urdU Prosody



This page is still in its formative stages and most of the information is not necessarily complete in itself ... I present a semi-formal, though non-historical, approach to urdU prosody - but it is by no means an expert's thesis. All this information is culled mostly from my memory and the meagre notes from having read Agha Sadiq's treatise on urdU prosody "nikAt-e fann" a long time ago. If you are reading all this for the first time, you can safely skip the "Historical Background" perspectives that I have interspersed in the actual text; especially because there exists the possibility of notational confusion if you read both the perspectives together.

If you are least interested in knowing the basis for generating the meters, you can, after having read the section on vazn, directly jump to the urdU Prosody (Meters and their Derived Forms) page.

vazn (weight)

To arrive at a semi-formal notion of vazn, let us assume that the symbols k, v, and V stand for a Consonant, Short Vowel, and Long Vowel respectively as used in urdU or hindI. Then, syllables may be classified as:


For the purposes of urdU prosody a distinction is not really made between the Closed and Open flavours. If at any point a distinction needs to be maintained, it will be explicitly mentioned at that point. Specifially, the closed syllable is indicated by a c subscript. Further, only the Short and Long forms are really "relevant" in urdU prosody. The Ultralong (kVk) is instead decomposed as Long + Short (kV + k). All other long syllables with trailing consonant clusters (e.g. kVkk, kVkkk etc.) are considered equivalent to kVk.

We denote the Long as L and Short as S and classify them as syllabic atoms. There is no notion of any equivalence or conversion relationship between the L and S atoms - and thus they both need to be treated independently as first class entities. There is however a "popular" notion of attributing the value 1 to S and the value 2 to L. This, in my opinion, is sadly misleading and seems to suggest that the vazn of an L atom is twice that of an S atom. Nothing could be more farther than the truth ... (this is notwithstanding the fact that there are indeed many urdU meters that allow a legal subsitution of an SS pair in place of an L at specific locations).

Definition: The vazn of a lafz (word) is defined as the syllabic equivalent defined in terms of the L and S. Syllabification in a lafz is defined in a manner such that the number of vowels in a lafz determines the number of syllables. While the first syllable in a lafz may start with a vowel (in which case the associated consonant is the void consonant), all following syllables begin with a single consonant. In other words, each consonant cluster in a word determines a syllable boundary, with the last consonant beginning the next syllable.

Thus, the lafz "par" has a vazn of L, while the lafz "pAr" has a vazn of LS. Similarly, the lafz "pardah" has a vazn LL.

Historical Background
Every word, or lafz, is made up of repeated instances of two basic atomic entities, namely
  1. mutaharrik
    movable, a consonant accented with the short vowels a, i, or u. Denoted by M. It is related to the word harakat which stands for motion. A harakat is then said to be present on the consonant.

  2. sAkin
    quiescent, a consonant not followed by a vowel. Denoted by S. It is related to the word sukUnat which stands for rest or tranquility. A sukUnat is then said to be following the consonant.
For instance, in the word "par'', p occurs as a M, while r occurs as a S. Similarly the word "kAr'' can be decomposed into ka[M]a[S]r[S]. Notice the splitting of the sound "kA'' into a MS pair.

Every lafz can thus be split into its constituent M and S parts. Any such combination of M and S defines the vazn, or weight, of the corresponding lafz. Loosely put, the vazn is a measure of the time taken in pronouncing the lafz.

Equivalence of vazn

The primary focus in fann-e `arUz, or the art of prosody, is matching up the vazn of two different lines. The basis for this is the definition of equivalence of vazn for two different alfAz (plural of lafz). Thus two alfAz are said to be hamvazn, or equal in vazn, iff their syllabic patterns defined in terms of the L and S atoms are identical.

Thus, "sabab" (sa[S] bab[L]) is hamvazn with "bahut" (ba[S] hut[L]).
However, "sabab" (sa[S] bab[L]) is not hamvazn with "ShAd" (ShA[L] d[S]) even though the number of L and S atoms in both the words is the same.

Historical Background
Two alfAz are said to be hamvazn, or equal in vazn, iff
  1. They have the same count of hurUf (letters from the Persian alphabet).

  2. The harakAt (phenomenon of having harakat) and the sakanAt (phenomenon of having sakUnat) of one of the alfAz can be exactly matched up with that of the other.
Note that the presence of fatha (the zabar or the vowel "a''), kasra (the zer or the vowel "i''), or zamma (the peSh or the vowel "u'') on a consonant does not modify the vazn. For instance, "pa'', "pi,'' and "pu'' are equivalent as far as the vazn is concerned.

To consider an example of hamvazn alfAz,
The lafz "sabab'' (sa[M]ba[M]b[S]) is hamvazn with "tarah'' (ta[M]ra[M]h[S])
But "sabab'' (sa[M]ba[M]b[S]) is NOT hamvazn with "saKht'' (sa[M]Kh[S]t[S])
As another example, notice that "GhAlib'' is NOT hamvazn with "hakIm'', as MSMS does not match up with MMSS, inspite of the fact that the sum durations of both the alfAz are the same! In the same vein "GAlib'' IS hamvazn with "nAsiKh'', as both are MSMS.

The vazn forms the sole foundation of the considerations that are based on metre. Hence to be able to judge the hamvazn-ness of two different lines, one should be aware of the correct duration of the pronunciations as employed in actual speech. Simply knowing the correct pronunciation is not enough. For instance, the sound "ko'' in "mujHkO'' and "kOrA'' are not equivalent in vazn! In the former case, depending on the use and position in the eventual metre, it is equivalent either to a M or MS, whereas in the latter it is equivalent to MS. In other words the hurdle towards learning to proof-read verse is knowing how to read it correctly in the first place! The only solution seems to be learning the correct durations by constantly listening to the masters in prosody.
As a consequence of this, note the following special cases
  1. "gA'o.n'' and words of that ilk should actually be matched with MSS as they are actually "gA.nv'' (ga[M]a[S]v[S]) etc..
  2. Similar to the one above are the words like "banA'o'' etc. which actually scan as "banAv'' (ba[M]na[M]a[S]v[S]).
  3. "dulhan'' actually scans as MMS since the "lh'' actually stand for the single sa.Nsk.rt consonant "lH''. Thus the breakup for "dulhan'' is du[M]lHa[M]n[S].
Similarly, there are a host of alfAz which have multiple auzAn (plural of vazn). I hope I will be able to come out with a list some day soon ...

taqtI` (scansion) and vazn

The notion of vazn assumes importance only in the context of the rules of scansion or taqtI`. A brief heuristic treatise on taqtI` is available for digression.

ajzA-e arkAn (elements of the feet)

A rukn (singular for arkAn) or the foot, is the primary entity using which a baehr (metre) can be defined. This immediately brings us to the topic of the ajzA (the elements or the basis) that constitute a rukn. For historical reasons, only specific syllabic patterns are classified as legal elements. These are enumerated below.

sabab-e KhafIf
These scan as L. For instance, "dil", "jA", etc.

sabab-e saqIl
These scan as SS. Note, this occurs only during tarkIb (composition). e.g. "gul-i Chaman" where the "gul-i" has the vazn SS. Otherwise, there is no urdU or fArsI word that contains two simultaneous short independent syllables. This is, for all practical purposes, an optional element since it occurs only via the form fAsila-e suGhra (see below). (Note, however, that some Arabic nouns loaned into urdU do admit such a syllablic structure.)

vatad-e majmU`
These scan as SL. For instance, "sabab", "Chaman", etc.

vatad-e mafrUq
These scan as LS. For instance, "bAt", or the "dAGh-e" in "dAGh-e dil" etc.

fAsila-e suGhrA
These scan as SSL. This is essentially a composition of SS and L. However, it has been retained for historical reasons - specifically since the feet are defined in terms of this element.

These scan as SSSL. Rarely, do such words occur in urdU or fArsI. Again, this is a composition of SS and SL. However, it also is retained for historical reasons and is, for all practical purposes, an optional element.

Note that a single S is not defined in this basis at all!

In actual practice, one needs to use only the following ajzA:

sabab-e KhafIf L
vatad-e majmU` SL
vatad-e mafrUq LS
fAsila-e suGhrA SSL

Historical Background
One harfI (lettered) alfAz
On what has been said about the lafz, it is obvious that any lafz comprised of only one harf (letter) will obviously start with S. This is not permitted in urdU, but instances can be found in other languages like sa.Nsk.rt or english. However, here the discussion being restricted to urdU, the scheme that is generated is specifically tailor-made for urdU poetry. As a consequence, a one harf long basis is not defined at all.

Two harfI alfAz
Since any lafz in urdU is not permitted to start with a S, there exist only two(!) kinds in this category.
This is the class of alfAz that scan as MS. For instance, "dil'', "jA'', etc..

The alfAz belonging to this class scan as MM. Note that this can occur only during tarkIb (composition). e.g. "gul-i-Chaman'', where the "gul-i'' has the weight MM.
Three harfI alfAz
Here again, only two cases are considered.
These scan as MMS. Thus "sabab'', "chaman'' etc. belong to this class.

These scan as MSM. This occurs during izAfat (annexation). Thus in "dAGh-e-dil'', "dAGh-e'' scans as da[M]a[S]Gh-e[M].
The other possible cases, but not recognized, are
  1. Those that scan as MSS. If such a lafz occurs in the middle of a misra` (line) the second S is converted to M, so that it will read MSM. However at the end of the misra` it retains its original form - MSS.
  2. Those that scan as MMM.
Four harfI alfAz
Here only one case is enumerated.
These scan as MMMS. e.g. chamanI (cha[M]ma[M]ni[M]i[S]).
This can be split up as MM + MS.
The other possible cases are,
  1. MMMM = MM + MM.
  2. MMSM
  3. MMSS In the middle of a misra` the second S is converted to M, so that it scans as MMSM. At the end of the misra` though it remains MMSS.
  4. MSMM = MS + MM
  5. MSMS = MS + MS
  6. MSSM = this necessarily gets converted to MSMM.
  7. MSSS In the middle of a misra` the last S is dropped and the penultimate S is converted to a M. Thus it will read MSM. However at the end of the misra` it is transformed into MSS by dropping the last S.
Five harfI alfAz
Rarely do such long words occur in urdU or fArsI. Here again only one case is enumerated.
These scan as MMMMS.
Notice that
fAsila-e-suGhrA <=> sabab-e-saqIl + sabab-e-KhafIf
fAsila-e-kubrA <=> sabab-e-saqIl + vatad-e-majmU`

arkAn-e `aShra: (The 10 feet)

As mentioned above, a rukn (foot) is primarily composed of the following individual juz (singular for ajzA):
sabab-e KhafIf L
vatad-e majmU` SL
vatad-e mafrUq LS
fAsila-e suGhrA SSL

Repeated application of these elements, taken singly or together, generate a rukn. In other words, a rukn is generated using the following productions

<rukn> ::= <juz><rukn> | <juz>
<juz> ::= L | SL | LS | SSL

The grammar defined above is not unambiguous since two different derivation trees exist for, say, LSL (LS + L or L + LS).

In principle, infinite arkAn can be generated. For practical purposes, however, ten basic ones are recognized. Each one has a name which is a "mnemonic" to the vazn that it encodes.

Tri-atomic arkAn

There are two of them.

fA`ilun (sabab-e KhafIf + vatad-e majmU`)
This scans as L SL.
e.g. "zindagI" (zin[L]da[S]gI[L]), "raOShnI" (raO[L]Sh[S]nI[L]), etc.

fa`Ulun (vatad-e majmU` + sabab-e KhafIf)
This scans as SL L.
e.g. "tamannah" (ta[S]man[L]nah[L]), "muhammad" (mu[S]ham[L]mad[L]), etc.

Quadra-atomic arkAn

There are six of them.

mafA`Ilun (vatad-e majmU` + sabab-e KhafIf + sabab-e KhafIf)
This scans as SL L L.
e.g. "razAmandI" (ra[S]zA[L]man[L]dI[L])

fA`ilAtun (sabab-e KhafIf + vatad-e majmU` + sabab-e KhafIf)
This scans as L SL L.
e.g. "AdmIyat" (A[L]d[S]mI[L]yat[L])

fA`i lAtun (vatad-e mafrUq + sabab-e KhafIf + sabab-e KhafIf)
Same as above, just a different scansion grouping LS L L. This is retained for historical reasons since the operations that generate the various meters and their derived forms rely on the actual manner in which the syllabic atoms are grouped together.

mustaf`ilun (sabab-e KhafIf + sabab-e KhafIf + vatad-e majmU`)
This scans as L L SL.
eg. "jAdUgarI" (jA[L]dU[L]ga[S]rI[L])

mustaf`i lun (sabab-e KhafIf + vatad-e mafrUq + sabab-e KhafIf)
Same as above, just a different scansion grouping L LS L. This is retained for historical reasons since the operations that derive the various meters and their flavours rely on the actual manner in which the syllabic atoms are grouped together.

maf`UlAtu (sabab-e KhafIf + sabab-e KhafIf + vatad-e mafrUq)
This scans as L L LS.
e.g. "raOShan fikra" (raO[L]Shan[L] fik[L]ra[S])

Penta-atomic arkAn

There are two of them.

mutafA`ilun (fAsilla-e suGhrA + vatad-e majmU`)
This scans as SSL SL.
e.g. "sitam-O jafA" (si[S]ta[S]m-O[L]ja[S]fA[L])

mafA`ilatun (vatad-e majmU` + fAsila-e suGhrA)
This scans as SL SSL.
e.g. "jafAtalabI" (ja[S]fA[L]ta[S]la[S]bI[L])

Historical Background
Five harfI arkAn
There are two of them
This scans as MS MMS.
e.g. "zindagI'' (zi[M]n[S]da[M]gi[M]i[S]), "rauShanI'' (rau[MS]Sha[M]ni[M]i[S]), etc..

This scans as MMS MS.
e.g. "tamannA" (ta[M]ma[M]n[N]na[M]a[S]), "mahammad'' (ma[M]ha[M]m[S]ma[M]d[S]), etc..
Seven harfI arkAn
There are eight of them
This scans as MMS MS MS.
e.g. "razAmandI'' (ra[M]za[M]a[S]ma[M]n[S]di[M]i[S])

This scans as MS MMS MS.
e.g. "AdamIyat'' (a[M]a[S]da[M]mi[M]i[S]ya[M]t[S])
This scans as MS MS MMS.
e.g. "jAdUgarI'' (ja[M]a[S]du[M]u[S]ga[M]ri[M]i[S])
This scans as MMMS MMS.
e.g. "sitam-O-jafA'' (si[M]ta[M]mo[M]o[S]ja[M]fa[M]a[S])
This scans as MMS MMMS.
e.g. "jafAtalabI'' (ja[M]fa[M]a[S]ta[M]la[M]bi[M]i[S])
This scans as MS MS MSM.
e.g. "rauShan fikra'' (rau[MS]Sha[M]n[S]fi[M]k[S]ra[M])
This scans as MSM MS MS. Notice that though this apparently reads the same as fA`ilAtun, the placement of the ajzA differ! The sum vazn is the same though, and this is thus a different form of fA`ilAtun.
This scans as MS MSM MS. This is the another form of mustaf`ilun, and hence scans differently from it.

aOzAn-e baehr (meters)

The 19 baehr (meters) that are in common use are listed below. It is not that one cannot have any other baehr. These ones have been more or less brought into common use for purely historical reasons. On a mathematical basis though infinite other baehr can be generated.

The baehr are generated using the production rule
<baehr> ::= <rukn>+
where each <rukn> is instantiated separately using any one of the ten arkAn defined above.

The list below enumerates the most commonly used ones. Note that these buhUr (plural of baehr) are considered the "base" forms. Further flavours are derived from these base forms via pre-defined operations on the rukn that make up the baehr. For the purposes of this section it is sufficient to enumerate the base forms only.
  1. mutaqArib
    fa`Ulun * 4
    Total span is SLL * 4

  2. hazaj
    mafA`Ilun * 4
    Total span is SLLL * 4

  3. ramal
    fA`ilAtun * 4
    Total span is LSLL * 4

  4. rajaz
    mustaf`ilun * 4
    Total span is LLSL * 4

  5. kAmil
    mutafA`ilun * 4
    Total span is SSLSL * 4

  6. vAfir
    mafA`ilatun * 4
    Total span is SLSSL * 4

  7. tavIl
    fa`Ulun mafA`Ilun * 2
    Total span is SLL SLLL * 2

  8. madId
    fA`ilAtun fA`ilun * 2
    Total span is LSLL LSL * 2

  9. basIt
    mustaf`ilun fA`ilun * 2
    Total span is LLSL LSL * 2

  10. munsareh
    mustaf`ilun maf`UlAtu * 2
    Total span is LLSL LLLS * 2

  11. maqtazab
    maf`UlAtu mustaf`ilun * 2
    Total span is LLLS LLSL * 2

  12. mazAre`
    mafA`Ilun fA`i'lAtun * 2
    Total span is SLLL LSLL * 2

  13. mujtas
    mustaf`i'lun fA`ilAtun * 2
    Total span is LLSL LSLL * 2

  14. KhafIf
    fA`ilAtun mustaf`i'lun fA`ilAtun * 1
    Total span is LSLL LLSL LSLL * 1

  15. sarI`
    mustaf`ilun mustaf`ilun maf`UlAtu * 1
    Total span is LLSL LLSL LLLS * 1

  16. mutadArik
    fA`ilun * 4
    Total span is LSL * 4

  17. qarIb
    mafA`Ilun mafA`Ilun fA`ilAtun * 1
    Total span is SLLL SLLL LSLL * 1

  18. muShAkil
    fA`ilAtun mafA`Ilun mafA`Ilun * 1
    Total span is LSLL SLLL SLLL * 1

  19. jadId
    fA`ilAtun fA`ilAtun mustaf`ilun * 1
    Total span is LSLL LSLL LLSL * 1

  20. rubA'I
    Derived from the hazaj meter, this is really a group of 24 meters all rolled into one four-line verse. Though not used in Ghazals, a discussion on the rubA'I is not out of place here.

Proceed to urdU Prosody (Modifiers on the Elements of the Feet) for notes on the various modifiers that are applied to the base forms of the meters to generate the derived forms.

Last updated: 28th May 2003
Created using and Optimized for Mozilla 1.3
© Roshan Kamath (roshbaby "at" hotmail "dot" com)

Valid HTML 4.01!(except ofcourse for the stupid server generated tags)