Phonostylistic description of Croatian oral literary rhetoric

U ovome doktorskome radu proučavaju se usmenoknjiževni žanrovi koji se u hrvatskoj znanosti o književnosti i folkloristici imenuju retoričkim ili govorničkim žanrovima. Kao njihove najvažnije značajke isticale su se govornost i uvjerljivost (persuazivnost), a sada se ti koncepti proširuju propitivanjem odnosa komunikacijskog učinka i performativnosti tih žanrova s njihovom fonetskom strukturom. Iznosi se problematika definiranja i klasificiranja hrvatskih usmenoretoričkih žanrova, a zatim povijest njihova zapisivanja te stručnog i znanstvenog proučavanja. Uz navođenje mogućnosti još barem dvaju teorijskih pristupa usmenoknjiževnoj retorici (retorički pristup i teorija govornih činova), teorijsku podlogu rada čine fonostilistika i fonosemantika (fonetski simbolizam) kao jezikoslovni koncept. U teorijskom dijelu disertacije iznesen je koncept fonetskoga simbolizma, povijest njegova proučavanja te njegova primjena na proučavanje književnosti. Opisan je i fonostilistički pristup (s kratkim pregledom dosadašnjih hrvatskih istraživanja) te srodni suvremeni koncepti ikoničnosti književnoga teksta. U analitičkom dijelu disertacije fonostilistički opis primijenjen je na četiri korpusa usmenoretoričkih žanrova (brojalice, brzalice, blagoslovi, kletve). Korpuse su činili zapisi iz Arhiva rukopisnih zbirki Katedre za hrvatsku usmenu književnost prikupljeni između 2000. i 2010. godine, a tekstovi su bili računalno obrađeni (fonološki transkribirani), automatski segmentirani na glasove i slogove u programskoj aplikaciji posebno izrađenoj za potrebe ovoga istraživanja. Analizirana je učestalost glasova i glasovnih skupina (koeficijent glasovnog ponavljanja) te stupanj eufonije (omjer samoglasnika i suglasnika, omjer otvorenih i zatvorenih slogova, koeficijent ponavljanja i koeficijent otežanog izgovora). Istraživanje je potvrdilo teorijski obrazložen prijedlog klasifikacije usmenoretoričkih žanrova na fonosemantične (brojalice, brzalice) i pragmasemantične žanrove (blagoslovi, kletve).In this doctoral dissertation the researcher investigates the oral literary forms which are defined as rhetorical or oratorical genres (Croatian: retorički/govornički žanrovi). Verbal expressiveness and persuasiveness have previously been described as the primary components of these oral literary genres. The theoretical framework of the dissertation is derived from the phonostylistic approach, which defines inherent qualities of oral literary discourse (rhythm, euphony, the phonemic structure and symbolism). This research intends to result in the first systematic and articulate description of Croatian oral literary rhetoric. The primary hypothesis of the doctoral thesis is that the communication effect of oral literary genres is directly achieved by their form. The hypothesis is supported by the results of the analysis of the four oral literary genres: counting-out rhymes, tongue-twisters, blessings and curses. There are actually two segments of the hypothesis: one group of texts achieves their direct communication effect by their phonetic structure, while in the other group of texts the direct communication effect is the result of specific semantic and pragmatic requests. This overall insight, obtained by a series of phonostylistic analyses, confirms the secondary hypothesis that there are two categories of oral literary genres: phonosemantic and pragmasemantic. The additional hypothesis predicting distinction between elementary and hybrid genres could not be confirmed in this research. The research corpora are records collected through fieldwork between 2000 and 2010, and they are representative both in terms of a large number of text records and in terms of the representation of certain subgenres. Phonostylistic analyses could not be thoroughly done without previous text preparation (consistent transcription of standard Croatian orthography to unigrams with the implementation of speech assimilation in the written form) and subsequent computational text analysis. The application for automatic segmental text analysis was created in the program language Python, specifically for this research, and it was able to measure the frequency of sounds (phonemes) and sound clusters and to consistently syllabify texts. Computational text processing enabled further summary interpretations of the results, which could not have been achieved using the classical approach. In the phonostylistic analysis emphasis is put on the frequency of sounds and sound clusters, sound repetition and euphony. The results are interpreted in accordance with the theoretical framework based on phonetic symbolism or phonosemantics. Oral literary rhetoric is certainly the least studied group of oral literary genres. Its distinctiveness was controversial for a long time, and there are also significant disputes concerning the classification of oral rhetorical genres. The aim of this thesis is to implement a single, phonostylistic approach to the analysis of the genres which could be classified (in accordance with the theoretical framework) into basic phonosemantic and pragmasemantic genres. Unlike the previous approaches which dealt with the problem eclectically or which focused on particular characteristics of certain genres, the approach adopted in this doctoral thesis dwells on the intrinsic traits of the texts which represent the basic oral rhetorical genres. The four analyzed genres are counting-out rhymes, tongue-twisters, blessings and curses; the only basic oral rhetorical genre not analyzed in this thesis are verbal charms because they are labeled as a transitory or mixed genre between phonosemantics and pragmasemantics. Although both the rhetorical and pragmalinguistic approach (the latter described through speech act theory) had been recognized as suitable for the theoretical framework, the research space inside the form of the doctoral dissertation was not sufficient for the complete triangulation of the results. The tripartite approach could have been used for a single genre but this would have disabled synthetic insights. The phonostylistic approach was recognized as the one which focuses most on the texts themselves and the collected records offered minimum insight into the context of oral performance. The texture of these records, which could be best described through the concept of spoken language qualities, was also not quite complete and that was the additional reason to choose phonostylistic methods. The lack of audio-visual records disabled analyses of intensity, the tempo of speech, gesture and facial expressions. It was possible to partially reconstruct the other three qualities, i.e. intonation, pause and the actual context. Therefore, the phonostylistic approach in this dissertation could be best described as phonological stylistics as opposed to speech stylistics. The linguistic framework for phonostylistic analyses was established inside the theories of phonetic or sound symbolism (the traditional term) and phonosemantics (a relatively new term). Since there was no previous systematic review in Croatian philology (both in linguistics and in literary scholarship), it was necessary to offer synchronic and diachronic perspectives of the theoretical framework. Closer attention was therefore directed to those segments of phonosemantics which had not been described satisfactorily (the types of phonetic symbolism, the experimental studies, the problem of phonetic symbolism in literature from antiquity to the modern period), and less attention was given to structuralism and related approaches (led by the fact that the basic principles of Saussurean linguistics are considered to be linguistic axioms). An attempt was made to present the status of phonostylistics among conventionalists and traditionalists in terms of their conflicting views and therefore there is no diachronic review which would place phonostylistic research in the overall history of stylistics. At the end of the theoretical chapter there is a short section on the semiotic, that is, iconic approach to the problem of literary language because it can be seen as complementary in many respects to the phonosemantic view of the problem of the relationship between sound and content. Adopting the hypothesis proposed by Matthew Williams in his doctoral thesis on the sound-content relationship in Latin poetry (2004) that both sound and content could be considered primary and active in poetry, it is possible to make a step forward in the description of counting-out rhymes and tongue-twisters. Sound in these genres can be considered primary and active; content, on the other hand, can be considered secondary and passive. Such a sound-content relationship, where content emanates from the phonetic (sound) structure of the genre (which in its turn is directly connected with the ultimate communication effect), justifies counting-out rhymes and tongue-twisters being classified as phonosemantic genres. In contrast to them, blessings and curses are classified as pragmasemantic genres because their content (connected with the pragmatic effect of delivering something good or bad) completely organizes the sound side of the message. Thorough phonostylistic analysis confirms this basic theoretic insight in three key aspects: sound (phoneme) frequency, the regularity of sound repetition and the euphonic status. The frequency of sounds and sound clusters is described most thoroughly by exact data. The results of computational analyses are interpreted in accordance with the described practice of the evaluation of the significance of stylistic deviation from the expected norm. The conclusions about phonosemantic and phonostylistic effects were reached primarily using the basic phonetic descriptions of Croatian language sounds. The results of all analyses, from the total frequency of vowels up to the ratio between voiceless and voiced consonants, confirm the hypothesis about essential structural differences between phonosemantic and pragmasemantic genres. Whenever blessings and curses (in the corpora) showed significant deviation from the neutral sound context, neither one of these deviations proved to be phonostylistically interesting or marked. These deviations are a result of grammatical and semantic demands set before these two genres. For instance, the increased frequency of the dental alveolar /l/ could be explained by the high frequency of the active past participle used optatively, whereas the decreased difference in the proportion of voiceless to voiced consonants could be explained by the increased frequency of motif words such as blagoslov (Engl. blessing), Bog (Engl. God), vrag (Engl. devil) or dabogda (Engl. may the God bring/inflict or I hope you…), which for the most part consist of voiced consonants. On the other hand, when phonosemantic genres showed deviation from the expected norm, then it was possible to draw conclusions based on the theoretical framework. The opposition between high and low vowels (close and open vowels in IPA terms), with the increased share of high vowels, proved to be important for counting-out rhymes because this opposition is almost universal in suggesting the small-big relation. In Croatian counting-out rhymes the high frequency of high vowels, combined with the high frequency of the dental affricate /c/ (with the high optimal frequency band), suggests the meanings such as small, lovely and darling. On the other hand, tongue-twisters consistently actualize their poetic principles by increasing the frequency of those sounds which require major articulation effort, such as vibrant (especially syllabic) and voiceless consonants. Tongue-twisters also have the lowest share of vowels, which ultimately contributes to their difficult pronunciation and results in speech errors. In general, it can be stated that the phonostylistic organization of counting-out rhymes contributes to effective rhythm, whereas with tongue-twisters the phonostylistic organization makes faster and multiple articulation very difficult for the speaker. Sound repetition (assonance and alliteration foremost and then poetic homophones) could not be analyzed so precisely, but the measure of repetition regularity proved as auxiliary. The comparison of repetition coefficients for unigrams (which is relevant for determining the frequency of the occurrence of assonance and alliteration) suffices to clearly distinguish phonosemantic from pragmasemantic genres. This is especially evident in comparisons of the repetition of homophonic sequences using the repetition coefficients for bigrams and trigrams. In this view counting-out rhymes and tongue-twisters appear as genres with a prominent poetic function, as defined by Roman Jakobson. The parallelism of patterns and the repetition of the same sounds or sound sequences transfer the rates of similarity from the selection axis to the combination axis. Contiguity in form thus becomes similarity in meaning and in the case of phonosemantic genres the phonetic structure is the entity which completely controls the meaning. The increased frequency of assonance in counting-out rhymes with a prominent rhythmic tendency extremely affects the text structure. The same thing happens with the high frequency of alliteration in tongue-twisters. Frequent alliterations in tongue-twisters, either through alliterative sequences or through the accumulation of the same or similar sounds in the nearest context, exclusively serve to create a speech error. This is in accordance with numerous experimental studies which proved that the high phoneme frequency in a sequence directly correlates to the speech error expectancy. The application of different analytical methods to determine the degree of euphony or anti-euphony was also helpful in establishing a clear dichotomy between phonosemantic and pragmasemantic genres. Analyzing the vowel-consonant ratio and the open-closed syllables ratio, especially consonant clusters in the syllable onset, and finally analyzing the results of the first attempts to apply mathematical formulas for determining the articulation difficulty coefficient (this was the first phase in the creation of the so-called phonotactical calculator), it is visible that pragmasemantic genres are neutral, while counting-out rhymes and tonguetwisters are at opposite extremes from each other. Counting-out rhymes rely on the positive correlates of euphony to achieve rhythmicity and melodiousness, while tongue-twisters rely on the negative correlates of euphony. These negative correlates are primarily the excessive repetition of the same sound sequences, and especially phonotactically marginal elements (sound sequences composed of voiceless consonants and the high frequency of syllabic vibrants). After the subject of the doctoral dissertation was conceived on the basis of reference literature from different areas of humanities, and after the methodology of research was continuously developed and tested, it can be said that this original research confirms many hypotheses, that it gives clear answers to many questions, but it also probably raises more additional questions. It has to be stated that during the process of the interpretation of particular results, especially those connected to the correlates of euphony, it became evident that there are many deficiencies in the methods employed for this purpose. The major deficiency is definitely the emphasis on quantitative parameters at the expense of qualitative ones. Nevertheless, this is the first major research which employs the developed computational method and probably the most important contribution is that all the results do show tendencies in ratios. If some correlates (for instance the degree of phonotactical proximity) were not measured most precisely, then they were consistently ruled out for all genres. This means that the conclusions should not necessarily be connected with particular numbers but with the ratios that are suggested by these numbers. After the theoretical and the experimental research it can be most certainly said that the impressionistic and intuitive differentiation of counting-out rhymes and tongue-twisters as opposed to blessings and curses is more than justified. This research tries to offer the widest range of possible analytical parameters, but it could not be equally focused on all aspects which were noticed as relevant during the research. The first aggravating factor was a complete lack of many basic phonetic studies of Croatian, primarily concerning phonotactical rules and then the lack of new and improved studies on sound/phoneme frequency based on as large corpora and as different speech situations as possible. The second factor which slowed down the research but also in some way decreased the qualitative homogeneity of the interpretative section of the dissertation was the lack of institutional, technical and professional support in carrying out the computational analyses. At the beginning of the research the corpora were in the basic archive form, there were no digitalized segments, and the computer application design required stronger commitment of the author and his co-workers, very often with very unclear outcomes. Nevertheless, despite all the shortages which the author is aware of, this doctoral dissertation can surely pave the way for new, better and more comprehensive studies. The dissertation offers a historical overview of the recording process of Croatian oral rhetoric but also an overview of the theoretical insights about the subject. A theoretical framework which has proved very reliable in the interpretation of results has also been established. It has the potential for wider application, primarily in poetry, both oral and written. Since no similar research on Croatian corpora had been done, and since the most similar foreign studies had not analyzed oral rhetorical genres, the research methodology was in many aspects completely authorial. The evaluation of the research results and especially further studies (which will test and improve the methods) will show how much useful and justified it was