In Canzhong Wu, Christian Matthiessen, and Maria Herke (eds.)
Proceedings of ISFC 2008: Voices Around the World. ISFC Sydney, July 2008: 7-12.
A Systemic Functional Micro-Grammar of Spanish Clitics
Víctor M. Castel
Consejo Nacional de Investigaciones Científicas y Técnicas
Universidad Nacional de Cuyo
The word order patterns and participant role distribution of Spanish clitics are two well-known
phenomena which have been thoroughly studied in Hispanic Linguistics from the perspective of
both formal and functional grammars. Arús (2006) is a Sydney Grammar approach to the semantics
of se but with no references to explicit realization rules or integration with the morpho-syntactic
and semantic distributional properties of other clitics. Thus, there are currently no descriptions
integrating these phenomena from the viewpoint of Systemic Functional Linguistics, let alone the
Cardiff Grammar framework and the specifics of River Plate Spanish (RPS). A set of data
illustrating the two phenomena is here accounted for adequately and elegantly in terms of the
Cardiff Grammar framework (Fawcett 2000, 2008), more specifically within the new
computational version of GeneSys, the Cardiff Grammar Generator of Fawcett & Castel (2006).
The paper presents a micro-grammar capable of generating representations that capture both the
patterns constraining word order and the expression of participant roles of RPS clitics.
The purpose of this paper is to specify a micro-grammar of River Plate Spanish clitics
(GSC) within GeneSys, the Cardiff Grammar (CG) environment for the development of
generation oriented grammars (Fawcett et al 1993, Fawcett & Castel 2006). CG is a
theoretical framework based on Halliday’s leading ideas on the systemic-functional
approach to natural languages (Fawcett 2000, 2008).
Since the CG framework has not yet been applied to Spanish (Fawcett 2008: 3), let
alone River Plate Spanish (RPS), this paper addresses the problem posed by the word
order patterns and participant role distribution of RPS clitics from the CG perspective.1
2 Word Order Patterns and Expression of Participant Roles
The word order constraints underlying the distribution of RPS clitics can be abbreviated
by the schemas (1)-(3):
(1) III II I A/D, (Modified version of Perlmutter (1972)’s surface structure filter.)
(2) III II I A/D + V [non-imperative finite form],
(3) V [imperative or infinitive or gerund] + III II I A/D,
where ‘III II I A/D’ is a growing monotonic sequence in which roman numbers stand for
clitic person values, ‘A/D’ is either an accusative (A) or dative (D) clitic, ‘V’ stands for
a verb, the expressions within square brackets refer to forms of such a verb, and the sign
‘+’ marks the relative order between the clitic sequence and ‘V’. The schemas capture in
an abbreviated way two aspects of word order: (i) patterns governing the occurrence of
1 Arús (2006) is a Sydney Grammar approach to the semantics of se but with no references to explicit
realization rules or integration with the semantic and morpho-syntactic properties of other clitics.
clitics with other clitics (cf. (1)), and (ii) patterns governing the occurrence of clitics
with their governing verbs (cf. (2)-(3)).
According to (1), the sequences in Table 1 are all the possible well-formed clitic
sequences that can occur in RPS clauses, i.e. any other clitic sequence is ill-formed:
se te me
se me lo/a(s) se te me le(s)
se te nos
se nos lo/a(s) se te nos le(s)
se te le(s)
te me le(s)
se te me lo/a(s)
se te lo/a(s)
te nos le(s)
se te nos lo/a(s)
se lo/a(s) me lo/a(s) se me le(s)
te nos lo/a(s)
nos lo/a(s) se nos le(s)
te me lo/a(s)
Table 1: Word Order Patterns among Clitics
Table 2 presents a few examples illustrating the well-formedness conditions
captured by the schemas (2) and (3) on the occurrence of clitics in relation to their
Clauses satisfying (2)
Clauses violating (2)
Clauses satisfying (3)
Clauses violating (3)
Me lo regalás.
*Me lo regalá.
Te las quieren regalar. *Quieren te las regalar. Quieren regalártelas.
*Quieren te las regalar.
Se estaban saludando. *Estaban se saludando. Estaban saludándose.
*Estaban se saludando.
Table 2: Word Order Patterns between Clitics and their Governing Verbs
The following table abbreviates a few core correlations between participant roles
(PR), and the clitics and verb endings that express them:
Me lo regalo/ás/a.
Me le regalo/ás/a.
mei, Af-Ca lon, Af-Po regaloi/ásk/ak, Ag mei, Af-Po lej, Af-Ca regaloi/ásk/ak, Ag mei, Af-Ca regaloi/ásk/ak, Ag
mei, Af-Po regaloi/ásk/ak, Ag
Te lo regalo/ás/a.
Te le regalo/ás/a.
tei, Af-Ca lon, Af-Po regalok/ási/ak, Ag
tei, Af-Po len, Af-Ca regalok/ási/ak, Ag
tei, Af-Ca regalok/ási/ak, Ag
tei, Af-Po regalok/ási/ak, Ag
Se lo regalo/ás/a.
*Se le regalo/ás. // Se le regala.
*Se regalo/ás. // Se regala.
i, Af-Ca lon, Af-Po regalo/ás/ak, Ag
sei, Af-Po len, Af-Ca regalai, Ag
sei, Af-Po regalai, Ag
sei, Af-Ca regalai, Ag
i, Af-Ca lon, Af-Po regalai, Ag
sei, Ag len, Af-Ca regalai, Ag
sei, Ag regalai, Ag
i, Ag lon, Af-Po regalai, Ag
Table 3: Correlations between PRs, and Clitics and Verb Endings
Note that no examples are given of clauses containing three or four clitic sequences,
for no provision is made here for ethical participants.
3 System Network and Realization Rules
The system network and the realization rules (RRs) needed to account for the data in §2
are given in Figures 1-7, and Figure 8, respectively. The RRs assume that the syntactic
unit clause (Cl) filling the text-sentence element ‘Σ’ is made up of a certain number of
places. When a ‘Cl’ is introduced into the linguistic representation being constructed, a
structure like (4) is defined:
Cl[1, …, 120]Cl]Σ
where ‘selection_expression’ is a variable ranging over semantic features selected from
the system network (cf. Figures 1-7), and ‘…’ ranges over 2 through 119.2 RRs are
defined for the elements Operator (O), Verb Root (M), and Verb Ending (Vnd) to
occupy places 99, 100 and 101, respectively:
(5) Cl[… 95 96 97 98 99 100 101 102 103 104 105 …]Cl
Clt3 Clt2 Clt1 CltA O M Vnd Clt3 Clt2 Clt1 CltA
The places 95-98 are reserved for the occurrence of clitics with non-imperative finite
verb forms (cf. schema (2)), and 102-105 for the occurrence of clitics with imperatives
(= proposal_for_action), infinitives and gerunds (cf. schema (3)). See lines L2-L3 of
RRs 20.11-12, 20 and 20.2-3 of Figure 8, responsible for the exponence of ‘CltA’,
‘CltD’, ‘Clt3’, ‘Clt1’, and ‘Clt2’ as lo/a(s), le(s), se, me/nos, and te, respectively. Note
that only the clitic sequences in Table 1 are allowed.
GSC generates text-sentences like (6), (8) and (10), after deletion of unused places
and stripping of the corresponding linguistic representations (7), (9), and (11),
(6) Lo saludás. (Cf. *Saludáslo. See second row in Table 2.)
(7) [[…, giver, …]
Cl[1 … 97 CltA/Af[lo] 99 M[salud] Vnd/Ag[ás] … 119 E[.]]Cl]Σ
(8) Saludálo. (Cf. *Lo saludá. See second row in Table 2.)
(9) [[…, proposal_for_action, …]
Cl[1 … 99 M[salud] Vnd/Ag[á] 102 103 104 CltA/Af[lo] … 119
(10) Se lo saluda. Impersonal se. (Cf. *Lo se saluda.)
(11) [[…, agent_and_affected, non_coref_rel_ag_af, agent_fcs, outsider_fcs,
low_deixis_fcs, outsider_low_deixis, outsider_fcs_recoverable, …, singular_fcs, …,
agent_stated, affected_stated, …, outsider_af_recoverable, …, singular_af, high_deixis_af,
least_active, outsider_high_deixis_least, non_feminine_af] Cl[1 … 94 Clt3/Ag[se] 96 97
CltA/Af[lo] 99 M[salud] Vnd/Ag[a] … 119 E[.]]Cl]Σ
RR 20.11 of Figure 8 is triggered by the feature outsider_high_deixis_least. This
feature corresponds to the meanings ‘other’, ‘high deixis’, and ‘least’ that García (1975)
assigns to the clitics lo/a(s). The system of Figure 5 introduces it as a gate on the basis
of the disjunctive features high_deixis_af and high_deixis_af_po subcategorizing
outsiders and a subset of addressees in the semantic specification of the PRs Affected
(Af) and Affected-Possessed (Af-Po), respectively. These PRs ‘Af’ and ‘Af-Po’ conflate
with the element ‘CltA’ depending on whether the clause selection expression contains
the feature high_deixis_af or high_deixis_af_po, respectively. Cf. L4-L5 of RR 20.11.
The element ‘CltA’ is expounded by the lexical item lo/a(s), independently of the
2 It is assumed here that GSC requires 120 places for the clause.
position it occupies with respect to the governing verb, and also independently of
whether it has been conflated with ‘Af’ or ‘Af-Po’. Cf. L6-L17 of RR 20.11. Whether
the exponence of ‘CltA’ is lo, la, los, o las depends on the selection of specific features
from the systems PERSON, NUMBER, GENDER, and TENOR. ‘CltA’ is expounded
by lo (cf. L6-L8) if the clause selection expression contains either (i) a singular outsider
non_feminine_af_po), or (ii) a singular addressee (singular_addressee_af or
singular_addressee_af_po), non-feminine (non_feminine_af or non_feminine_af_po),
and the tenor is formal. Mutatis mutandis, ‘CltA’ is expounded by la (Cf. L9-L11), los
(Cf. L12-L14), and las (Cf. L15-L17).
The occurrence of se in (10) results from the application of RR 20, which is called
by the feature outsider_low_deixis introduced by the rule of Figure 7. The features
low_deixis_fcs, low_deixis_af, low_deixis_af_po, low_deixis_af_ca subcategorize
outsiders and a subset of addressees in the semantic specification of the PRs ‘Ag’, ‘Af’,
‘Af-Po’, and ‘Af-Ca’, respectively. The rule in Figure 7 reduces the disjunction of
features to the gate outsider_low_deixis and thus it captures García (1975)’s proposal
that the meaning of se is made up of ‘other’ and ‘low deixis’, fulfilling in (10) a
The paper has presented the system network and realization rules of a generation
oriented micro-grammar of RPS clitics. The specification is based on Castel (2007)’s
rule typology of the Cardiff Grammar framework (Fawcett 2008, 2000; Fawcett et al
1993). GSC is written within GeneSys, the development environment of the Cardiff
Grammar Generator (Fawcett & Castel 2006). GSC accounts for core phenomena of
RPS clitics in simple clauses: the word order patterns governing the occurrence of clitics
with other clitics and with their governing verbs, and the distribution of participant
Arús, Jorge (2006) “Perspectiva sistémico-funcional de los usos de se en español”, Signos 39 (61): 131-
Castel, Víctor M. (2007) “Rule Types in a Systemic Functional Grammar: An XML Definition of the
Cardiff Lexicogrammar Generator”. In Barbara, L. y T. Berber Sardinha (ed.) Proceedings of the 33rd
International Systemic Functional Congress. San Pablo: PUCSP.
Fawcett, Robin P. (2000) A Theory of Syntax for Systemic Functional Linguistics.
Amsterdam/Philadelphia: John Benjamins Publishing Company.
Fawcett, Robin P. (2008) Invitation to Systemic Functional Linguistics. The Cardiff Grammar as an
extension and simplification of Halliday’s Systemic Functional Grammar. London: Equinox.
Fawcett, Robin P., G.H. Tucker y Y.Q. Lin (1993) “How a systemic functional grammar works: the role
of realization in realization.” En Horacek, H. y M. Zock (eds.) New Concepts in Natural Language
Generation. London: Pinter.
Fawcett, Robin P. and V. M. Castel (2006) Software for GeneSys: Prototype Generator 3. Cardiff UK:
Computational Linguistics Unit, Cardiff University.
García, Érica C. (1975) The Role of Theory in Linguistic Analysis. The Spanish Pronoun System.
Amsterdam/Oxford: North-Holland Publishing Company.
Perlmutter, David M. (1972) Deep and Surface Structure Constraints in Syntax. New York: Holt and Co.
o Fig .
low_deixis _fcs ( sp23
high_deixi s_fcs ( sp24)
o Fig .2
outsider_fcs ( sp31 )
o Fig. 5
p rformer_ f
i cludes_p r
a dressee_ f
i cludes_a d
d essee_af T
o Fig. 4
h gh_deixis_least ( 20.11)
a _recover b
a le_from_ r
p evious_u i
I _LEAST_G T
outsider - af_recove a
r ble_from c
_ urrent_u i
a _recover b
a _specifi d
h gh_deixis_less ( 20.12)
I _LESS_GA E
→ outsider_ o
l w_deixis ( 20)
Figures 1-7: Systems Relevant for Clitic Realization Rules
L1 20.11 : outsider_high_deixis_least :
L1 20.2 : referent_includes_performer :
L2 [if not (proposal_for_action or infinitive or gerund) then CltA @ 98,
L2 [if not (proposal_for_action or infinitive or gerund) then Clt1 @ 97,
L3 else CltA @ 105],
L3 else Clt1 @ 104],
L4 [if high_deixis_af then Af by CltA],
L4 [if referent_includes_performer_af or (coref_rel_ag_af and
L5 [if high_deixis_af_po then Af-Po by CltA],
L5 referent_includes_performer_fcs) then Af by Clt1],
L6 [if (singular_af or singular_af_po or (formal and (singular_addressee_af
L6 [if referent_includes_performer_af_po or
L7 or singular_addressee_af_po))) and (non_feminine_af or
L7 (agent_and_affected_possessed_coref and
L8 non_feminine_af_po) then CltA < "lo"],
L8 referent_includes_performer_fcs) then Af-Po by Clt1],
L9 [if (singular_af or singular_af_po or (formal and (singular_addressee_af
L9 [if referent_includes_performer_af_ca or
L10 or singular_addressee_af_po))) and (feminine_af or
L10 (agent_and_affected_carrier_coref and referent_includes_performer_fcs)
L11 feminine_af_po) then CltA < "la"],
L11 then Af-Ca by Clt1],
L12 [if (plural_af or plural_af_po or plural_addressee_af or
L12 [if singular_performer_af or singular_performer_af_po or
L13 plural_addressee_af_po) and (non_feminine_af or
L13 singular_performer_af_ca or (coref_rel_ag_af and
L14 non_feminine_af_po) then CltA < "los"],
L14 singular_performer_fcs) or (agent_and_affected_possessed_coref
L15 [if (plural_af or plural_af_po or plural_addressee_af or
L15 and singular_performer_fcs) or (agent_and_affected_carrier_coref
L16 plural_addressee_af_po) and (feminine_af or feminine_af_po)
L16 and singular_performer_fcs) then Clt1 < "me"],
L17 then CltA < "las"].
L17 [if plural_performer_af or plural_performer_af_po or
L18 plural_performer_af_ca or (coref_rel_ag_af and
L1 20.12 : outsider_high_deixis_less :
L19 plural_performer_fcs) or (agent_and_affected_possessed_coref
L2 [if not (proposal_for_action or infinitive or gerund) then CltD @ 98,
L20 and plural_performer_fcs) or (agent_and_affected_carrier_coref
L3 else CltD @ 105],
L21 and plural_performer_fcs) then Clt1 < "nos"].
L4 [if high_deixis_af_ca then Af-Ca by CltD],
L5 [if (singular_af_ca or (formal and singular_addressee_af_ca))
L1 20.3 : singular_addressee :
L6 and not outsider_af_po_recoverable then CltD < "le"],
L2 [if casual then [if not (proposal_for_action or infinitive or gerund)
L7 [if (plural_af_ca or plural_addressee_af_ca) and not
L3 then Clt2 @ 96, else Clt2 @ 103],
L8 outsider_af_po_recoverable then CltD < "les"].
L4 [if singular_addressee_af or (coref_rel_ag_af and
L5 singular_addressee_fcs) then Af by Clt2],
L1 20 : outsider_low_deixis :
L6 [if singular_addressee_af_po or
L2 [if not (proposal_for_action or infinitive or gerund) then Clt3 @ 95,
L7 (agent_and_affected_possessed_coref and
L3 else Clt3 @ 102],
L8 singular_addressee_fcs) then Af-Po by Clt2],
L4 [if low_deixis_fcs then Ag by Clt3],
L9 [if singular_addressee_af_ca or
L5 [if low_deixis_af then Af by Clt3],
L10 (agent_and_affected_carrier_coref and
L6 [if low_deixis_af_po then Af-Po by Clt3],
L11 singular_addressee_fcs) then Af-Ca by Clt2],
L7 [if low_deixis_af_ca then Af-Ca by Clt3],
L12 Clt2 < "te"].
L8 Clt3 < "se".
Figure 8: Clitic Realization Rules