نشریه زبان و زبان شناسی
پیاپی 11 (بهار و تابستان 1389)

Khwadje Nasir Tusi (the Persian thinker of the 13th century AD) raises a question in his mejar-ol-'ashar, which seems to be unanswered so far. The question may be re-expressed in modern linguistic terms as follows: why does a specific string of one- and two-moraic syllables (ex. u - - u - - u - - u -) in a specific meter (ex. moteghareb-e mosamman-e mahzuf or maghsur) have just one correct scansion (in this case: faulon faulon faulon faal = u - -/ u - -/ u - -/ u -), and other logically possible scansions (ex. mafailo mostaflon faelon = u - - u/ - - u -/ - u -, or faulon mafailo mostafelon = u - -/ u - - u/ - - u -) are all incorrect (Khwadje Nasir 1369/1991: 35-36)? He himself cleverly evades giving a direct answer to this, but as a matter of fact his question must be considered as a key one in Persian classical meter. Some scholars believe that the question itself is based on a wrong presupposition because there is no metrical foot in this kind of versification at all (Elwell-Sutton 1976: 85). Others believe that the question is not at all wrong, but some of the traditional metrical feet have to be changed anyway (Khanlari1373/1995: 155; Utas 1994). Taking a position in line with the second group,the present paper argues that according to a metrical analysis of the words of standard Persian, the feet of this language are right dominated, bounded and quantity sensitive (Hayes 1981) or iambic with an “End Rule Final” rule (Hayes 1995). It is very interesting that by analyzing the words of any Persian misra' (hemistich) into its feet, one reaches its natural rhythmical elements traditionally called atanin. Interestingly enough, any Persian speaking child can easily analyze a misra' into its atanin simply after seeing just a few examples. For instance the meter of moteghareb-e mosamman-e mahzuf atanin (" faulon faulon faulon faal") is analyzed into its atanin as "ta tan tan ta tan tan ta tan tan ta tan". The paper employs this finding as a basis both to re-describe the metrical feet of Persian versification and to answer Khwadje Nasir Tusi's question.

Spirantization is a categorical phonological pattern in which affricates are produced without closures. The acoustic analysis of spirantization of the Persian affricates showed that, in connected speech, the friction of affricates is licensed by cue (Steriade (1997, 2001 and Flemming 2002) and not by prosody (Nespor and Vogel 1986). In order to minimize the articulatory efforts and costs in clusters composed of affricates, Persian speakers do not implement the most salient perceptual correlate of affricate manner, i.e. silence duration, in the waveform. Spirantization is a gradient phonetic weakening in which closure of affricates reduces in time and at the same time affricate-fricative opposition is maintained.

This paper investigates the raising of the low back vowel /ɑ/ to high back vowel [u] in the context of nasal consonant –n within the experimental phonology framework of Beddor (1983). There exists an inconstancy in Persian. The raising process happens in some cases but not in others. Investigating the source of this inconsistency, the paper attempts to represent a phonetic explanation for the phenomenon. Focusing on the acoustic parameters of nasalization and their effects on vowel height the phonological pattern of this phenomenon was determined. The pattern proved to be a style–dependent process which takes place only in colloquial Persian. Two kinds of speech data were acoustically analyzed: (1) Four words in formal style including the sequence [ãn] and the same four words in colloquial style including the sequence [ũn]; (2) Three sustained speech sounds [a], [u] and [n]. All the words and sounds were pronounced in isolation by five male Persian speakers. The results showed that according to parameters F1, F2 and F2-F1, while nasalization has no decreasing effect on the F1 of the low back nasalized vowel, it increases the F1 of the high back nasalized vowel compared to its corresponding oral vowel. In addition, the low back nasalized vowel appears to be further front than its corresponding oral vowel. Finally, since the value of acoustic salience for [ũn] was significantly higher than the same value for [ãn], it was concluded that this style-dependent phonological raising makes the sequence more perceptible in the colloquial style.

Vowel harmony is a phonological process during which vowels acquire the feature(s) of nearby vowel(s). In the first part of the present article, patterns of Persian vowel harmony are discussed. The paper shows that vowel harmony occurs within morphemes and across morpheme boundary in this language and certain consonants are transparent, while others are opaque in different harmony patterns. It is further argued that when vowel harmony occurs within a morpheme, the target is in most cases [e,æ,o] and the trigger is [i,u,A], and when vowel harmony occurs across morpheme boundary, all the six vowels of Persian can be the trigger, whereas only [e] and [æ] are targets. In the second part of the article, patterns of Vowel-to-Vowel coarticulation in Persian are investigated. V-to-V coarticulation is a phonetic process during which vowels in a string of segments affect one another, but the influences are too subtle for hearers to detect. V-to-V coarticulation can only be detected through instrumental investigation. The acoustic investigation of the laboratory speech of two adult male speakers of Persian showed that in this language V-to-V coarticulation is sensitive to vowel quality, the voicing characteristics of the intervocalic consonant and its place of articulation.

In this paper, a continuous speech recognition system for the Persian language is introduced and the roles of acoustic and language models are examined. Context-independent and context-dependent acoustic models are used in the system and the results of their employment are presented. Moreover, word-based, POS-based and class-based triphone language models are extracted using Persian text corpus and incorporated in the speech recognition system. In addition, a grammatical language model based on GPSG is implemented in the system and is used in combination with the statistical language model. Experimental results demonstrated hat as expected, context-dependent phonetic models show the best performances. Also, the word-based triphone language model showed superiority over other statistical language models. Moreover, the combination of grammatical language models with statistical ones proved to lead to better recognition results. The system introduced in this paper is the first Persian speech recognition system capable of practical usage and is based on numerous research works performed for its design and implementation.

The present paper focuses on three distinct types of full reduplication processes in Persian (namely 'pure full reduplication', 'medial full reduplication' and 'final full reduplication') and examines their syntactic and semantic properties within the framework of Morphological Doubling Theory (MDT) (Inkelas and Zoll, 2005), an alternative approach aiming at resolving the problems with mainstream phonologically-based theories of reduplication. The question to be addressed here is: do the existing patterns of Persian full reduplication support Inkelas and Zoll's Morphological Doubling Theory? The data gathering process involved a library search of the existing Persian morphological studies(such as those by Kalbasi 1992 and Shaghaghi 2000). The patterns in question were then analyzed using the MDT constructional schemas methodology. Among the most important results obtained, which are mostly in agreement (and partially in disagreement) with the predictions made by the MDT approach, the following are significant: 1) patterns of Persian full reduplication are not merely limited to the morpheme level, but rather cover a full range of linguistic expressions from a single word to an entire syntactic construction; 2) depending on the RED pattern involved, semantic features of the output may vary on a relative continuum ranging from purely iconic to totally idiomatic/metaphorical meanings and, in some cases, they are affected by contextual parameters; 3) patterns of Persian full reduplication are sometimes of stylistic significance and are subject to certain register restrictions; and 4) some patterns of Persian full reduplication obviously challenge the MDT analytical methodology with regard to its semantic component.

Speech has both segmental and suprasegmental features. Suprasegmental features are more important in speech since without them speech becomes mechanical and machine-like. Suprasegmental features include stress, intonation, juncture, pause and the like. Among these phenomena, pause is the one that has received less attention because most people equate it with juncture and use the two terms interchangeably. The present research was designed to examine the acoustic correlate, position and frequency of pause in terms of sociolinguistics variables. Data were collected from formal and informal speech styles and analyzed with the help of Praat software.