Master of Science / Food Science Institute / Edgar Chambers IV / Sorghum (Sorghum bicolor [L.] Moench), is an ancient grain that possesses many health and economic values. Understanding the differences in the flavor profiles among cultivars is essential to increase the usage of the grain for food application purposes. The main objectives of this study were to 1) develop a model system to taste grain varieties and 2) develop a sensory lexicon to describe the flavor profile of sorghum grains. Fifty-seven sorghum cultivars including US commercially available samples as well as important breeding lines from around the world were included and investigated in the study. After thorough investigations, this study developed a model which includes cooked grain and cookie applications to facilitate flavor characterization among different cultivars. The cooked grain and cookie recipes developed in this project are neutral and consistent. Therefore, they can serve as systems prototypes for identification of a grain’s flavor profile and support descriptive analysis studies of flavor among grain cultivars. The developed model has been applied in sorghum and wheat to successfully characterize the flavor profile among different cultivars. Moreover, a sensory lexicon with 28 descriptive terms was developed by the trained panel to describe the flavor of sorghum in grain form and in finished products. Some of the terms in the lexicon were starchy, beany, cardboard, oil-heated, brown-sweet, buttery, umami, overall green, musty dusty, woody, wheat like, sweet, salty, sour, bitter and metallic. Each attribute has a descriptive definition to describe the term and two to three descriptive references. The panel validated the effectiveness of the developed lexicon using a set of 20 sorghum cultivars, which were presented to the panelists in two forms: cookies and cooked grains. The descriptive analysis results were analyzed using PCA to produce sensory maps with key attributes associated with each sample. The generated maps for both cooked grain and cookies application showed clear differentiation in flavors among sorghum cultivars indicating that the lexicon can be used to effectively characterize sorghum’s flavors in multiple applications. Such understanding will help to support researchers, food producers, food manufacturers and contribute to promoting the use of sorghum grains in food applications.

Master of Science / Department of Human Nutrition / Koushik Adhikari / The cosmetic industry is growing rapidly, and one popular category is lip products. Women consider lip products a necessity and many would not leave home without it. Though the bigger cosmetic companies may have internal cosmetic lexicons, they are not available in the public domain. The purpose of this study was to develop a lexicon for descriptive sensory testing of lip products. Lip balms, lip glosses, and lipsticks were tested in this study. In part 1 (EXPERIMENT 1), two focus groups were conducted to understand women's perceptions of lip products, and also to elicit desirable and undesirable characteristics in the products. The women's idea of a perfect lip product was: clear/sheer/neutral color, smooth, not sticky, moisturizing and flavorless/tasteless. In part 2 (EXPERIMENT 2), a lexicon was developed for the lip products. Attributes were categorized under "Initial Texture", "Initial Appearance", "After Appearance" and "After Texture." In part 3 (EXPERIMENT 3) of the study, the lexicon was validated by testing various lip products using lexicon developed in part 2. The analysis of variance (ANOVA) results by product type indicated that the lexicon was able to differentiate among the lipsticks, lip balms and lip glosses. The lexicon was further able to show similarities and differences within a product type. Principal component analysis and cluster analysis, which are both multivariate techniques, validated the inferences from the univariate analysis (ANOVA). The two panelist groups (three panelists from the lexicon development panel – group 1, and three new panelists – group 2) showed no differences (P > 0.05) in attribute evaluations for all the samples tested. The lexicon developed in this study could be used to identify similarities and differences in other lip products such as lip plumper, lip liners and multi-use products. The authors hope that this research is extrapolated to other aspects of the personal care industry, such as hair care and skin care, and can aid in product development, product optimization, and claim substantiation.

Analytic knowledge about the systematic variation in a language has an important place in the description of the language. Such knowledge is interesting e.g. in the language teaching domain, as a background for various types of linguistic studies, and in the development of more dynamic speech technology applications. In previous studies, the effects of single variables or relatively small groups of related variables on the pronunciation of words have been studied separately. The work described in this thesis takes a holistic perspective on pronunciation variation and focuses on a method for creating general descriptions of phone-level pronunciation in discourse context. The discourse context is defined by a large set of linguistic attributes ranging from high-level variables such as speaking style, down to the articulatory feature level. Models of phone-level pronunciation in the context of a discourse have been created for the central standard Swedish language variety. The models are represented in the form of decision trees, which are readable for both machines and humans. A data-driven approach was taken for the pronunciation modelling task, and the work involved the annotation of recorded speech with linguistic and related information. The decision tree models were induced from the annotation. An important part of the work on pronunciation modelling was also the development of a pronunciation lexicon for Swedish. In a cross-validation experiment, several sets of pronunciation models were created with access to different parts of the attributes in the annotation. The prediction accuracy of pronunciation models could be improved by 42.2% by making information from layers above the phoneme level accessible during model training. Optimal models were obtained when attributes from all layers of annotation were used. The goal for the models was to produce pronunciation representations representative for the language variety and not necessarily for the individual speakers, on whose speech the models were trained. In the cross-validation experiment, model-produced phone strings were compared to key phonetic transcripts of actual speech, and the phone error rate was defined as the share of discrepancies between the respective phone strings. Thus, the phone error rate is the sum of actual errors and discrepancies resulting from desired adaptations from a speaker-specific pronunciation to a pronunciation reflecting general traits of the language variety. The optimal models gave an average phone error rate of 8.2%. / QC 20100901

L’objectif de cette recherche est d’étudier le développement du vocabulaire en production et en compréhension orales en turc (L1) et en français (L2) des enfants bilingues franco-turcs âgés de 6 à 10 ans en France en comparaison aux enfants monolingues français et turcs. Le test allemand « Wortschatz- und Wortfindungstest für 6- bis 10-Jährige » (Glück, 2011) a été utilisé. Celui-ci a permis d’analyser les stratégies de réponses en fonction de six types de questions posées, les différentes catégories lexicales (1 : vie quotidienne / famille ; 2 : non spécifique et 3 : école), les phénomènes sémantiques, les items connus et non connus des enfants ainsi que leur répertoire linguistique. Le corpus est constitué de 180 enfants dont 60 bilingues, 60 monolingues français et 60 turcs fréquentant le CP, le CE2 et le CM2. Les résultats révèlent un meilleur niveau en L1 en CP puis une amélioration en L2 en CE2 et un écart rapproché en CM2. Les questions difficiles sont les Types 1, 2, 4 et 5 dans les deux langues pour les CP ; 1, 4 et 6 pour les CE2 en L2 et 1, 2, 4 et 5 en L1 alors que pour les CM2, il s’agit du Type 1. Les catégories sémantiques les mieux réussies en L2 sont les 1 et 3 alors qu’en L1, il s’agit de la 2. Les sous-extensions sont nombreuses en CE2 et CM2 dans la L2 alors que les CP ont tendance à sur-extendre. Les bilingues font davantage appel à des stratégies de compensation comme la catégorisation, l’approximation ou la traduction. Le retard constaté pour le vocabulaire total chez les CP est vite rattrapé en CE2. De fortes et positives corrélations (Pearson et al., 1993) ont été constatées entre les langues : plus la L1 est maîtrisée, plus la L2 l’est aussi. / This research examine lexical development in receptive and expressive skills in Turkish (L1) and French (L2) of French-Turkish bilingual children aged between 6 and 10 years in France compared to French and Turkish monolinguals. The German test scale “Wortschatz- und Wortfindungstest für 6- bis 10-Jährige” (Glück, 2011) allowed to analyze spoken vocabulary, production and comprehension strategies according to 6 question types, different categories (1: everyday life/family; 2: general; 3: school), semantic phenomena, known and unknown items and their linguistic repertoire. The sample consists of 180 children, including 60 bilinguals, 60 French and 60 Turkish monolinguals composed of first, third and fifth grade pupils. Results show that bilingual have better vocabulary level in their L1 in the first grade and they made good progress in L2 in the third grade and the lexical gaps are significantly reduced in the fifth grade. The most problematic clauses are Types 1, 2, 4 and 5 in the first class; 1, 4 and 6 in L2 and 1, 2, 4, 5 in L1 in the third class whereas they are Type 1 in the fifth class. The most successful semantical category in L2 are the 1 and 3 whereas they are the 2 in L1. The under-extension is more in L2 in the third and fifth classes while the over-extension is more for the first class. Bilinguals make more use of compensation such as categorization, approximation and translation. Bilinguals’ total vocabulary is lower than monolinguals’ however the delay is regain in the third grade. According to the correlation (Pearson et al., 1993), there is a strong and positive correlation between two languages: the more we promote L1 development, the more we promote L2. / Bu çalışma Türkiye ve Fransa’daki 6 ile 10 yaş arası çift dilli çocukların tek dilli çocuklara kıyasla anadilleri olan Türkçe ve ikinci dilleri olan Fransızcada anlama ve kullanma becerilerindeki kelime gelişimini araştırmaktır. Kullanılan Alman testi « Wortschatz- und Wortfindungstest für 6- bis 10-Jährige » (Glück, 2011), 6 soru türüne göre sözcük stratejilerinin, 3 farklı kategorinin (1: günlük yaşam/aile; 2: genel; 3: okul), çocukların anlamsal olgularının, bilinen ve bilinmeyen öğelerinin ve dil repertuvarlarının analizlerini yapmamızı sağlamıştır. Derlemimiz 180 çocuktan oluşup 60’ı çift dilli, 60’ı tek dilli Fransız ve diğer 60’ı ise tek dilli Türk birinci, üçüncü ve beşinci sınıf öğrencileridir. Çift dillilerin ilkokul birde ana dillerinde iyi bir sözcük seviyesine sahip olduklarını, üçte ise ikinci dillerinde ilerleme kaydettiklerini ve beşinci sınıfa gelince ise sözcük yetisi farkının azaldığını ortaya koymaktadır. En zor soru türleri birler için 1., 2., 4. ve 5. olurken, üçler için Fransızcada 1., 4. ve 6. Türkçede ise 1., 2., 4. ve 5. olmaktadır. Beşler için her iki dilde 1. soru türüdür. Fransızcadaki en iyi bilinen tümceler 1. ve 3. kategorilerine aittir, oysaki anadilde 2. kategoridir. Anlam daralması üçüncü ve beşinci sınıfta daha fazladır. Birinci sınıfta anlam genişlemesi daha çoktur. Çift dilliler kategorilere ayırma, tahmin etme, çeviri gibi telafi stratejilerine başvurmaktadırlar. Kelime hazinesi daha zayıf öngörülen birinci sınıf çift dilliler, bu farkı üçüncü sınıfta telafi etmekte. Her iki dil arasında müspet ve güçlü bir bağlantı (Pearson et al., 1993) olduğu tespit edilmiştir: anadile ne kadar hakim olunursa ikinci dile de o kadar hakim olunur.