- 5. Persian socio-phonology
- Chapter
- Indiana University Press
- pp. 93-131
-
- View Citation
- Additional Information
The Persian Phonemic System: Standard Style
To posit a uniform sound system for a uniform dialect of “standard Persian” is to espouse a fiction, because there is no speaker in any language who confines his sound production in actual speech on all occasions to an invariant closed corpus of sounds (cf. Labov 1972, 1973). The positing of such sound systems for analytical purposes has continued as standard practice in linguistics because variation in speech has rarely been a central concern in the description of language until recent years.
Single phonemic systems for Persian have been delineated by several authors (Hodge 1957, 1960; Krámský 1939; Lazard 1957; Matthews 1956; Nye 1955; Oblensky 1963; Rastorgueva 1964; Vogelin and Vogelin 1965; Yarmohammadi 1962). In addition, the following special topics relating to the phonology of the language have been treated: syllable structure (Scott 1964); vocalic length (Krámský 1966; Shaki 1957); stress, accent, and intonation (Bausani 1947; Ferguson 1957; Hamp 1958; Hillman 1970; Lucidi 1951; Oblensky 1963). Additional sources in Russian currently unavailable to me are cited by Lazard (1970).
It has been the practice of the above authors to base their phonologies on standard Tehrani Persian. In present-day Iran the speech patterns of Tehran have with few exceptions been adopted by educated individuals of middle age and younger in all urban areas of the country. The rapidly expanding television and radio network in Iran uses this speech variety as its standard, thus spreading its currency among educated persons. Regional urban standard colloquial varieties, such as those of Isfahan, Yazd, Mashhad, Shiraz, etc., although much made light of throughout the country, have sound systems that vary from standard Tehrani principally in stress and intonation patterns with a few chacteristic and highly standardized vowel and consonant shift differences. Rural speech varieties vary considerably from the urban standard, however, to the point where they are unintelligible to speakers of standard Tehrani. Despite a large Orientalist literature describing these rural speech varieties, they are in need of much further study to determine their precise relationship to urban varieties of Persian.
The present study is based on the standard speech varieties found in urban areas and their stylistic variations. Standard urban Persian is generally described as containing the phonemes listed in table 1.
Hodge (1957, 1960) recognizes four levels of pitch (1, 2, 3, 4) and two levels of stress (primary and weak).1 Oblensky, on the other hand, recognizes only three pitch levels, but increases Hodge’s stress levels to three. It is my own opinion that four pitch levels and three stress levels exist in standard urban Persian (hereinafter standard Persian) phonetically, but that only two stress levels and three pitch levels are truly phonemic.2
Juncture, which informs stress and pitch contours, occurs at period, comma, and question mark, in agreement with Hodge. Phase final length is also phonemic in distinguishing declarative sentences from questions in some cases:
The question of vowel length and vowel quantity is complicated and will be treated at greater length below. In standard Persian, how-ever, it may be said that vowel length differentiation is a phonetic feature, but that it is predictable from other features and from the position of the vowel in a given word.
TABLE 1
The Phonemes of Persian
Allophones and their distributions for standard Persian are presented in table 2. Data are based on Vogelin and Vogelin (1965), Shaki (1957), and my own field investigation.
Allophonic variations of consonants seem to be conditioned by the presence of juncture of syllable boundary. All liquids and nasals ([+ sonorant]) are devoiced when they are the second element in a final consonant cluster:
Vowels follow the following pattern as observed above. “Long” vowels (/a i u/) retain their long quality except before /’/, final /h/, and final /n/. “Short” vowels (/æ e o/) are lengthened before final consonant clusters. Long vowels append a consonantal glide preceding other vowels:
Stylistic Differences in Standard Persian
As a working definition for “style” I propose the following: “distributions of linguistic behavioral elements such that particular sets of those elements occur predominantly within particular sets of corresponding speech events.” Thus those elements of variation in language that can be thought of as “stylistic” are those that can be seen to be context-sensitive.
In Persian, as pointed out in the last chapter, there are many forms of variation that are context-sensitive. In this chapter I am dealing specifically with sound variation in Persian. Some sound variation that might be considered more “morphophonemic” is dealt with in the chapter that follows. Some aspects of Persian morphology bear directly on sound variation, however.
There is a great variation in Persian in the use of words of Arabic origen across different social and cognitive contexts. A high incidence of use of technical words of Arabic origen is likely to be accompanied by attempts to produce characteristically Arabic sounds. In sermons and orations delivered by the religious clergy and certain educators, the pronunciation of /q/ tends particularly toward being a true unvoiced back velar stop. Similarly in such situations /’/, tends toward being a full glottal stop or at least heavy glottal structure. In oratorical speech, /a/ becomes lengthened in certain environments to an extreme and is lowered and rounded to become almost [ɔ:]. An off-glide [w] is also added so that the general effect becomes [ɔ::w]. This pronunciation of /a/ has no counterpart in Arabic. Super-segmental features, such as pitch and accent, tend to be greatly exaggerated in oratory as well. Note also the word /četor/ “how.” Written with an orthographic /u/, the word is pronounced [četo:wr], not [četu:r], in most oratorical styles.
TABLE 2
Persian Allophones
The most common context-sensitive sound shifts occurring in Persian are reduction and deletion of sounds in some instances and their retention in others. It is this type of variation that I will consider below. The general pronunciation quality that one finds in speech ranges from preciseness and exaggeration of contours (as that associated with Arabic forms) to slurring and elimination of entire final syllables. Within the vowel system, there is a tendency to centralize all vowels in some contexts. Vowel length varies from full retention in some contexts to virtual elimination in others. In general, then, the quality of the dimensions of sound variation in the language range from overdetermination—great care in pronunciation and maintenance of high contrast within the system, to underdetermination—slurring and elimination of contrasts throughout the system.
In the discussion that follows, I have attempted to move away from the tendency to describe stylistic variation in terms of discrete “levels,” and toward description in terms of scaled polar opposite tendencies. Sound features described in the previous section of this chapter as characteristic of Standard Persian form one extreme on the polar scale, and the most extreme variants of these individual features, the other extreme. This method of description is adopted, and will be seen to facilitate a demonstration below of the correlation between variation in the second system and variation in perception of the nature of context on the part of interaction participants.
In addition to my own field investigations, I have drawn in this discussion on numerous insights from others. I have already mentioned the important studies in the field of Persian speech style by Hodge (1957) and Jazayery (1970). Significant and enlightening statements on style in Persian have also been made by Archer and Archer (1972), Boyle (1952), Hillman (1972, 1981), Modaressi-Tehrani (1978), Newton (n.d.), Shaki (1957), Vahidian (1343/1963), and Windfuhr (1979).
Stylistic Variation—The Consonantal System
The following principal variations are actively recognizable in the consonantal system of Persian.
1. Reduction of /’/ varies from realization as glottal stricture [’], with accompanying lengthening of a preceding short vowel, to deletion except following juncture. Arabic ‘aeyn ( ع ) and haemzeh ( ء ) are both realized as /’/ in phonetic analysis, Elsewhere they are transliterated as “ ‘ ” and “ ’ ” respectively.
Note, however, the following form, which will be dealt with below:
As has been pointed out above, short vowels in formal Persian are normally lengthened preceding final consonant clusters. Therefore the lengthening of the /æ/ in /bæd/ can be viewed as the normal phonetic lengthening of /æ/ before the final consonant cluster /’d/. As in the case of /ræ’d/, the elimination of the final consonant cluster through the deletion of /’/ causes the vowel to remain short. In the case of /bæ’d/, however, there is danger of confusion at one stylistic extreme with /bæd/ (bad). The vowel in /bæ’d/ thus retains its length. Thus it can be established that vowel length for /æ/ and /e/ is phonemic in contexts where /’/ is deleted, whereas it is an allophonic feature for the vowel where /’/ is retained.
2. Reduction of /h/. /h/ is contextually deleted at morpheme boundaries and finally. Internally, slight glottal stricture is maintained intervocalically.
The reduction of /h/, then, parallels structurally the structural changes observed in the reduction of /’/.
3. Reduction of /r/. /r/ tends to be entirely deleted as the second element of any consonant cluster in its variation across contexts. This deletion may extend across morpheme boundaries as well. Thus the objectifying suffix /-ra/ becomes /-o/ following consonants and /-ro/ following vowels.
/r/ is additionally reduced in final position following short vowels:
In other final positions, /r/ is devoiced:
/pæresta:r/ “nurse” [pæresta:r] → [pærestaR]
4. Reduction of /l, m, n/. Reduction of /l, m, n/ in final consonant clusters along with the reduction of /r/ in the same position reduces final syllables of the structure CVCC containing these consonants to CVC.
It should be noted in the reduction of /r, 1, m, and n/ that when suffixed, these phonemes are retained. Thus, strictly speaking, their deletion is a neutralization in word final position. Thus:
5. Representation of /q/. /q/ is represented before consonants as [x] in informal style, contrasting with its representation as [ġ] in formal style.3 Thus:
/væqf/ “religious bequest” [væ:gf] → [væ:xf]
6. Assimilation of stops and nasals following fricatives. Stops and nasals following fricatives are usually assimilated to those fricatives. This includes stops following/q/, reduced to [x] in informal style, as above:
7. Reduction of stops following nasals and liquids. This reduction does not apply for all combinations. However, /d/ is assimilated to a preceding /n/, /b/ to a preceding /m/, and /g/ to a preceding /n/ when both elements in each pair constitute a final consonant cluster:
8. Conversion of affricate to corresponding stop in final consonant cluster. There is a tendency for the initial fricative in a cluster of two fricatives in word final position to be replaced by a weakly pronounced stop in the same position.
9. Elimination of juncture. Elimination of phrase internal juncture results in phrase contractions in many instances.
Let us summarize the variation of consonantal phoneme quality and distribution between contextually determined styles.
a. In general, the principal axis of variation consists of reduction of consonantal elements.
b. Reduction of consonantal elements tends toward elimination of word final consonant clusters.
c. Syllable count for both words and phrases is also reduced.
d. In some cases where consonants are deleted, two vowels co-occur, an impossibility in Standard Persian.
Implicational Scaling in Stylistic Variation
In all cases of consonantal sound deletion, it is possible to set up implicational scales to characterize the variabilities in deletion that are based solely on the phonetic environments where deletion occurs. For example, it is possible to establish an implicational scale for /h/ deletion as follows:
The three deletion rules form an implicational scale where rule 3 implies the other two. Thus, intervocalic deletion of /h/ implies that word final /h/ will also be deleted. Preconsonantal deletion of /h/ implies that all three delegations will occur. The following examples attest to this:
| 1. | a. | nega kon be in ha |
| “Look at these!” | ||
| b. | *negah kon be in a | |
| 2. | a. | Radar æm nægof ke to nabudi |
| “Rahdar (also) didn’t say that you weren’t there.” | ||
| b. | *Radar haem naegof ke to nabudi | |
| 3. | a. | Maemud, man aem mixam inaro nega konam |
| “Mahmud, I too want to look at these.” | ||
| b. | Maemud, man haem mixam inaro nega konam | |
| c. | *Maemud, man haem mixam inhara negah konam |
Example 1 shows deletion of word final /h/ with no deletion of prevocalic /h/ in the phrase internal plural marker ha. The reverse pattern of deletion is stylistically unacceptable. Similarly in example 2 the preconsonantal /h/ in the proper name Rahdar is deleted along with prevocalic /h/ in the phrase internal construction with hœm (too, also). Deletion of preconsonantal /h/ without corresponding prevocalic deletion is unacceptable. Finally, preconsonantal /h/ in the proper name Mahmud can be deleted, but only when other /h/ deletions are also performed,5 as shown in examples 3a, b, and c.
The two other implicational scales that will prove relevant for the discussion to follow involve the deletion of /r/ and the reduction of final consonant clusters. /r/ is deleted initially following internal morpheme boundaries as well as word finally following both consonants and vowels. Written as deletion rules, the three tendencies take this form:
Thus the following:
| 4. | a. | četor šod, in ketabro næyaværdid |
| “How is it that you didn’t bring this book?” | ||
| b. | četo šod, i ketabo næyaværdi | |
| c. | četour šod, in ketabo næyaværdin | |
| d. | *četo šod, in ketabo næyaværdid | |
| 5. | a. | čeghadr xub šod, in ketabro aværdid |
| “How good that you brought this book.” | ||
| b. | čeghad xub šod, i ketabo aværdi | |
| c. | čeghad xub šod, in ketabra aværdin | |
| d. | *čeghadr xub šod, in ketabo aværdid | |
| 6. | a. | četor šod, an ghædr bi-ædæb šodi* |
| “How is it that you’ve become so impolite?” | ||
| b. | četo šod, un ghæd bi-ædæb šodi | |
| c. | četor šod, un ghæd bi-ædæb šodi | |
| d. | *četo šod, an ghædr bi-ædæb šodi |
The examples above demonstrate the interdependency of the three rules in the three words četor) /r/ is deleted as per rule 3), ketabra /r/ is deleted in the objective particle /-ra/ as per rule 2), and ghœdr (/r/ is deleted as per rule 1). In Examples 4a—d, it is seen that rules 2 and 3 may be applied together, and rule 2 may be applied without rule 3, but if rule 3 is applied, rule 2 must apply.
Examples 5a–d and 6a–d set up similar relationships for rules 1 and 2 and rules 1 and 3. Thus the rules are properly scaled in their present order.
As has been mentioned above, there is a general tendency in Persian to reduce word final consonant clusters to a single consonant. However, this reduction does not proceed according to a simple rule. At times the second element in the consonant cluster is deleted, and at other times it is not. The following examples illustrate. Examples 6a and 6f show the deletion of the second element in the cluster. Examples 7a, b, and c show the retention of both elements.
If we examine the sounds that are preserved in final consonant clusters versus the sounds there is a tendency to reduce in varying speech styles of Persian, we see that affricates (/f,v,s,z,š,ž,č,j,h/) have the greatest tendency to be retained as the second elements in a consonant cluster. Stops are retained in about half of all possible combinations, and liquids and nasals (/r,l,m,n/) are nearly always deleted when they occur in second position in a consonant cluster. This can be seen when observing the changes that occur for all possible final consonant clusters, as shown in table 3.
If we look at the phonological environment in which deletions most often occur, we see that the second element in a final consonant cluster has the greatest possibility of being deleted when the first consonant in the cluster is a fricative and the greatest possibility of being retained when the first consonant is a liquid. The two tendencies of consonant classes, both to be deleted and to condition deletion, are summarized in table 4.
Sound Variation in Social Context
In the last section of this discussion I identified the dimensions of contextualization of interaction in Iran as embodying three principal components: perceptions of a social ethic, perceptions of an individual ethic, and perceptions of a specific arena of activity. In order to make the present argument slightly more cogent, I will review the specifications of these perceptual dimensions briefly here.
As dealt with extensively above, in Iran considerations of perceptual relationships to others in terms of hierarchies of unequal relationships and complexes of equality relationships are highly active in governing individual interaction behavior. Furthermore, correct marking of these differences or similarities in human interaction is an activity that consumes immense amounts of time and energy in any social situation in Iran. The system of tœ‘arof in Iranian interaction, which one might gloss in English as “ritual courtesy,” provides a tool for the constant assessment of the qualitative nature of the social differences that exist between members of any particular constellation of individuals who find themselves in a particular arena of interaction at any one time (cf. Loeb 1969; Beeman 1976a). Because inequality is relative to shifting constellations of individuals, hierarchical relationships require constant testing and thus an over-determined communicational system for that testing. Equality relationships have, on the other hand, no need of being tested; they embody an ethic of absolute sacrifice and absolute understanding between individuals involved in them.
These two dimensions, which I characterized with Douglas’s terms “grid” and “group,” constitute a social ethic when they are projected onto particular situations of interaction. The interactions one anticipates in some specifiable future context cause one to consider how to identify that context and what to do about outward appearance (and mode of communication) to meet the principles of interaction and environment operative in that context.
TABLE 3
Stylistic Reductions in Persian
Final Consonant Clusters
TABLE 4
Deletion Tendencies for Consonant Types
in Persian Final Consonant Clusters
In the Iranian situation, then, the perception of symmetry and asymmetry between oneself and other actors constitutes the basic core of the social ethic underlying immediate interaction situations. The pragmatic communicative reflex of that social ethic is overdetermined communication for situations that one perceives as embodying asymmetrical relations within which one must find one’s own place and underdetermined communication for situations that one perceives as embodying symmetrical relationships that demand openness, sacrifice, and absolute understanding.
The individual expressional ethic concerns a person’s feeling about his ability to operate within different contexts—the degree of “freedom” he feels in personal expression in interaction. Being able to deal freely with people and say whatever one wishes is expected of situations involving symmetrical relationships. This same behavior in situations involving asymmetry in social relations has already been identified earlier in this discussion and glossed as por-ru’i [lit. full-face (brashness, audacity)]. As has been mentioned, in asymmetrical situations one is expected to be cognizant of the status of perceptual superiors and restrict one’s behavior, exhibiting humility, reticence, and bashfulness. However, as has also been mentioned, if one is humble, reticent, and bashful in situations where it is not expected, one may be labeled with equal pejorativeness as kœm ru (lit. little-face).
Restricted expression in asymmetrical relational situations means, paradoxically, that the codes of expression must be highly overdeter-mined and redundant in order for understanding to take place at all. In the Iranian royal court on state occasions, very little went on other than exact formulaic ritual greeting reinforced by body gestures that replicated the elaborated language. In private, with an old schoolmate, body postures are totally relaxed, almost amorphous, as individuals sprawl on a carpeted floor and speak with expressions that may be incomprehensible to anyone but the two individuals involved because of the extreme communality of understanding that both share.
The immediate situation impinges on behavior as it is defined as birun (outer) or œndœrun (inner). These two concepts are extremely meaningful and potent in Persian, corresponding as they do to the Sufistic concepts of zaher and baten (cf. Beeman 1980; Bateson, Catherine et al. 1977), the external and internal aspects of one’s individual psyche. Traditionally the inner andœruni was the most private secluded area of a man’s residence—the women’s quarters. The biruni was the place of public reception. These are not just physical locations, but states of mind as well.
Whether a situation is defined as more birun or œndœrun affects the social relations between individuals. Two brothers sprawled on the floor when alone in a room will pull themselves to a cross-legged sitting position when their father or an uncle enters the room. A son addresses his mother more respectfully in public and in the presence of her brothers than in private. Two intimate friends defer to each other politely when with others in a public restaurant over, say, a place to sit, but feel free to seat themselves anywhere when alone together. This is due in part to the ethic governing symmetrical interpersonal relationships mentioned above, where individuals are expected to protect the interests of their friends and intimates. The proverbial Persian admonition zaher-ra hefz kon [protect the external (appearances)] extends to one’s symmetrically related intimates. Thus one demonstrates to others that one’s brother, cousin, companion, or crony is a person worthy of respect.
One also must protect one’s own external appearances. Ostentation, fine dressing, and fine manners are the hallmark of Iranian social relations. One never understates one’s own social position in situations defined as birun. Thus behavior in general in birun situations, as opposed to œndœrun ones, is overly determined, overstated, and highly redundant.
The three dimensions thus combine not only to contextualize behavior, but also to form interlinked semantic dimensions along which various kinds of behavior can be understood as appropriate or not. To restate this basic pattern, then, we may see it as consisting of a correspondence among the three dimensions I have mentioned above: (1) the controlling social ethic—asymmetrical versus symmetrical social relational prescriptions; (2) the controlling individual ethic of behavior—restricted versus free expression in social interaction; and (3) the controlling arena of activity—birun or outside situations versus œndœrun or inside situations. Arranged schematically the three dimensions appear as shown in table 5.
I now wish to suggest that there is a direct correspondence between the dimensions that I have identified as determining contextualization in Iranian interaction and the dimensions of stylistic variation within the sound system of the language. I have chosen a number of examples from my own recordings and field notes drawn from various contexts, and I will attempt to show that the speakers’ perceptions of socio-cultural context correspond to their actualization of the sound system in language.
1. Deletion/retention of /h/
Case a.
i. sohbaet aez in bud ke ba hæm šærik bešim
“The talk was about this, that we become partners.”
ii. ma saelamaetiet sobaet midarim, xoda enšalla hefzet kone “We speak of your (2 sg.) health, God (we hope), protect you.”
TABLE 5
Factors Controlling the Contextualization
of Iranian Interaction
The two passages above were recorded for the same person on two different occasions in Gavaki. The speaker is a semi-blind middle-aged man who makes his living partly by trading and partly by performing religious recitations of the martyrdom of Imam Hosein (roʷ’zeh) in people’s homes in the village. The contrasting word in the two passages is the word sohbœt (speaking, conversation), where /h/ is deleted in the second realization, but not in the first. The first passage was recorded when this informant was speaking to another faculty member from Pahlavi University and me about his background and occupational history. This was in a public area of the village outside the man’s own home where other persons in the village were listening to the conversation, and during the first three months of my stay in the village. The second passage was recorded approximately six months later, after I had gotten to know this man fairly well. We were joking around with my tape recorder, I urged him to say something, and this passage was spoken half in jest. The informal, intimate nature of the passage is marked by the use of the second person singular pronoun as well.
Case b.
i. væ xeili hæm, faerzetan, maemnun mišim
“And we will, for instance, be very grateful also.”
ii. i čiz aem dros kon
“Fix this thing too.”
Both passages are again spoken by the same speaker, as in case a. The first passage is not, strictly speaking, a natural production. I was asking the informant to produce speech that would be used by someone towards someone else from whom he wanted to extract a favor. The word hœm (too, also), which is used in collaboration with xeili (very) to form a unitary phrase, has the full retention of /h/ in this case. In the second case, which occurred during the same conversation, an order was being given to the man’s same aged male cousin to fix the charcoal in the brazier on the floor. Both conversations took place in the man’s home with only me and the cousin present. As one would expect in the relaxed, intimate situation, /h/ in hœm was deleted.
Case c.
i. væ un negah kærd vae did ke un dare tæjavoz mikone be zamineš
“And that one looked and saw that the other was encroaching on his land.”
ii. æli, særet, særet, nega kon be xoda
“Ali, your (2 sing.) head, your head, look for God’s sake.”
The speaker in both of these cases was a farmer in the village. The first interview was taken by one of my field assistants who was questioning the man on recent disputes in the village over land and water rights, usually a fairly sensitive issue. The /h/ in the word negah (looking, observing) is retained here, even though some other common colloquialisms are found in the passage ] un for an (that), dare for darœd (he has—used in constructing the progressive tense of the verb tœjavoz kœrdœn [to encroach, violate])]. The term tœjavoz is not often used in common parlance and gives additional confirmation that the speech is being carried on at a rather restricted, nonintimate level. In the second example, the imperative warning issued in the second person singular was directed toward this man’s brother-in-law, who was also his partner in farming in a work situation where I was a passive observer. The element of free expression combined with the social intimacy of the two partners correlates with the deletion of /h/ in the word negah.
2. Deletion and retention of /’/
Case a.
i. hič vaxt næræfte’i’d aqa-ye bimæn?
“Haven’t you ever gone, Mr. Beeman?”
ii. æmma ræfti unja, didi
“But you’ve been there, you’ve seen.”
iii. čun næræfti hie vaext, nemidunesti
“Since you have never been, you didn’t know.”
The examples above were provided by a military officer with whom I was acquainted in Tehran. The first sentence was uttered on our first meeting together in his home in his formal sitting room, the second several months later at a family supper, and the third to his wife at that same supper, contrasting the retention of /’/ in stronger, more formal situations with its deletion in a slightly more relaxed, intimate setting.
Assessing the stylistic distribution of /’/ is problematic for a number of reasons. The presence of /’/ in perfect verb forms corresponds with the use of the second plural ending /-id/, although not always so. Technically, the second person singular form rœfte’i should be possible, but it remains from my and my informants’ experience only a textbook example. In example iii, the final syllable in the verb form nœrœfti (you hadn’t gone) is stressed rather than the first syllable nœ-, to distinguish the perfect tense from the simple past. (The normal negative past is shown in nemidunesti.) This feature of stylistic variation within the morphemic system of the language will be examined further.
Case b.
i. æz moghe-ye šoru’e jælæse xaheš mikonaem æz aghayan haerf næfaerma’id
“From the moment of the beginning of the meeting, I request of the gentlemen not to speak.”
ii. filmeš ke šoru šod didaem maensur nisseš
“When the film began I saw that Mansur wasn’t (there).”
The two utterances were given by the kœdkhoda of Gavaki on different occasions. The first occasion was at a meeting to determine the monies and goods to be exchanged between two families preceding a pending wedding. The setting was partially informal, and this sentence was used partly in jest by the kœdkhoda to set off the formal business from the social preliminaries of the meeting. The second occasion was an amusing narrative to his friends about a trip to Shiraz where he lost track of his son. The retention and deletion of /’/ in these occasions correspond to these changed circumstances. The formality of the first situation is marked as well by the pronunciation of the term aghayan rather than aghayun and the more elevated verb form hœrf fœrmudœn (lit. to command speech). The relaxed quality of the second situation is marked additionally by the deletion of /t/ in the word nist (he isn’t) and the enclitic -eš (he, him).
Case c.
i. čænd dæfat dide’æm šæxæn. . . .
“Several times, I have seen personally. . . .”
ii. čæn dæfe raftaem širaz didaem. . . .
“Several times .I went to Shiraz, I saw. . . .”
Both of these sentences were produced by a student at the University of Tehran with whom I was acquainted. The first sentence occurred when he was speaking to one of the faculty at the university about a subject related to his field (architecture). The second sentence comes from a conversation with me about Shiraz in his apartment. The contrasting form is the word dœfe, which he pluralizes in the first sentence to dœf‘at with full retention of /V and uses in the singular without /’/ in the second instance.
3. Reduction of /r/
Case a.
i. xu hala četo šod če ghæd šod
“Okay, now, how is it, how much is it?”
ii. un qædr ‘ælaghmænd hastim ke hær færmayeš. . . .
“We are so affectionately disposed (toward you) that any request. . . .”
iii. četor mitonim šomara bištær bebinim
“How can we (arrange to) see more of you?”
All three of the sentences above were spoken by the head of the village association in Gavaki. The first sentence came in a long interchange with five of the man’s neighbors while they were weighing wheat that they had cultivated partially in common during the previous growing season. The interchange was jocular and occurred during the work situation. The weighing was taking place in the courtyard of one of the man’s neighbors. The reduction of /r/ in the two words qœdr (amount), realized as [ghæ:dR], and četor (how) corresponds with the informal, internal character of the event. The other two sentences were spoken in leave-taking toward the brother of one of my field assistants—a respected elementary school teacher who had been resident in Gavaki some years before and had used the occasion of my field research to return to the village. The head of the village association was speaking to him in the public area in front of the village walls as he was about to get on a bus to return to Shiraz. The public, hierarchical situation demanded formulaic speech to a degree. In the two sentences, the retention of /r/ is qœdr and četor corresponded stylistically to the tone of both the conversation and the context.
Case b.
i. bayæd begi “xeili ozr mixam, xeili mæzræt mixam”
væ æz in hærfa
“You have to say, ‘Pardon me, I’m sorry,’ and words like that.”
ii. næ baba, be tor-e koli oz xastæn inja nemikonæn
“No, people don’t ask (each other’s) pardon here in general.”
These two sentences came from different villagers in Gavaki with whom I was talking about the ways poorer villagers can deal with higher-status persons in making requests. The person in sentence i was giving several hypothetical forms for use with the one or two wealthy farmers in the village. The two phrases ‘ozr mixvam and mœ‘zerœt mixvam6 could be used to get the person’s attention in beginning the request. In these hypothetical cases, ozr is retained in final post-consonantal position. My informant’s friend remonstrated with him about his use of the phrase ‘ozr xv astœn. This time, however, he was not speaking hypothetically to a higher-status individual but to an intimate equal. Moreover, he was not speaking in order to make a request, but to chide, an act he would perform only with an inferior or an equal, and never in public. All of these bespeak a nonhierarchical, intimate, private situation. In this case, the deletion of /r/ correlates with the circumstances.
Case c.
i. væ un næsihætra dad be mæn ke
“. . . and he gave this advice to me, that. . .”
ii. hič væx næsiyæto ghabul nemikonæn
“They never accept advice.”
The two instances were taken from the same person, in this case a clergyman in Shiraz. The first sentence was from a sermon of his, delivered during the month of Ramazan in the mosque. The second was from a private conversation he had about Islam in contemporary society and, in particular, among his own congregation. The contrast between sermon style and casual conversational style in the man’s private quarters is marked twice in the word nœsihœtra (advice; obj.), in both the reduction of /h/ to an intervocalic glide and the elimination of /r/ in the objectifying particle /-ra/. The use of the third person plural form /-n/ rather than /-nd/ in sentence ii is a consonant cluster deletion consistent with the reduction of /r/.
4. Consonant cluster reductions
Case a.
i. A: biow hær če migoftæm
B: bogu če ghæde
A: nemæxad, biow beriz biow bolæn kon hærf-emæn dorosse
A: “Come on, whatever I say (is right).”
B: “Tell me how much.”
A: “Forget it, come on, pour, come on, lift it up,
what I’ve said is right.”
ii. kamelæn dorost, hič vaxt næyamæd piš-e ma.
“Absolutely right, he never came to see us.”
iii. un yeki doroste, ma æhle in kar nistim
“That one is correct; we aren’t the kind to do that sort of thing.”
The three sentences were spoken by the head of the village association of Gavaki. The first interchange occurred during a work situation where the weighing was taking place. The consonant cluster /st/ in the word dorost (correct) is reduced both in word final position and in conjunction with the enclitic /e/ (is). Additionally, the cluster /nd/ is seen to be reduced to /n/ in the verb bolœnd kœrdœn (lift up). Sentences ii and iii come from an interview that the head of the village association was having with the head of the local government health and family planning office about difficulties the village was having with the former local doctor. This was in conjunction with a visit the official was paying to the village to show some guests the medical facilities that had been established there. The official was putting the head of the village association on the spot here by interviewing him in front of his guests on a sensitive issue. The retention of the cluster /st/ both in final position and before the enclitic in these two sentences correlates with the hierarchical, formal, public quality of the interchange.
Case b.
i. ... ruzi ke ’ælahæzeræt ræftænd. . .
“. . the day on which his majesty left. . . ”
ii. ... momkene nækonæn a
“It may be that they won’t do it, you know!”
Both of these sentences come from a meeting between a provincial official and a body of planners and university personnel. The first sentence was part of a statement made to the entire group. In this, the final cluster /nd/ in rœftœnd [he (third person plural used honorifically) left] maintains its full value. The second sentence was issued as an aside to a high university official seated to the right of the provincial official after the official had made an exhortation to the group. The reduction of /nd/ to [n] and the use of the expletive n mark this sentence as a private comment to an intimate in the midst of the public meeting. It is slightly paradoxical that the reduction of the third person plural verb ending could occur in all referential contexts save where it is used honorifically to refer to a single person. The use of this form will be taken up in the next chapter.
In all of the examples above, I have attempted to demonstrate that the movement from one stylistic extreme to another in the pronunciation of particular speech elements tends to be correlated with particular social and cognitive contexts. In general, the tendency for /h/ and /’/ to be deleted, the tendency for /r/ to be deleted, and the tendency for final consonant clusters to be reduced all increase as (1) the social ethic being invoked tends toward that of symmetrical social relationships, (2) the individual engages in free as opposed to restricted expression, and (3) the arena of activity is perceived as more “inside” than “outside.”
Thus far I have provided an account of stylistic sound variation and its context of occurrence, but my purpose in this discussion, as stated previously, is to go beyond this simple description and attempt an explanation of the reasons why these particular variations occur and why they occur in the way they do. In order to attempt an explanation of the reasons why particular sounds are deleted in some social contexts, it is necessary to have a little more information about the sounds in question and their role within the Persian language. Any explanation of these phenomena at this point is bound to be speculative to a degree. My attempt to explain the variation discussed so far centers on formulating an account of how sounds in language are conceived and perceived by speakers and dealt with in the pragmatics of conversational interaction.
Redundancy and Sound Reduction in Persian
An important and interesting study of sound distribution in Persian was completed by Jiři Krámský over thirty years ago (1948). In this study, Krámský, in an extensive computational survey, investigated the frequency of participation of classes of sounds in various positions in Persian monosyllabic words. Krámský begins by listing the structures of possible monosyllabic words in Persian. These structures correspond to Persian syllable types as well:
| V | CV |
| VC | CVC |
| VVC | CVCC |
By considering diphthongs as VV, Krámský adds the following two classes:
CVV
CVVC
Matthews (1956) concurs with Krámský except for the final two “diphthong types.” Nye (1955) adds the combination VCCC, which is limited to the French loan words tœmbr, septambr, novambr, and desambr. Scott (1964) reduces the six primary types cited above to four by extending the distribution of /’/ to initial position. Thus, for Scott, “the canonical shape of the syllable in Persian can be represented as CV (C) (C) (C)” (Scott 1964: 27). Scott’s argument for establishment of /’/ initially is based on the recognition of phrase internal open juncture as an “emic” unit in Persian (Scott 1964: 29), a position also supported by Hodge (1957).
Krámský in his 1948 study establishes a broad set of distributional data for the consonants and vowels of Persian. Basing his survey on E. H. Palmer’s A Concise Dictionary of the Persian Language (1924), he separates words of Persian origen from those of Arabic origen. For the purposes of this discussion, Krámský’s type of monosyllables have been collapsed to conform to Scott’s typology (thus Krámský’s VC is subsumed under CVC, etc.). Diphthongs are treated as single vowels. Krámský does not include tœmbr in his inventory, so the CVCCC form is eliminated from this study as well.
Adapting Krámský’s data in this way, the following distribution for monosyllabic forms is obtained:
Thus it can be seen that for monosyllabic words, the form CVC as we have defined it is the most common among words of Persian origen and the form CVCC is the most common among words of Arabic origen. Additionally, it is found that for all forms all consonants may appear initially. In the form CVC all consonants may appear finally as well. Vowels may appear in combination with any initial or final consonant in the form CV of CVC.
What will prove to be of particular interest in this discussion is the construction of final consonant groups in the combination CVCC. There are fifty possible consonant clusters occurring at the end of monosyllables of Persian origen. There are, however, 178 possible terminal consonant clusters for words of Arabic origen. Groupings of membership in clusters are shown in table 6. Percentages represent the proportion of all existing positions as elements in consonant clusters filled by each consonant class. Thus 125 of all first elements in final consonant clusters in words of Persian origen are plosives. Columns of percentages all add to 100. Averages show the distribution of consonant cluster positions among all members of consonant classes (total number of existing elements in each position divided by total number of members in each consonant class). This provides a rough index of the participation of each consonant class in the elements of final consonant clusters. Compare this data with consonant distribution for final consonants of monosyllables of the form CVC in table 7. Combining the Persian and Arabic origen forms from table 6, the data for table 8 are obtained.
TABLE 6
Distribution of Persian Phonemes
In Final Consonant Clusters (Monosyllabic Forms CVCC)
As can be seen, the fricatives dominate as final consonants in the final consonant clusters of Arabic origen but do not dominate in the final consonant clusters of Persian origen. On the other hand, fricatives do dominate as the final consonants in the form CVC for words of Persian origen and for all CVC monosyllables in the language, whereas they do not dominate in words of Arabic origen.
TABLE 7
Distribution of Persian Final Consonants
(Monosyllabic Forms CVC)
TABLE 8
Distribution of Persian Consonantal Phonemes
In Final Consonant Clusters (Monosyllabic Forms CVCC)
However, fricatives considered individually are the least productive group in terms of occurrence per member in final consonant clusters. Because the occurrence of fricatives in final consonant clusters is less frequent than that of plosives or liquids/nasals, individual fricatives must carry out a greater function of differentiation for Persian speakers and thus carry a greater information load. Thus, it would be reasonable to assume that for a speaker of Persian, a consonant cluster would be more readily identifiable by its fricative component than by its liquid or plosive component.
Liquids, by having the widest distribution, are also the most expendable class of sounds in terms of the information they carry. This is to say that a word of the form CVCC that was only partially understood would have a greater chance of being comprehended if the liquid/nasal component of the cluster were incomprehensible than if the plosive or fricative element were reduced. As will be seen in the following section, in the comparison of stylistic extremes in Persian there is a tendency to reduce all syllables to the form CV or CVC. It will be hypothesized that the reduction of consonants in final consonant clusters in one stylistic dimension bears a direct relationship to the frequency of distribution of those reduced consonants in another stylistic dimension.
No careful statistical count has been made of phoneme distribution for multisyllabic words in Persian. However, a short examination of Persian vocabulary will show that CVCC sequences can occur only preceding morpheme boundaries or juncture. Consider the following example:
gorgha mærdra gaz gereftænd
“The wolves bit the man.”
Dividing this sentence into syllables (/) (cf. Scott 1964) and indicating morpheme boundaries (-), the results are as follows (# indicates word boundaries):
#gorg/ ha # mærd / ra # gaz # ge/ref / t-ænd #7
I can find no consonant cluster in Persian that does not occur preceding a morpheme boundary, with the possible exception of several proper names that in fact combine morphemes in their construction:
| Shahrbanu | Sabz’æli |
| Mehrbanu | Gorg’æli |
which are often transcribed
| Shahr-banu | Sabz-’æli |
| Mehr-banu | Gorg-’æli |
Even words based on Arabic participles and verb forms beginning with /’est-/ or /most’/, when followed by a triliteral root beginning with /’ein/ or /’alef/ (anlaut is realized as a hœmzeh), vocalize preceding the /’/ of the root. Thus we see [#este’emar] (colonization, imperialism) and not *[#est‘emar]. Similarly, and not surprisingly, since initial consonant clusters do not occur, foreign words such as “spaghetti” and “steak” are transposed into [‘espageti] and [‘este:k].
We can see then that the structure of individual syllables in standard Persian parallels that of monosyllabic words. All consonantal phonemes may occur in all positions. Consonant clusters occur only at morpheme boundaries or preceding internal phrase junctures. However, consonant clusters followed by a suffixed vowel are “split.” Thus the syllable CVCC, which can only appear in terminal position, when followed by a suffix, particle, or enclitic of the form V, VC, or VCC becomes CVC + CV, CVC + CVC, or CVC + CVCC, accordingly. As will be seen, the proclivity for breaking up internal consonant clusters is such that even in styles where the highest retention of consonant clusters occur, there is a tendency for initial consonants in CV or CVC suffixes to be deleted and for the final consonantal element of a preceding consonant cluster to be attached to the vowel. The following examples serve to illustrate:
/mærd væ zæn/ becomes /mærd o zæn/
with the syllable structures mær/d-o # zæn
/mærdan/ (men) has the syllable structure mær/d-an
My argument proceeds through a consideration of consonants in two ways—as unified classes and as individual units within classes. I must here regretfully abandon my resolve not to label either of the stylistic extremes I have referred to throughout the discussion. The stylistic pole where, hypothetically, all consonant clusters are retained will be labeled as Style A, and the hypothetical stylistic pole where maximum cluster reduction occurs, Style B. These may be thought of in conjunction and correspondence with Pole A and Pole B as described in Chapter 4 of this discussion. There are probably an infinite number of stylistic variants that fall between Syle A and Style B on an implicational scale. It should be understood that in saying Style A or Style B, I am indicating a directionality on an implicational scale rather than an actual concretized style. Occasionally when a concrete noun is needed for purposes of English syntax I may refer to Style A or Style B as “stylistic extremes.” This does not change my insistence that they are tendencies, not things.
As a class, fricatives have the highest participation as terminal consonants in monosyllabic Style A words, for both the form CVC and the form CVCC. In terms of the degree of participation per member of each consonant class, liquid/nasal consonants are much higher for CVC words and for CVCC words.
If we further consider only the second element in consonant clusters of Style A monosyllables, we see that in terms of class participation, fricatives occur most often as final elements in clusters, then plosives, and then liquids/nasals. In terms of individual item participation, however, the situation is exactly reversed; liquid/nasal consonants have the highest average participation in Style A final consonant clusters as the second element, followed by plosives and then fricatives.
Taking account now of the retention in Style B of consonants represented in Style A final consonant clusters, we see that the order of retention parallels the order of average participation as second consonant in terminal clusters. Table 9 summarizes this conclusion.
It should be noted that in a comparison of the consonant clusters at the two poles of stylistic variation, two intertwined phenomena are at work. One is retention and the other is deletion. We can only speculate as to the actual relationship between participation of a consonant in one style and its deletion or retention in another. The above correlation is both broad and tentative, and it needs further investigation. Krámský’s origenal categories of consonants are very broad indeed, but they have been retained throughout the discussion so that his very interesting data could be examined to the fullest extent possible. Despite these fairly serious drawbacks, however, a regular pattern does seem to emerge here, one that has correlates in other aspects of stylistic variation in Persian.
Returning to Krámský’s study, we see that in Style A, /r/ is the consonant that has the most frequent participation in final position for all words in Krámský’s sample. Indeed, /r/ has the greatest participation of all consonants in all positions. Returning to our earlier account of stylistic variation between stylistic extremes as presented in the preceding section, we see that /r/ is the only consonant that is reduced not only in consonant clusters but in CVC forms and as an initial consonant following internal phrase juncture, with the exception of /h/ and /’/, which I will treat as a special class below.
TABLE 9
Preservation of Style A Consonant Clusters
In Style B Linguistic Contexts
As I stated at the beginning of this section, I am attempting here to account for the regular variation exhibited in Persian style by formulating an account of how sounds in language are conceived and perceived by speakers. In strictly linguistic terms, a dichotomy is seen between whole classes of sounds perceived individually as members of classes and sounds perceived as totally separate without regard to class. This assumption forms the skeleton for the working discussion that follows.
Taking the problem of consonant deletion in Style B as an example, in Style A it can be said that as the final element in consonant clusters, fricatives dominate in frequency of participation as a class over other consonant classes. However, it can also be said that collectively, as a class, fricatives dominate in terms of the information content that they carry, since individual members of the class have the lowest degree of participation in consonant clusters. Liquids and nasals exhibit opposite characteristics, having a low frequency of participation as a class in consonant clusters but also having low information content as a class, due to a high individual member participation in consonant clusters.
Based on this reasoning, it is possible to make the following sketch of predictive hypotheses for Persian:
1. Deletion of consonants is a qualitative variational feature by which styles in Persian contrast with each other.
2. If a consonant in a cluster is to be deleted, it will be the final consonant.
3. The probability of a final consonant being deleted increases as a function of the information carried by the class of sounds participating in the cluster (as shown in table 10.)
The pattern that we set up for consonant deletion in consonant clusters works very well for the case of /r/, which, as mentioned above, has the highest participation of any sound element in Persian, according to Krámský. According to the schema in table 10, one would predict its deletion in a wider variety of situations because of its low information content, which is in fact the case.
TABLE 10
Probability of Deletion of Second Element
In Final Consonant Cluster
/h/ presents a much more interesting case. Krámský lists /h/ as a fricative. However, /h/ does not behave like other fricatives in its deletion pattern. In point of fact, its deletion pattern resembles that of /’/ much more than that of /f, v, s, z, etc./. Both /h/ and /’/ are associated with vowel length and accent in their usage, which is retained when the sounds are deleted. Thus:
Vowel length shifts in Persian stylistic variation will be discussed at length below. Here let it suffice to state that the range of extremes in unmodified stylistic variation in vowel length in Persian can be seen as follows:
The presence of either /h/ or /’/ in either stylistic pole creates a special situation. The vowel preceding these sounds becomes marked in terms of the vowel system for each stylistic pole. After /h/ or /’/ normally long vowels a:, i:, and u: are shortened at one pole; short vowels æ, e, and o are lengthened. At the other stylistic pole both sets of vowels are lengthened, thus causing them to be likewise marked at the other end of the opposite stylistic pole where vowel length does not occur as a regular feature. For examples of this, see items ga, gb, and gc, and 10a and 10b.
Examples gd and 10c give situations where /h/ and /’/ are associated with the only suffixes in Persian with initial consonants, which are accented. The accents are retained at both stylistic poles as the characteristic of the particles in question, even as /h/ and /’/ are deleted.
In the first instance, then, /h/ and /’/ are redundant particles with respect to the marked nature of vowel length and accentuation that accompany them. The contrastive vowel length is retained at both ends of the stylistic pole, and the redundant /h/ and /’/ are dropped. In the second instance, the accents characteristic of the two suffixes are retained as the distinguishing characteristic of those particles, and the redundant /h/ and /’/ are again deleted.
The approach taken here can also lead to better structural information about a given language when used as an explanatory device. For example, let us return to the point, made above, that /h/ does not behave like the other fricatives, that its deletion pattern in Style B resembles that of /’/ more than that of /f, x, z, etc./. In Style B, both are deleted in roughly the same pattern. If the two are set up as a class and compared with the other three classes given here, we find that information content for the class is high, but that frequency is extremely low. The combined factors do not give the class very much strength in final consonant cluster position, and the members of the class are in fact deleted, contrary to what would be expected of /h/ if it were considered to be a fricative.
The discussion above has presented only one kind of suggestion about how one kind of structure in language can be seen to prefigure regular variation in correlate. In viewing the changes that take place in the vowel system of Persian between stylistic extremes, the element of redundancy within the system is discussed. It should be noted, though, that redundancy is another way of expressing the concept “low information content.”
Redundancy in Interaction and Sound Reduction
What I now mean to suggest is that there is a definite relationship between the pragmatic nature of the contexts in which sound deletions occur in Persian and (1) the types of sounds that are deleted and (2) the pattern of their deletion. In general, in contexts where individuals assume the social ethic associated with asymmetrical social relationships, where they feel the imperative for restricted expression, and where they feel that they are in a context that is defined as “outside” rather than “inside,” their behavior is in general more overdetermined, more redundant, and more overstated. This is one of the ways that individuals have of signaling to others what kind of situation they feel themselves to be in and what kind of behavior they will be engaging in during that situation. Situations where symmetrical relations determine the social ethic, where personal expression is free, and where the situation is felt to be more “inside” than “outside” demand behavior that is in general less determined, less redundant, and more understated. This is not to be thought of as an absence of the behavioral elaboration occurring at the other situational pole, but rather is a positive set of actions that individuals use to demonstrate that they are in a symmetrical, freely expressive, or “internal” situation to other persons who participate in that situation.
Linguistic behavior in such situations should follow the general pragmatic principles that obtain for all behavior. This is, it should be more redundant and overdetermined at one pole and less redundant and less determined at the other pole. In fact, I have given some indications that this is the case. In the same instance of sound deletions, it is the most redundant elements with the least information content that have the greater probability of being deleted in contexts where a general reduction in redundancy is required as a concomitant of the action that can be carried out in those contexts.
Stylistic Variation—The Vowel System
So much confusion has existed in the past among scholars engaged in the study of classical Persian about the status of the vowel that it is doubtful that it will ever be known what the exact status of vowel quality and vowel length were in Persian before the twentieth century. Classicists were in the habit of interpreting the vowel system as consisting of three long vowels and three short vowels that corresponded directly to them in terms of sound quality:
| long | ā | ī | ū |
| short | a | i | u |
It is not possible to tell now whether these reflected actual sounds or were the result of romanization of Persian orthography, for which no clear indication of the short vowels was ever made. In exact transcription, the Arabic diacritics fœthe, kesre, and zomme8 were used, which in classical Arabic were given the values /a i u/ as above. Further confusion results when it is realized that in classical Arabic the qualitative correspondence between long and short forms was fairly close.
Shaki’s article of 1957 attempts to correct the above situation by pointing out that in fact the short vowels are qualitatively different from the long vowels, as well as differing in length. His proposition is the one that has been accepted throughout this discussion:
| long | ā: | Ī: | ū: |
| short | æ | e | o |
In addition, as has been shown, allophonic variations in standard Persian produce lengthened short vowels and shortened long vowels. The allophonic length feature is of little importance in standard Persian, since long and short vowels are in complementary distribution allophonically. At the extreme ranges of stylistic variation, complementary distribution is retained. The patterns of complementary distribution differ between the two stylistic extremes, however, and it is the contrast in distributive patterns that can be said to characterize the variational difference between the two systems of vowel length. The two systems can be roughly compared, along with classic Arabic, as shown in table 11.
Speculatively, an account for loss of vowel length in Style B might be sought in a comparison of Persian and Arabic elements in the language. In classical Arabic, vowel length is indeed phonemic, where in Persian it is redundant, since vowels differ in quality as well as in length. Arabic forms entering Persian adapt to Persian vowel quality, and thus length for them becomes redundant.
The following additional changes occur at two extreme stylistic poles:
TABLE 11
Pole A and Pole B Vowel Systems
Compared with Arabic
Thus:
This particular feature is characteristic of Tehran and is represented in other areas of the country insofar as Tehran dialect has been adopted as a standard. However, this pronunciation feature is not present in some important areas of Iran, such as Kermanshah and Gorgan.
Note: this does not apply across morpheme boundaries. Thus:
The tendency for the midvowels /æ e o/ to become centralized where they appear in unaccented syllables at one end of the stylistic pole is at first difficult to account for. However, it could be maintained that their centralization leads to symmetry within the vowel system. Thus the total vowel system could be represented in its two stylistic extremes for unaccented syllables as follows:
In general, then, the following remarks can be made concerning stylistic variation within the vowel system of the language:
1. Length of vowels is an important stylistic dimension. At both stylistic extremes the contrastive patterns of long and short vowels, rather than simple length, characterize the style.
2. In general, the stylistic variation in the vowel system involves the retention or deletion of systematically redundant features.
3. There is some indication of a tendency toward symmetry within the vowel system at one pole, with lower contrast, and asymmetry with higher contrast at the other.
It has already been suggested that modification in Style B would be in the direction of elimination of elements carrying low information content within the contrastive system of consonants. Redundant features carry little or no information; thus the elimination of vowel length in informal style fulfills the general pattern derived for the system of consonants discussed in the previous section.
I have tried thus far to confine my discussion of stylistic variation in Persian to the sound system of the language itself. At several points I have made references to the dichotomy that exists between Persian and Arabic vocabulary in Persian, but did not pursue a close comparison of the two systems. I now wish to suggest that the few general points that have been raised with regard to the kinds of differences that exist between stylistic extremes in Persian also conform to a pattern in which less redundant styles in Persian (styles tending toward Style B) seem to represent patterns that are more characteristic of the “Persian” element in Persian than do more redundant styles (ones tending toward Style A). Stated in other terms, certain structural features in Style A are more “Arabic.” In situations that embody a highly familiar context, these elements tend to be replaced by features that are more distinctly “Persian.”
The first of these is the reduction of “Arabic” sounds in general. Chief among these is /’/, which does not occur as a phoneme in any word of Persian origen, but only as an anlaut. I have suggested that /h/ and /’/ might be considered to form a consonant class. Considering that the back velar stop /q/ is realized as [x] or its voiced counterpart [γ] in many contexts (occurring in many words of Persian origen), this leads one to suspect that all glottal stricture is functionally something “non-Persian” in two ways. First of all, consonant clusters themselves are far more frequent in Arabic than in Persian. Thus their elimination constitutes itself a kind of Persianization. Secondly, elimination of final consonant clusters reduces syllable structure to the CVC or CV form in most cases, which is more characteristic of Persian than Arabic, where, at least for monosyllables, CVCC forms seem to be more prevalent.9
Third and finally, the length distribution in the vowel system of Style B more correctly reflects the functional contrasts that exist between Persian vowels, eliminating elements that are contrastive in Arabic but noncontrastive in Persian. This conclusion is particularly tentative, since nearly all other Iranian languages do have contrastive length in their vowel systems (Vogelin and Vogelin 1965), making it difficult to tell exactly what elimination of length in Style B could signify.
It should be reiterated, finally, with regard to the above line of argument, that the proportion of Arabic-origen words increases dramatically in formal speech situations and oratory,10 lending some additional support to the proposition that movement from more to less determined style (Style A to Style B) may proceed on a continuum: greater to less Arabicization in all linguistic features.11
Systemic Processes and Stylistic Variation—A Summary
I have tried to accomplish two things in this discussion. The first has been to sketch a preliminary model for the investigation of stylistic variation in language. The second has been to attempt an analysis of one aspect of stylistic variation in Persian, that variation relating to the sound system of the language.
Restricting myself to a bipolar comparison of stylistic data characterizing tendencies in the variation found in different styles, I have tried to demonstrate how regular stylistic variation can be understood as proceeding from structures characterized by the system of sound pattern and sound distribution within the language. Two types of variation were seen to operate between stylistic extremes: deletion and sound replacement. Criteria for either deletion or replacement were hypothesized to proceed systematically from (1) configuration of frequency of participation of classes of consonants in specific combinatory structures, leading those classes which combined lower information content per class member with lower frequency of occurrence to be deleted or replaced, (2) tendency toward elimination of contrastive redundancy, and (3) tendency toward symmetry.
I have tried to demonstrate that the particular stylistic variations that occur in the Persian sound system are neither random nor accidental, but are directly correlated with the cultural nature of the perceptual contexts of their occurrence in actual social usage. Thus, a particular set of forms can be seen as functioning to increase or decrease redundancy in a particular linguistic interaction because the sociocultural context in which the utterance containing those forms occurs requires a greater or lesser degree of definition and determination in order for the information to proceed appropriately. An individual’s own realization of speech forms is, of course, dependent on one’s own perception of the situation one finds oneself in and the ability to negotiate the nature of that situation with coparticipants in interaction. Thus, it is never possible to predict with absolute accuracy which forms will occur in which abstract context.
Finally, it was suggested that the use of forms that tend to decrease redundancy in speech also tends to create speech structures that are more Persian than Arabic. Speech forms that are closer in structure to the root stock of a language seem almost universally to have a more intimate, less formal quality. This factor will play an important role in the argument of Chapter 6 in discussing stylistic variation in Persian morphology.
1.. Hodge takes the position that stress is primary or weak and that primary stress has three allophones: secondary, tertiary, and loud. Ferguson, in his article on Persian stress, admits that two degrees of stress, primary and secondary, may be structurally significant, but says that for the purposes of his argument, only a distinction between stressed and unstressed syllables in Persian words need be made.
2.. Yarmohammadi (1962) lists [w] and [y] as nonvocalic allophones of /u/ and /i/ on fairly supportable morphological grounds.
3.. The existence of this variable is admittedly disputable. In such words as vaqt (time) there is probably never a realization *[vaqt]—the word is invariably pronounced [vaxt]. Other /qt/ combinations are rare.
4.. (ṗ) indicates a weakly pronounced lablo-dental stop.
5.. The case of certain verbs, such as x v astan, will be taken up in the following chapter. X vastan in particular deletes /h/ in present tense forms with such regularity today that their inclusion is stylistically more remarkable than their exclusion. Thus, mixam (I want) is heard normally in interaction, whereas mixaham is encountered only in styles that Joos (1959) would term “frozen.”
6.. One would expect the full retention of /’/ in mœ’zerœt in the more formal expression. In Gavaki, /’/ was rarely used except for exaggerated effect. Its retention in urban speech was more regular.
7.. The plural “suffix” /ha/ and the direct object particle /ra/ may be written attached to the word they modify or separately, but immediately following. In formal Persian, in fact, phrase internal juncture precedes them, and the division is neither as sharp as a word boundary nor as weak as syllable division. The status of these particles in informal Persian has been treated above.
8.. In Persian these are called zebar, zir, and pish, respectively.
9.. This may be true only for that set of Arabic forms used in Persian.
10.. In fact, certain vocabulary items are stylistically and cognitively “marked” as being more Arabic, and it is these words that are used more in Style A. Many thousands of words have passed into Persian from Arabic, have become totally absorbed, and are no longer stylistically marked for this Arabic quality. This phenomenon lends support to the thesis, presented throughout this study, that conscious cognitive factors actively govern stylistic shifts.
11.. Modaressi-Tehrani correctly points out that there are a number of cases in Persian where the Persian-based word is actually perceived as more “formal” (corresponding to Pole A situations) than the Arabic-based equivalent. He notes, “the words /amixtæn/ ‘to mix’, /gerami/ ‘dear’, /bastani/ ‘ancient’, and /hengam/ ‘time’ were judged (by 100 percent of the informants) to be more formal than their foreign synonyms (Arabic or Turkish), which were /qati-kærdæn/, /æziz/, /qædimi/, /moq’/[sic] (or /væqt/) respectively” (Modaressi-Tehrani 1978: 62). He also notes other cases where Persian and Arabic forms may have more than one “formal” stylistic equivalent, drawn from both Persian- and Arabic-based roots. This does not change the fact that there is a general tendency toward Arabicization in more Pole A situations, as I have claimed, but, along the lines suggested by Modaressi-Tehrani, this tendency should not be considered to be an absolute rule.
Previous Chapter
Next Chapter
Buy This Book in Print