Truncation vs Reduction in development.

Adriana Belletti – University of Siena

June 2021

1. Introduction: The issue.

In early stages of their linguistic development, young children are known to undergo a period

in which they may make use of somewhat incomplete clausal forms to express contents which,

in the adult grammar, would be expressed through a full clause, with all relevant inflectional

features and functional positions present and properly realized, according to the general

morphosyntactic properties of the languages involved.

A much-studied stage of this type is the Root Infinitive/RI stage (up to around 3 years of age),

discussed in Rizzi 1993/94 and much related work (e.g. Wexler’s 1994 Optional Infinitives).

Following Rizzi’s original proposal and its later developments, the RI stage may be

characterized as resulting from a truncation operation of the syntactic tree of the clause,

available to the developing child. Such truncation option should be favoured to some extent in

early stages as it allows for an overall less complex syntactic computation of the clause, less

costly for the child’s developing grammar (Rizzi 2006): a smaller portion of the complete

clause structure is computed by truncating it at the CP level.1 In accord with continuity

assumptions, comparable truncated structures are available in particular constructions in the

adult grammar of different languages (as in e.g. German Topic-drop constructions). Thus,

truncation constitutes a possible parametric option that young children (may) exploit during

their development.

It can also be the case that in early stages, the syntactic tree be only partly accessible to the

developing child. Thus, only those syntactic constructions and computations that involve the

portion of the syntactic tree structure available to the child at that given point would be possible,

1
In a cartographic articulated CP, truncation may take place at different levels. The point is addressed below.

1
whereas those that involve bigger portions of the tree, would be impossible, hence absent. This

is the ‘growing tree’ logic, recently presented in Friedmann, Belletti and Rizzi (2021). As

discussed there, ‘growing tree’ and ‘truncation’ define different stages, as the latter process

optionally applies on an already fully-grown syntactic tree, possibly to allow for a more

parsimonious computation for the developing system, in the terms described. Truncation is

thus interpreted as a reduction operation, taking place on the completely grown syntactic tree.

Further reduced constructions of various types are also present in different languages (beside

e.g. the aforementioned Topic-drop construction). Adult grammars handle such constructions

in different ways, in compliance with general properties of the specific languages. Among the

reduced structures present in several languages, Reduced Relative Clauses/RRCs are the case

which will be the focus of the discussion here, with special reference to Italian. Reduced

relative clauses are (past) participial clauses available in various languages, including Italian.

In (1) two examples illustrate the construction in Italian and in the English equivalent

translation:

(1) a Il libro letto <__> dallo studente

The book read by the student

b Il bambino abbracciato <__> dalla mamma

The child hugged by the mother

The position corresponding to the thematic interpretation of the noun phrase head of the

reduced relative clause is indicated with <__>, highlighting the fact that the head of the relative

corresponds to the internal argument of the verb, expressed in the past participle form in the

reduced relative. Such past participle corresponds to a passive past participle as witnessed by

the (possible) presence of a by-phrase expressing the external argument of the verb.

The term ‘reduced’ refers here to the intuitive analysis that any native speaker would informally

and naturally provide, according to which the sentences in (1) are the reduced form of complete

relative clauses such as those in (2), in which the deleted part is not pronounced:

2
(2) a Il libro che è (stato)/viene letto <__> dallo studente
The book that is (has been) read by the student

b Il bambino che è (stato)/viene abbracciato <__> dalla mamma

The child that is (has been)/comes hugged by the mother

Both structures mentioned, RI and RRCs, are instances of clausal constructions that use a

limited portion of the clause structure, be this the result of a truncation operation on the already

grown tree2, or else the output of a different type of reduction operation as the one informally

pictured in (2). In what follows some items of the empirical background that led to truncation-

type proposals in the late 19-nineties will be briefly reviewed in the aim of proposing a

comparison with the different reduction operation occurring in reduced relative clauses of the

type in (1). The comparison is inspired by developmental results over a long period of time

concerning on the one side the truncation stage, leading to use of Root-Infinitives in early

grammars, as has been documented in several languages, and, on the other side, more recent

experimental and spontaneous production results on the production of object A’-dependencies,

such as object relative clauses and RRCs in particular. It appears that the incomplete clausal

structures are not treated in the same way in the two cases during development: truncated

structures belong to an early stage, whereas RRCs belong to a later stage. The early vs late

appearance of the different expressions both using limited portions of the clause structure

strongly indicates that something more than just the ‘small size’ of the clause must be at stake

and play a crucial role. What the relevant factor(s) may be is the focus of the following pages.

Section 2 provides the essential background defining the truncation stage and its interaction

with growing tree; in section 3 the further theoretically relevant background on the locality

account in terms of featural Relativized Minimality/fRM in object A’-dependencies is

2
Or possibly in very early stages, it could be the consequence of the incomplete growth of the syntactic tree, as
mentioned.

3
illustrated. In Section 4 the main empirical observation from developmental results is described

concerning the late appearance of RRCs; a proposal is outlined of the main features of the

assumed analysis of RRCs from which their fundamental different status w.r.t. truncated

structures clearly emerges. This difference may be at the core of the reason why the two

incomplete structures have a different developmental time line. Some related consequence of

the proposal are sketched out in section 5. Some reflections will conclude this work focusing

on the (ir)relevance of complexity measures that limit themselves to pre-theoretical

quantitative considerations on the number of items involved in syntactic computations.

2. One starting observation and some background considerations: the truncation

stage in language development (and the growth of the syntactic tree)

As mentioned in the introductory section 1, young children typically undergo a stage in their

development known as the Root Infinitive stage in which they produce root declarative

descriptive clauses with no specification of tense nor expression of features of finite agreement

inflection (i.e. including the person feature). (3) below is an example from French where the

verb appears at the infinitive form:

(3) voir l'auto papa

see the car daddy (Rizzi 1993/94: 374, ex.1)

This stage can be described as one in which a somewhat impoverished sentence structure is

consistently properly mastered by young children, in contrast with one expressing a full-

fledged functional structure. Cross-linguistic differences have been detected in the way the

stage can manifest itself in different languages (e.g. for Italian through use of past participles

or other not richly inflected forms such as the imperative, Antinucci & Miller 1976; Salustri &

Hyams 2003, 2006). In all cases the impoverished expression of phi-features characteristically

associated with finiteness (i.e. in particular person) is a most salient property.

4
Following Rizzi’s original proposal and its earlier and later developments, the RI stage may be

characterized as resulting from a truncation operation, available to the developing child (a

similar operation of pruning has been assumed to occur in the domain of aphasia, as proposed

by Friedmann 2002). Such truncation option is assumed to be to some extent favoured in early

stages as it allows for an overall less complex syntactic computation, less costly for the child’s

developing grammar (Rizzi 2006): A smaller portion of the complete clause structure is

computed by truncating it. Truncation may occur at the CP level, as illustrated in the sketchy

diagram in (4); in (4) the possibility that truncation may occur at a lower level e.g. vP, is also

illustrated:

(4)
CP
1
1
C TP
1
1
T vP
5

As Rizzi’s term RI emphasizes, the impoverished structure is solely present in the children’s

productions of this stage in Root declarative contexts. Specifically, it is never found in

embedded contexts nor in contexts in which a richer structure and its components are present,

including those of the CP layer as in e.g. wh-questions, as illustrated in (5):

(5) Oú elle va maman?

Where she goes mummy (Crisma 1992, quoted in Rizzi 1993/94)

In (5) and in all embedded clauses clearly containing the CP level, which are produced by

children during the RI stage, the verb always appears fully inflected in its finite form; the

subject is also consistently overtly realized in the non-null subject French. This RI option

correlates with the concomitant Child Null Subject stage, in which the specifier of the root, i.e.

Spec/TP in (4), can remain unpronounced. The Child Null Subject stage in turn is known to

5
also be a root phenomenon, as confirmed by the overt realization of the subject in (5).3

Crucially, productions like (5) in the RI infinitive stage indicate that children during this stage

do have access to the complete functional structure of the clause. They may just not use it in

root declarative contexts. This would allow them to perform a less complex computation

including a smaller structural portion of the clause, by truncating it at the root.

In a cartographic articulated CP as in the map in (6) from Rizzi and Bocci (2017), truncation

may in principle obtain at different levels:

(6) Force > Int > Top > Q/Foc > Mod > Fin > IP

Topic drop constructions mentioned earlier, of the type found in German (Cardinaletti 1990,

1990a for a first discussion), can be analysed (Rizzi 2006) as one such instance of truncation

available in adult grammars, in which the operation occurs at the TopP level. Spec/TopP thus

becomes the specifier of the root and as such it can remain unpronounced (dropped, as in the

descriptive name of the construction).4

Thus, in compliance with continuity assumptions, truncation constitutes a possible parametric

option that young children (may) exploit during an early stage of their development, which,

according to the proposal just reviewed, may be favoured by complexity considerations.

2.1 An aside on the relation between growing tree and truncation

According to the growing tree logic recently presented in Friedmann, Belletti and Rizzi (2021),

the syntactic tree of the clause grows incrementally during development. The main

consequence of this stage-by-stage growth is that the clausal structure can be only partly

3
This is the crucial property differentiating the null subject of this stage from the null subject of null subject
languages, a consequence of the positive setting of the null subject parameter (Rizzi 1993/94 reconsidering the
first grammatical approach to this important developmental stage presented in Hyams’ 1986 first pioneering and
influential work).
4
An instance of CP projection up to the FocP layer may be instantiated by the case of the CP of cleft sentences
(Belletti 2008, 2012), which will be taken up again in section 5.1 See footnote 15 also relevant to the point made
in the text.

6
accessible to the developing child in earlier stages. The growing tree approach is a recent

revisitation of a family of proposals formulated during the 19-nineties5, grounded on the results

from work in syntactic cartography, which provides refined maps and detailed levels of the

functional structure of the clause. This drives detailed investigations, which allow for subtle

discoveries that may otherwise remain unnoticed through less fine-grained lenses. As the

analysis of the acquisition of the Hebrew left periphery has revealed, the acquisition of the

clausal structure progresses in three fundamental stages, with different constructions involving

the relevant portion of structure appearing at different pace. Each stage only allows for those

syntactic constructions and computations that involve the portion of the syntactic tree available

to the child at that point, and excluding those that involve bigger, i.e. higher, portions of the

tree, not available yet.

The growing tree logic is crucially implicational: If the tree has grown up to CP, all structures

belonging to the lower TP (and vP) level will also be available. In the cartographically analyzed

CP the implications are detailed: set of constructions realized in low positions of the clause

belong to a certain Stage1, and thus all appear during this stage, whereas constructions

involving higher positions of the left periphery belong to the later Stage2 and are not available

yet in Stage1, i.e. they do not appear in children’s productions. In contrast, if structures of Stage

2 are present, structures of Stage 1 are all also available and thus appear in children’s

productions. In a simplified representation as (4), the implicational chain implies that if the CP

level is available, the TP level must be available as well (and a fortiori vP as well). The precise

manifestation of the stages in the acquisition of the Hebrew left periphery are presented in

detail in Friedmann, Belletti and Rizzi (2021). Here, it should just be kept in mind that the stage

at which relative clauses are acquired is the last: the stage in which the highest positions of the

5
Dating from Radford (1990) on; see Friedmann, Belletti and Rizzi (2021) and references cited there in this
connection.

7
CP space are present. Given the map in (6), the left periphery is present up to the highest ForceP

level.

The implicational flavor of growing tree also characterizes the truncation stage.

Thus, similarly to what just noted in terms of the growing tree, truncation in an already grown

tree can occur at the CP level, hence the available structure will contain both the TP and the vP

layers; if it occurs at the TP level the available layer will be vP. Crucially, truncation cannot

only affect the TP level without also affecting the higher CP. As observed in the introduction,

work on truncation has determined that this reduction operation takes place on completely

grown syntactic trees. This is so as truncation appears to be an option that children may exploit

in Root clauses, in the same stage in which they master constructions involving full-fledged

clauses, with the expression of the complete functional structure up to the CP layer, such as

wh-questions as in the example in (5).6

3. Further background and other findings from language development.

3.1. Further background: fRM, intervention and feature inclusion

A rich bulk of experimental findings coming from a mass of work on the acquisition of A’-

dependencies (e.g. Relative clauses), that appeared over the last ten years or so, has reached a

number of robust conclusions schematically summarized in (7):

(7) i. children can master complex structures such as relative clauses from relatively early
on (3;6-4)
ii. Subject Relatives/SRs are mastered better and earlier than Object Relatives/ORs
(confirming results from classical literature).
iii. not all ORs are hard for children (a finding also holding in types of impairment).

As for (7)iii, , the main finding has been that the only hard ORs, in both development and

impairment in fact (Friedmann, Yachini, Szterman 2015), are those showing a situation of

6
By truncating the whole CP layer, the remaining structure would be a (non-finite) TP projection, whence the
RI stage.

8
featural intervention expressed according to the featural Relativized Minimality/fRM

grammatical principle (Friedmann, Belletti and Rizzi 2009). These are the hardest structures,

which are not mastered by children until late in development (not up until around age 9).

Specifically, according to the system developed in Friedmann, Belletti and Rizzi (2009 and

and related subsequent literature) the hardest ORs (and object A’-dependencies more

generally) are those in which a feature inclusion situation is created between an intervening

lexical subject and the lexical head of the relative clause. Assuming a nominal feature dubbed

[+NP] to be among the features to which the principle fRM is sensitive, inclusion of the

nominal feature [+NP] in a situation of structural intervention is a configuration that cannot be

computed by young children (in the relevant age range). This holds crosslinguistically, as

should be expected given the principled nature of the account. The account is schematized in

(8):

(8) i. The principle:

fRM: X … Z ….. Y

*
The relation between X (target) and Y (origin) cannot be established if there is
a Z that structurally intervenes and X and Z share relevant features.
(adapted from: Rizzi 1990, 2004; Starke 2001; Friedmann, Belletti and
Rizzi 2009)

ii. The inclusion relation:

The lion that the elephant wets ___

+R, +NP +NP
X Z Y

where feature inclusion of [+NP] in the intervention configuration is

hard/impossible to compute for young children, in both production and
comprehension.

In elicited production experiments, various ways have been found that children adopt to handle

the hard object dependency: they systematically tend to resort to alternative constructions

which do not involve the intervention configuration in (8). They do so either maintaining the

9
target meaning or just producing non-target sentences (most typically: a SR in place of the

elicited OR; a declarative sentence; a simple DP etc…). As a matter of fact, experimental

results over the years have shown that adults as well tend to avoid the hard intervention

configuration illustrated in (8). This point is taken up in the following subsection which

compares children and adults in this domain, with a focus on Italian.

3.2.1 Other findings from development: The case of RRCs and the comparison with
adults.

Results from production experiments in Italian have shown that both adults and children tend

not to produce elicited ORs containing the hard intervention configuration of feature inclusion.

Children and adults, however, do so in different ways.

The first crucial observation is that adults tend to produce passive past participial Reduced

Relative Clauses/RRCs; i.e. structures of the type presented in (1) in the introduction. Indeed,

they do so in an ample proportion. Considering the results presented in Contemori and Belletti

(2014) for instance, in about half of their answers in the passive, adults resort to RRCs (48%);

the other half of their productions are full relative clauses using copular or venire passives. As

in the quoted references and related ones, I refer to this type of productions as PORs – Passive

Object Relatives –, i.e. relative clauses in the passive answering a question eliciting an active

object relative. PORs may come in various guises: the reduced form is the one largely exploited

by adults (a result also found in Belletti & Chesi 2014). In contrast, in the same experimental

conditions, children tend to produce different types of structures. They also resort to some

PORs, and more and more so as they grow older and the passive computation is better mastered

by them. Interestingly, however, the passive clauses that they use as their PORs are not RRCs,

which are virtually absent from their productions. Their PORs are clearly unreduced clauses.

Moreover, most of their passives are ‘longer’ passive as in the case of si-causative passives, a

10
much-preferred structure by children from their early productions.7 The comparison between

children and adults in this respect is discussed in detail in Belletti (2017). Here the focus is

going to be on the absence of RRCs in children production. Specifically, the following new

question is raised by the results just reviewed:

Q: What makes the reduced relative clause structure of RRCs somewhat hard,
so that children do not appear to be ready to master it and they do not resort to
RRCs in their PORs, differently from what adults do in the same elicitation
contexts?

The question is taken up in the following section 4. It will be addressed by viewing it from the

special perspective of establishing a comparison between the reduction implemented in RRCs

with the smaller structure that young(er) children access, in contrast, through the truncation

operation described in 1 and 2 in their early stages of development. 8

4. Reduction in RRCs and the comparison with truncation: New questions from the
perspective of language development.

The question that adults’ PORs realized as RRCs raise can in turn be addressed from the angle

of the issue of what reduction amounts to in RRCs, in particular in light of the insights coming

from the RI stage and the truncation analysis with which our discussion began.

The relevant question from this angle can be phrased as Q’:

Q’: What makes the reduction implemented in passive past participial RRCs harder
and different from the reduction derived through truncation that young(er) children
often select in early(-er) stages of their development?

7
Results from Contemori and Belletti (2014): in the age range 3;4-8;10 up to 40% of children’s PORs are realized
with si-causative passives in one experiment, and up to 35% are realized as copular/venire passives in another. In
all tasks the si-causative passive is the first and only one type of passive to appear in the youngest groups.
8
A complement question is the following: Why do children opt for an apparently more complex structure (like a
full passive or a si-causative passive); what makes it more readily accessible to them? Some possible hypotheses
to answer this question are presented in Manetti and Belletti (2015), Belletti (2020); for space reasons they will
not be addressed here. The attention here is rather on the lack of children’s use of RRCs. It is worth noticing that
since the experimental conditions are always very informal, the fact that adults resort to RRCs in their answers to
the elicitation questions suggests that such structure is productive and not limited to special formal registers in
standard Italian.

11
From this angle, the reduction occurring in the POR realized as RRCs must not be simple for

the young children; in fact, it appears to be harder than the computation of some complete full

clauses, such as the PORs in the si-causative passive or even in the copular/venire passive,

which children produce to some extent in the elicitation experiments reviewed (footnote 7).

We note first of all that the difficulty with RRCs cannot be a consequence of not having yet

available the relevant position in the CP involved in relativization, in a growing tree type spirit.

The same young children are able to compute subject relatives at the same time in which they

do not resort to RRCs (up to more than 80% SRs in the same experimental results from

Contemori & Belletti 2014 are produced by the youngest group, 3;4-3;11). And some complete

PORs are also produced, as mentioned. Hence, it cannot be an issue having to do with the

process of relativization per se, in general and specifically in an incomplete tree. The syntactic

tree is fully available to them as is the relativization process (and passive as well, to some

extent).

As discussed in previous work (see also Belletti 2014), PORs in which relativization is

performed in a passive clause, constitute a way to implement a derivation whereby the

relativization of the object can be performed in a structure in which intervention is not

disturbing anymore, precisely due to passivization. The hypothesis is based on the assumption

that the derivation of passive involves smuggling (Collins 2005; Belletti 2014; Belletti and

Collins 2021), moving a chunk of the verb phrase containing the verb at the past participle and

the Internal argument pass the External argument sitting in the specifier of the highest vP.

Form this smuggled position, the object moves into the position of the left periphery hosting

the relative head, Spec-Force as in Friedmann, Belletti and Rizzi (2021)9; in this step of the

9
The relative CP is embedded in a D projection, as in Bianchi’s (1999) analysis, assumed in Friedmann, Belletti
and Rizzi (2021).

12
derivation, it does not encounter any intervener on the way. The assumed derivation of a POR

is schematized in (9):

(9) Relativization step

Il bambino che [(__) è [VP pettinato <il bambino>] da [vP la mamma <VP>]

Passivization step/smuggling
The child that (__) is combed <the child> by the mother

(8) schematically illustrates the derivation of a full POR, namely a POR with a complete clausal

structure and a full-fledged morphological manifestation of tense, agreement and the passive

voice, including the inflected passive auxiliary, the past participle and the by-phrase. Given

this background analysis, the question to ask is: how does it compare to the analysis of a

reduced POR, or, more generally of a RRC?

4.1. Outline of the analysis of RRCs

RRCs are small clauses that contain the CP layer, hosting the position into which the noun

phrase head of the relative clause is attracted; the complement of such CP is an impoverished

clausal projection. It is a non-finite complement, containing the past participle and other

components of the passive morphology (Siloni 1995; Harwood 2018, for proposal along similar

lines; Sleeman 2017 for overview). The CP layer is not a complete left periphery in turn: it

contains the higher Force projection hosting the relative head in its Spec, but does not contain

the lowest Fin head, consistently with the non-finite nature of its small clausal complement.

Lack of Fin is the way in which absence of the overt realization of the complementizer in RRCs

can be expressed, under the assumption that the finite complementizer always originates in the

lowest Fin head of the structure in (6) and then raises up into the Force head in full embedded

clauses, such as complement clauses and relatives (see 5.1 for more on this point). The ForceP

layer takes as its complement (at least) a past participial projection in the passive voice,

13
possibly embedded under an aspectual head.10 This analysis is illustrated in (10)a, b. (10)a

presents the low portion of the functional structure of past participial RRCs in the passive voice

indicating the smuggling step of the derivation; in (10)b the high portion of the functional

structure of the past participial RRC is presented in which the relativization step occurs:

(10)

a AspP The smuggling step

2
byP
2
2
by vP
2
DP/EA 2
v PstPrt
2
2
-t 2
V D-NP/IA

b 3 The relativization step

D ForceP
3
3
Force/R AspP
3
byP
3
3
PstPrt by vP
2 3
2 DP/EA 3
-t 2 v <___>
V NP/IA

10
This analysis may also extend to past participial small clauses in general, such as the Absolute small
clauses/ASC discussed in Belletti (1992); Cecchetto and Donati (2020) for recent revisitation. One salient
difference between RRCs and the ASCs is lack of the relativization step in the latter.

14
 4.2. The source of the difficulty with RRCs in (early stages of) development

Given the analysis of RRCs sketched out in (10), a natural reason is provided for the late

appearance of passive past participial RRCs in the detected developmental stage discussed: It

is the consequence of the difficulty in computing a non-canonically grown sentence structure,

in a stage in which the syntactic tree is fully grown, up to the highest Force layer. In particular,

since past participial RRCs are subordinate clauses, they imply presence of the CP level up to

Force. A fully-grown syntactic tree should then contain all positions lower than the highest

Force. However, the structural representation in (10) precisely realizes a syntactic structure in

which the highest position is present, but neither lower portions of the Left periphery nor

positions of the TP area are fully present. In other words, the derivation in (10) occurs in a

structure that is incompatible with the incremental growth of the syntactic tree as well as with

the discussed implicational logic of both growing tree and truncation, according to which if a

higher structural position is present, all positions below it are present as well. Indeed, the

truncation option only operates at the edge of the clausal projection. As discussed, truncation

cannot affect intermediate layers of the structure. This is the crucial property illustrated in (4).

In contrast, the clausal representation of RRCs precisely instantiates a structure that cannot be

obtained through truncation. It is natural to hypothesize that, at the stage of development

spotted by the results, which indicate total lack of RRCs in the elicited productions of young

children in contrast with adults, since such structures cannot be derived through some form of

truncation, they simply cannot be derived at all: a fully projected CP up to Force, should

express a fully projected sentence structure at this stage11. Note that this is consistent with the

results from children’s elicited productions, in which, in the same experiment, when they

resorted to PORs, children always produced full clausal unreduced PORs, as mentioned in

11
This is an updated way of phrasing Rizzi’s principle according to which all clauses are CPs (Rizzi 1993/94).

15
3.2.1. The reduction instantiated by RRCs is thus not a parsimonious computation at young

ages, in contrast with truncation, which may be accessed (also) in very early stages, truncating

the sentence structure at the edge.12

5. Some related consequences

5.1. Lack of Fin and the absence of the complementizer in past participial RRCs

According to the proposed analysis, the reduced structure of past participial RRCs only

contains the upper Force layer of the CP left periphery. Specifically, the lowest Fin head

closing the CP domain is not present (Rizzi 1997 and structure in 6).

This expresses the non-finite/untensed nature of past participial RRCs.13 If the finite

complementizer che originates in Fin the absence of an overt complementizer in RRCs is

expected (and similarly in past participial Absolute small clauses, see footnote 9).

The idea that the finite complementizer originates in Fin, and then raises up to Force – thus

satisfying selection in case of complement clauses – is not new (see e.g. Belletti 2008, 2013

and references cited therein). Here I will just mention the case presented in the quoted previous

work providing evidence for the low location of the finite complementizer, lower than the

higher left peripheral (corrective) focus position. The relevant evidence comes from clefts

sentences in Italian (and also the finite complement of perception verbs in pseudorelatives)

where location of the finite complementizer che in Fin is visible. This is illustrated in (11)a, b

(for an object cleft), where the complementizer displays a distribution parallel to that of the

12
The more general issue of the acquisition of small clauses in development goes beyond the scope of the present
discussion. The prediction is that small clauses could be early acquired only if their analysis is compatible with
the incremental growth of the syntactic tree or with truncation at the edge. Hence, embedded small clauses may
in fact be late acquired. A result in need of closer investigation, but which is consistent with results on the late
acquisition of raising (Mateu and Hyams 2021 and references cited therein; also considering other properties at
play interacting with locality) and also on the late acquisition of ECM/Raising to Object, on which see section
5.3.
13
The proposal shares insights with Pesetsky’s (2021) Exfoliation proposal, according to which non-finiteness
correlates with smaller clausal structure.

16
Japanese complementizer no, the lowest complementizer of the Japanese right periphery, also

appearing in Japanese clefts, as discussed in Saito (2012):14

(11) a be [FocP/contr/corr…[ FinP che [TP S … …O(/PP)]]]]

(adapted from Belletti 2012, 2013, 2015)15

b È UN LIBRO che Maria ha comprato (non un giornale)

it is A BOOK that Maria bought (not a newspaper)

Thus, in RRCs lack of the Fin layer, leads to no overt expression of the complementizer.

5.2. Past participial adjectival modification in Italian child spontaneous speech and in

elicited production

Belletti and Chesi (2014) reported that some past participial modifications are present in young

children’s spontaneous productions. The corpus study undertaken in that work (files from

CHILDES database), has revealed the presence of expressions like the following:

(12) “mamma io ho le mani occupate”/lit: I have the hands occupied

(Camilla 3;4)

These expressions should not be considered RRCs but are rather best analyzed as

corresponding to forms of adjectival modification, occurring within the nominal projection. As

such, they do not involve the kind of clausal reduction that young children do not appear to be

able to compute at this early age otherwise.

Similarly, recent experimental findings in Martini (2020), have documented apparent past

participial reduced PORs in children’s answers to eliciting questions (in both French and

Italian). Most of these productions (87%), however, do not contain the overt expression of the

by-phrase, suggesting again an adjectival nature. Moreover, the age factor may also play a role

as these elicited productions mainly came from older children (e.g.: Quella sgridata/ the one

14
In Japanese a dedicated form of the low complementizer, i.e. no, is available, differently from Italian where the
same form che also used in full declaratives appears in clefts. See references quoted for further details.
15
In (11) the complement of the copula is a FocP, due to the selecting properties of be; hence the structure of
the CP does not reach the Force layer; the complementizer may be assumed to remain in Fin in this construction.
See references quoted for further details.

17
punished. Quella punita dalla maestra/ the one scolded by the teacher), suggesting a gradual

development of access to the reduced structure of RRCs16.

5.3 Comparative cross-linguistic developmental considerations

Similar conclusions on late access to reduced structures can be drawn from results on the late

acquisition of ECM/Raising to Object-RtO (and also some Object control structures) in

European Portuguese/EP, discussed in Santos, Hyams and Gonçalves (2016). According to the

results presented, young children (3 to 5 yo) do not properly master use of reduced (defective)

sentential complements in RtO until relatively late. Interestingly, as the authors note,

sometimes they also over-extend use of Inflected infinitival complements in contexts in which

these forms are not possible in the adult target grammar (e.g. in object control), as illustrated

in the following example:

(13) *... ensinou [os gansos saltarem]. (4;05,12)

taught the geese jump.INF.3PL (vs. … ensinou os gansos a saltar(em))
(Santos, Hyams and Gonçalves 2016: ex.55a, p.24)

In the authors’ words, children have an early preference for ‘complete functional

complements.’(34, p.11). This is especially significant in the (impossible) case in (13), in

which they opt for a structure including a complete rich person agreement. These results from

EP are also in line with the observed later acquisition of RtO with the verb want in English

presented in Landau & Thornton (2011). This observation is especially significant when

compared the one on subject control sentences, which are assumed to involve a more richly

articulated clausal structure, and which, all things being equal, appear to be more easily

acquired from early on.

16
Ages tested 3 to 9.yo Distribution of the relevant productions is as follows: 1 case at 5yo; 9 cases at 7yo, of
which only 1 with by-phrase; 14 cases at 8yo; 25 cases at 9yo, of which 5 with by-phrase.

18
 6. General concluding remarks

The developmental data reviewed, strongly suggest an overall general conclusion, which is

particularly significant if one looks at it from the perspective of the issue of complexity. The

conclusion is that young children do not necessarily favor shorter expressions involving ‘less’

words, affixes…., over longer ones, and they do so sometimes in contrast with what adults

favor instead. We have seen this happen in the Italian example reviewed concerning the late

access to RRCs. In the same elicitation condition, the typical answer provided by adults is a

‘shorter’ RRC. Children’s answers are different and, when realized in the passive, they are

typically ‘longer’. This is a type of result that also the cross-linguistic findings briefly

mentioned in the previous section indicate, in different domains. Then, the conclusion is that

it all depends on how the ‘size’ of the clause structure is obtained. And it may so happen, as

in the case reviewed, that shorter does not count as simpler for the developing grammar. It can

in fact count as harder, thus in fact more complex at the appropriate level of analysis. This in

turn provides a reason as to why the construction appears later in development. The proposal

here has been that the reduction of the sentential structure of RRCs may be one such case, as

its late mastering suggests, in clear contrast with the early process of truncation at the root,

which young(er) children can exploit on complexity grounds. According to the proposal

developed here, the difficulty with RRCs comes from the incomplete projection of the syntactic

structure in the intermediate area between the CP and the TP layers. The proposal here has

been that this type of reduction of the clause structure is at odds both with the incremental

growth of the syntactic clausal structure and with a truncation process that could only occur at

the edge of the clausal projection, not inside it.

A theoretically driven analysis thus turns out to be of help in understanding seemingly

unexpected behaviors during development such as the one emerged from the reviewed data

discussed in these notes, whereby adults opt for ‘less’ where children go for ‘more’. In turn,

19
developmental stages may be illuminating of relevant steps of linguistic analyses and the

theoretical assumptions behind them, thus contributing to reveal their deepest significance.

These notes are thus a contribution to the productive ongoing dialogue between formal

linguistic theory and language acquisition studies.

References

Antinucci, F. & R. Miller. 1976. How children talk about what happened. Journal of Child
Language 3. 167–189.

Belletti, A. 2020. (Reflexive) Si as a route to passive in Italian. in L.Franco & P.Lorusso

(eds.), Linguistic Variation: Structure and Interpretation. Berlin: De Gruyter, 73-86.

Belletti, A. 2017. On the acquisition of complex derivations with related considerations on

poverty of the stimulus and frequency. In E.Di Domenico ed., Syntactic Complexity
from a Language Acquisition Perspective, Cambridge Scholars Publishing, 28-48.

Belletti, Adriana. 2015. The Focus map of clefts: Extraposition and Predication. In: U.Shlonsky
ed. Beyond Functional Sequence, Cartography of Syntactic Structures volume 10,
Oxford University Press, 42-59

Belletti, Adriana. 2014. Notes o Passive Object Relatives. In P.Svenonius ed. Functional
Structure from Top to Toe, New York:Oxford University Press, 97-114

Belletti, A. 2013. On Fin: Italian che, Japanese no, and the selective properties of the copula in
clefts, in Miyamoto Y. et al. eds. Deep Insights, Broad Perspectives. Essays in Honor
of Mamoru Saito, Kaitakusha Co., Ltd, Tokyo, p 42-55

Belletti, A. 2012. Revisiting the CP of Clefts. In: Discourse and Grammar. From Sentence
Types to Lexical Categories. Berlino:de Gruyter, 91-114

Belletti, A. 2008. The CP of Clefts. RGG. Rivista di Grammatica Generativa, vol. 33, 191-204

Belletti, A. 1992. Agreement and Case in Past Participial Clauses in Italian. In: T.Stowell
and E.Wehrli eds. Syntax and Semantics, vol.26, p. 21-44, New York: Academic Press

Belletti, A. & C. Chesi 2014. A syntactic approach toward the interpretation of some
distributional frequencies: Comparing Relative clauses in Italian corpora and in elicited
production. In Rivista di Grammatica Generativa, 36, 1-28

Belletti, A. & C. Manetti 2018. Topics and passives in Italian speaking children and adults.
Language Acquisition, (2019, 26:2, 153-182) DOI: 10.1080/10489223.2018.1508465

Belletti, A. & C. Collins 2021, Smuggling in Syntax. OUP

20
Bianchi, V. 1999. Consequences of antisymmetry: Headed relative clauses. Berlin: Mouton de
Gruyter.

Cardinaletti, A. 1990. Pronomi nulli e pleonastici nelle lingue germaniche e romanze,

Doctoral dissertation, University of Venice.

Cardinaletti, A. 1990a. Subject/object asymmetries in German null-topic constructions and

the status of specCP, in: Mascaró, J. & M. Nespor (eds.), Grammar in Progress,
Dordrecht: Foris, 75–84.

Cecchetto, C, and C. Donati 2020. Relabeling participial constructions, In: Linguistic

Variation: Structure and Interpretation (SGG, 132), L. Franco and P. Lorusso (eds.),
De Gruyter Mouton, 149-162

Collins, C. 2005. A smuggling approach to the passive in English. Syntax 8(2). 81–120.

Contemori, C. & A. Belletti 2014. Relatives and Passive Object Relatives in Italian speaking
children and adults: Intervention in production and comprehension. Applied
Psycholinguistics 35(6), 1021-1053

Crisma, P. 1992. On the Acquisition of W/i-Questions in French. Geneva Generative

Papers/GenGenP 1-2, 115-122.

Friedmann, N. 2002. Question production in agrammatism: The Tree Pruning Hypothesis.

Brain and Language 80, 160-187. doi:10.1006/brln.2001.2587

Friedmann, N., A. Belletti, L.Rizzi 2021. Growing trees: The acquisition of the left periphery.
accepted for publication in Glossa.

Friedmann N., A. Belletti, L.Rizzi 2009. Relativized Relatives: Types of intervention in the
acquisition of A’-dependencies. Lingua 119: 67-88

Friedmann, N., Yachini, M., and Szterman, R. 2015. “Relatively easy relatives: children with
syntactic SLI avoid intervention,” in Structures, Strategies and Beyond. E. Di
Domenico, C. Hamann, and S. Matteini eds.Amsterdam: John Benjamins; Linguistik
Aktuell series, 303–320.

Harwood, W.2018. Reduced Relatives and Extended Phases: A Phase-Based Analysis of the
Inflectional Restrictions on English Reduced Relative Clauses. Studia Linguistica,
72(2), 428-471

Hyams, N. 1986. Language acquisition and the theory of parameters. Dordrecht: D. Reidel

Landau, I. and R. Thornton 2011. Early Child Control. Lingua 121, 920-941.

Manetti, C. & Belletti, A. 2015. Causatives and the acquisition of the Italian passive. In
Language Acquisition and Development. C.Hamann and E. Ruingendijk eds.,
Cambridge Scholars Publishing, 282-298

Martini, K. 2020. Aspects of the acquisition of locality. Comprehension and production studies

21
 in Italian and French. Doctoral dissertation, University of Geneva

Mateu, V. & N. Hyams 2021. On children’s late acquisition of raising seem and control
promise: Is a unified account possible? In: C.Collins and A.Belletti eds. Smuggling in
Syntax ,Oxford Studies in Comparative Syntax, OUP, 222-254

Pesetsky, D. 2021 Exfoliation: Towards a derivational theory of clause size.

lingbuzz/004440

Radford, A. 1990. Syntactic theory and the acquisition of English syntax: The nature of early
child grammars of English. Blackwell: Oxford.

Rizzi, L. 2004. Locality and Left Periphery. In: Structures and Beyond – The Cartography of
Syntactic Structures, vol 3, A. Belletti ed., Oxford University Press, Oxford, New
York, 223-251

Rizzi, L. 2006. Grammatically-Based Target-Inconsistencies in Child Language. In Deen,

K.U., J. Nomura, B. Schulz and B.D. Schwartz (eds), Proceedings of the Inaugural
Conference on Generative Approaches to Language Acquisition–North America,
Honolulu, HI. University of Connecticut Occasional Papers in Linguistics, 4, 19-49.

Rizzi, L.1993/94. Some Notes on Linguistic Theory and Language Development: The Case of
Root Infinitives. Language Acquisition, 4 (3), 371-393.

Rizzi, L. 1990. Relativized Minimality. MIT Press

Rizzi, L. & G.Bocci 2017. The left periphery of the clause – Primarily illustrated for Italian.
In: The Wiley Blackwell Companion to Syntax, Second Edition, M. Everaert & H. C.
van Riemsdijk eds., John Wiley & Sons, Inc. 2017, DOI:
10.1002/9781118358733.wbsyncom104

Saito, M. 2012. Sentence type and the Japanese right periphery. In E.Zimmermann and
G.Grewendorf eds. Ten years after. Studies in Generative Grammar 112, De Gruyter

Salustri, M. & N.Hyams 2003. Is there an analogue to the RI stage in the Null Subject
languages? Proceedings BUCLD 29. Sommerville: Cascadilla Press. 692-703

Salustri, M. & Hyams, N. 2006. Looking for the universal core of the RI stage. In V. Torrens
& L. Escobar, eds. The Acquisition of Syntax in Romance Languages, John Benjamins,
Amsterdam

Santos A.L., Gonçalves A. & N. Hyams 2016. Aspects of the acquisition of object control
and ECM-Type verbs in European Portuguese, Language Acquisition, 23:3, 199-233.

Siloni, T. 1995. On participial relatives and complementizer D°: A case study in Hebrew and
French. Natural Language and Linguistic Theory, 13(3), 445-487.

Sleeman, P. 2017. Participial Relative Clauses, Oxford Research Encyclopedia in Linguistics.

Starke, M. 2001. Move dissolves into merge: A theory of locality. Geneva, Université de

22
 Genève dissertation.

Wexler, K. 1994. Optional Infinitives, head movement and the economy of derivations in child
grammar. In D. Lightfoot & N. Hornstein (Eds.) Verb movement. Cambridge University
Press: 305-350

23