is advantageous for acoustic communications, especially for human language, which requires expression of various phonemes realized by a flexible maneuver of the vocal tract configuration. Based on this idea, the articulatory phonetics focuses on the positions of the vocal organs to describe the produced speech sounds. The source-filter theory elucidates the mechanism of “resonance tuning,” that is, a specialized way of singing. To increase efficiency of the vocalization, soprano singers adjust the vocal tract filter to tune one of the resonances to the vocal pitch. Consequently, the main source sound is strongly amplified to produce a loud voice, which is well perceived in a large concert hall over the orchestra. It should be noted that the source–filter theory is based upon the assumption that the source and the filter are independent from each other. Under certain conditions, the source and the filter interact with each other. The source sound is influenced by the vocal tract geometry and by the acoustic feedback from the vocal tract. Such source–filter interaction induces various voice instabilities, for example, sudden pitch jump, subharmonics, resonance, quenching, and chaos.
Pier Marco Bertinetto
Speech rhythm is a popular research topic but a still poorly understood phenomenon. A critical assessment of the algorithmic tools developed in the last two decades to analyze rhythm in natural languages shows that they can at best lead to a topological arrangement of the languages to be compared, with no ambition to actually offer objective and absolute measures. Besides, all available tools are heavily influenced by any source of variability, in particular: speech rate, speech style (most notably, spontaneous vs. read), and even speaker identity. Although this shows their high sensitivity to the input details, it raises severe doubts as for the actual relevance of the comparative results obtained in the study of different languages. Future research will have to learn to overcome these weaknesses. Most importantly, readers should be alerted to the false idol of a common Romance rhythmic footprint. Close inspection of the prosodic characteristics of the main Romance languages indicates that the differences are indeed remarkable and likely to feed diverging rhythmical behaviors. Besides, one should take into account the vast intrafamily variability, up to the tiniest local vernaculars, which often diverge in extraordinary ways from the ‘roof’ language supposed to constitute a sort of common denominator.
Romanian stands out from its sister Romance languages through the conditions of its historical evolution. It has developed in isolation from the other Romance languages, and in cultural and linguistic contact with various non-Romance populations. The history of writing in Romanian, and the earliest preserved texts, dating from the 16th century, also reflect this rather unique heritage. The main dialectal division is marked geographically by the Danube river. The variety developed north of the Danube forms the Daco-Romanian group, while the variety developed south of the Danube includes Aromanian and Megleno-Romanian. The most characteristic changes affecting consonants in the development of Romanian include several patterns of palatalization (with or without affrication, depending on the segments’ place and manner of articulation), the emergence of labial-coronal clusters as part of a more general preference for labials, and rhotacism, a major feature of nonstandard varieties. Major vocalic changes include patterns of diphthongization, vowel raising before nasals and in the context of trills, which led to the development of two phonemic central vowels, /ɨ/ and /ʌ/. Many of these patterns show variation among different varieties. In all varieties of Romanian, vowel alternations are involved in morpho-phonological alternations. The stress pattern of modern Romanian follows the stress pattern of Balkan Romance. The standard and nonstandard varieties differ with respect to their intonation patterns, particularly in the case of yes/no questions.
A number of recent developments in phonological theory, beginning with The Sound Pattern of English, are particularly relevant to the phonology of compounds. They address both the phonological phenomena that apply to compound words and the phonological structures that are required as the domains of these phenomena: segmental and nonsegmental phenomena that operate within each member of a compound separately, as well as at the juncture between the members of compounds and throughout compounds as a whole. In all cases, what is crucial for the operation of the phonological phenomena of compounds is phonological structure, in terms of constituents of the Prosodic Hierarchy, as opposed to morphosyntactic structure. Specifically, only two phonological constituents are required, the Phonological Word, which provides the domain for phenomena that apply to the individual members of compounds and at their junctures, and a larger constituent that groups the members of compounds together. The nature of the latter is somewhat controversial, the main issue being whether or not there is a constituent in the Prosodic Hierarchy between the Phonological Word and the Phonological Phrase. When present, this constituent, the Composite Group (revised from the original Clitic Group), includes the members of compounds, as well as “stray” elements such as clitics and “Level 2” affixes. In its absence, compounds, and often the same “stray” elements, are analyzed as a type of Recursive Phonological Word, although crucially, the combinations of such element do not exhibit the same properties as the basic Phonological Word.
The Romance languages, despite their overall similarity, display interesting internal diversity which can be captured only very partially by looking at the six major standard languages, as typological databases often do. This diversity spans over all the levels of linguistic analysis, from phonology to morphology and syntax. Rather than making a long list of features, with no space to go much beyond their mere mention, the article focusses on just four main areas in a little more detail, trying to develop, if minimally, a discussion on their theoretical and methodological import. The comparison with the full-world typological background given by the WALS Online shows that the differences within Romance may reach the level of general typological relevance. While this is probably not the case in their rather mainstream segmental phonology, it surely holds regarding nominal pluralization and the syntax of negation, which are both areas where the Romance languages have often distanced themselves quite significantly from their common ancestor, Latin. The morphological marking of nominal plural displays four values out of the seven recorded in WALS, adding a further one unattested there, namely subtraction; the negation strategies, although uniformly particle-like, cover all the five values found in WALS concerning linear order. Finally, Romance languages suggest several intriguing issues related with head-marking and dependent-marking constructions, again innovating against the substantially dependent-marking uniformity characteristic of Latin.
Brazilian Portuguese is the native language of more than 200 million people living in Brazil. Spoken in South America since around the year 1500, Brazilian Portuguese has peculiar phonological traits, many of them variable. The extensive language contact that has taken place in Brazil caused Brazilian Portuguese to break up into regional dialects. Various phonological processes affect Brazilian Portuguese at the segmental and suprasegmental levels. Some of the processes target consonants, such as the regressive palatalization of /t, d/, the fricatization of /r/ in syllabic onset; some processes target vowels, such as the raising and lowering of unstressed /e, o/ vowels; others target the intonation of utterances, such as the rising of the nuclear stress of yes–no questions. The results of several empirical studies on varieties of Brazilian Portuguese show that not all of the processes correspond to change in progress in Brazilian Portuguese; some of them are stable variables. They also show that not every variable is present in all dialects and that some variables are socially salient and stigmatized. Compared to present European Portuguese, the phonology of Brazilian Portuguese seems to be conservative in some aspects, such as in the raising of vowels in unstressed, word-final syllables; innovative in others, such as in the vocalization of /l/ in syllabic coda.