\textbf\textscVowels and Consonants** The past sixty years have seen increasingly rapid advances in the field of phonetics and speech acoustics, which established the acoustic properties of vowels and consonants. Many studies within this research paradigm were guided primarily by the urge to identify the invariant features of speech that realize speech sounds. The invariant features were considered as the actual’’ properties of vowels and consonants whereas the non-invariant features were thought to be insignificant. For instance, vowels were identified by the first two—and in some cases three—formant frequencies, which were measured once in the middle of each formant frequency. Along with this line of research, however, there has been an increasing interest over the effects of factors, such as speaker physiology, sociolinguistic properties, emotional state etc. on the acoustic structure of vowels and consonants. Following this line of research, in a number of studies, I attempt to determine how vowel and consonant spectra convey both linguistic and non-linguistic information. To understand these patterns, I employ methods from signal processing and machine learning. Especially, I am extremely interested in machine learning methods, such as deep Neural Networks, SVMs, HMMs, which are behind current impressive achievements in text-to-speech, speech perception, machine translation, image description generation, and semantic interpretation. I believe that these methods can offer insights into human perception and cognition including the processing and accessing information.

Specifically, in an article, which has been published in Speech Communication (Themistocleous, 2017), I provide a classification model of two Modern Greek dialects, namely Athenian Greek and Cypriot Greek, using information from formant dynamics of $F1, F2, F3, F4$ and vowel duration. The measurements were employed in classification experiments, using three classifiers: Linear Discriminant Analysis, Flexible Discriminant Analysis, and C5.0. The latter outperformed the other classification models, resulting in a higher classification accuracy of dialects, namely Athenian Greek and Cypriot Greek. C5.0 classification shows that the duration and the zeroth coefficient of $F2$, $F3$, and $F4$ contribute more to the classification of the dialect than the other measurements; it also shows that formant dynamics are important for the classification of dialect. An important contribution of this work is that it demonstrates how acoustic properties are fined-tuned to convey both linguistic and non-linguistic information.

In an article, which was published in Phonetica (Themistocleous 2017b), I investigated the acoustic properties of vowels of Standard Modern Greek (SMG) and Cypriot Greek (CG). The study shows that the two varieties differ in their vowels. The findings show that Greek stressed vowels are more peripheral than unstressed vowels, (2) SMG unstressed /i u/ vowels are more raised than CG vowels, (3) SMG unstressed vowels are shorter than CG unstressed vowels, and (4) SMG /i o u/ are more rounded than the corresponding CG vowels. Moreover, it shows that variation applies to specific subsystems, as it is the unstressed vowels depending on the dialect whereas the stressed vowels display only minor differences between varieties. The sociolinguistic implications of these findings with respect to vowel raising and vowel reduction are discussed in the paper. That study is the first to provide acoustic evidence of Cypriot Greek vowels and the first to provide acoustic evidence on the sociophonetic differences of southern Greek dialects.

\beginfigure \includegraphics[width= 1\linewidth]./images/dynamics.jpg Mean relative positions of SMG and CG vowels in the F1 x F2 plane and their formant movement measured at seven equidistant points (20-30-40-50-60-70-80\%) across the vowel. Shown are stressed and unstressed variants. Arrows are located at the 80\%-point of the formant movement.

In a follow up article to Phonetica (submitted). I am exploring the dynamics of vowel formants and how they change during articulation. The findings from vowel dynamics are impressive as they show that although Modern Greek vowels are considered relatively monothoptongal, their formants change during articulation significantly (see Figure \refdynamics). Low and middle vowels shift from low to high; the high vowel [i] remains relatively static and the vowel [u] becomes more fronted. Most importantly vowel quality, stress, and dialect affect significantly formant dynamics. The study argues that the dynamic approaches can constitute a significant methodological improvement over static approaches of vowels.

In an article published in the Journal of the Acoustical Society of America, I investigated the effects of dialect on the spectral properties of stop bursts. An impressing finding is that besides linguistic information, i.e., the place of articulation and the stress, bursts, which are extremely short segments of speech, can also encode dialectal information. A classification model using decision trees showed that skewness and standard deviation enable the classification of bursts across dialects.

vowel Coarticulatory effects of Cypriot Greek fricatives on the F1 and F2 of the following vowel.

Several studies have explored the acoustic structure of fricatives, yet there has very little acoustic research on the effects of dialects on the production of fricatives. In an article that was submitted this June (June 2017) at the Frontiers of Psychology, I investigate the effects dialect on the temporal, spectral, and coarticulatory properties of fricatives and aim to determine the acoustic properties that convey information about the dialect (Athenian Greek and Cypriot Greek). Productions of voiced and voiceless labiodental, dental, alveolar, palatal, and velar fricatives were extracted from a speaking task from typically speaking female adult speakers (25 Cypriot Greek and 20 Athenian Greek speakers). Measures were made of spectral properties, using a spectral moments analysis. The formants of the following vowel were measured and second degree polynomials of the formant contours were calculated. The findings show that the two dialects differed in their alveolar fricatives. Duration, spectral moments, and the starting frequencies of $F1$, $F2$, $F3$, and $F4$ contributed the most to the classification of dialect. The measures showed effects of the dialect on the acoustic structure of fricatives and on the coarticulation of fricatives with the following vowel. These findings provide a solid evidence base for the manifestation of dialect information in the acoustic structure of fricatives. The study argues that these patterns of features provide the foundation for the emergence of sociophonetic shibboleths distinguishing the speakers of one variety from the speakers of another variety.