Modal syntax cuts short the claim that modern Persian lacks

Modal syntax cuts short the claim that

modern Persian lacks apocopated inﬁnitives

Setayesh Dashti

University of Oxford

Ash Asudeh

University of Rochester

LFG 2023

University of Rochester

July 23, 2023

1 Background

• Persian is an SOV Indo-European language with ‘pro-drop’.

• Verbal morphology follows a two-stem system, traditionally called

1. Present stem: no overt present tense marker; and

(1) xor ‘eat’

2. Past stem: modulo suppletive patterns, the past tense is regularly marked with -d and its allomorphs

(Anoushe 2018).

(2) xord ‘eat’

• The present stem always occurs with either aspectual or mood markers; mi- for imperfective aspect (3a) and

be- for subjunctive mood (3b).

2,3

• The unpreﬁxed past stem with agreement sufﬁxes is used to show the perfective aspect (3c).

• Past imperfective, progressive and perfect are also derived from the past stem with agreement sufﬁxes; for

example, past imperfective is formed with the same preﬁx as present imperfective, mi- (3d).

(3) a. Nika

Nika

madrese

school

mi-rav-ad.

IPFV-go.PRES-3SG

‘Nika goes to school.’

b. Nika

Nika

ayad

may

madrese

school

be-rav-ad.

SBJV-go.PRES-3SG

‘Nika might go to school.’

c. Nika

Nika

madrese

school

raf-t.

go-PAST.3SG

‘Nika went to school.’

d. ba

ce-h

child-PL

har

every

ruz

day

madrese

school

mi-raf-t-and.

IPFV-go-PAST-3PL

‘The kids used to go to school every day.’

The dialect reported on here is colloquial spoken Persian, not the written standard.

Glosses are abbreviated as follows: AUX–auxiliary, COP–copula, EZ–ezafe (nominal linker), IPFV–imperfect, INF–inﬁnitive, NEG–

negation, PP–past participle, PRES–present tense, PAST–past tense, SBJV–subjunctive mood, SG–singular, PL–plural, DO–direct object.

We use the hyphen (-) to indicate an afﬁx boundary and an equal sign (=) to indicate clitic attachment.

What we have glossed as IPFV–imperfective is sometimes glossed as DUR–durative. We prefer to gloss it based on its morphological

form rather than its typical morphosyntactic function.

Past imperfective also functions as a fake past to convery counterfactuality, regardless of tense (Bjorkman and Halpert 2017).

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 2

• Persian contains several adverbial and complex predicate modals, but there are two main simplex verbal

modal auxiliaries, b

ayestan (necessity/) and

sodan (possibility/♦).

• These modals always appear in the default third person singular form: b

ayad (.PRES)/b

ayest (.PAST) and

mi-

se (IPFV-♦.PRES)/mi-

sod (IPFV-♦.PAST).

• They can either occur with:

1. a ﬁnite complement (4), marked with subjunctive mood in present tense (4a) or imperfective aspect in past

tense (4b); or

2. a nonﬁnite complement (5)

• In the second case, the verb in the complement has a simple past stem, which resembles the third per-

son singular past inﬂection, but is historically an apocopated inﬁnitive (short inﬁnitive); importantly, it is

interpreted as an impersonal (5).

(4) a. b

ayad

.PRES

xune

home

be-rav-am.

SBJV-go.PRES-1SG

‘I have to go home.’

b. b

ayad

.PRES

ceh

child-PL

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

‘The children had to go home.’

(5) b

ayad

.PRES

zood

early

xune

home

raf-t.

go-??

‘It’s necessary to go home early.’/

‘One must go home early.’

• When the modal occurs with a ﬁnite complement, it is possible to topicalize the embedded subject to the

left:

(6) a. b

ayad

.PRES

ceh

child-PL

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

‘The children had to go home.’

b. ba

ceh

child-PL

ayad

.PRES

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

∼‘As for the children, they had to go home.’

• A verb that works very similarly to the simplex modal verbs is be nazar

amad-an (lit. ‘to opinion come-

INF’/‘to seem’), which is the equivalent of seem.

• Unlike b

ayad (/necessity), which never inﬂects for agreement, some speakers allow both the non-agreeing/default

form (7b) and the agreeing form (7c); all speakers allow the ﬁrst, non-agreeing form:

(7) a. be

nazar

opinion

mi-

a-d

IPFV-come.PRES-3SG

that

ce-h

child-PL

xaste

tired

so-d-an.

become-PAST-3PL

‘It seems that the children have gotten tired.’

b. ba

ce-h

child-PL

nazar

opinion

mi-

a-d

IPFV-come.PRES-3SG

that

xaste

tired

so-d-an.

become-PAST-3PL

‘As for the children, it seems that they have gotten tired.’

c. % ba

ce-h

child-PL

nazar

opinion

mi-

a-n

IPFV-come.PRES-3PL

that

xaste

tired

so-d-an.

become-PAST-3PL

‘The children seem to have gotten tired.’

• Note that in (7) we are emphasizing the colloquialness of the reported example by using certain spoken-only

forms, such as the contracted 3PL in (7c): -(a)n rather than -(a)nd.

There is some debate over the status of the modal

ayestan. Some literature, such as Karimi (2005) and Taleghani (2008), treats it

as another modal auxiliary, while other literature, such as Labbafankhosh and Darzi (2015), treats it is a modal adverb.

The subjunctive in the past tense has the same form as the imperfective.

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 3

2 Puzzles/questions

1. How should we account for the complement in (5)?

(5) b

ayad

.PRES

zood

early

xune

home

raf-t.

go-??

‘It’s necessary to go home early.’/ ‘One must go home early.’

Is it a past tense form or a short inﬁnitive (synchronically as well as diachronically)?

2. How can we capture the impersonal and personal readings of modals like (5) vs. (4)?

(4) a. b

ayad

.PRES

xune

home

be-rav-am.

SBJV-go.PRES-1SG

‘I have to go home.’

b. b

ayad

.PRES

ceh

child-PL

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

‘The children had to go home.’

3. What is the syntactic structure of simplex modal constructions?

4. How should the variable agreement displayed in (7) be explained?

(7) b. ba

ce-h

child-PL

nazar

opinion

mi-

a-d

IPFV-come.PRES-3SG

that

xaste

tired

so-d-an.

become-PAST-3PL

‘As for the children, it seems that they have gotten tired.’

c. % ba

ce-h

child-PL

nazar

opinion

mi-

a-n

IPFV-come.PRES-3PL

that

xaste

tired

so-d-an.

become-PAST-3PL

‘The children seem to have gotten tired.’

3 The syntax of Persian modals

• The main challenge is the construction with an impersonal interpretation, as in (5), shown here with the

complement structure made explicit, or the similar example (8):

) b

ayad

.PRES

[zood

[early

xune

home

raf-t].

go-??]

‘It’s necessary to go home early.’/ ‘One must go home early.’

(8) b

ayad

.PRES

[

sab-h

night-PL

eight

a’at

hour

ab-id].

sleep-??

‘It’s necessary to sleep for eight hours a night.’/‘One must sleep for eight hours a night.’

• This is often called the impersonal construction (IC; e.g. Karimi 2008).

• Let us assume that the complement is a clause, because there is apparent inﬂection on the embedded predicate.

• The question is: what kind of clause?

• Is it ﬁnite, as indicated by its shared form with the past tense (see, e.g., Karimi 2008)?

• Is it inﬁnitival, despite the shared form (Karimi 2005, Samvelian 2018)?

• Perhaps it is not a clause at all,

but presumably some kind of nominal?

For example Darzi and Kwak (2015: 2) write, “This language, like Greek and unlike English, lacks nonﬁnite clauses.”

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 4

3.1 Inﬁnitives in Persian

• The morphosyntactically distinguished Persian inﬁnitive is formed from the past stem and the sufﬁx -an.

(9) a. dav-id-an

run-PAST-INF

bar

aye

for

sal

amati

health

moﬁd-e

beneﬁcial-COP.3SG

‘Running has health beneﬁts.’

b. Nika

Nika

dav-id-an=ro

run-PAST-INF=DO

dust

friend

ar-e

have-PRES.3SG

‘Nike likes running.’

• Contrast these with similar impersonal complements:

(10) a. b

ayad

.PRES

bar

aye

for

sal

amati

health

dav-id

run-??

‘One should run for health reasons.’

b. b

ayad

.PRES

dav-id-an=ro

run-PAST-INF=DO

dust

friend

ash-t

have-??

‘One must like running.’

3.1.1 The syntax of Persian inﬁnitives

• The forms in (9) and (10) do not have the same distribution.

• There is, indeed, a case for arguing that the long inﬁnitives are nominals, whereas the other forms are

short/apocopated inﬁnitives forming nonﬁnite verbal clauses.

1. This immediately suggests a function for the sufﬁx -an: it is a nominalizer.

2. Long inﬁnitives can take modiﬁers with the nominal linker -e (ezafe). The inﬁnitive in (11a) has a genitive

modiﬁer, the one in (11b) has an adjectival one.

(11) a. xand-id-an-e

laugh-PAST-INF-EZ

nowz

baby

ame

sbax

s-e

calming-COP.3SG

‘Babies’ laughing is calming.’

b. Ali

Ali

dav-id-an-e

run-PAST-INF-EZ

arum=ro

slow=DO

sen

swim

kar-d-an-e

do-PAST-INF-EZ

sari’

fast

tarjih

preference

mi-d-e

IMPF-give.PRES-3SG

‘Ali prefers running slowly to swimming fast.’

3. It explains the fact that a) unbounded dependencies can be formed into the impersonal complement marked

by the short inﬁnitive, as shown in (6b) repeated here as (12a), since it is a clause, b) whereas unbounded

dependencies cannot be formed into long inﬁnitives as putative complex nominals (a):

(12) a. ba

ceh

child-PL

ayad

.PRES

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

∼‘As for the children, they had to go home.’

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 5

(13) a. Sarina

Sarina

tanh

alone

baby

sir

milk

a-d-an=ro

give-PAST-INF=DO

dust

friend

na-d

are

NEG-have.PRES.3SG

‘Sarina doesn’t like to breast-feed the child alone.’

b. * tanh

alone

Sarina

baby

sir

milk

a-d-an=ro

give-PAST-INF=DO

dust

friend

na-d

are

NEG-have.PRES.3SG

‘Sarina doesn’t like to breast-feed the child alone.’

c. * be

baby

Sarina

tanh

alone

sir

milk

a-d-an=ro

give-PAST-INF=DO

dust

friend

na-d

are

NEG-have.PRES.3SG

‘Sarina doesn’t like to breast-feed the child alone.’

• In sum, we have presented syntactic arguments for why the long inﬁnitive is a nominal and for why the form

in the impersonal construction is not.

• This already suggests that the impersonal complement is a nonﬁnite clause, as this is the obvious alternative

to its being a nominal.

• In the next section, we will show that in fact the form in the impersonal is nonﬁnite, since it cannot take

tense/aspect/mood-marking without losing its impersonal interpretation.

• The assumption that the form in the impersonal is a short inﬁnitive that heads a clause explains why it does

not take the -an sufﬁx: this sufﬁx is a nominalizer and the short inﬁnitive is not a nominal.

• The fact that this looks superﬁcially like the past form is a fact of morphosyntactic syncretism and nothing

more, because the short inﬁnitive does not function as a past form or have past meaning, as we will show

next.

3.1.2 The function and interpretation of Persian inﬁnitives

• The past ﬁnite complement of the modal should bear imperfective marking, but adding this marking to the

sort of complement under discussion renders an impersonal reading unavailable and requires it to have a

personal reading. Contrast (14a), repeated from (8) above, with (14b):

(14) a. b

ayad

.PRES

sab-h

night-PL

eight

a’at

hour

ab-id.

sleep-??

‘It’s necessary to sleep for eight hours a night.’/‘One must sleep for eight hours a night.’

b. b

ayad

.PRES

sab-h

night-PL

eight

a’at

hour

mi-x

ab-id.

IPFV-sleep-PAST.3SG

#‘It’s necessary to sleep for eight hours a night.’/‘One must sleep for eight hours a night.’

X‘pro.3SG had to sleep for eight hours a night.’

• Persian is sometimes assumed to lack a nonﬁnite clause (Darzi and Kwak 2015), exactly because of the

similarity in morphological form between the third singular past form, which is unmarked for agreement

morphology (e.g., raf-t go-PAST.3SG) and the simple stem form in question (e.g., raf-t go-??).

• But, as we have just seen, this does not account correctly for the impersonal readings.

• We instead assume that this verbal form is inﬁnitival and thus unmarked for TENSE/ASPECT/MOOD.

Sentence (14b) can have another interpretation in which the subject of the verb is pro-dropped, which will translate to ‘pro.3SG

should have slept eight hours a night’. This is a different construction than the one in question; most importantly, the alternative

construction is never impersonal.

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 6

• The future construction, shown in (15), provides further evidence for nonﬁniteness of this verbal form, now

glossed INF.

• This builds on Lowe’s (2019) claim that nonﬁnite forms generally appear in periphrastic constructions as the

lexical content of the clausal predicate.

(15) Ali

Ali

farda

tomorrow

madrese

school

ah-ad

want-3SG

raf-t.

go-INF

‘Ali will go to school tomorrow.’

• In the analysis section §4, a template (Dalrymple et al. 2004, Asudeh et al. 2013) is used to generate this

defective/inﬁnitival verbal form.

Summary

• The ﬁrst question posed in §2 above was whether the morphologically ambiguous form in the impersonal

modal construction is a past tense form or a short inﬁnitive (apocopated inﬁnitive).

• We argued that the form in question, just as in (15), is an inﬁnitival form synchronically, and that the apoco-

pated inﬁnitive is morphologically formed by referral to the past stem, which explains their identity.

• However, the agreeing past tense form and the non-agreeing short inﬁnitive have different functions and

interpretations.

4 An LFG analysis of Persian modal syntax

• LFG assumes a separation of syntax into two levels, c(onstituent)-structure and f(unctional)-structure.

• C-structure represents syntactic distribution, via categories, constituency, hierarchy, and linear order.

• F-structure represents relational aspects of syntax, such as grammatical functions, agreement, case-marking,

as well as local (control/raising) and non-local (unbounded dependencies) relations.

• The following illustrate the c-structure position of the modal and the general structure of the CP and IP:

(16) a. Mariam

Mariam

[

goft

said

[

ke]

that

[

kodoom

which

ketab-ha-ro

book-PL=DO

[

bayad]

must

[

ce-ha

child-PL

be-xun-and

SBJV-read-3PL

]]]]]]

‘Mariam said that the children must read WHICH BOOKS?’

b. Mariam

[

goft

[

ke]

that

[

kodoom

which

ketab-ha-ro

book-PL=DO

[

ce-ha

child-PL

[

bayad]

must

[

be-xun-and

SBJV-read-3PL

]]]]]]]

‘Mariam said that the children must read WHICH BOOKS?’

c. Mariam goft [

[

ke]

that

[

kodoom

which

ketab-ha-ro

book-PL=DO

[

ce-ha

child-PL

[

xun-d-and

read-PAST-3PL

]]]]]]

‘Mariam said that the children read WHICH BOOKS?’

d. Mariam goft [

[

ke]

that

[

ce-ha

child-PL

[

kodoom

which

ketab-ha-ro

book-PL=DO

[

xun-d-and

read-PAST-3PL

]]]]]]

‘Mariam said that, as for the children, they read WHICH BOOKS?’

The insight that the so-called past stem in these constructions is the apocopated inﬁnitive is not novel (especially in the context of

the future construction; Windfuhr 1979), but the theoretical literature seems largely to have taken it to be the PAST.3SG form of the verb

(for instance, Karimi 2008, Mirrazi 2022).

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 7

• Example (16a) shows that there is a position for the top of an unbounded dependency below C, since the C

position is occupied by an overt complementizer.

• We assume that this position is an IP-adjunct, since otherwise the wh-phrase would be in regular subject

position in SpecIP.

• Example (16b) shows that there is a position for an internal topic below this IP-adjunct position.

• We postulate that this is an I

-adjunct.

• Thus, in (16b), ba

ceha is in a non-agreeing topic position, reﬂected by the lack of plural agreement on the

modal (which is not possible).

• Example (16c) shows that when an agreeing subject is present, in a simple case without a modal, it can be

assumed to occur in the standard SpecIP subject position.

• Example (16d) shows the IP-adjunction in (16c) can be reversed freely, with a topicalized ba

ceha occurring

adjoined to an IP that itself contains an IP-adjoined wh-phrase.

• The following rules license the left periphery in the c-structures in (16):

(17) a. CP → XP

(↑ DIS) = (↑ DISPATH)

↑ = ↓

b. C

→ C

↑ = ↓

c. IP → XP

(↑ SUBJ) = ↓

↑ = ↓

d. I

→ I

↑ = ↓

e. IP → XP

(↑ DIS) = (↑ DISPATH)

↑ = ↓

f. I

→ XP

(↑ DIS) = (↑ DISPATH)

(↑ DIS)σ ∈ (↑σι TOPIC)

↑ = ↓

• We assume the following lexical entry for b

ayad (.PRES):

(18) bayad I (↑ PRED) = ‘musthCFiSUBJ’

(↑ TENSE) = PRES



@EXPL-SUBJ

(↑ COMP MOOD) =

SUBJUNCTIVE



(↑ SUBJ) = (↑ XCOMP SUBJ)



• This lexical entry is for both the personal and impersonal present modal construction, so some information

is shared, but the information needs to diverge at some point.

• The modal in both constructions occupies an identical position, hence the category I is shared.

• The modal in both constructions is present tense, hence the speciﬁcation of [TENSE PRES].

• The two modals are also forms of the same basic predicate, so have the same PRED value.

• The distinction is that the personal construction takes a closed sentential complement, COMP, which can

realize its own subject, whereas the impersonal is a kind of subject raising construction, taking an open

sentential complement, XCOMP, which cannot realize its own subject.

• The personal construction also requires that its complement independently have subjunctive mood.

The equation regarding DIS in (17e) connects the top and bottom of the unbounded dependency in the corresponding f-structure

(Dalrymple et al. 2019: 39ff). The grammatical function DIS is a way of unifying the previously distinguished f-structural functions of

TOPIC and FOCUS, which are properly elements of i-structure rather than f-structure, as an overlay function that captures the abstract

f-structural role of the top of unbounded dependencies; this was originally proposed by Asudeh (2004), where the function was named

UDF, for unbounded dependency function, The set statement regarding DIS and TOPIC in (17f) states that the top of the unbounded

dependency encodes a TOPIC at i(nformation)-structure (Dalrymple and Nikolaeva 2011).

The fact that the subjunctive mood requirement is information that is checked by the modal, rather than information that is actually

contributed by it, is modelled by the constraining equation, marked =

rather than simply =.

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 8

• The lefthand side of (18) calls a template, EXPL-SUBJ.

• A template call is marked by @.

• The semantics of template invocation is very simple (Dalrymple et al. 2004): the template just deﬁnes

a bundle of lexical information and gives it a name; when the template is invoked, the corresponding

information it encodes is substituted in.

• Note that a template may call other templates, so there may be multiple such substitutions; this is also

exempliﬁed by EXPL-SUBJ.

(19) EXPL-SUBJ := ¬(↑ SUBJ PRED)

@SUBJ-3SG

(20) SUBJ-3SG := (↑ SUBJ PERS) = 3

(↑ SUBJ NUM) = SG

• The righthand case in (18) is for the nonﬁnite-complements containing apocopated inﬁnitives, e.g. (5).

• We deﬁne the following templates for apocopated inﬁnitives:

(21) APINF(P) := (↑ PRED) = P

@NO-TAM

@IMPERS-SUBJ

(22) NO-TAM := ¬(↑ TENSE)

¬(↑ ASPECT)

¬(↑ MOOD)

(23) IMPERS-SUBJ := (↑ SUBJ PRED) = ‘pro’

(↑ SUBJ PRONTYPE) = IMPERSONAL

@SUBJ-3SG

• Note that the APINF template is one that takes an argument: whatever is passed in as the argument

becomes the value of PRED.

• The lexical entry for a sample apocopated inﬁnitive, raft (‘go’), is:

(24) raft V @APINF(‘gohSUBJ,OBLi’)

• The f-structures for examples (5), (6a), and (6b) respectively are shown in (25)–(27); the corresponding

examples are repeated in (28)–(30).

12,13

(25) (26) (27)

(28) b

ayad

.PRES

zood

early

xune

home

raf-t.

go-INF

‘It’s necessary to go home

early.’/

‘One must go home early.’

(29) b

ayad

.PRES

ceh

child-PL

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

‘The children had to go

home.’

(30) ba

ceh

child-PL

ayad

.PRES

xune

home

mi-raf-t-and.

IPFV-go-PAST-3PL

‘The children had to go

home.’

Note that the imperfective and the subjunctive are syncretic in the past tense. Therefore, although for consistency we have always

glossed mi- as imperfective (IPFV), we assume that it can convey subjunctive mood and hence satisfy the constraining equation in (18).

We do not attempt to account for this syncretism here.

See footnote 10 regarding the function DIS in (27).

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 9

Interim summary

• We are now in a position to answer the second and third questions in §2.

• The second question asked, What is the syntactic structure of simplex modal constructions in Persian?

• Persian modals occupy the category I; this is unsurprising from an LFG-theoretic perspective, since modals

in general are base-generated in this category or C (depending on distribution).

• This interacts with the general structure of the left periphery that we have provided—see (16) and the

c-structure rules in (17)—such that all and only the valid orderings are captured.

• The third question asked, How can we capture the personal and impersonal readings of modals like (4a) vs.

(5)?

(4) a. b

ayad

.PRES

xune

home

be-rav-am.

SBJV-go.PRES-1SG

‘I have to go home.’

) b

ayad

.PRES

zood

early

xune

home

raf-t.

go-INF

‘It’s necessary to go home early.’/

‘One must go home early.’

• The lexical entry for the modal b

ayad (.PRES) in (18) explains the differences by treating the personal

as a subcategorized subjunctive COMP and treating the impersonal as a raising predicate which allows the

requirements of the apocopated inﬁnitive, as captured in template (21), to control the reading, with the

modal simply wrapping necessity around this.

4.1 Capturing the variation

• The fourth question in §2 asked, How should the variable agreement displayed in (7) be explained?.

(7) b. ba

ce-h

child-PL

nazar

opinion

mi-

a-d

IPFV-come.PRES-3SG

that

xaste

tired

so-d-an.

become-PAST-3PL

‘As for the children, it seems that they have gotten tired.’

c. % ba

ce-h

child-PL

nazar

opinion

mi-

a-n

IPFV-come.PRES-3PL

that

xaste

tired

so-d-an.

become-PAST-3PL

‘The children seem to have gotten tired.’

• Our proposal may have been anticipated by now:

1. Speakers who only allow the non-agreeing form (7b) maintain an analysis of the preposed nominal, ba

ceh

(‘children’), as a TOPIC.

• It is a general fact about Persian (and perhaps universally), that topichood is not sufﬁcient to directly

trigger agreement.

2. Speakers who do allow the agreeing form have instead analyzed the preposed nominal as a SUBJ, which

robustly triggers agreement in Persian.

• The light verb,

amadan (‘to come’), in this construction, unlike the modals, is a fully agreeing form

(mi

an), which allows for (7c).

• For these speakers, be nazar

amadan ‘seems’, when it shows agreement with a preposed element, is

akin to English copy raising (Rogers 1973, Postal 1974):

(31) Harry seems like he is tired.

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 10

• However, since Persian is pro-drop, the embedded pronominal does not surface.

• When it does not show agreement, as in (7b), it is akin to English seems that with topicalization; i.e.,

there is an (in Persian, unrealized) expletive subject with the bare-topicalized nominal occurring in only

apparent subject position:

(32) As for Harry, it seems that he is tired.

Summary

• We hope to have shown that a fairly simple LFG analysis of Persian modal syntax is possible using standard

tools of the framework.

• This analysis lends further support to the view that synchronic Persian grammar indeed does contain an apoc-

opated inﬁnitive, and that this short inﬁnitive’s formal resemblance to the past stem/zero-marked PAST.3SG

form is misleading.

5 Conclusion

• We have now answered all of the questions that we posed in §2:

1. How should we account for the complement in the impersonal modal construction?

⇒ It is a (short/apocopated) inﬁnitival which is formally but not functionally identical to the past stem.

The formal identity can be captured by standard means, such as rules of referral (Stump 2016) or their

alternatives in other frameworks.

For example, in L

FG (Melchin et al. 2020, Asudeh and Siddiqi 2022, among others; lrfg.online), the

required statement could look like this:

(33) h[v],@AP-INF( )i

−→ ν h[v],@PASTi

The statement in (33) is intended as a rule schema or meta-rule. Its lefthand side is underspeciﬁed for the

parameter of the AP-INF template and can thus match any instance of it in the Vocubulary. The schema

states that the exponent of this vocabulary item is the exponent of the PAST template, which controls

contribution of [TENSE PAST] to the f-structure. As mentioned above, the past tense exponent is -d and

its allomorphs (Anoushe 2018). Thus, the schema in (33) elegantly captures the fact that the apocopated

inﬁnitive form of ‘eat’ is xord, as in Bayad xord (‘One must eat.’), that the one of ‘go’ is raft, as in (28)

above, and that the one of ‘sleep’ is x

abid, as in (14a) above. But it does so without ever referring to any

particular form.

2. How can we capture the personal vs. impersonal readings of the modals?

⇒ The distinction is governed by the lexical entry for the modal and the templates that it uses.

3. What is the syntactic structure of the simplex modal constructions?

⇒ The modal is in I. There is a topic position above this, but below C.

4. How should the variable speaker agreement displayed for the subject of the raising/perception verb be

nazar

amadan (lit. ‘to opinion come’/∼‘to seem like/that’) be captured?

⇒ The light verb that anchors this predicate,

amadan, is a fully agreeing predicate, unlike the modals.

Some speakers have reanalyzed the preposed topic as a subject, since the position it occupies is in many

cases string-identical to subject position. On this analysis, the verb must agree with the subject, as is the

case overall in Persian grammar. However, the other analysis, in which the preposed nominal is actually a

topic, is also available, but does not trigger agreement. Therefore, these speakers display variation in their

utterances.

In fact, one could possibly get it to surface given enough discourse support, but it is difﬁcult because of opposing discourse forces.

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 11

References

Anoushe, Mazdak. 2018. A Revision of Persian Past Tense Inﬂection: A Distributed Morphology Approach.

Language Related Research 9(1): 57–80.

Asudeh, Ash. 2004. Resumption as Resource Management. Ph.D. thesis, Stanford University.

Asudeh, Ash, Mary Dalrymple, and Ida Toivonen. 2013. Constructions with Lexical Integrity. Journal of

Language Modelling 1(1): 1–54.

Asudeh, Ash, and Daniel Siddiqi. 2022. Morphology in LFG. To appear in Dalrymple (2022).

Bjorkman, Bronwyn M., and Claire Halpert. 2017. In an imperfect world: Deriving the typology of counter-

factual marking. In Ana Arregui, Mar

ıa Luisa Rivero, and Andr

es Salanova, eds., Modality Across Syntactic

Categories, 157–178. Oxford University Press.

Dalrymple, Mary, ed. 2022. The Handbook of Lexical Functional Grammar. Language Science Press. Forth-

coming.

Dalrymple, Mary, Ronald M. Kaplan, and Tracy Holloway King. 2004. Linguistic Generalizations over De-

scriptions. In Miriam Butt and Tracy Holloway King, eds., Proceedings of the LFG04 Conference, 199–208.

Stanford, CA: CSLI Publications.

Dalrymple, Mary, John J. Lowe, and Louise Mycock. 2019. The Oxford Reference Guide to Lexical Functional

Grammar. Oxford: Oxford University Press.

Dalrymple, Mary, and Irina Nikolaeva. 2011. Objects and Information Structure. Cambridge: Cambridge

University Press.

Darzi, Ali, and Saera Kwak. 2015. Syntax and semantics of subjunctive clauses in Persian. Lingua 153: 1–13.

Karimi, Simin. 2005. A Minimalist Approach to Scrambling: Evidence from Persian. Berlin: De Gruyter

Mouton.

—. 2008. Raising and Control in Persian. In Simin Karimi, Vida Samiian, and Don Stilo, eds., Aspects of

Iranian Linguistics., 177–208. Cambridge Scholars Publishers Ltd.

Labbafankhosh, Zahra, and Ali Darzi. 2015. On the Syntactic Analysis of Persian Modals:

ayad ‘Perhaps’

and B

ayad ‘Must’ [In Persian]. Journal of Linguistic Research 5(2): 97–112.

Lowe, John J. 2019. The Syntax and Semantics of Nonﬁnite Forms. Annual Review of Linguistics 5(1): 309–

328.

Melchin, Paul B., Ash Asudeh, and Dan Siddiqi. 2020. Ojibwe Agreement in Lexical-Realizational Functional

Grammar. In Miriam Butt and Ida Toivonen, eds., Proceedings of the LFG20 Conference, 268–288. Stanford,

CA: CSLI Publications.

Mirrazi, Zahra. 2022. Tense in Conditionals: Ins and Outs. Ph.D. thesis, University of Massachusetts Amherst.

Postal, Paul M. 1974. On Raising: One Rule of English Grammar and its Theoretical Implications. Cambridge,

MA: MIT Press.

Rogers, Andy. 1973. Physical Perception Verbs in English: A Study in Lexical Relatedness. Ph.D. thesis,

UCLA.

Dashti · Asudeh Modal syntax . . . Persian . . . inﬁnitives LFG 2023 · 12

Samvelian, Pollet. 2018. Speciﬁc Features of Persian Syntax: The Ez

afe Construction, Differential Object

Marking and Complex Predictaes. In Anousha Sedighi and Pouneh Shabani-Jadidi, eds., The Oxford Hand-

book of Persian Linguistics, 226–269. Oxford University Press.

Stump, Gregory T. 2016. Inﬂectional Paradigms: Content and Form at the Syntax-Morphology Interface.

Cambridge: Cambridge University Press.

Taleghani, Azita H. 2008. Modality, Aspect and Negation in Persian. Amsterdam: John Benjamins.

Windfuhr, Gernot L. 1979. Persian Grammar: History and State of Its Study. The Hague: Mouton.