Search

Links are needed to bridge the gap between the analysis of speech as a set of discrete, ordered but durationless linguistic unit and analyses of the continuously changing acoustic signals, defined along a time axis. Current recognition and synthesis devices do not make good use of the structure imposed by speech production processes on the mapping between an allophone sequence and the many possible associated speech signals. A quantitative, flexible articulatory time framework has been developed as a contribution to the new kinds of phonetic descriptions needed. Units of articulation for allophones of the phonemes of British English and methods for linking adjacent allophones are proposed. Tentative specifications for a sub-set are offered, based on a review of published findings for natural speech. Articulatory schemes are taken to be organised with reference to particular events E . Pairs of events need to be appropriately coordinated in time. The two events may relate to inter-articulator coordination between two different quasi-independent articulators or to the durational extent of a statically maintained state for a single articulator. The coordination between the two events is expressed through the duration D of the time interval between them. Six examples are given of the construction of a complete articulatory time plan for an English sequence. This forms the first stage for a computer-implemented model of the articulatory, aerodynamic and acoustic processes of speech production. The synthetic speech output from the model is given acoustic variations intended to mimic those arising in natural speech due to a speaker's choice of options, including a change in rate of speech. This is achieved in the modelling by altering one or more D values in the articulatory time plan and by dispensing with some optional actions. The variability of multiple repetitions by a real speaker can be introduced into the synthetic speech by perturbing the D values. The model needs to be matched to specific real speakers in order to assess the extent to which it is realistic in its simulation of the variation and variability of acoustic pattern features for natural speech and the extent to which covariations can be predicted with it.

Two sets of stimuli, /χi:d/–/ħi:d/ and /ʁi:d/–/ʕi:d/, were synthesized to investigate the role of F1 in distinguishing the uvular and pharyngeal fricatives, /χ/ vs. /ħ/ and /ʁ/ vs. /ʕ/, in Arabic. Another two sets of stimuli, /χi:d/–/ħi:d/ and / ʁ i:d/–/ʕi:d/, were constructed to find out whether F1 plays a role in the distinction between emphatic and nonemphatic uvular and pharyngeal fricatives, /χ/– /ħ/ and / ʁ /–/ʕ/. Identification functions for ten native speakers of Jordanian Arabic showed that, with raised F1, listeners moved from uvulars to pharyngeals in a discrete manner. The distinction between emphatic and non-emphatic items is mainly related to lower F2 in the emphatic allophone, but this is also accompanied by some F1 raising. The results of these tests agree at the perceptual level with the model proposed by Stevens and Klatt relating the values of the lower four formants to the place and degree of constriction in the back cavity during the production of back consonants. The smaller and further back the constriction, the lower F2 and the higher the F1 values.

Three experiments measured tongue position (electropalatography) and jaw movements (strain gauge) during the production of nonsense phrases composed of repetitions of the syllable “da”. In experiment 1, jaw acceleration was found to vary with allophonic variation rather than with stress level. Experiment 2 found that four acoustic patterns corresponded to several jaw displacement and acceleration patterns. Experiment 3 examined the relationship between tongue palate contact, jaw movement and the acoustic wave. The pattern of tongue-palate contact was related to jaw movement and strongly related to acoustic pattern. Listeners judged specific acoustic patterns to be roughly associated with the production of the phonemes /d/, flap and tap. An additional allophone of /d/, a short /d/ was often utilized for unstressed syllables during rapid speech. This allophone was not perceptually distinguished from the normally long /d/.

A language’s use of the phonetic vowel space depends not only on how many vowel phonemes the language has, but on how each phoneme varies allophonically across contexts. This study tests the hypothesis that Japanese vowel allophones, measured from a wide range of contexts, will not fill the vowel formant space. This was predicted because Japanese has few vowel phonemes, distributed unevenly in the vowel space, and has no obvious processes of vowel reduction. Formant frequencies of vowel tokens from word lists and from read texts were compared for 7 speakers. These data show that, in Japanese, vowel allophones in prose fill in the vowel formant space more than allophones in word lists do, mainly as a result of centralization of the prose tokens. The use of the total formant space is determined in part by the distribution of phonemes, and in part by allophone centralization.

Much has been accomplished in the area of speech technology in order that machines can talk to people. Various stringing techniques such as phoneme, allophone and diphone have been demonstrated to give those machines virtually unlimited vocabularies. The subject to be covered in this paper is not a new speech construction technique, but how a construction technnque can be combined with a speech synthesizer to allow the machine to sing. The system which will be discussed uses linear predictive coding (LPC) as its synthesis method and allophiones as its constructoin technique. In this paper, each segment of the system will be discussed in some detail.

0. Introduction. Tuyuca evidentials are a mandatory feature of the independent verb, combining information about person and tense with statements about the evidence on which the report is based. Tuyuca speakers must indicate that they obtained their information in one of the following ways: visually; through a sense other than the visual; through evidence of the state or event; by being told about the state or event; or by assuming what happened. Moreover, these five evidential categories are arranged hierarchically according to speaker preference. The following discussion demonstrates the use of these evidential categories, as well as describing a feature analysis of them. First, however, the reader deserves a brief introduction to the nature of the Tuyuca language. Tuyuca is the mother tongue of approximately 700 people who live along the Papuri and the Tiquie rivers and their tributaries in Colombia and Brazil. It is classified by Castellvi (1962:241-43) as a Tucanoan language and by Waltz and Wheeler (1972:128) as a member of the Central Eastern Tucanoan branch of that language family. The Tuyuca phonemes are /p t k b d g s r w y h ii u e a o/ plus /~/ (nasalization) and / '/ (stress). Syllable types are V and CV. Nasalization in the Tucanoan languages does not operate solely on the vowel, but on the syllable or morpheme, with the result that [m] is an allophone of /b/, [n] of /d/, [j] of /g/, and [f] of /y/. In Tuyuca, nasalization is a feature of the morpheme, so that every morpheme is either entirely nasalized or entirely oral. Stress is concurrent with high pitch and occurs once per word. The basic word order in Tuyuca is Subject-Object-Verb, but other orderings occur as discourse considerations demand. Short sentences of

Eight normally developing preschool children manifesting incomplete mastery of /r/ articulation repeated three times at four week intervals sentences containing the allophones [ r ] embedded in various consonantal contexts. Two judges evaluated /r/ allophone production as correct or incorrect. Children whose /r/ production improved showed greater success with [ r] articulation than with []. Results indicate that normally developing children may be distinguished from more slowly developing children on the basis of differential success with production of various /r/ allophones.

The present paper makes a case for retention of the (Revised) Alternation Condition in Lexical Phonology, a theory in which any single rule which presebts beytralizing, lexical effects restricted to derived forms along with allophonic, derivationally unterstricted dffects is cominally impossible. However, Korean obstruent palatalization does display both of these properties, whereby /t, th/ neutralize with /ĉ, ĉh/ before [i], but only if the [i] occurs in another morpheme (cf. /path + i/→ [paĉhi] ‘field-SUBJ’ vs. monomorphemic [pathi] ‘endure’), whereas / s / acquires the palatal allophone [∫[ before [i] both within ([∫i] ‘poem’) and between (/os + i/→ [o∫i] ‘cloth-SUBJ’) morphemes. The Revised Alternation Condition alone imposes just this restriction on a single palatalization rule functioning both lexically and post-lexically in Korean, which suggests that its removal from the theory is premature.

Taking them one by one, it is generally agreed that [] are the allophones of /w r j/ occurring only before nasal vowels. Ward (1952) noticed the nasalization of these approximants before nasal vowels but was rather confused by their orthographic forms. We have observed that all instances of [ɲ] occur when [j] is followed by a nasal vowel. This non-syllabic palatal nasal is therefore an allophone of /j/ and it is different from the syllabic one which is homorganic with the following consonant. Examples of this non-syllabic [ɲ] can be seen in the following words

I Cheyenne, a member of the Algonquian language family, is spoken in central Oklahoma and on the Northern Cheyenne Reservation in southeastern Montana. The phonemes of Cheyenne are: p, t, k, ?, s, s, h, m, n, v, a, e, and o. /e/ varies phonetically from [i] to [e], centering on [t]. / o/ varies from [u] to [o]. /t/ assibilates to [ts] preceding /e/. /h/ has the allophone [s] between /e/ and /t/, and [s] between /e/ and /k/. [x] is an allophone of /g/ and /h/; see Leman (1980a). In this article, Cheyenne transcriptions are "orthographic" (a practical transcription very close to a classical phonemic level), unless otherwise noted. A dot over a vowel indicates that it is voiceless. For descriptions of Cheyenne phonological rules see Frantz (1972a) and Leman (1979). This study, particularly in the basic analysis of the pitch sandhi, has benefited from discussions with Richard Rhodes. I am indebted to him and to several Cheyenne speakers who somehow survived hours of my efforts to hear pitches on words they patiently pronounced.

The presence or absence of the dental allophone of /n/ in /nθ/ sequences was examined under conditions of varying juncture boundaries falling between /n/ and /θ/. Six subjects read a list of sentences in which the /nθ/ sequence was embedded such that juncture boundaries of seven different ranks occurred between /n/ and /θ/. Results indicate that dental coarticulation, as indicated by dental contact during /n/, is not impeded by the interposition of most junctures. On occasion, when the juncture was marked by a long pause, the dental contact was first made, broken, and re-established after the pause. It would appear that absence of dental coarticulation does not usefully mark most higher-level junctures studied.

Cashibo is a member of the Pano family of languages, centered in the jungle region of Peru, and extending over the border into Brazil, and Bolivia. The Cashibo speakers live in Peru along the Aguaitia River and some of its tributaries, including the Shambuyacu, the location in which the author did her field work. The Cashibo tribe is comprised of a few hundred families sporadically scattered along the rivers. The data for this paper were gathered first at a military school near Lima from Gregorio Estrella, a Cashibo youth, and later supplemented by three field trips to the tribe totaling about eighteen months. The field work was done in collaboration with Gloria Gray, under the auspices of the Summer Institute of Linguistics. One of the pan-Pano linguistic features seems to be a nasal allophone actualized with a stop off-glide. See Henry Osborn for Amahuaca, IJAL 14.188-90 (1948). Compare also non-Pano language with similar allophone, Jules Henry for Kaingang of southern Brazil, IJAL 14.194-204 (1948). 1.1 The consonant phonemes may be presented in four series in terms of articulation from front to back

1. There are twenty six consonant phonemes in Ocaina which contrast as to stops, fricatives, assibilants and nasals. The nasals, perhaps, exhibit the most interesting sound features of the language in that they contrast as to fortis and lenis in their three points of articulation. The fortis feature may be actualized either as length or as tense articulation or as both. (The fortis phonemes will be symbolized with capital letters.) /1/ and /f/: /ptfsoNo/ you are saving; /plsofio/ you are digging. /N/ and /n/: /haV?Na/ he spoke; /ha?'na/ hurry. /M/ and /m/: /baxoMa/ remedy; /oh

The following three tales in the dialect of L&wak (i.e. Klawock on Prince of Wales Island) were dictated to me by Mr. Lester Roberts, the native informant who told me the two Linglt stories previously published in this journal.1 Boas's transcription of Tlingit,2 which has been employed3 in this as well as in the earlier paper, is entirely phonemic-with the exception of the use of the symbol o, which stands for a comparatively rare allophone of /a/ or /a/ after labials. The phoneme inventory of Tlingit comprises 8 vowels and 41 consonants. The vocalic phonemes can be symmetrically arranged as follows.4

Phonemes are quite abstract units. The criteria for establishing them are primarily distributional and articulatory, not acoustic. A unit intermediate between a single speech‐event‐segment and a phoneme is the allophone. An allophone is a statistical average, with constant, identifiable acoustic characterization. For many purposes of engineering, psychology, pedagogy, and information theory, the allophone may be a more suitable unit than the phoneme.

1. Edward Sapir's The psychological reality of phonemes is still read as a classic of the American linguistic heritage.1 In it Sapir presents a brief description of the stop consonant allophones of Southern Paiute, a Shoshonean language spoken in southwestern Utah. His immediate concern is with the intervocalic allophone [IB] of /p/, but he touches upon a host of other Southern Paiute phenomena which cannot be clearly understood without reference to his larger study, The Southern Paiute language, upon which the materials in the article are based.2 The purpose of this note is to formulate a restatement of the phonological system underlying the Southern Paiute materials of Sapir's psychological reality paper as well as certain other interesting aspects of the full phonology, especially the phenomena of quantity, stress, and voicing. Sapir recognized the following segmental phonemes: /? (h) pt 6 kkwmnq w y a o ii u/. He is reluctant to set up a phoneme /h/ (SP 48-9), since only a few initial occurrences were observed, as in [hai-] to say and [ha&i-] bow, and its occurrence

I N MODERN colloquial Japanese, the patterns of occurrence of the velar nasal [ij] are sufficiently skewed to require or elicit special techniques of phonological description. The symmetry of the Japanese phonemic system and the vague Systemzwang that can be imputed to the phonology of the language suggest an / m, n, Ij / nasal series parallel to the series of voiced and voiceless stops. Still, even a gross statistical estimate shows that the [g]i phone is limited to particular phonetic contexts and bears a very limited structural load. As a matter of phonemic theory, it is by no means unusual to find individual phonemes with infrequent occurrences in special environments in various languages, and, when synchronic and diachronic linguistic techniques have been profitably combined, such situations often indicate the emergence of an allophone to phonemic status.

This study was designed to determine if training on any one specific /r/ allophone would result in transfer to other /r/ allophones without specific training. Twelve grade school children who were unable to imitate correctly the /r/, /ɝ/, and /ɚ/ allophones in 40 stimulus items during baseline served as subjects. A multiple baseline was used; six subjects served as controls while six were trained, after which the first six were trained. Each subject received training on a randomly assigned syllable representing a specific allophone of /r/. After training, generalization to the untrained /r/s was assessed. Since most subjects increased the number of correct responses to items in several allophonic categories regardless of the specific allophone taught, the different allophones of /r/ may be members of the same response class rather than independent of one another.

High speed cineradiograms were used to examine the articulatory characteristics associated with selected allophones of American English /r/. Only one of the three speakers produced a ‘retroflexed’ /r/ with a raised tongue apex. The other speakers produced a ‘bunched’ /r/ characterized by a raising of the tongue dorsum. Two basic types of /r/ were observed for all three speakers: (1) prevocalic /r/, which initiated a syllable, and (2) post-vocalic /r/, which terminated a syllable or formed a syllabic nucleus. The prevocalic allophone was characterized by greater lip rounding, a more advanced tongue position, and less tongue dorsum grooving. The two allophones also were associated with different variability and movement attributes. The differences in /r/ allophones were found to be similar to previously reported differences in American English /1/ allophones.

In his paper ‘The problem of serial order in behavior’ Karl Lashley (1951, p.113) points out that ‘language presents in a most striking form the integrative functions that are characteristic of the cerebral cortex’ adding ‘... the problems raised by the organization of language seem to me to be characteristic of almost all other cerebral activity’. Some idea of the complexity of the integrative processes involved in speech can be gained from the fact that the adult speaker’s ability to produce syllables at an average speed of 210 to 220 a minute (or roughly 14 phonemes per second) means individual muscular events occurring throughout the speech apparatus at a rate of several hundred every second; in the case of some phonemes the total time required to activate the muscles involved in their production being as much as twice as long as the duration of the sound itself. Not very much is known at present about what this involves on the neuronal level, where the rate at which individual events occur must be greater by a large factor, but it is a point of considerable interest that there is at least some evidence to suggest that in some instances the order of neuronal events might be different from that of the muscular events with which they are correlated.* The point Lashley is making in his paper is that any form of behaviour revealing this degree of complexity in its organization cannot be analysed as an associative chain of reflexes. But, as he points out, in the case of speech the evidence against the associative chain hypothesis is particularly compelling. This arises from considerations of two kinds. The first is the fact that the character of certain sounds is determined not only by the sounds that precede them but also by those that follow them. The second is the fact that the character of certain sounds is determined not only by the sounds in their immediate environment but also by the position they occupy with respect to the syntactic structure of the utterance. To take just one example, the speech of Standard English speakers contains at least twelve varieties (allophones) of the phoneme t . But whenever this is the first sound in a word and is immediately followed by a vowel they will always use the aspirated allophone never any of the others. This is clear evidence that in producing utterances speakers follow out principles of organization relating to syntactic structure. To produce a plausible model for speech we have to postulate not only principles of organization more complex than the Markov processes of associative chain theories but hierarchies of organization, elements on one level corresponding to what Lashley calls ‘generalized schemata of action' and Miller, Galanter & Pribram (1960) call ‘plans’ which are carried out on the level below. Evidence in favour of such a model can be obtained from a study of speech disorders, ranging from the transpositions occurring in the speech of a tired or nervous speaker to remarks of aphasics indicating that although for the most part they can only produce strings of unintelligible sounds they still ‘know what they want to say’. All these disorders can be viewed as involving in some degree a breakdown in integrative functions, an inability to carry out successfully plans for utterances.

In this paper we present new results related to our research work for building an accurate performance speaker independent recognizer for Classical Arabic, particularly Quran recitation; it is a part of the "Computerized Teaching of the Holy Quran" project funded by King Abdulaziz City for Sciences and Technology (www.kacst.edu.sa). In an early stage of this project, our efforts have been directed to propose a new labeling scheme covering all the Arabic sounds and their phonological variations. Since annotated speech corpus of the Classical Arabic sounds was not available yet, we next focused on building a well-designed sound database from Quranic recitations. Ten reciters have been recorded in an appropriate environment memorizing a part of the Quran resulting in almost nine hours of speech. We then concentrated on its manual segmentation and labeling on three levels: word, phoneme, and allophone in order to be accurate as much as possible due to the specificity of the Classical Arabic pronunciation and particularly the holy Quran recitation. This paper describes the development of a recognizer for the allophonic sounds of Classical Arabic (language of the holy Quran) based on Quranic recitations. This recognizer is built using the Cambridge HTK tools. Each allophonic sound is modeled by an acoustic Hidden Markov Model (HMM) with 3-emitting states. A continues probability distribution using 16 Gaussian mixture distributions is used for each emitting state. Results show that recognition rates reached a good level of accuracy (88% average) without using any specific language model, which is very promising and encouraging.

Etant tous.tes les deux ��tudiant.es issu.es de la migration et pour qui la langue fran��aise constitue une seconde langue (L2), nous nous sommes int��ress��.es �� travers ce m��moire aux trajectoires langagi��res des enseignant.es(-stagiaires) allophones et issu.es de la migration en lien avec leurs motivations �� apprendre et �� s���approprier une nouvelle langue dans un contexte migratoire en Suisse romande. Les moteurs d���apprentissage de la L2 pour ces apprenant.es s���av��rent ��tre nombreux. Ils ��manent �� la fois de ressources internes et de projections du soi r��v�� en lien avec la L2 et d���attentes externes plac��es sur l���apprenant.e. De plus, les processus qui poussent �� l���apprentissage de la L2 ou qui freinent cette action sont ��galement impact��es par les repr��sentations langagi��res (n��gatives ou positives), par le fait d���avoir v��cu l���immigration comme un processus volontaire ou involontaire, par l���accueil offert aux personnes migrant.es allophones et la fa��on dont ces derniers.��res le per��oivent, par l���attractivit�� ou non de l���endroit o�� la L2 est la langue officielle, par le contexte g��opolitique et ��conomique favorable �� une projection et �� une installation �� long terme dans le pays d���accueil ou encore par des facteurs psychologiques ou affectifs comme le fait d'associer la langue �� apprendre �� une personne bien aim��e ou �� des autrui significatifs (repr��sentant des mod��les positifs dans lesquels les apprenant.es allophones peuvent se reconnaitre et/ou �� qui iels d��sirent ressembler ou non).

Dans l’optique de répondre à la question de recherche suivante, Quels modèles sont mis en oeuvre afin de favoriser l’intégration scolaire et sociale des élèves allophones primoarrivants au secondaire 1 et 2 ?, ce travail propose un état des lieux précis des structures qui aident à l’intégration des élèves allophones primo-arrivants dans l’enseignement secondaire 1 et 2 au niveau national, cantonal et local (établissement primaire et secondaire de Crissier, ainsi que les gymnases de Beaulieu, de Chamblandes et de Renens). Afin d’offrir un cadre conceptuel précis, les termes d’allophonie, de primo-arrivant et d’intégration seront définis et un ancrage contextuel politico-législatif sera proposé au lecteur en préambule. Par la suite, des entretiens réalisés avec différents responsables et intervenants (enseignants et doyens) des structures locales susmentionnées ont été menés dans le but de prendre en compte les représentations et pratiques des professionnels du métier. Des éléments saillants ressortent de cette étude, tels que le flou législatif qui entoure les structures qui reçoivent les élèves allophones, la diversité des pratiques pédagogiques mises en place, l’importance de l’intégration scolaire et sociale des jeunes fréquentant ces structures, ainsi que la nécessité de développer la collaboration au sein de ces dernières- Finalement, suite aux analyses menées, ce mémoire propose différentes pistes d’action concernant l’accueil de ces élèves dans le système d’enseignement obligatoire et postobligatoire, lesquelles pourraient servir de base de travail pour des réflexions futures. Ce travail met en avant principalement la nécessité de proposer une centralisation des dispositifs ainsi qu’une collaboration plus intensive entre les différents membres de ces institutions. Il propose aussi de créer des dispositifs de parrainage entre les élèves allophones et les autochtones en visant en filigrane, au travers de l’intégration scolaire, l’intégration sociale des premiers.

Ce travail porte sur le thème de la transmission linguistique et plus spécifiquement sur le rôle de l’aîné allophone dans la transmission de la langue de scolarisation, dans le rapport entre famille et école. Nous nous demandons : comment l’aîné allophone entre dans la langue française ? Quels moyens didactiques utilisent les enseignantes de l’accueil (CIF, intégration, spécialisée) pour aider les élèves allophones dans l’apprentissage du français ? Comment l’aîné transmet la langue à sa famille ? Est-ce que la place comme aîné est importante ? Est-ce que la place de l’aîné allophone est prise en compte par les enseignants au sein de l’établissement ? A quelle condition devrait-elle être prise en compte par les enseignants ? Nous avons démarré notre travail en nous basant sur la recherche de Christiane Perregaux, suite à laquelle nous avons amené des éléments nouveaux en cherchant à avoir le regard des enseignantes de l’accueil. Voici brièvement les points clés de notre méthodologie : nous avons effectué des entretiens avec quatre enseignantes de l’accueil en utilisant une grille. Ces entretiens nous ont permises de créer des vignettes afin de faire des entretiens avec cinq élèves aînés allophones choisis parmi nos classes de stage. Puis, finalement un dernier entretien avec une ancienne élève allophone aînée. Les résultats de notre recherche mettent en avant le manque de prise en compte de la place de l’aîné au sein de l’école ainsi que ses différentes fonctions.

International audience; On sait désormais que les régions françaises ultramarines contribuent largement à la diversité linguistique de la France. On sait moins que dans ces mêmes régions - en tout cas pour ce qui concerne la Guyane - une partie non négligeable de la population adulte ne parle pas ou très peu français. L'État et ses institutions communiquent dans la langue officielle, le français, qui est loin d'être partagée par tous les individus vivant sur le territoire français. Que se passe-t-il lorsque la population soignée par un hôpital est non francophone à plus de 80%? À l'hôpital comme dans d'autres institutions en Guyane, il n'y a pas de politique linguistique explicite. On note un certain nombre de pratiques plurilingues mais celles-ci semblent plus le fait de volontés individuelles, voire de décisions ponctuelles, que d'une politique concertée. Elles sont dictées par la réaction à l'urgence, sans réflexion préalable basée sur un diagnostic de la situation sociolinguistique locale - en particulier de l'environnement linguistique des patients et des usagers - à laquelle il s'agirait de s'adapter. Elles permettent certes une communication minimale pour le geste médical mais la question de la prise en charge du patient dans sa totalité reste, pour sa part, entière.

The objective of this investigation was aimed to study the mineral composition of soils developed on dacito-andesite and their volcanic tuffs from Radocelo and Takovo and zeolitised andesite tuffs in village Igros in Zupa-Aleksandrovac. A selection of samples was collected from the area designated as Eutric Cambisol (Gajnjaca), Brown Forest and Brown Soils on the pedological maps. After collection the samples were air-dried and sieved to pass a 2 mm sieve and were used for further analyses. Soil reaction was determined in water as well as in 1M KC1 and in 1M NaF suspensions. Selective dissolution of Al, Fe and Si in non-crystalline materials was performed as follows: acid ammonium oxalate and sodium pyrophophate, as well as phosphate retention, determined according to BLACKMORE et al. (1987). The minerals in soil samples were studied by a combination of methods. The mineralogy of bulk soil samples was obtained by polarized microscopy and X-ray diffraction analysis (XRD). Minerals of clay fraction were also characterized by infrared spectroscopy and the XRD analysis. The XRD analyses were made on the clays lt 2 μm specimens, as follows: air dried, ethylene glycol solvated, and progressively heated at 300°C and 500°C. Mineral composition of fresh volcanic rock is as follows: feldspars, zoned plagioclase, hornblende, pyroxenes and biotite, which are found either as phenocrysts or cryptocrystals in groundmass, in association of minor to traces of tourmaline, sphene, apatite and ore minerals. All ferromagnesian minerals are altered in various degrees to chlorite and clay minerals. Groundmass comprises of feldspar microlites in dark-grey volcanic glass and minor amounts of other minerals. The clay fraction of soils is composed mainly of halloysite/kaolinite, illite, chlorite, feldspars, and zeolite and poorly-ordered mineral-allophane and CT-opal. Only in the clay fraction from soil developed on a layer of zeolitised andesite tephra in Igros, heulandite-clinoptilolite-mordenite type zeolite was found as dominant resistant mineral associated with minor amounts of smectite and halloysite. Phosphate retention varies between 23.0 and 48.5% and the oxalate exchangeable active Alox + 1/2 Feox component varies between 0.50 and 1.58 as well as presence of the volcanic glass and about 9 to 11% of allpohane in the Bt horizon of the soils investigated have pointed to vitric-andosol type, with a lot of skeletal material. Andosoli predstavljaju poseban tip u klasifikaciji zemljišta. Obrazuju se na vulkanskim stenama i njihovim tufovima. U cilju poznavanja mineralnog sastava zemljišta, razvijenog na andezitimna i njegovim tufovima, na području Radočela, na području kod sela Igroš (Aleksandrovačka Župa) i na području kod Takova, prikazane su njihove najbitnije karakteristike. Dacito-andazitski i labrador andazitski tufovi na planini Radočelo i kod Takova sastavljeni su od krupnih odlomaka kristala feldspata: sanidina, zonarnih plagioklasa sa 20-45% anortita, zatim biotita, muskovita, hornblende i retkih kristala augita, koji se pojavljuju kao fenokristali. Kao sastojci osnovne mase javljaju se turmalin, sfen, apatit i cirkon, kao i retki metalični minerali (magnetit i ilmenit). Kvarc se pojavljuje manjim količinama, kao sitna korodovane individua. Osnovna masa izgrađena je od mikrolita feldspata, vulkanskog stakla i ređe amfibola. Zemljišta sadrže značajne količine vulkanskog tufa kao skeleta u raznim fazama površinskog raspadanja. Frakcija gline sadrži mešavinu kaolinita i halojzita, ilita, i manjih količina smektita i mešovito slojevitih silikata po tipu ilit - smektit kao i tragova feldspata, getita i amfibola. Zemljište na Radočelu sadrži još i: kristobalit (0.405 nm), tridimit (0.411 nm) i opal, koji je zapažen po difuznom difrakcionom piku sa maksimumom oko 0.41 nm, što je indikativno je za CT - opal, neuređen i nisko temperaturni silicijski hidroksid, koji se nalazu u vulkanskim tufovima. Mineralni sastav zemljišta iz atara sela Igroš (Aleksandrovačka Župa), je drugačiji jer je matični supstrat zeolitisani tuf. Zeolitisani tuf se sastoji od devitrifikovanog vulkanskog stakla koje pokazuje različite stadijume zeolitizacije. Zeolit je iz grupe heilandit-klinoptilolit (Ca-Na zeoliti). U sastav tufa ulaze još feldspati i kvarc, praćeni manjim količinama hornblende, piroksena i biotita, zatim apatit i opake minerale. Zemljišta pokazuju distrični karakter, umereno ili slabo kiselu reakciju. Reakcija zemljišta u KC1 generalno je za čitavu jedinicu niža nego pH u vodi. Reakcija zemljišta u NaF je bila 10.07 do 10.71, odnosno značajno veća od 9.5 u dubljim horizontima, što je indikativno za andični material. Retencija fosfata je važan dijagnostički kriterijum za vitrične-andosole. Ispitivana zemljišta su pokazala varijaciju u retenciji fosfata između 23 i 48.5% što ih svrstava u vitrični-andosol. Sadržaj Alox + 1/2 Feox varira između 0.50 do 1.58 % što potvrđuje prethodne rezultate i ukazuje na vitrični-andosol sa tendencijom ka tipičnim andosolima na Radočelu. Nešto niže retencije fosfata mogu biti posledica intenzivnog đubrenja jer se parcele-njive koriste za poljoprivrednu proizvodnju. Sadržaj Siox varira između 0.17 to 1.46 %, pa zemljišta u Bt horizontu sadrže od 1.3 do 11.0% alofana. Prisustvo značajnih količina kaolinita/halojzita, kao i alofana (11.%), zatim vulkanskog stakla (15-20%), tipična asocijacija minerala u frakciji lt 2 mm, indikativno je za zemljišta tipa vitrični-andosol, koja su nastala na starim izlivima neutralnih i bazičnih stena.

Cette étude a pour objet la relation que des gymnasiens allophones entretiennent avec leur langue première et les langues de l’école. Afin de collecter des données, un questionnaire a été distribué auprès des élèves de quatre classes 1M avec option italien. Parmi les 72 élèves interrogés, 63 se sont prêtés à l’exercice. Sur la base de leurs réponses, il a été possible de connaître leur nationalité, leur répertoire langagier et leur mobilité entre cantons et depuis l’étranger. Puis lors d’entretiens compréhensifs, 13 élèves arrivés en Suisse entre 10 et 16 ans se sont exprimés sur le processus d’accueil mis en place dans l’école obligatoire et au gymnase, ainsi que sur les difficultés qu’ils rencontrent actuellement dans la maîtrise de la langue de scolarisation. Par ailleurs, ils ont été questionnés sur l’éventuelle prise en compte par l’école de leur langue première et de leur culture. Enfin, ils ont fait part des motifs qui les ont poussés à choisir l’option italien pour leurs études gymnasiales., Mémoire professionnel, Diplôme d'enseignement pour le degré secondaire II

A text-to-speech synthesis system receives digital code representative of characters from a local or remote source, and converts those character codes into speech. A set of allophone rules is contained in a memory and each incoming character set is matched with the proper character set to describe the sound of that particular character set. A microcontroller is dedicated to the comparison procedure which provides allophonic code when a match is made. The allophonic code is provided to a speech producing system which has a system microcontroller for controlling the retrieval, from a read-only memory, of digital signals representative of the individual allophone parameters. The addresses at which such allophone parameters are located are directly related to the allophonic code. A dedicated microcontroller concatenates the digital signals representative of the allophone parameters, including code indicating stress and intonation patterns for the allophones. An LPC speech synthesizer receives the digital signals and provides analog signals corresponding thereto to a loud speaker to produce speech-like sounds with stress and intonation.

Ohman's 1966 investigation of the acoustic correlates of coarticulation in Swedish and English VCV sequences [J. Acoust. Soc. Amer. 39, 151–168 (1966)] indicates that terminal formant transition frequency values are strongly influenced by the nature of the transconsonantal vowel. However, the literature reflects no attempts to discover whether these coarticulation effects are perceptible, e.g., whether in a VCV sequence the initial VC carries significant perceptual cues as to the quality of the following V. The present experiment was designed to investigate the latter question. Forty‐eight VCV utterances involving the vowels /i ae a u/ and the consonants /p t k/ were recorded and either the initial or final vowel removed by splicing techniques. The resulting tapes were played to listeners who were asked to indicate which vowel had been deleted. Results were random indicating that coarticulation effects are not sufficient cues to the perception of the vowels which occasion them. These results are counterevidence for the claim by Wickelgren [Psychol. Rev. 76, No. 1, 1–15 (1969)] that the “context sensitive” allophone may be a basic cue in speech recognition. The results also suggest that some coarticulation effects may not have to be taken into consideration in speech synthesis.

Preliminary outlines for the exploitation of natural redundancy in phonetically encoded speech messages are introduced. The techniques are specific to the Votrax SC-01A speech synthesiser but can be applied to any phoneme or allophone based devices.

Preliminary estimates of the sizes, shapes, and locations of perceptual target zones (PTZ) of ten vowels, six dipthongs, a twenty‐nine consonants have been made. The PTZ's are defined as enclosed volumes in the space x = log(PF3/PF2), y = log(PF1/PR), and z = log(PF2/PF1), where PR is the perceptual reference in Hertz and PF1, PF2, and PF3 are the center frequencies of the perceptual formants. The perceptual formants are estimated from spectral envelopes as guided by rules for distinguishing burst‐friction spectra and glottal‐source spectra and for naming and numbering formants based on their logarithmic relations to the value of the reference. The estimation of PTZ's will be illustrated for at least one allophone from each of the following classes: vowels, diphthongs, approximates, nasals, voiceless fricatives, and voiced fricatives. These preliminary estimates indicate that PTZ's: (1) are large, thus encompassing a considerable range of spectral variants as might be associated with coarticulatory effects and articulatory style; (2) have irregularly shaped, abutting boundaries; and (3) for the most part, exhibit little overlap. [NINCDS grant R0 I‐NS21994‐01.]

The basic goal for the described project is selective and interactive production of synthetic speech from a large data base such as a training manual. In a training environment at a sophisticated computer control console, an operator should learn more effectively and efficiently if both aural and visual senses are used and if his eyes and hands are free to concentrate on the panel before him. The system described includes an operator‐directed control strategy to selectively retrieve that part of text which is to be spoken, a considerable amount of text preprocessing in non‐real time to obtain synthesizer control parameters, and a microcomputerized cascade terminal analog speech synthesizer. The preprocessing of the English text utilizes an extensive set of translation rules, dictionary lookup of complete words, and morph lexicon lookup of parts of words to produce 123 distinct allophone designators with corresponding stress, pitch, and duration designators. Storage requirements for the preprocessed speech...

General rules for production of English phonemes may be studied very effectively by statistical methods. The rules have numerous exceptions, however, involving special treatment of sometimes infrequently occurring conditions. The problem is to identify these exceptions, and then to find enough examples to characterize them. Speech synthesis is especially suitable for this kind of exploratory study. When synthesized utterances are substantially correct, exceptions stand out clearly, as mispronounced words. In our synthesis, a program for regular timing and allophone selection is followed by a set of rewriting rules to handle exceptions. In using the system, when utterances are found for which the rules are inadequate, the program allows quick examination of similar phrases, and interactive editing at the segment level to evaluate potential rule changes. New rules are easily added, or existing ones further qualified. Using this method, we have developed quantitative rules for nasal assimilation, flaps, chan...

Theoretical proposals for the representational unit used in recognizing spoken words include the phoneme, allophone, and position‐specific phoneme. The present experiment was devised to differentiate among these proposals. Subjects heard natural CVC stimuli in prime‐target pairs and named the second item (target) as quickly as possible. The phonetic overlap between prime and target was varied to create five different intrapair relations such that the expected pattern of priming effects across the set of relations was different for each of the three phonetic representations. The obtained pattern is inconsistent with an abstract phonemic unit and points to either a position‐specific phonemic or allophonic unit. This pattern was found for both word and nonword stimuli. Thus these priming results appear to tap the recognition process prior to lexical activation. Priming effects were eliminated when stimuli within a prime‐target pair were produced by different voices. The implications of the findings for refining models of spoken word recognition will be discussed. [Work supported by NINCDS.]

It is well known that the acoustic manifestation of any phone is, to varying degrees, a function of its acoustic environment. We have tried to evaluate this explicitly. Our data consisted of all common fricative and stop consonant allophones embedded in carrier phrases, each rendered by three speakers (two male, one female). Time‐domain segmentation and analysis, based on individual homogenous segments of frication and release‐aspiration, from the fricatives and stops, respectively, and characterized these by six parameters. For each phoneme and speaker, each allophone was normalized with respect to an appropriate neutral referent allophone. Pairwise phoneme recognition tests tried to distinguish to which of 2 phonemes belonged an unknown allophone of one of these. A modified Euclidean distance measure compared the acoustic characteristics of this allophone with the predicted allophone effects derived from all allophone, belonging to /k/ or /p/, was assigned to one of these on the basis of allophone characteristics derived from /g/, /b/, /d/, and /t/, by the same speaker. The ranges of scores for the 3 speakers, for different and same places of articulation, were 71%–86% and 44%–69% for stops, and 83%–95% and 75%–83%, for fricatives, respectively. [Part of work supported by ARPA (♯F44620‐73‐C‐0074), at Carnegie‐Mellon University.]

Recent analyses of the allophonic variants of phonemes in particular environments often seem to assume one or more of the following: (1) The proportions of variants encountered in a multispeaker sample represent an “irreducible” statistical component of phonology; (2) these proportions predict the likelihood of encountering these same allophones in new material; (3) the probability of encountering a particular allophone of some phoneme is independent of the observed allophones of other phonemes nearby. These related assumptions are questioned on theoretical and practical grounds, using transcribed data from 630 speakers reading two sample sentences. The frequencies of occurrence of all the allophones of certain phoneme tokens in the sample sentences were measured across all speakers; then the conditional co‐occurrence of all the allophones of certain phoneme pairs in the sample sentences were analyzed. For example, first the occurrence of a flap in the words “suit in” was counted across all speakers; then the occurrence of a flap in “suit in” was counted for only those speakers who deleted /t/ in “don't ask,” and so on. Comparing these two kinds of analysis has implications for theories of variation. The application co‐occurance analysis in speech recognition will be illustrated. [Work supported by DARPA.]

A new theoretical procedure was developed for optimizing the acoustic characteristics of hearing aids. The algorithm increases predicted speech intelligibility while keeping the estimated loudness of amplified speech at a predetermined level. The intelligibility criterion was the mutual information between spoken phonetic messages and the corresponding streams of auditory patterns in a psychoacoustic model of the listener's auditory system. The speech signal was modeled as a sequence of allophone transitions. Each transition was represented by three auditory patterns, each containing ten cepstrum coefficients, one voice and one duration parameter. The shape of optimized hearing‐aid frequency responses depended strongly on the desired loudness of amplified speech. When normal loudness was desired for speech at 65 dB SPL, and the optimization was initialized with the NAL prescription, hearing‐aid gain was reduced, but predicted speech intelligibility was nearly unchanged. Predicted intelligibility of amplif...

Synthesizing speech by concatenating allophones is a new application of linear predictive coding (LPC) that provides an unlimited vocabulary. Since the stored units are individual speech sounds, any English word can be synthesized by concatenating the appropriate sounds. Our inventory contains 59 speech sounds and five pauses. Each allophone was created by extracting it from the digital waveform of a word, and then synthesizing it. The sounds are called allophones because some phonemes have two or three versions for different environments. For example, /k/ and /g/ each have three allophones: one to be used before front vowels, one before back vowels, and one in final position. In contrast, research revealed that the same /p/ can be used in all environments. Isolating final /r/ from preceding vowels was impossible because of important transitional information. Therefore there are five phonemically unique V + /r/ combinations. Due to the omission of many transitions, allophone synthesis suffers from reduced...

In a constructive synthesis environment, lack of natural prosody detracts both from the naturalness and from the intelligibility of synthetic speech. It is time consuming and frustrating for the linguistically untrained user of a constructive‐synthesis system to assign pitch patterns and unit durations by mere guesswork. A semi‐automatic prosody assignment scheme has been designed here which takes the majority of the prosody‐assignment burden off the user and which is capable of assigning fairly natural‐sounding pitch patterns to a constructive synthesis string of allophones. The user is required only to mark the primary and secondary stress locations in the text or in the allophone string, plus an indication of whether the constructed phrase should rise or fall in pitch at the end. These stress marks serve as anchor points for the system to compute both an intonation contour and a timing‐adjustment contour for the whole phrase. Thus, with a minimum of effort and linguistic knowledge, the user may achieve a relatively natural‐sounding constructive speech phrase.

A number1 of linguists have attempted to establish a traditional phonemic system for Modern Faroese,2 but they have uniformly come to grief over the vowel segments. Since there are only a very few minor problems involving the phonetic facts,3 it is clear that the difficulties lie not in our knowledge of Faroese but rather in the linguists themselves and most particularly in their insistence on honoring the invariance condition. The invariance condition is the demand that every occurrence of [xi] be considered an allophone of /x/ and that [xl x2 x3...], allophones of /x/, share at least the defining features implied in /x/. This restriction does not allow the assignment of some [xi] to /x/, some to /y/, etc., nor may [x y z ...] be derived from /x/ by the change of some feature implied in /x/.4 In what follows I have not sought to satisfy this condition. I have instead tried to establish the simplest distinctive-feature matrix that minimally distinguishes the vowel morphophonemes of Faroese and that through the application of the simplest set of ordered rules can be mapped into the actually occurring phones.5 The phones that I wish to account for are the following (brackets omitted):6

In a recent article in LANGUAGE,1 Charles F. Hockett has presented an analysis of the stressed syllabics of the Vespasian Psalter and Hymns.2 Hockett opposes the traditional view which interprets ea, eo, io, and ie as short diphthongs, and offers a solution of his own which within the limitations of one MS is in part convincing. In general, he equates each vowel letter or digraph with a phoneme. In agreement with us, with Moss6, and with Daunt, but in contradiction of the established tradition, he regards the digraphs ea, eo, io, and ie as spellings for monophthongs; he believes, however, that these short digraphs represent separate short vowel phonemes; we believe that in early OE they represented allophones of the front vowels /i/, /e/, and /ase/; we believe that eo, io, and ie represented allophones ONLY in the earlier period. These allophones later became phonemes, whereas the allophone represented by ea did not. Hockett's speculations about the motivations of the scribes who first selected these particular digraphs in these particular values are appealing. Especially ingenious is his theory about the origin of the letter y in OE, which he considers to have been originally a digraph (591-4). Although we agree to a limited extent with the conclusions of Hockett's article (which might better be entitled The Stressed Syllabics of the Vespasian Psalter and Hymns), we wish to reaffirm the stand taken in our monograph and our subsequent article,3 in particular with regard to me-ea, but also with regard to the structure of OE vowels in general. Concerning the long diphthongs and long vowels we are more nearly in agreement with the position Hockett took in the book that appeared before this article :4 'there were complex nuclei consisting of one or another of the eleven vowels [we do not think there were as many as eleven vowels] plus a glide element. The details are not clear.' There are, however, other basic, more axiomatic assumptions with which we do not agree at all. They are discussed below.

The frank withdrawal of several claims and statements originally made by Stockwell and Barritt has answered some of our objections; but, for the most part, their rebuttal (The Old English short digraphs: some considerations) has not only failed to convince us of the soundness of their general position, but has provided further evidence of its unsatisfactory nature and basis. We shall be as brief as possible, and for ease of reference, our comments are so arranged and labeled as to correspond with the numbered sections of the rebuttal.' We begin with the revisions and emendations which Stockwell and Barritt propose to make in their first article.2 ?2.12. In their interpretation of the phenomenon of 'slur', they go far beyond anything that we can find in Martin Joos's monograph.3 We suggest that slur alone cannot account for the combinative sound changes which gave rise to OE ea, eo, io, ie. Breaking, to take the simplest example, affected only certain vowels. Slur, according to Joos, affects all of the speech sounds. Breaking occurred only in certain environments, although the resulting sound might later be transferred analogically to a limited number of other environments. Slur takes place in any environment, the phoneme being modified by any sound that immediately precedes or follows it. In the words of Joos, 'The speaker has, in the articulatory and acoustic aspects, a whole vowel-phone system for the context [d--d], another system for the context [s-p], and so on. And in each context he articulates the appropriate allophone' (108). Breaking must have involved consciousness. If the Anglo-Saxons had been unconscious of the difference between simple vowels and the sounds represented by ea, eo, io, ie, they could not have distinguished them in spelling, e.g. xrn:earn. When, through phonetic change, they lost the distinction in sound (e.g. /i/ and /ie/ in one stage of West Saxon, cf. our fn. 4), they also lost the ability to make the orthographical distinction. Slur, on the other hand, is automatic and unconscious. Breaking could lead to phonemic change, as Stockwell and Barritt observe (?7). Slur (if we assume that the phenomenon is phonetic and not purely acoustic) produces a multitude of allophones, but Joos

0. Introduction. In his valuable pioneering study The Phonemes of Gothic, Lg. 24.76-86 (1948), William Moulton suggests that Gothic /w/ has two allophones: [i] and [w]. The first of these is spelled with u, e.g. auk, 'iup; the second with w, e.g. wigs, gaidw, or with hv (interpreted as /hw/), e.g. 7vas, or with q (interpreted as /kw/), e.g. qam. He regards the letter u in other occurrences as (1) designating a short syllabic /u/, (2) designating a long syllabic /u:/, and (3) supplying the second half of the compound symbol au-a means of writing additional short and long syllabic phonemes. He assumes that the letter w also has an alternative reference, a short syllabic found only in foreign words, e.g. Lwstrws. Moulton's analysis is based on an interpretation of Gothic orthographic practices that has been widely accepted. It is unlikely that a more consistent way of organizing the material can be found, so long as this or a similar interpretation is used as the starting point. Since Moulton's article was written, renewed efforts have been made to find solutions for the orthographic and phonological problems raised by the Gothic labiovelars. The arguments that have been adduced seem to indicate that in several instances reappraisal of the traditional interpretation is needed. The present paper deals with implications of the new theories insofar as they bear on the question of nonsyllabic references for the Gothic w symbol. 1. Phonemic transcription-native words. If the Gothic alphabet came into existence as reported in fifth-century sources,1 it must have supplied a good fit for the phonemes of Wulfila's dialect. In deciphering such an alphabet each operation should start with the assumption that the manuscript spellings reflect the phonological patterns of the language. Any system of segmentation that assigns more than one symbol to a phoneme-or more than one phoneme to a symbol-is prima facie questionable and needs careful scrutiny.2 Phonetic and phonemic inferences that help to eliminate formulations of this sort deserve the most serious consideration. A case in point is interpretation of the compound symbol au. If it is conceded that this digraph had monophthongal references in all positions,3 the allophone

Lateral cinefluorographic films were taken of three subjects at a speed of 150 frames/sec. The subjects, natives of the midwestern U. S. and judged to possess a dialect typical of that region, produced words in which an /l/ allophone was a part of a pre‐ or postvocalic consonant cluster. Articulatory movements were analyzed by a frame‐by‐frame tracing of the structural contours through the phonetic sequences under investigation. In prevocalic clusters, articulation of the [l] was shown to coarticulate with the articulatory movement associated with the phonetic element immediately preceding the [l]. The articulation of the phonetic element immediately following the postvocalic [l] coarticulated with the movements associated with the [l]. The degree of overlap or the timing of the sequence of articulatory movements in the clusters was variable.