Upload
others
View
66
Download
1
Embed Size (px)
Citation preview
Urdu Grammar Report
Urdu Grammar Group
Konstanz
ParGram 2008, California
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 1 / 34
Overview
Project Status
Have applied for a project (January 2007).
Decision still pending, looks like might end up being a resubmit.
In the meantime, patched together and Hiwi-Funding.
Managed to get some university-internal funding.
(German ParSem efforts now seriously under way, are to include Urdu,Spanish, German and English — possibly Turkish.)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 2 / 34
Overview
Bigger Progresses
Transliterator
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 3 / 34
Overview
Bigger Progresses
Transliterator
Urdu Ezafe (LFG08)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 3 / 34
Overview
Bigger Progresses
Transliterator
Urdu Ezafe (LFG08)
Tense/Aspect (perennial problem)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 3 / 34
Transliterator
Transliterator
Goal: Keep the morphological analyzer in ASCII so we can do bothUrdu (Arabic-based script) and Hindi (Devanagari) in the long run.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 4 / 34
Transliterator
Transliterator
Goal: Keep the morphological analyzer in ASCII so we can do bothUrdu (Arabic-based script) and Hindi (Devanagari) in the long run.
To this end, need a Transliterator that will take us from scriptweirdnesses to ASCII (and back).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 4 / 34
Transliterator
Transliterator
Goal: Keep the morphological analyzer in ASCII so we can do bothUrdu (Arabic-based script) and Hindi (Devanagari) in the long run.
To this end, need a Transliterator that will take us from scriptweirdnesses to ASCII (and back).
Other efforts have also realized this, so now several sites working onthis.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 4 / 34
Transliterator
Transliterator
Goal: Keep the morphological analyzer in ASCII so we can do bothUrdu (Arabic-based script) and Hindi (Devanagari) in the long run.
To this end, need a Transliterator that will take us from scriptweirdnesses to ASCII (and back).
Other efforts have also realized this, so now several sites working onthis.
We’d like the Transliterator to be written in XFST, for a cleaninterface with the morphology.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 4 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
We specified the transliteration scheme (mainly Tina).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
We specified the transliteration scheme (mainly Tina).
They do the following:
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
We specified the transliteration scheme (mainly Tina).
They do the following:
Use Input with or without diacritics.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
We specified the transliteration scheme (mainly Tina).
They do the following:
Use Input with or without diacritics.Normalize the Input to contain all the diacritics.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
We specified the transliteration scheme (mainly Tina).
They do the following:
Use Input with or without diacritics.Normalize the Input to contain all the diacritics.Transliterate this.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Transliterator
Transliterator — CRULP
Current: Have farmed out a work package to CRULP (Center forResearch on Urdu Language Processing).
We specified the transliteration scheme (mainly Tina).
They do the following:
Use Input with or without diacritics.Normalize the Input to contain all the diacritics.Transliterate this.
Work is supposed to be done by the end of September. First resultslook good.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 5 / 34
Urdu Ezafe
Urdu Ezafe — Background
We implemented an analysis of Urdu ezafe:
1 to deal with a new phenomenon in Urdu (no other existingliterature/implementation to date)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 6 / 34
Urdu Ezafe
Urdu Ezafe — Background
We implemented an analysis of Urdu ezafe:
1 to deal with a new phenomenon in Urdu (no other existingliterature/implementation to date)
2 Discuss and illustrate issues of the morphology-syntax-prosodyinterface — this is important for PR and discussions of architectureissues outside of ParGram (and so we want your help with it).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 6 / 34
Urdu Ezafe
Urdu Ezafe - History
The Ezafe is a loan construction from Persian
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 7 / 34
Urdu Ezafe
Urdu Ezafe - History
The Ezafe is a loan construction from Persian
It originates in an Old Iranian relative pronoun, specializing as adevice for nominal attribution in Middle Iranian
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 7 / 34
Urdu Ezafe
Urdu Ezafe - History
The Ezafe is a loan construction from Persian
It originates in an Old Iranian relative pronoun, specializing as adevice for nominal attribution in Middle Iranian
The Ezafe tends to be part of the high/literary language since Persianwas the language of the Moghul court and mostly influenced thelanguage of the courtiers and poets.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 7 / 34
Urdu Ezafe
Urdu Ezafe - History
The Ezafe is a loan construction from Persian
It originates in an Old Iranian relative pronoun, specializing as adevice for nominal attribution in Middle Iranian
The Ezafe tends to be part of the high/literary language since Persianwas the language of the Moghul court and mostly influenced thelanguage of the courtiers and poets.
It remains productive today, but modern speakers show a tendency toleave it out.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 7 / 34
Urdu Ezafe
Urdu Ezafe - The Modifier
Expresses a dependency between the head noun and its modifierwithin the NP:
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 8 / 34
Urdu Ezafe
Urdu Ezafe - The Modifier
Expresses a dependency between the head noun and its modifierwithin the NP:
Modifier can be a Noun
hukuumat=e paakistaangovernment-Ez Pakistan‘the government of Pakistan’ Schmidt (1999:246) Urdu
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 8 / 34
Urdu Ezafe
Urdu Ezafe - The Modifier
Expresses a dependency between the head noun and its modifierwithin the NP:
Modifier can be a Noun
hukuumat=e paakistaangovernment-Ez Pakistan‘the government of Pakistan’ Schmidt (1999:246) Urdu
Modifier can be an Adjective
sadaa=e bulandvoice-Ez high‘a high voice’ Delacy (2003:100) Urdu
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 8 / 34
Urdu Ezafe
Urdu Ezafe — Headedness
The Ezafe-Construction is weird, as Urdu is generally a head-finallanguage.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 9 / 34
Urdu Ezafe
Urdu Ezafe — Headedness
The Ezafe-Construction is weird, as Urdu is generally a head-finallanguage.
Compare the functionally (but not syntactically) similar genitiveconstruction.
pAkistAn=kI hukUmatPakistan=Gen.F.Sg government‘Pakistan’s government‘ Schmidt (1999:246) Urdu
hukUmat=e pakistAngovernment=Ez Pakistan‘The government of Pakistan‘ Schmidt (1999:246) Urdu
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 9 / 34
Urdu Ezafe
Urdu Ezafe — Clitic Status
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 10 / 34
Urdu Ezafe
Urdu Ezafe — Clitic Status
The Ezafe has scope over coordination
[mAl O daulat]=e dunyAmaterial and wealth=Ez world‘The material and wealth of the world‘
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 10 / 34
Urdu Ezafe
Urdu Ezafe — Clitic Status
The Ezafe has scope over coordination
[mAl O daulat]=e dunyAmaterial and wealth=Ez world‘The material and wealth of the world‘
The Ezafe never attaches to the single constituents of thecoordination.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 10 / 34
Urdu Ezafe
Urdu Ezafe — Clitic Status
The Ezafe has scope over coordination
[mAl O daulat]=e dunyAmaterial and wealth=Ez world‘The material and wealth of the world‘
The Ezafe never attaches to the single constituents of thecoordination.
Ezafe also takes scope over other phrasal material(quantifier/determiners).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 10 / 34
Urdu Ezafe
Urdu Ezafe — Clitic Status
The Ezafe has scope over coordination
[mAl O daulat]=e dunyAmaterial and wealth=Ez world‘The material and wealth of the world‘
The Ezafe never attaches to the single constituents of thecoordination.
Ezafe also takes scope over other phrasal material(quantifier/determiners).
We analyze the Ezafe as a clitic (similarly to case markers in Urdu).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 10 / 34
Urdu Ezafe
Urdu Ezafe - Prosody
From the prosodic point of view:
The Ezafe depends on a host.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 11 / 34
Urdu Ezafe
Urdu Ezafe - Prosody
From the prosodic point of view:
The Ezafe depends on a host.
It attaches to the prosodic word to its left.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 11 / 34
Urdu Ezafe
Urdu Ezafe - Prosody
From the prosodic point of view:
The Ezafe depends on a host.
It attaches to the prosodic word to its left.
sAhib=e takHt (‘owner of throne’)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 11 / 34
Urdu Ezafe
Urdu Ezafe - Prosody
From the prosodic point of view:
The Ezafe depends on a host.
It attaches to the prosodic word to its left.
sAhib=e takHt (‘owner of throne’)
Easily verified if construction is pronounced
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 11 / 34
Urdu Ezafe
Urdu Ezafe - Analysis
To summarize the facts:
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 12 / 34
Urdu Ezafe
Urdu Ezafe - Analysis
To summarize the facts:
The Ezafe triggers a head-initial construction.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 12 / 34
Urdu Ezafe
Urdu Ezafe - Analysis
To summarize the facts:
The Ezafe triggers a head-initial construction.
It is prosodically attached to its host on the left, but demands acomplement to its right, with which it forms a constituent.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 12 / 34
Urdu Ezafe
Urdu Ezafe - Analysis
To summarize the facts:
The Ezafe triggers a head-initial construction.
It is prosodically attached to its host on the left, but demands acomplement to its right, with which it forms a constituent.
This constituent modifies the NP to its left.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 12 / 34
Urdu Ezafe
Urdu Ezafe - Analysis
What should the analysis be?
and
How is the Ezafe clitic =e best analyzed with respect to themorphology-syntax-interface?
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 13 / 34
XLE-solution
Our solution
Our analysis does not place the Ezafe-clitic within the morphology(contra Samvelian 2005 and HPSG, who see it as a phrasal affixlicensed by a lexical rule)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 14 / 34
XLE-solution
Our solution
Our analysis does not place the Ezafe-clitic within the morphology(contra Samvelian 2005 and HPSG, who see it as a phrasal affixlicensed by a lexical rule)
Instead, the analysis is a cooperation between c-, f- and p-structure
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 14 / 34
XLE-solution
The c-structure - ”lion of Punjab”
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 15 / 34
XLE-solution
The c-structure - ”lion of Punjab”
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
The Ezafe is the head of an EzP-constituent and licenses a complement toits right (panjAb).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 15 / 34
XLE-solution
The c-structure - ”lion of Punjab”
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
The Ezafe is the head of an EzP-constituent and licenses a complement toits right (panjAb).
The EzP constituent forms the modifier of the NPez on the left, whichcontains the head of the entire NP.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 15 / 34
XLE-solution
The f-structure - “lion of Punjab”
"sher e panjAb"
'sher'PRED'panjAb'PRED
countCOMMONNSEMcommonNSYN
NTYPE
GEND masc, MOD-TYPE ezafe, NUM sg, PERS 330
MOD
+_EZAFECHECKcountCOMMONNSEM
commonNSYNNTYPE
GEND masc, NUM sg, PERS 31
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 16 / 34
XLE-solution
The f-structure - “lion of Punjab”
"sher e panjAb"
'sher'PRED'panjAb'PRED
countCOMMONNSEMcommonNSYN
NTYPE
GEND masc, MOD-TYPE ezafe, NUM sg, PERS 330
MOD
+_EZAFECHECKcountCOMMONNSEM
commonNSYNNTYPE
GEND masc, NUM sg, PERS 31
The head of the NP is sher
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 16 / 34
XLE-solution
The f-structure - “lion of Punjab”
"sher e panjAb"
'sher'PRED'panjAb'PRED
countCOMMONNSEMcommonNSYN
NTYPE
GEND masc, MOD-TYPE ezafe, NUM sg, PERS 330
MOD
+_EZAFECHECKcountCOMMONNSEM
commonNSYNNTYPE
GEND masc, NUM sg, PERS 31
The head of the NP is sher
panjAb is in the complement-position of the Ezafe and functions as modifierof the head of the NP
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 16 / 34
XLE-solution
The f-structure - “lion of Punjab”
"sher e panjAb"
'sher'PRED'panjAb'PRED
countCOMMONNSEMcommonNSYN
NTYPE
GEND masc, MOD-TYPE ezafe, NUM sg, PERS 330
MOD
+_EZAFECHECKcountCOMMONNSEM
commonNSYNNTYPE
GEND masc, NUM sg, PERS 31
The head of the NP is sher
panjAb is in the complement-position of the Ezafe and functions as modifierof the head of the NP
We analyze it as a MOD because the ezafe-construction is functionally alsosomewhat similar to compounding.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 16 / 34
XLE-solution
The f-structure - “lion of Punjab”
"sher e panjAb"
'sher'PRED'panjAb'PRED
countCOMMONNSEMcommonNSYN
NTYPE
GEND masc, MOD-TYPE ezafe, NUM sg, PERS 330
MOD
+_EZAFECHECKcountCOMMONNSEM
commonNSYNNTYPE
GEND masc, NUM sg, PERS 31
The head of the NP is sher
panjAb is in the complement-position of the Ezafe and functions as modifierof the head of the NP
We analyze it as a MOD because the ezafe-construction is functionally alsosomewhat similar to compounding.
However, neither f-structure nor c-structure model Ezafe’s prosodicattachment to the prosodic word on the left.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 16 / 34
XLE-solution
The p-structure - “lion of Punjab”
The proper place for the prosodic attachment would seem to be a prosodicprojection(Butt and King 1998).
CL-FORM ezafe, DOMAIN P-WORD, P-FORM sher-1P-PHRASEDOMAIN83
DOMAIN P-WORD, P-FORM panjAb-2P-PHRASEDOMAIN125
P-PHRASEDOMAIN127130
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 17 / 34
XLE-solution
The p-structure - “lion of Punjab”
The proper place for the prosodic attachment would seem to be a prosodicprojection(Butt and King 1998).
CL-FORM ezafe, DOMAIN P-WORD, P-FORM sher-1P-PHRASEDOMAIN83
DOMAIN P-WORD, P-FORM panjAb-2P-PHRASEDOMAIN125
P-PHRASEDOMAIN127130
In the p-structure, the Ezafe can be grouped with the prosodic host to itsleft, namely sher, the head of the NP
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 17 / 34
prosodic architecture
workbench
How did we accomplish this?
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 18 / 34
prosodic architecture
workbench
How did we accomplish this?
We actually (and unfortunately) had to use a hack
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 18 / 34
prosodic architecture
workbench
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 19 / 34
prosodic architecture
workbench
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
The Ezafe -e receives a Check-feature in its lexical entry:(ˆ CHECK Ezafe) = +
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 19 / 34
prosodic architecture
workbench
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
The Ezafe -e receives a Check-feature in its lexical entry:(ˆ CHECK Ezafe) = +
The N under NPez checks for the Check-feature (ˆ CHECK Ezafe) =c +
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 19 / 34
prosodic architecture
workbench
CS 1: NP
NPez
NPez_
N
sher
EzP
EZ
e
N
panjAb
The Ezafe -e receives a Check-feature in its lexical entry:(ˆ CHECK Ezafe) = +
The N under NPez checks for the Check-feature (ˆ CHECK Ezafe) =c +
If the feature is present, the CL-form Ezafe is included into the prosodicword of the N: (p::* $ CL-FORM) = ezafe
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 19 / 34
prosodic architecture
workbench
However, the linguistically correct version would be that the Ezafeitself asks to be attached to its prosodic host on the left.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 20 / 34
prosodic architecture
workbench
However, the linguistically correct version would be that the Ezafeitself asks to be attached to its prosodic host on the left.
I.e., the clitic should be able to specify that it would like to attach toa host to its left (something like left sister).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 20 / 34
prosodic architecture
workbench
However, the linguistically correct version would be that the Ezafeitself asks to be attached to its prosodic host on the left.
I.e., the clitic should be able to specify that it would like to attach toa host to its left (something like left sister).
Question: Can this be done in XLE?
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 20 / 34
prosodic architecture
workbench
However, the linguistically correct version would be that the Ezafeitself asks to be attached to its prosodic host on the left.
I.e., the clitic should be able to specify that it would like to attach toa host to its left (something like left sister).
Question: Can this be done in XLE?
If not, next question: could this be done in XLE in the future?
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 20 / 34
prosodic architecture
The Larger Picture
To be in principle able to deal with such a construction in a linguisticallyreasonable manner in XLE is important because:
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 21 / 34
prosodic architecture
The Larger Picture
To be in principle able to deal with such a construction in a linguisticallyreasonable manner in XLE is important because:
There seems to be a fundamental misunderstanding of grammararchitecture with respect to the interplay of morphology, syntax andprosody (cf. discussions within LFG on morphology, HPSGapproaches to clitics, Anderson’s books on morphology, etc.)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 21 / 34
prosodic architecture
The Larger Picture
To be in principle able to deal with such a construction in a linguisticallyreasonable manner in XLE is important because:
There seems to be a fundamental misunderstanding of grammararchitecture with respect to the interplay of morphology, syntax andprosody (cf. discussions within LFG on morphology, HPSGapproaches to clitics, Anderson’s books on morphology, etc.)
Samvelian (2005), for example, in an HPSG analysis of Persian Ezafeconcludes that the Ezafe is a phrasal affix (a “special” clitic becauseit is not promiscuous) simply because she does not seem to be able tointegrate prosodic postlexical information into the architecture —everything has to be done prelexically within the morphology.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 21 / 34
prosodic architecture
The Larger Picture
To be in principle able to deal with such a construction in a linguisticallyreasonable manner in XLE is important because:
There seems to be a fundamental misunderstanding of grammararchitecture with respect to the interplay of morphology, syntax andprosody (cf. discussions within LFG on morphology, HPSGapproaches to clitics, Anderson’s books on morphology, etc.)
Samvelian (2005), for example, in an HPSG analysis of Persian Ezafeconcludes that the Ezafe is a phrasal affix (a “special” clitic becauseit is not promiscuous) simply because she does not seem to be able tointegrate prosodic postlexical information into the architecture —everything has to be done prelexically within the morphology.
It would therefore be nice to have an alternative implementationdemonstrating the underlying architecture.
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 21 / 34
Tense/Aspect in Urdu
Tense/Aspect System of Urdu
1 large range of fine-grained temporal and aspectual distinctions
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 22 / 34
Tense/Aspect in Urdu
Tense/Aspect System of Urdu
1 large range of fine-grained temporal and aspectual distinctions
2 history: from 200 BCE to 1000 CE, a large part of the inflectionalmorphology got lost
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 22 / 34
Tense/Aspect in Urdu
Tense/Aspect System of Urdu
1 large range of fine-grained temporal and aspectual distinctions
2 history: from 200 BCE to 1000 CE, a large part of the inflectionalmorphology got lost
3 today: small number of morphemes in combination with only a fewauxiliaries and aspectual verbs
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 22 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 canonical uses of present and past tenses are not easy to identify
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 23 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 canonical uses of present and past tenses are not easy to identify
2 only one verb with a present and past paradigm (‘be’)
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 23 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 canonical uses of present and past tenses are not easy to identify
2 only one verb with a present and past paradigm (‘be’)
3 this morphology on other verbs is interpreted as subjunctive
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 23 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 canonical uses of present and past tenses are not easy to identify
2 only one verb with a present and past paradigm (‘be’)
3 this morphology on other verbs is interpreted as subjunctive
4 therefore: combinations of main verb with the ‘be’ auxiliary to expresstense
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 23 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 canonical uses of present and past tenses are not easy to identify
2 only one verb with a present and past paradigm (‘be’)
3 this morphology on other verbs is interpreted as subjunctive
4 therefore: combinations of main verb with the ‘be’ auxiliary to expresstense
5 odd situation in comparison to English, German, etc
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 23 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 Present Reading: main verb in the imperfect + auxiliary hOnA in thepresentnAdyA yasIn=kO mArtI hE
Nadya.F.Sg.Nom Yassin.Masc.Sg=Acc hit.Impf.F.Sg be.Pres.Sg’Nadya hits Yassin.’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 24 / 34
Tense/Aspect in Urdu
Present and Past Tense Reading
1 Present Reading: main verb in the imperfect + auxiliary hOnA in thepresentnAdyA yasIn=kO mArtI hE
Nadya.F.Sg.Nom Yassin.Masc.Sg=Acc hit.Impf.F.Sg be.Pres.Sg’Nadya hits Yassin.’
2 Past Reading: main verb in the perfectnAdyA=nE yasIn=kO mArA
Nadya.F.Sg.Nom Yassin.M.Sg=Acc hit.Perf.M.Sg’Nadya hit Yassin.’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 24 / 34
Tense/Aspect in Urdu
Present Tense Reading
CS 1: ROOT
S
KP
NP
N
nAdyA
KP
NP
N
yasIn
K
kO
VCmain
Vmain
V
mArtI
AuxTense
Vtns
hE
"nAdyA yasIn kO mArtI hE"
'mAr<[1:nAdyA], [16:yasIn]>'PRED
'nAdyA'PRED
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE nom, GEND fem, NUM sg, PERS 31
SUBJ
'yasIn'PRED
obl_NMORPHCHECK
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE acc, GEND masc, NUM sg, PERS 316
OBJ
infl_MTYPE_VMORPH
-_RESTRICTEDCHECK
+AGENTIVELEX-SEM
ASPECT impf, MOOD indicative, TENSE presTNS-ASP
CLAUSE-TYPE decl, PASSIVE -, VFORM impf, VTYPE main33
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 25 / 34
Tense/Aspect in Urdu
Perfect
1 standard present perfect reading: main verb in the perfect and ’be’ inthe presentnAdyA=nE yasIn=kO dEkHA hE
Nadya,F.Sg=Erg Yassin.M.Sg.=Acc see-Perf.M.Sg be.Pres.3.Sg’Nadya has seen Yassin.’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 26 / 34
Tense/Aspect in Urdu
Perfect
1 standard present perfect reading: main verb in the perfect and ’be’ inthe presentnAdyA=nE yasIn=kO dEkHA hE
Nadya,F.Sg=Erg Yassin.M.Sg.=Acc see-Perf.M.Sg be.Pres.3.Sg’Nadya has seen Yassin.’
2 past perfect reading: the main verb in the perfect and ’be’ in the pastnAdyA=nE yasIn=kO dEkHA tHA
Nadya.F.Sg=Erg Yassin.M.Sg=Acc see-Perf.M.Sg be-Past.M.Sg’Nadya had seen Yassin.’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 26 / 34
Tense/Aspect in Urdu
Continuitive Reading
1 intense use of various ways of expressing a continuing action
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 27 / 34
Tense/Aspect in Urdu
Continuitive Reading
1 intense use of various ways of expressing a continuing action
2 complex continuitive, iterative and habitual expressions which seem tobuild on and interact with one another in a compositional manner
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 27 / 34
Tense/Aspect in Urdu
General Progressive Reading
1 progressive is a relatively recent innovation of Urdu
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 28 / 34
Tense/Aspect in Urdu
General Progressive Reading
1 progressive is a relatively recent innovation of Urdu
2 general progressive reading: main verb stem + progressive rah ’stay’+ present or past auxiliary
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 28 / 34
Tense/Aspect in Urdu
General Progressive Reading
1 progressive is a relatively recent innovation of Urdu
2 general progressive reading: main verb stem + progressive rah ’stay’+ present or past auxiliary
3 rah still means ’stay’ or ’live’, however in this case it is analyzed as anaspectual auxiliarynAdyA yasIn=kO mAr rah-I
Nadya.F.Sg.Nom Yassin.M.Sg=Acc hit stay-Perf.F.Sg
hE
be.Pres.3.Sg’Nadya is hitting Yassin.’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 28 / 34
Tense/Aspect in Urdu
General Progressive Reading
CS 1: ROOT
S
KP
NP
N
nAdyA
KP
NP
N
yasIn
K
kO
VCmain
Vmain
V
mAr
AuxAsp
AuxProg
Vprog
rahI
AuxTense
Vtns
hE
"nAdyA yasIn kO mAr rahI hE"
'mAr<[1:nAdyA], [16:yasIn]>'PRED
'nAdyA'PRED
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE nom, GEND fem, NUM sg, PERS 31
SUBJ
'yasIn'PRED
obl_NMORPHCHECK
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE acc, GEND masc, NUM sg, PERS 316
OBJ
infl_MTYPE_VMORPH
_AUXASP +, _RESTRICTED -CHECK
+AGENTIVELEX-SEM
ASPECT prog, MOOD indicative, TENSE presTNS-ASP
CLAUSE-TYPE decl, PASSIVE -, VFORM bare, VTYPE main33
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 29 / 34
Tense/Aspect in Urdu
Iteration and Longer Continuation with ’go’ and ’walk’
1 the main verb jA ’go’ normally functions as a main and a light verb ora passive auxiliary, here: jA functions as an aspectual auxiliary
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 30 / 34
Tense/Aspect in Urdu
Iteration and Longer Continuation with ’go’ and ’walk’
1 the main verb jA ’go’ normally functions as a main and a light verb ora passive auxiliary, here: jA functions as an aspectual auxiliary
2 the action is considered to continue over a longer time spannAdyA mArtI jAtI hE
Nadya.F.Sg.Nom hit-Impf.F.Sg go-Impf.F.Sg be.Pres.3.Sg’Nadya keeps on hitting (over a very long time).’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 30 / 34
Tense/Aspect in Urdu
Iteration and Longer Continuation with ’go’ and ’walk’
CS 1: ROOT
S
KP
NP
N
nAdyA
VCmain
Vmain
V
mArtI
AuxAsp
AuxCont
Viter
jAtI
AuxTense
Vtns
hE
"nAdyA mArtI jAtI hE"
'mAr<[1:nAdyA]>'PRED
'nAdyA'PRED
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE nom, GEND fem, NUM sg, PERS 31
SUBJ
infl_MTYPE_VMORPH
_AUXASP +, _RESTRICTED -CHECK
ASPECT impf, CONT +, ITERATIVE +, MOOD indicative, TENSE presTNS-ASP
CLAUSE-TYPE decl, LEX-SEM unerg, PASSIVE -, VFORM impf, VTYPE main16
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 31 / 34
Tense/Aspect in Urdu
Iteration and Longer Continuation with ’go’ and ’walk’
But it gets even better......
1 You can also add the auxiliaries rah and cal to the constructionnAdyA yasIn=kO mArtI
Nadya.F.Sg.Nom Yassin.M.Sg=Acc hit-Impf.F.Sg
calI jA rahI hE
walk-Perf.F.Sg go stay-Perf.F.Sg be.Pres.3.Sg’Nadya keeps on hitting Yassin (over a very long time).’
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 32 / 34
Tense/Aspect in Urdu
Iteration and Longer Continuation with ’go’ and ’walk’
CS 1: ROOT
S
KP
NP
N
nAdyA
KP
NP
N
yasIn
K
kO
VCmain
Vmain
V
mArtI
AuxAsp
AuxCont
Vcont
calI
Viter
jA
Vprog
rahI
AuxTense
Vtns
hE
"nAdyA yasIn kO mArtI calI jA rahI hE"
'mAr<[1:nAdyA], [16:yasIn]>'PRED
'nAdyA'PRED
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE nom, GEND fem, NUM sg, PERS 31
SUBJ
'yasIn'PRED
obl_NMORPHCHECK
namePROPER-TYPEPROPERNSEM
properNSYNNTYPE
+SPECIFICSEM-PROP
CASE acc, GEND masc, NUM sg, PERS 316
OBJ
infl_MTYPE_VMORPH
_AUXASP +, _RESTRICTED -CHECK
+AGENTIVELEX-SEM
ASPECT prog, CONT +, ITERATIVE +, MOOD indicative, TENSE presTNS-ASP
CLAUSE-TYPE decl, PASSIVE -, VFORM impf, VTYPE main33
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 33 / 34
Tense/Aspect in Urdu
Future Work
1 work is being continued in order to explore the entire feature system
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 34 / 34
Tense/Aspect in Urdu
Future Work
1 work is being continued in order to explore the entire feature system
2 also need a DURATION feature and need to explore full space ofcombinatorial possibilities (a first survey is in Butt 2008).
Urdu Grammar Group (Konstanz) Urdu Report ParGram 2008, California 34 / 34