Unrounded Vowels Research Articles

Factors Underlying the Articulatory and Acoustic Structure of Vowels

Three-mode (PARAFAC1) factor analysis was performed on the tongue shapes and mel formant values of ten English vowels as spoken by five subjects. Vocal tract cross-dimensions were measured from cine x rays, along lines perpendicular to the midline of the tract and spaced equidistantly at approximately 1-cm intervals. Deviations from “neutral” average-vowel values were computed for both formants and cross-dimensions. Only two factors were found, and these accounted for roughly 90% of the articulatory and 95% of the acoustic data variance. Factor 1 produces a predominately antisymmetric perturbation of the vocal tract and thus would have a greater acoustic effect on the unrounded vowels (since their vocal tract termination impedance is lower). It presumably is due to the position of the jaw and action of the genioglossus muscle. It mainly affects the second formant. Factor 2 is both less stable across analyses and less clearly defined. It produces a mainly symmetric perturbation and would thus be more effective during rounded vowels. It can be interpreted as resulting in part from the action of the styloglossus muscle. It affects both formant frequencies. PARAFAC1 analysis of formant data for Swedish, Danish, and Twi vowels shows similar factors, indicating a consistent pattern of interpersan variations along these two dimensions.

Crosslanguage study of vowel perception.

This study examines the discrimination and identification of synthetic rounded and unrounded vowels by speakers of two languages (English and Swedish). The unrounded vowels are phonemic in both languages, whereas the rounded vowels are phonemic only in Swedish. A subsidiary aim of the study is to compare the perception of the synthetic vowels with that of synthetic consonant-vowel syllables in which the consonants are stops arranged along a continuum from /b/ to /d/ to /g/. The data indicate that the ability of subjects to discriminate between the vowels is relatively independent of their linguistic experience: Swedish and American English subjects exhibit similar performance in the discrimination tests, though they have somewhat different identification functions. The discrimination functions are characterized by peaks and valleys, suggesting that listeners can discriminate given shifts in the vowel formant frequencies more readily in some vowel regions than in others. Comparison of the data on stop-consonant and vowel perception is consistent with earlier findings: the number of discriminable tokens along the stop-consonant continuum is roughly equal to the number of absolutely identifiable items (three in this case) ; on the other hand, the number of discriminable vowels is much greater than the number that can be absolutely identified. The data are in accord with the view that a human listener uses different modes in the perception of steady-state vowels and stop consonants.

Spectral form and duration as cues in the recognition of English and German vowels.

Pairs of short and long vowels, such as English / / and /3: / as in shut vs. shirt, differ from each other in both spectral form and duration. Experiments were carried out on two such pairs of English vowels and two pairs of German vowels, using synthetic speech, in order to assess the relative importance of the two cues in the discrimination of the vowels. The results confirmed the hypothesis that the importance of the duration cue is inversely proportional to the distance between the qualities of a given pair of vowels. The results showed in addition that spectral form is in general more important than duration in vowel recognition in both English and German, since it is only when two vowels are very close in quality that the duration cue is more important for their discrimination. To examine the possibility that the different linguistic background of speakers of English and German might influence the amount of attention they devote to spectral form and duration, English and German subjects were presented with a test in which they had to discriminate a pair of back, unrounded vowels, [ ] and [ :], that occur neither in English nor in German. For the English subjects the duration cue was slightly more important than the spectral form cue. The Germans, on the other hand, discriminated the vowels largely on the basis of the spectral form cue. These results are in accordance with predictions that might be made from consideration of the phonological systems of the two languages.

Crosslinguistic Study of Vowel Discrimination

The objective of this experiment is to compare the identification and discrimination of a series of rounded and unrounded vowels by Swedish and American English listeners. All stimuli were produced with the OVE II synthesizer at the Royal Institute of Technology in Stockholm. The unrounded vowels are in the phonemic system of both Swedish and American English, and encompass the three vowels /i/, /ɪ/, and /ɛ/. The rounded vowels encompass the Swedish vowels /i/, /y/, /u/, but are not in the phonemic system of American English. Each series consists of 13 vowels, for which the first three formant frequencies vary in approximately equal logarithmic steps through the indicated range. By discrimination tests showed that both series of vowels gave essentially identical discrimination data for both groups of listeners. Identification functions for the unrounded vowels were similar for the two groups of listeners, but American English listeners were less consistent than Swedish listeners in the identification of the rounded vowels. Implications of the results with regard to theories of speech perception are discussed. [This study supported in part by the National Science Foundation.]

Identification and Discrimination of Rounded and Unrounded Vowels

The aim of this experiment was to examine the identification and discrimination of a series of rounded and unrounded vowels by a group of listeners (Swedish) who were familiar with both types of vowels. The following groups of stimuli were assembled using the Stockholm OVE II speech synthesizer, and were arranged in the form of identification tests and ABX discrimination tests: (1) a series of thirteen vowels in which the first three formants varied in approximately equal logarithmic steps through three unrounded vowels (roughly /i/, /ɪ/, and /ɛ/); (2) a similar series in which the second and third formants varied in approximately equal logarithmic steps from the unrounded /i/ through the rounded /y/ to the rounded /u/; and (3) a series of thirteen vowels constructed by taking /i/ as a center and then including the first six steps along the unrounded continuum and the first six steps along the rounded continuum. The discrimination functions showed small peaks near the phoneme boundaries; i.e., discrimination between adjacent stimuli on the vowel continuum was better when the stimuli were near phoneme boundaries. Discrimination was at least as good for the rounded series as for the unrounded series, and seemed to be especially sharp in the region where rounding first became apparent.

A Note on the Phonemic Status of Russian High Unrounded Vowels

A Note on the Phonemic Status of Russian High Unrounded Vowels

The Dialect Geography of hast, hat in Swiss German

In The pronunciation of English in the Atlantic States (Ann Arbor, 1961), Hans Kurath and his collaborator, Raven I. McDavid Jr., have given us the first detailed study in structural dialectology that has ever been made of a large geographical area. In the present article, offered in Kurath's honor, I shall try to make a similar contribution in miniature: a structural analysis of the vowel of a single pair of words, involving part of the short-vowel system of part of German-speaking Switzerland. Following Kurath's example, I shall examine the vowel phonically, phonemically, and historically. The vowel which I have chosen for analysis is that of the present-tense verb forms (du) hast, (er) hat. It is an interesting topic for a study in structural dialectology because it shows how misleading a simple phonic presentation can be if it is not accompanied by phonemic and historical presentations as well. The data are taken from the field records of the Sprachatlas der deutschen Schweiz, in Zurich; they were made available to me through the kindness of Rudolf Hotzenkocherle, the director and editor of the Atlas.l In gathering the Atlas materials, the field workers did their recording impressionistically, using a highly flexible phonetic alphabet to note every shade of sound they were able to distinguish. Of interest to us here is the treatment of the front unrounded vowels. The symbol e was used for a mean-mid vowel, and diacritics were added for closer and opener values: a single subscript dot for close e, a double subscript dot for over-close e, a single subscript hook for open e, and a double subscript hook for over-open e (much like English a in bat). Intermediate values were indicated by parenthesizing the diacritics; a vowel even opener than over-open e was written with the ligature a?. If all of these phonic data for the words hast, hat were reproduced in map form, the resulting picture would be so detailed as to defeat the very purpose of cartographic presentation. In making Map 1, I have therefore combined some of these values, as follows: (1) plain and parenthesized close e are symbolized by a vertical line (no examples of over-close e were found in the data for hast, hat); (2) mean-mid e and plain and parenthesized open e are symbolized by a diagonal line; (3) x and plain and parenthesized over-open e are symbolized by a horizontal line. Using IPA symbols, I have written these three values as [e], [e], and [ae] respectively. A few remarks need to be made about the base map, which is taken from that of the Atlas. The area covered includes 425 of the 573 points in the Atlas,2

Germanic Umlaut Anew

1.1. A realistic reinterpretation of the Old English spellings ea, eo, io has been suggested by Charles F. Hockett, who approached the problem de novo by investigating the evidence in the Vespasian Psalter and Hymns (Anglian dialect of the first half of the 9th century) and by interpreting this evidence in the light of modern descriptive techniques.l Hockett finds that the three digraphs in question represented independent phonemes (not allophones of /ae/, /e/, /i/) at a time prior to the writing of the VP, although 'Sound change and dialect mixture had probably led to a state of affairs in which every form pronounced with io had a by-form pronounced with eo (though not conversely), and in which every form pronounced with ea had a by-form pronounced with a or with eo (though again not conversely). These conclusions account both for the spelling habits which the Vespasian Scribe inherited and for the particular way in which he modified and varied those habits' (590). In determining the phonetic values to be ascribed to ea, eo, io before the onset of the secondary modifications evident in VP, Hockett turns to the testimony of the Icelandic First Grammarian (ca. 1150). A bilingual person attempting to represent Old English sounds was faced with the paucity of vowel symbols in Latin. To find new symbols, the Old English vowels were analyzed with respect to the five Latin vowels, as was done for Icelandic by the First Grammarian. The latter, for instance, analyzed the sound represented by 0 as 'made up from the sounds of e and o, spoken with the mouth less open than for e and more than for o, and therefore [is] written with the cross-bar of e and the circle of o'2 Hockett suggests, therefore, that those who established the Old English orthography chose ea (older spellings aea aeo) to represent the height of /ae/ and the retracted tongue position of /a/, i.e. [a]; eo to represent the height and lip position of /e/ and the tongue position of /o/, i.e. [a]; and io (older iu) to represent the height and lip position of /i/ and the retracted tongue position of /u/ or /o/, i.e. [i]. He then makes a theoretic pronouncement of great import for the interpretation of mutation in Germanic: 'From the point of view of realism in phonetic change, particularly in assimilations, it is certainly as likely that a back-umlauting of front unrounded vowels should produce unrounded back vowels as it is that a front-mutating of back rounded vowels should produce front rounded vowels. The same applies to the modification of vowel color by a following consonant: breaking is a loaded term stemming from the habit of talking about letters instead of sounds, and prejudices reinterpretation' (595).3