You are on page 1of 15


University of Leiden, November 28-29

Verb particles in Medieval French: a syntactic account

Michelle Troberg
University of Toronto Mississauga

1. Introduction
Medieval French particles look much like their contemporary Romance counterparts, which are still very
productive in Northern Italian varieties and present (to a lesser extent) in others such as Spanish, Catalan,
Romanian, and so on. Particles with directional or aspectual interpretations are not, however, present in modern
French; see Iacobini & Masini (2007); Burnett & Tremblay (2009); Mateu & Rigau (2010); Iacobini (2015).12

Latin Old French Italian Catalan Spanish French

‘up’ sursum sus, amont, contremont su amunt arriba ---
‘down’ deorsum jus, aval, contreval giù avall abajo ---
‘out’ foras fors fuori fora fuera ---
‘in’ intus ens dentro dins dentro ---
‘forward’ abante avant avanti endavant adelante ---
‘back’ *ad retro arriere indietro en(da)rere atras ---
‘away’ via --- via --- --- ---

Table 1: Directional/aspectual verb particles

Old French verb particles are well known for their function in specifying the direction of a change of location
event, as shown in (1).
(1) et qu'il alast hors et parlast à luy (Reg. crim. Chât., I, 382)
‘and that he would go out and speak to him’

But they can also play a role in the aspectual reading of an event. For example, they often emphasize an
existing telic interpretation, as shown in (2).
(2) li ont jus la tieste copée (Mousquet, Chronique, 22705)
‘they cut his head down/off’

Goals for the talk:

à To provide a basic description of the particles in order to show the kinds of meanings they can contribute to
an event and the remarkable range of verbs with which they occur in the MedFr corpus. I will also attend to the
kinds of constructions in which they are not attested.

à To offer a syntactic account of the particles. The challenge will be to account for the range of contexts in
which the particles can occur and yet to constrain the system so that it does not generate unattested

Rhaeto-Romance presents a rich system of particles as well. These are not discussed in the present study.
Modern French has no directional verb particles. French has a series of so-called intransitive prepositions, dessus, dessous,
dedans, and possibly dehors, but Zribi-Hertz (1984) and later Authier (2014) argue convincingly that these are prepositions
with null referential complements.
Going Romance, Leiden University November 28-29, 2019

constructions. Central to the analysis will be the typological question of where Medieval French particles fit in
Talmy’s two-way typology; Talmy 1985, 1991, 2000.

à To reflect on the implications of these particles for how we understand contemporary Romance particles and
how we understand what verb-framedness and satellite-framedness are.

Road map

1. Introduction
2. Medieval French verb particles: a description
3. Medieval French verb particles: a syntactic account
4. Conclusion, implications, questions

2. Medieval French verb particles: a description

This description presents the particles in light of their semantic contribution and the range of verbs with which
they are attested.3 Many of the examples demonstrate possibilities that have scarcely been touched on in the

2.1 Directional interpretation

MedFr particles are most frequently found in combination with a verb whose denotation entails a change of
location or with a verb that is often used in contexts where the activity can incite or cause something to change
location. In all cases, the particle specifies or emphasizes the result state of the entity undergoing the change of
(3) Orientation with change of location verbs
a. et mesire Gauvains vient avant et li deslace son hiaume. (Artu, p. 106; Burnett et al. 2010)
‘and lord Gauvains comes forward and takes off his helmet.’
b. Atant vint d'une chanbre fors la pucele (Chevalier au Lion, 94d.3949)
‘Then the girl came out from a bedroom’
(4) Pleonastic with directional change of location verbs
a. Del munt u furent sunt aval avalé (Guillaume, 25)
‘From the mountain where they were, they went down’
b. par les degrez descendent jus (Marie de France, Lais (Lanval), 96)
‘they go down via the stairs’
c. Sa lance lessa jus cheoir (Brut, partie Arthurienne, 124)
‘His lance (he) let fall down’
d. le font monter par les degrez amont an la sale (Conte du Graal, 367e. 1757)
‘they had him go up the stairs into the room’
e. il ne puissent monter sus. (Roman de Thèbes, 104)
‘they are unable to go up.’
f. Sus le voldrent faire lever en la roche (VieSGrég1, ms. A2, 2411)
‘They want to have him lifted up onto the rock’
h. et pour chou retournoit ariere en Babilone (Beaudouin de Sebourc, 133)
‘and for this he returned back to Babylon’

The examples are either taken from the literature or from a current study interrogating Textes du Français Ancien (TFA :
1100-1450), a subset of the texts found in Frantext Moyen Français (FMF : 1300-1549), and ARTFL-FRANTEXT (1500-
1699), yielding a total of 216,510,377 words.

Going Romance, Leiden University November 28-29, 2019

(5) Translative motion verbs (sometimes only implied) with strong manner component
a. Uns messages arriere cort (Chrétien de Troyes, Charette, 186)
‘A messenger runs back’
b. li aigue en coula jus (Aiol, 89)
‘the water flowed down (from it)’
c. tantost les gettent et vomissent hors, (Purification, p.59)
‘immediately they hurl and vomit them out’
d. Fort se teneit a la pere Que nel rosast le unde arriere (St. Brandan, v.1125)
‘He held fast to the rock so that the waves wouldn’t wash him over’
e. il escouoit la lance jus (Olivier de La Marche, Mémoires II, 132)
‘he shook the lance down’
f. Devant ses piés le fait jus craventer! (Moniage Rainouart I, 259)
‘He had him crushed/knocked down before his feet!’
g. il le vonche hors et le giete de son cors (Queste del Saint Graal, 163)
‘he summons it out and throws it from his body’
h. Cil apelet Brandan avant (St. Brandan, v.1484)
‘The aforementioned called Brandan forward’
i. Et de mon cuer hors les bany (Miracle de saint Panthaleon, 319)
‘And from my heart banished them out’

2.2 Aspectual interpretation

2.2.1 Completive
Similar to the examples involving verbs of change of location, MedFr particles often occur with causative
change of state verbs to emphasize the result state of the internal argument, such as terdre ‘wipe’ and vider
‘empty’ in (6a-b), but we also find such particles with other transitive verbs having a telic reading such as
paiier ‘pay’ and boire ‘drink’, in (6c-d), and unergatives such as pleurer ‘cry’ whose object gives a creation
(6) a. Le sanc jus de ses plaies tert (Charette, 89)
‘He wipes the blood off of his wounds’
b. Fait hors vuidier Toute sa chambre (Pisan, Fortune t.3, 41)
‘(He) has his whole room emptied out’
c. Enz vit une chandoille esprise Par l'uis qui est overz arriere (2ième cont. de Perceval, ep.34, p.494)
‘Inside he saw a torch through the door which is opened back’
c. Et ne furent pas adont tout hors paiiet (Froissart, Chroniques, 153)
‘and they were not then all paid out.’
d. si le but toute hors. (Beaudouin de Sebourc, 186)
‘and (she) drank it right up.’
e. quant lermes de pitié Vit que ploroit hors par compassion (Miracle de Théodore, 132)
‘when he saw that he was crying out tears of pity out of compassion’
2.2.2 Restitutive
In addition to its spatial meaning, the particle arriere ‘back’ can contribute a restitutive interpretation to an
event, whereby an entity regains a former state. We find this kind of interpretation with unaccusative

Going Romance, Leiden University November 28-29, 2019

directional verbs (7a), change of state (7b), activity verbs (7c), and in predicational contexts: (7d-e) occur with
the copula, for example.
(7) a. ou revient arriere hors pour scouter et regarder et soy ressuier. (Phébus, Chasse, 161)
‘or (the deer) comes back out in order to listen and watch and to dry off.’
b. et puis leissié refroidir et arriere bouillir (Phébus, Livre de chasse, 117)
‘then let (it) cool and then come to a boil again’
c. arriere les voit consillier (Belinagr. 95; Burnett et al. 2005)
‘he sees them back giving counsel’
d. Vus fuissiez mielz ariere es trés (Wace, Roman de Brut, 614)
‘You would be better back at the tents’
e. Et serons chi arrière dedens un mois (Froissart, Chron. L., III, 99)
‘and we will be back here within a month’

2.2.3 Continuative
In addition to its spatial meaning, the particle avant ‘forward/forth’ can contribute a continuative interpretation
to an event, whereby an activity or state continues beyond a presupposed or expected endpoint. We find this
interpretation with unergatives (8a-b), and even subject experiencer verbs (8c).

(8) a. ne jeo n’en sai avant cunter (Marie de France, Lais (Lanval), 112)
‘nor am I able to go on telling about it’
b. si garda avant devant lui (Graal, p.372c.2989)
‘and (he) looked on ahead of himself’
c. Jo ne la dei amer avant, ne haïr ne la dei par tant; (Thomas, Tristan, 25)
‘I mustn’t continue to love her, but I should not hate her for as much;’

2.2.4 Inchoative
Avant can also contribute an inchoative interpretation whereby the beginning of the event is emphasized.
(9) a. Un sage home i out ki parla, Ki la parole avant mustra: “[…]” (Roman de Brut, 329)
‘There was a wise man who spoke, who spoke forth/who began to speak: “…”’
b. il s'entrecommencent a regarder et semont li uns l'autre de parler avant. (Artu, 134)
‘they start to look at each other and encourage one another to begin speaking.’
c. Le chanp ont avant recovré (Wace, Roman de Brut; 126)
‘They started to retake the field’4

2.4 Unselected objects are unattested

These data reveal a particle system that permits a remarkable range of verbs, some of which have scarcely been
reported or discussed. Still, certain types of verb-particle combinations such as those involving unergative verbs
with unselected objects as illustrated in (10), permitted in the Germanic languages, for instance, have not been
attested anywhere to my knowledge.

(10) Unselected objects

a. You can run me up/over/down to the market.
b. He slept off his hangover.
c. Mary sang the night away.

The Bretons were invigorated/encouraged by the arrival of Yder, and so began to fight again.

Going Romance, Leiden University November 28-29, 2019

3. Medieval French verb particles: a syntactic account

3.1 Particles are phrasal

It has been well-established in the literature that MedFr directional particles are phrasal. They commonly undergo
movement operations such as scrambling and topicalization and are often modified; see (11a-c) and discussion
in Burnett & Tremblay (2009), Zaring (2011), among others.

(11) a. Jus descendirent (Moniage Guillaume 1, 29) (topicalised)

‘Down they went’
b. Sa lance lessa jus cheoir (Brut, 1155, 124) (scrambled)
‘His sword he let fall down’
c. il se traient adont un poi ariere (Trispr, p.39; Burnett et al. 2010) (modified)
‘they then moved a bit back’

3.2 Particles as elements of a resultative secondary predicate

A constellation of facts suggest that the particle is first merged within the secondary predicate. In (12), the vP
remnant, which includes the internal argument, the particle, and the PP expressing the Ground (source or goal),
raises to the CP domain. This is a common topicalization pattern which reflects the vP-internal order predicted
in an extended PP, where PathP dominates PlaceP. See 3.4 for a derivation.

(12) a. [Le sanc jus de ses plaies] tert (Charette, p.89)

‘He wipes the blood off of his wounds’
b. [Tes ennemis hors d'entour toy] Chace et deboute. (Miracle de Barlaam et Josaphat, 253)
‘He chases and pushes out your enemies from around you’

Auxiliary alternations provide further evidence for this. When a particle occurs with motion verbs in a
compound tense, the être (be) form of the auxiliary is selected, signalling unaccusativity (13a). This contrasts
with the same verb used to express an activity, for which the avoir form of the auxiliary is selected, signalling
unergative syntax. The systematic cooccurrence of certain particles and the auxiliary être in these contexts
suggests that the particle and the subject of the clause are in a predication relation within the VP. 5

(13) a. Se ele an ert alee fors, (Le Chevalier au Lion, p.92, f.3509)
‘And she went out (of it)’
b. Tant ont alé qu’il […] (Aiol, 8176)
‘So much did he go about that …’

In order to capture the Medieval French facts, I adopt an extended PP resultative secondary predicate following
Svenonius (2010), 6 also compatible with a somewhat sparser representation for secondary predicates as proposed
in Ramchand (2008). The basic ingredients are shown in (14).

See Dufresne & Dupuis (2010), Dupuis & Dufresne (2012), and Caudal, Burnett & Troberg (2016) for a discussion of the
auxiliary être as a marker of unaccusativity in Old French.
Building on Koopman (2000) and den Dikken (2010).

Going Romance, Leiden University November 28-29, 2019

(14) vP
v DirP
PP 3
Path PlaceP
DP 3

PlaceP contains the locative predicate which situates the Figure in relation to the Ground. Within this
framework, PlaceP is shorthand for series of functional heads that build, compositionally, the denotation of the
region in which the Figure is located. Path encodes transition semantics, while Direction restricts the set of
possible paths. In English, the morphemes to and from have been argued to lexicalize Path. Svenonius (2010)
proposes that directional particles like up, down, over, back, out, etc. merge as the specifiers of DirP.

3.3 Path and the satellite vs. verb-framed distinction

Since Talmy (1985) particles have been associated with satellite-framed languages. For example, Iacobini &
Masini (2007), among others, have proposed that Italian verb particles reveal satellite-framed constructions
within what should otherwise be a verb-framed grammar. In response to these kinds of claims, Mateu & Rigau
(2010) argue convincingly that Romance particles do not pattern like Germanic satellite-framed particles. They
point out that Romance particles are restricted to verbs that select an internal argument; unergative verbs are
thus ruled out and we therefore do not find unselected objects in Romance verb-particle constructions. For
instance, English and Italian allow verb-particles with manner of motion verbs, as shown in (15), but the crucial
contrast is demonstrated in (16), where English allows particles with unergative verbs, but Italian does not.

(15) a. John ran away

b. Gianni è corso via.

(16) a. John danced/drank/sang the night away.

b. *Gianni ha danzato/bevuto/cantato via tutta la notte.

Most formal approaches to this issue converge on the treatment of languages like English as building Path
semantics within a secondary predicate and thus independently of the meaning of the main verb. Such satellite-
framed languages allow, as a consequence, pure Manner roots to lexicalize the verbalizing head. Various
formalizations have been proposed. It has been characterized as a syntactic compounding in the work of Mateu
& Rigau (2002, 2010), McIntyre (2004), Zubizarreta & Oh (2007), Mateu (2008, 2011), and, in a similar way,
Folli & Harley (2016) treat it as a case of adjunction followed by M-merge.

In English, for instance, the presence of the Path elements to and from allow a manner root to lexicalize the
main verb, resulting in well-known goal-of-motion constructions such as “Mary limped to the store”. On the
other hand, in languages like Italian that ostensibly do not have such Path elements, a goal-of-motion reading is
unavailable when the main verb expresses pure manner of motion.

(17) a. John walked to the beach

b. *Gianni è camminato in spiaggia. (OK for locative reading)

In order to account for the additional fact that English commonly allows a goal interpretation with simple
locative prepositions such as “John limped under the bridge”, a null TO has been proposed (Svenonius 2010) so

Going Romance, Leiden University November 28-29, 2019

that transition semantics arises compositionally: “John limped TO under the bridge”. English further allows a
particle to combine with goal-of-motion constructions to produce “John limped down under the bridge”, shown
in (18a). In contrast, only verbal roots can lexicalize Path in Italian, formalized in (18b).7

(18) a. John limped down under the stairs b. Gianni è corso via.

vP vP
2 2
2 DirP 2 PathP
limp ØGO 2 cors- ØGO 2
down 2 <cors-> PlaceP
Ø PathP 2
2 Gianni 2
ØTO PlaceP via
John 2
under the stairs

3.4 Medieval French particles are verb-framed

What kind of elements lexicalize Path in Medieval French? Given the broad distribution of MedFr particles,
Troberg & Burnett (2017) propose a satellite-framed analysis, but this account can’t be right. It hinges on a null
Path element with TO semantics, and such an element would predict forms of resultative secondary predication
for which we simply have no evidence.
For example, there is no evidence that particles occur with unselected objects. Similar facts hold for complex
adjectival resultative constructions, also found in Medieval French (see Troberg & Burnett 2014); only the
weak variety are attested, which is to say that they, too, are restricted to verbs that have Path semantics (see
Washio 1997).
In light of these facts, I outline an analysis whereby all MedFr particles are derived within a verb-framed
grammar. This type of analysis requires that in each case the verbal element merges as an element of the
extended PP predicate. My account for Medieval French therefore aligns very closely with the work of Mateu
& Rigau (2010) and Mateu (2011) for Modern Romance and Folli & Harley (2016) for Modern Italian in that
Path always raises to v. This raises the question of how to formalize the difference between the Medieval
French particle system those in Modern Romance. I discuss this in section 4.

The topicalization patterns illustrated in (12) provide strong evidence that Medieval French particles merge as
specifiers of DirP, much like English particles. PlaceP introduces the Figure argument and the result: the goal
or source of movement, and the verb is merged first as a Path element. To illustrate, a derivation of example
(12a) is given below.
[CP [vP le sanc jus de ses plaies] tert [TP pro … [ vP [le sanc] tert [DirP jus Ø[PathP tert [PlaceP[le sanc] de ses plaies] ]]]]]]
Particles with directional and completive readings are easily derived within a verb-framed system. They merge
as specifiers of DirP, while PlaceP introduces the Figure argument and the result: the goal or source of movement,
shown in (19).

Various accounts exist for why Path is required to incorporate into v in Modern Romance. Mateu (2011)’s approach
reduces it to a general fact about the licensing of null v and the lexical items that instantiate Path0, shown here. Acedo-
Matellán (2010, 2016) describes it as a result of PF operations while Real-Puigdollers (2010) attributes it to the defectivity
of Path within a particular version of phase theory. Recently, Folli & Harley (2016) formalize it in terms of a head-raising
parameter. See Section 4 “Conclusions & Implications” where I briefly outline my analysis of Modern Romance, which
involves a bundled v-Path category.

Going Romance, Leiden University November 28-29, 2019

(19) a. Sus le… lever en la roche (see ex. 4b) b. Le sanc jus de ses plaies tert (see ex. 12a)

vP vP
2 2
2 DirP 2 DirP
2 ØGO 2 2 ØGO 2
lev- ØDir sus 2 terd- ØDir jus 2
<ØDir> PathP <ØDir> PathP
2 2
<lev-> PlaceP <terd-> PlaceP
2 2
le 2 le sang 2
en la roche de ses plaies
Some completive readings occur with verbs that lexicalize the Ground itself, denoting the result state of the
Figure. This kind of derivation has already been proposed in the literature for Romance, but what has not been
discussed is what lexicalizes Path. To account for the Medieval French facts, I propose that a null verbal element
lexicalizes this head. Crucially, this vocabulary item is not prepositional, as has been proposed for English. Again,
if this were the case, we would predict a cluster of other constructions that are unattested in MedFr. While null
TO of English (or Latin) would remain low and permit manner modification, the null verbal element of Medieval
French must raise to v along with a lower Ground element that encodes result. Manner modification is thus ruled

(20) a. arriere bouillir (see ex. 7b) b. si le but toute hors. (see ex. 6d)

vP vP
2 2
2 DirP 2 DirP
4 ØGO 2 4 ØGO 2
bouill- arriere 2 boi- hors 2
<ØDir> PathP <ØDir> PathP
2 2
<Ø- > PlaceP <Ø- > PlaceP
2 2
pro 2 toute pro 2
<Ø> <bouill-> <Ø> <boi->

The primary challenge for this analysis is to account for particles that occur with verbs that have no inherent or
pragmatically induced transition semantics, namely verbs denoting pure activities and states. The interpretation
of the particles is always aspectual in these contexts. I’ve repeated a few examples of this from above.

(21) a. semont li uns l'autre de parler avant. (Artu, 134; Burnett et al. 2010)
‘(they) encourage one another to begin speaking.’
b. Jo ne la dei amer avant, ne haïr ne la dei par tant; (Thomas, Tristan, 25)
‘I mustn’t continue to love her, but I should not hate her for as much;’

Overt verbal elements are prevented from merging at v since this would leave the element(s) in Path as bare roots (as
discussed in Embick 2010). An analysis involving a null verbal element could naturally align with serial verb
constructions where the Path element remains uninflected.

Going Romance, Leiden University November 28-29, 2019

c. arriere les voit consillier (Belinagr. 95; Burnett et al. 2005)
‘he sees them back giving counsel’
d. Vus fuissiez mielz ariere es trés (Wace, Roman de Brut, 614)
‘You would be better back in the tents’
The most natural assumption would be to treat these as satellite-framed constructions. In fact, Acedo-Matellán
& Mateu (2013), citing parler avant, take this pattern to be satellite-framed, a vestige of similar Latin
constructions. Troberg and Burnett (2017) also treat such uses as satellite-framed, that is, one in which the
verbs are merged above the secondary predicate and in which a null prepositional element with TO semantics
lexicalizes Path.
I nevertheless propose that these MedFr constructions can be accounted for within a grammar that is not
satellite-framed and that this is desirable in light of the general restriction on Manner modification in MedFr.
For activity verbs such as parler, consillier, aimer, etc., I propose that they are derived – coerced – in these
contexts as denominal verbs within a construction that denotes transition as shown in (22).

(22) vP
v DirP
Ø 3
avant 3
arriere Dir0 PathP
Ø 3
Ø- PlaceP
DP 3
ØWITH parl- (parole)
am- (amour)
à These denominals are derived as locatum verbs in the sense of Hale and Keyser (2002), so that aimer, for
example–normally a subject experiencer verb–, would be interpreted as an activity that causes someone to be
with love. Evidence that this kind of interpretation arises in other contexts comes with the use of denominal verbs
with aspectual prefixes, which, following Troberg & Burnett (2017), lexicalize Path. While deparler and debaver
are agentive, enaimer would have to be derived as an unaccusative.

(23) enamer ‘fall in love’; deparler ‘speak badly of someone’; debaver ‘cover in drool’

3 PathP
4 ØGO 3
deparl- <de- > PlaceP
debav- <en- > 3
enam- FIGURE 3
<parl-> (parole)
<am-> (amour)

Going Romance, Leiden University November 28-29, 2019

Turning finally to restitutive arriere occurring in predicational contexts, I propose that the particle in this context
lexicalizes a deictic head, Deix. Following Svenonius (2010), particles that modify locative PPs introduce
viewpoint relative to some logophoric centre. This is common in MedFr; the example in (24) shows jus
expressing spatial deixis. Jus presupposes that the region in which the handsome knight is situated is lower than
the window out of which the woman is looking.

(24) Por la froidor son cief bota en la fenestre. .I. chevalier vit jus aval sor le gravier qui mout ert
acesmés et biax
‘For the coolness, (she) placed her head in the window and saw a knight down below on the gravel who
was very elegantly dressed and handsome’
(Vengeance Raguidel, p. 4)

Restitutive arriere would work the same way. Like spatial arrière, which presupposes that the Figure was
previously located in the region of the Ground (A messenger ran back - see 5a), temporal arrière presupposes
that Figure has returned to a state it once occupied in the past. It is thus deictic in that the point of view of the
present event is considered relative to a previous event.

(25) Vus fuissiez mielz ariere es trés (see 21d) 9

fuissiez pP
vous 3
p DeixP
en 3
arriere 3
Deix DP
Ø les tentes
Note that derivations involving DeixP do not involve Path and should therefore be independent of the satellite-
vs. verb-framed distinction. They should, in principle, be possible in Modern French.10

4. Conclusion, Implications, Questions

4.1 Conclusions
àI have synthesized old and new evidence from Medieval French for a description of the distribution of verb
particles, demonstrating how they contribute directional and various aspectual meanings to events with a
surprising range of verbs. I have proposed that they merge as specifiers of DirP.
àContrary to previous accounts of these particles, I have proposed that the range of uses can be derived within
a verb-framed grammar, which means that the verbal root always lexicalizes Path.
àDerivations in which the verbal root begins as a Ground element (change of state, denominal) require a null
element to lexicalize Path. I have suggested that a null verbal element what characterizes Medieval French and
distinguishes it from a language like English, which instead has a null TO that doesn’t require raising to v.
àThe only uses of particles that are not dependent on a Path projection are those that have a deictic meaning.
These would be elements of Place P.

The conjoined P-D complex (en les à es) suggests that arriere moves leftward, possibly to a low aspectual field.
A lexical explanation for this lacuna is entirely plausible: arrière is no longer of the category P; sus, jus, aval, amont,
also used deictically, fell out of use as particles–and as lexical items–by the end of the 16th century.

Going Romance, Leiden University November 28-29, 2019

4.2 Implications
4.2.1 Medieval French particles in relation to Contemporary Romance particle systems
The distribution of MedFr particles differs sharply from most contemporary varieties: the range of verbs and
meanings is much broader. If Medieval French particles are reflexes of a verb-framed grammar, the difference
between them and other contemporary systems such as Northern Italian, Catalan, Spanish, etc. must be properly
formalized. I have proposed elsewhere (Troberg 2018, 2019) that v and Path are a bundled category in
contemporary Romance concluding, as does Mateu & Rigau (2010, ex. 23), that contemporary Romance particles
are merged within PlaceP. Such an analysis is compatible with more fine-grain accounts of the very productive
Northern Italian particles according to which they raise from a VP-internal position to positions within a lower
aspectual field (Tortora 2002; Quaglia & Trotzky 2017).11

(26) a. Gianni è corso via. (Mod. Italian)

b. Jean descend à la cave. (Mod. French)
v-Path PlaceP
v-Path PlaceP
cors- 2
descend- 2
Gianni 2
Jean 2
à la cave

A bundled v-Path analysis for Modern Romance avoids the need to propose obligatory Res-to-v head
movement in contemporary Romance (Folli & Harley 2016) or, put another way, to require Path to incorporate
into v as a result of a manner conflation parameter disallowing the syntactic compounding of a manner root
with a phonologically null verbalizing head with the meaning of GO (see Mateu & Rigau 2002, 2010; McIntyre
2004; Zubizarreta & Oh 2007; Mateu 2008).

Even more serious, however, such accounts of a v-framed grammar in Modern Romance over-generate:
o nothing rules out the merger of particles within DirP
o there is no account for why weak adjectival resultative constructions are no longer permitted
o there is no explanation as to why the range of verbs permitted in goal-of-motion constructions is so
The bundled v-Path proposal that I have advanced can account for all of these phenomena. The bundling itself
rules out DirP, while the lexical effects of the bundling had profound effects on the way verbal roots are now
specified. Elements that lexicalize events involving transition are currently specified as v-Path (aller,
descendre, courir, laver, etc.), whereas many examples from Medieval French show much more verbal
elasticity. Any verb having some implication of transition appears to have been able to lexicalize Path.
Systems like that of contemporary Romance could thus be characterized as strong verb-framed, or even radically
verb-framed because v and Path are one head. On the other hand, systems like that of Medieval French, which
are more permissive and, on the surface, quite similar to satellite-framed systems, might be called weak verb
framed. This gradience of discrete verb-framed grammars would be similar to Acedo-Matellán (2010, 2016)’s
classification of satellite-framed systems, where languages like Latin and Slavic are weak, while Germanic
languages are strong. Table 2 proposes Talmy’s typology anchored on each side by the strong-framed grammars,
and allowing weak versions of each as intermediate systems. The major points of variation are formalized in
terms of the lexical items that instantiate Path. In strong s-framed languages, Path is a free morpheme while in
strong v-framed languages, Path and v are a single category.

More investigation is required to determine the cooccurrence patterns of Northern Italian verb particles and what seem
to be Ground PPs and if some N. Italian verb particles are base-generated as aspectual elements (magnàr fò, pagar fora,

Going Romance, Leiden University November 28-29, 2019


English Latin Medieval Modern

French/Romance French/Romance
Path0 to, from prefixes: a-, de-, e-, etc. v-Path0
ØTO (raising from Place0) Ø-TO (raising from Place0) Ø (raising from Ground0)
unselected ✓ ✓ ✗ ✗
directional ✓ ✓ (limited) ✓ ✓ (limited)
✗ (Fr)
adjectival ✓ ✗ ✓ ✗
goal-of-motion ✓ ✓ ✓ ✓ (limited)
Table 2: Intermediate (weak) stages of the satellite- vs. verb-framed typology
4.2.2 Romance particles in a diachronic perspective
In considering the evolution of the particle systems through Romance, it is crucial to be able to account for the
complete loss of particles in French. The present analysis does so in a clear way: particles that merge within DirP
are ruled out categorically once v and Path become a bundled category. This is supported by quantitative studies
that show a unified loss of directional/aspectual particles in French (see Burnett et al. 2005, 2010, among others).
Frequencies begin to decline in the 13th century, and particles are obsolete by the 16th.

Romance particles hold particular interest because their robust presence in early texts and subsequent contraction
correlates with the well-known typological shift whereby satellite-framed Latin developed into verb-framed
Romance (Talmy 2000). Properly understanding their properties can thus shed light on the intermediate grammars
that arose as the typological shift played out. There has been debate about the exact properties, but the prevailing
assumption has been that Old Romance particles are qualitatively identical to those of contemporary Romance.
This assumption has made it difficult to develop a persuasive formal account of the role of particles in the
typological shift.

As a pan-Romance phenomenon, verb-particle constructions have received a uniform treatment from a

typological standpoint. This may seem natural given that the lexical stock of particles, their distribution, and the
range of verbs that they occur with are strikingly similar through both time and space. For example, verb particles,
particularly productive in Northern Italian varieties, appear to share basic syntactic properties with Old French.
They both behave as phrases (as opposed to heads) demonstrated through modification (27) and lack of strict
adjacency with the verb (28).

(27) a. Il cameriere porta [ancora più avanti] il carrello. (Italian)

‘The waiter pushes the trolley even further forward.’ (Quaglia & Trotzke 2017, fn.3)

b. li sans en chiet a le terre [tout jus] (Old French)

‘the blood fell right down to the ground’ (Beaudoin de Sebourc 18587)

(28) a. Tira {sempre/ ancora/ già} fuori la lingua. (Italian)

‘He {always/again/already} sticks out his tongue.’ (Cordin 2011, fn 20)

b. il se traient adont un poi ariere (Old French)

‘they then moved a bit back’ (Trispr, p.39; Burnett et al. 2010)

Going Romance, Leiden University November 28-29, 2019

As I mention in section 3.3, Mateu and Rigau (2010) point out that the range of verbs that are permitted in Italian
is limited to those that encode Path–true of Old French as well. Particles are ungrammatical with pure manner
verbs such as dance.

(29) a. John danced/twirled out.

b. * Gianni è ballato fuori.
c. * Jean a dansé hors. (unattested)

Still, Northern Italian varieties have a rich and productive system of particles that fall into several types as
described in Cordin (2011) for Trentino and more generally in Quaglia (2016): pleonastic (uscire fuori ‘to exit
out’), locative (andare fuori ‘to go out’), idiomatic (fare fuori ‘to kill’), and aspectual (lavare via ‘to wash away’).
As I have shown, these types of particle use are also present in Medieval French. However, the use of aspectual
avant and arriere with denominal verbs (see ex. 22) and the copula (ex. 25) sharply distinguish Medieval French
from Northern Italian varieties.

The restrictions vis-à-vis the verb for Italian verbi sintagmatici point to an analysis in line with Mateu & Rigau
(2010) that contemporary Romance verb particles are not Path denoting (as has been proposed for English, for
example), but rather are generated within a verb-framed grammar. Qualglia and Trotzke (2017) account for this
and the constellation of distributional facts by proposing that particles in Italian raise to the aspectual field above
VP, sketched below for vola sempre fuori ‘(it) always flies out’.12

(30) [TP [pro] vola j [AspP sempre Ø [PartP fuorii Ø [vP t j t i]]]]

This analysis cannot possibly hold, however, for Old French. The grammar that generates the Italian verbi
sintagmatici is in principle entirely compatible with Modern French, and yet the verb-particle system was
abruptly lost from French between the 14th and 16th centuries. 13 This fact alone justifies a different formal
treatment such as the one I argue for here.

The implication of these facts is that the grammar of verb particles in Proto Romance would have been closer to
Medieval French than to Modern Italian varieties. While the verb-particle system was lost from French it
appears to have undergone a straightforward reanalysis–a rebirth of sorts–in Italian. The particles as specifiers
of DirP–and which often raised to higher positions in the clause–were reanalysed within a verb-framed
grammar as prepositional elements that raise out of the VP to the aspectual field. By way of illustration, the
example below frames a comparison of the particle ‘out’ in Old French and Modern Italian in terms of
diachronic reanalysis: vole fors à vola fuori.

(31) [TP vol- i [vP t i [DirP fors Ø [PathP t i …]]]] à [TP vol- j [PartP fuorii Ø [vP t j t i]]]

If the present account is on the right track, then the rich and productive verbi sintagmatici of Northern Italian
varieties is not a unique innovation among the Romance varieties, as suggested in Mateu & Rigau (2010) and
claimed in Iacobini (2015). Instead, these varieties seamlessly evolved from a proto-system of a very similar
richness but which is closer in approximation to a satellite-framed grammar, what I have dubbed a weak verb-
framed grammar.

The authors claim that the particle saturates the goal argument of a verb, yet a particle and PP goal seem to easily co-
occur. What would the implications be if the particle is in some cases first merged in the aspectual field? This would be a
common grammaticalization path from a diachronic point of view.
Existing sociolinguistic (Iacobini & Masini 2006) and functional (Foulet 1958) explanations fail to account for quantitative
studies that track the loss of particles in French (Burnett et al. 2005, 2010, inter alia).

Going Romance, Leiden University November 28-29, 2019

5. Selected References

ACEDO MATELLÁN, V. 2016. The Morphosyntax of Transitions. OUP.

ACEDO MATELLÁN, V. AND J. MATEU. 2013. Satellite-framed Latin vs. verb-framed Romance: A syntactic
approach’, Probus 25: 227-265.
BURNETT, H, G. GAUTHIER, & M. TREMBLAY. 2010. La perte des particules arrière et avant en français
medieval: étude quantitative. In F. Neveu et al. (eds.) Congrès Mondial de Linguistique Française 2010.
BURNETT, H., K. PETRIK & M. TREMBLAY. 2005. La grammaire des particules en ancien français: sémantisme,
distribution et perte de productivité. Proceedings of the 2005 annual conference of the CLA.
BURNETT, H. & M. TREMBLAY. 2012a Change in the encoding of Direction in the history of French: A
quantitative approach to argument structure change. In Nynke de Haas & Ans van Kemenade (eds.)
Historical Linguistics 2009. Amsterdam: John Benjamins, 333-353
BURNETT, H. & M. TREMBLAY. 2012b Directionalité et aspect en ancien français: l’apport du système
prépositionnel. in Mario Barra-Jover et al. (eds). Études de linguistique gallo-romane. Presses de
l’Université de Vincennes, 217-232.
BURNETT, H. & M. TREMBLAY. 2009. Variable-behaviour Ps and the location of PATH in Old French, in E.
Aboh et al. (ed), Romance Languages and Linguistic Theory 2007. Benjamins, 25-50.
CAUDAL, P, H. BURNETT & M. TROBERG. 2016. Les facteurs de choix de l’auxiliaire en ancien français: étude
quantitative. In S. Prévost and B. Fagard, Le français en diachronie: dépendances syntaxiques,
morphosyntaxe verbale, grammaticalisation, 218-237. Bern: Peter Lang.
CORDIN, P. 2011. Le costruzioni verbo-locativo in area romanza. Dallo spazio all’ aspetto. Berlin/New
York: Mouton de Gruyter.
DEN DIKKEN, M. 2010. ‘On the functional structure of locative and directional PPs’, in G. Cinque and L. Rizzi
(eds), Mapping spatial PPs: The cartography of syntactic structures Vol. 6. New York: OUP, 74-126.
DUFRESNE, M, F. DUPUIS, & C.-M. LONGTIN. 2001. Un changement dans la diachronie du français: La perte de
la préfixation aspectuelle en a-. Revue québécoise de linguistique 29(2): 33-54.
DUFRESNE, M, F. DUPUIS, & M. TREMBLAY. 2000. The Role of features in the historical change. In S. Dworkin
and D. Wanner (eds.) New Approaches to Old Problems: Issues in Romance historical linguistics.
Amsterdam: John Benjamins, 129-149.
DUFRESNE, M., F. DUPUIS, & TREMBLAY, M. 2003. Preverbs and particles in Old French. In Booij, G. and A.
van Kemenade (eds), Yearbook of Morphology 2003, Kluwer, 30–60.
DUFRESNE, M. & F. DUPUIS. 2010. Les structures inaccusatives en français médiéval. In F. Neveu, V. Muni
Toke, T. Klingler, J. Durand, L. Mondada & S. Prévost (eds.), Actes du 2ème Congrès Mondial de
Linguistique Française (CMLF 2010), 159–176.
DUPUIS, F. & M. DUFRESNE. 2012. La télicité : un aspect des constructions inaccusatives en français médiéval.
Constructions en changement: hommage a Paul Hirschbühler, 55–106. Québec: Presses de l’Université
FOLLI, R. & H. HARLEY. 2016. Against deficiency-based typologies. Manner-alternation parameters in Italian
and English. In E. Carrilho, A. Fiéis, M. Lobo, & S. Perei (Eds.), Romance Languages and Linguistic Theory
10. Selected papers from 'Going Romance' 28, 103-120. Amsterdam/Philadelphia: John Benjamins.
IACOBINI, C. 2015. Particle verbs in Romance. In P. Müller et al. Volume 3 Word-Formation. An International
Handbook of the Languages of Europe. Berlin, Boston: De Gruyter Mouton, 627-659.
IACOBINI, C. & B. FAGARD. 2011. ‘A diachronic approach to variation and change in the typology of motion
expression. A case study: From Latin to Romance’, Cahiers de Faits de Langue 3: 151-172.
IACOBINI, C. & F. MASINI. 2006. ‘The emergence of verb-particle constructions in Italian: locative and actional
meanings’, Morphology 16: 155-188.
KOOPMAN, H. 2000. ‘Prepositions, postpositions, circumpositions, and particles: The structure of Dutch PPs’,
in H. Koopman (ed), The Syntax and specifiers and heads: Collected essays of Hilda J. Koopman. London:
Routledge, 204-260.
MATEU, J. 2011. Conflation and incorporation processes in resultative constructions, in V. Demonte & L.
McNally (eds), Telicity, Change, and State: A Cross-Categorial View of Event Structure, 252-277. Oxford:
MATEU, J. & G. RIGAU. 2010. Verb-particle constructions in Romance: a lexical-syntactic

Going Romance, Leiden University November 28-29, 2019

account. Probus 22(2). 241-269.
QUAGLIA, S. 2016. Particle Verbs in Italian. PhD dissertation, University of Konstanz (KOPS).
Available at:
MCINTYRE, A. 2004. Event paths, conflation, argument structure, and VP shells. Linguistics 42(3): 523-571.
QUAGLIA, S. & A. TROTZKY. 2017. Italian verb particles and clausal positions, In Bar-Asher Siegal, Elitzur A.
(ed), Proceedings of IATL 2015. Cambridge, MA : MITWPL 82: 67-82.
REAL PUIGDOLLERS, C. 2010. A microparametric approach on goal of motion constructions: properties of
adpositional systems in Romance and Germanic. Catalan Journal of Linguistics 9: 125-150.
SVENONIUS, P. 2010. Spatial P in English, in G. Cinque and L. Rizzi (eds), Mapping spatial PPs: The
cartography of syntactic structures Vol. 6. New York: OUP, 127-160.
SVENONIUS, P. 2012. ‘Aktionsart and aspect in verb-particle constructions’, Workshop on Verbal Periphrasis,
UMR 7023, CNRS/Paris 8.
TALMY, L. 2000. Typology and Process in Concept Structuring. Cambridge (Massachusetts): MIT Press.
TROBERG, M. & H. BURNETT. 2014. Le prédicat résultatif adjectival en français médiéval. Linguisticae
Investigationes 37(1) :152-176.
TROBERG, M. & H. BURNETT. 2017. From Latin to Modern French: A punctuated shift. In E. Mathieu and R.
Truswell From Micro-change to Macro-change, OUP.
TROBERG, M. 2018. ‘Diachronic reanalysis and the satellite- vs. verb-framed distinction’, Talk given at the 48th
LSRL, York University.
WASHIO, R. 1997. Resultatives, compositionality and language variation. Journal of East Asian Linguistics 6:
ZUBIZARRETA, M L. & E. OH. 2007. On the syntactic composition of manner and motion. Cambridge, MA:
MIT Press.