Professional Documents
Culture Documents
2018d Four Directionalities For Grammat
2018d Four Directionalities For Grammat
2018d Four Directionalities For Grammat
EDITORIAL
ARTICLES
1. Introduction
Directionality and grammaticalization have been closely associated from the very
beginning of grammaticalization’s relevance on the grammatical scene as a com-
mon framework for the analysis of diachronic processes, starting in the 1980s.
https://doi.org/10.1075/jhl.17032.com
Journal of Historical Linguistics 8:3 (2018), pp. 356–387. issn 2210-2116 | e‑issn 2210-2124
© John Benjamins Publishing Company
Four directionalities for grammaticalization 357
I am aware that the table above poses some theoretical questions. One question is
whether directionality C, usually known as “neutral” or “neither down nor up,” is
an actual instance of directionality. I include it in the table and consider it as a type
of directionality, because change may establish diachronic paths with no need to
move from one level of grammar to another. The second question is why direc-
tionality D is not the sum of directionalities B + A – in other words, why it would
not be more desirable to decompose directionality D into two minor directionali-
ties? As I argue in Section 5, directionality D requires its own status, distinct from
the sum of A and B, due to its specific source and due to the fact that its reinser-
tion into the grammar is in a specific category. It has its own path and distribution
as well; its circular path is characteristic and innovative.
Directionality D is a type of directionality practically unknown in the special-
ized literature, described only in Octavio de Toledo y Huerta (2002) and Com-
pany’s papers presented in 2016 and 2017 (cf. Acknowledgments). I call this fourth
type round trip directionality.
Before entering into the analysis of diachronic data, it is necessary to define
grammar and its basic domains, at least operatively, because the four types of
directionality involve diachronic movement across different grammatical
domains. Grammar is simply the combinatory ways in which words bind together
and the meanings that such combinations produce; rules, principles, and patterns
determine the possible word combinations in a language (Real Academia
Española & Asociación de Academias de la Lengua Española 2009: 4).
A grammar is comprised of two different subsystems, which are separated
by a diffuse border. These subsystems are sentence grammar (SG) and periphery
grammar (PG). Sentence grammar is, essentially, the domain of core syntactic
relations, constituency, and (quasi-)obligatory distributions within a sentence,
whether simple or complex; it is a subsystem of obligations and dependencies,
allowing a lower level of order variability. The second subsystem is the domain of
extra-clausal constituency, no constituency, and weak or loose syntactic relations.
Within this subsystem, more variability in order is permitted, although in many
cases the linguistic forms in periphery grammar are placed at the extremes of an
utterance and separated by pauses from the rest of the utterance. This domain is
determined by the situation of discourse and by the speaker’s point of view. Both
grammatical subsystems are complementary and equally important for the con-
strual of utterances.
The distinction between SG and PG – or at least the distinction between
these two subsystems under a diverse set of labels – is a traditional one; papers
on grammaticalization usually describe this grammatical dichotomy, although
often without explicitly defining the two subsystems. Kaltenböck, Heine & Kuteva
(2011: 849–852) reviewed the differences between these two domains. The authors
differentiated between two systems: “sentence grammar,” also named “anchor”
and “core grammar”; and “thetical grammar.” These integrate into “discourse
grammar,” which “is composed of all the linguistic resources that are available
for constructing spoken, written (or signed) texts.” Essentially, they use the labels
“sentence,” “thetical,” and “discourse” to describe what I call “sentence,” “periph-
ery,” and “grammar,” respectively. Certainly, the discussion of whether a given lin-
guistic form is “in” the sentence grammar or not depends upon the researcher’s
conception of what grammar is.
Downgrading changes, as I said above, are diachronic paths that feature instances
of the following: lexical form > grammatical form; free forms > bound forms;
phrase/content word > functional word/morpheme; optional use > obligatory use;
peripheral grammar > sentence grammar; syntax > morphology; loose parataxis
> tight syntax.
The existential Spanish construction ha + ý > existential verb hay exemplifies
this first type of directionality paradigmatically.2 This directionality corresponds
to the traditional definition of grammaticalization.
The Old Spanish adverb ý < Lat. ībi ‘there’ became a bound form -y of the pre-
sent indicative existential verb hay ‘there is/there are’.3 In Old Spanish, the con-
servative-etymological existential verb was ha < Lat. habet. The form ha lacks any
affix, as (1) shows. (2) displays the typical context whereby the grammaticalization
of ha-y started. In this latter sentence, there is a remarkable locative redundancy
because, besides the locative adverb ý, there are other locative complements: en
la montaña do yo moro ‘on the mountain where I live’ encodes a scene in which
the direct object (do), un lago muy grande ‘a very big lake’, is located; this do also
has locative meaning. The locative adverb y functions as a full anaphor, which
retrieves the locative do previously mentioned. (3) shows both the etymological-
conservative ha and the innovative-grammaticalized ay coexisting in the same
sentence. This example demonstrates the typical contexts that favor each existen-
tial verb form at the beginning of the grammaticalization process. The conserva-
tive ha usually appears with a generic non-specific do, as with peçes e agua ‘fishes
2. Old Spanish examples are taken from the Corpus Diacrónico del Español (CORDE) of Real
Academia Española (www.rae.es), and Corpus Diacrónico y Diatópico del Español de América
(CORDIAM) of Academia Mexicana de la Lengua (www.cordiam.org). Modern Spanish exam-
ples come from Corpus del Español Actual (CREA) of Real Academia Española (www.crea.es),
from Google Books and from spontaneous Mexican speech.
3. This change is examined at length in Company (2012).
and water’ in (3), while the innovative form ay subcategorizes a concrete, definite
specific do, as in the case of un galápago mi amigo ‘a turtle who is my friend’.4 In
(4), the free, lexical, stressed adverb ý may appear in any position in the sentence;
this is seen post- and preverbally in (4a) and (4b), respectively. It can also appear
with tenses other than the present and can co-occur with verbs other than haber,
as in (4b) and (4c), showing that ý was a full, free adverb in Old Spanish.
(1) Ca en las cosas en que tan gran mal ha, que se non pueden
because in those things in which so big evil have.3sg that refl neg can.3pl
cobrar si se fazen.
avenge if refl make.3pl
‘Because in those things that are greatly evil, that they cannot be avenged if
they are embarked upon.’ (CORDE: 14th c., Don Juan Manuel, Conde Lucanor)
(2) En la montaña do yo moro ay un lago muy grande
on the mountain where I live.1sg have.3sg a lake very big
‘On the mountain where I live there is a big lake.’
(CORDE: 14th c., Anonymous, Caballero del Cisne)
(3) E yo se de un lugar apartado e muy viçioso do ha peçes
and I know.1sg of a place faraway and very nice where have.3sg fish
e agua, e ay un galapago mi amigo.
and water and have.3sg a turtle my friend
‘I know a nice, faraway place where there are fish and water, and there is a tur-
tle who is my friend.’ (CORDE: 13th c., Anonymous, Calila e Dimna)
(4) a. seyendo la tierra de suso sana e entera, que nunqua ouiera
be.ger the land by itself healthy and entire that never have.3sg
ý poblança alguna.
there settlement any
‘The land being healthy and entire by itself, never there was any settle-
ment.’ (CORDE: 13th c., Alfonso X, General estoria. Segunda parte)
4. In Spanish, as in many Romance languages, the noun phrase following the existential verb
ha, today hay, is a direct object, because the etymon is the transitive Latin verb habeo. The proof
of this noun phrase being do is that it always phrase pronominalizes in accusative form: hay un
lago ‘there is a lake’ > lo hay ‘there is it-acc.m’; hay una montaña ‘there is a mountain’ > la hay
‘there is it-acc.f’. From the 16th century on, a reanalysis of the existential verb as intransitive
one took place place; the consequence has been that the verb displays a plural inflectional mor-
pheme while the do is being reanalyzed as a subject, generating agreement with the verb: ha
habido problemas ‘there is problems’ > han habido problemas ‘there are problems’. The change is
already in progress (Hernández Díaz 2006).
Upgrading changes, as I said above, are diachronic paths that feature instances of
the following: sentence grammar > periphery grammar; forms having syntactic
grammatical status > forms having less grammatical status; bound forms > free
forms; grammatical words > lexical words; univerbation > deverbation; morphol-
ogy > syntax; syntax > discourse; narrow scope > wide scope.
The change from full movement verb andar > subjective/intersubjective dis-
course marker of exhortation and confirmation ándale exemplifies this second
type of directionality paradigmatically.6 Examples (7)–(11) show the diachronic
continuum through which the movement verb andar gradually evolved into a
discourse marker. The process is a change by (inter)subjectification. The change
exemplifies an exit from the sentence grammar and an entrance into the periphery
grammar.
In (7a)–(7b) andar ‘to walk’ is a full movement verb; it takes animate human
subjects with an agentive role, such as Blasillo in (7a) and the native mining men
in (7b); these subjects displace to a locative goal that is an argument of the verb,
such as a la escuela ‘to school’ in (7a) and a ellas ‘to them [mines]’ in (7b).
5. Company & Espinosa (2014) analyze the relationship of this change to the general structura-
tion of demonstrative adverbs in Old Spanish.
6. This change and the next one are analyzed extensively, for other theoretical purposes, in
Company (2006, 2008).
The example in (8) shows that the directive locative goal has been weakened and
replaced by a stative location, aquí ‘here’, lessening the meaning of movement or
displacement towards a goal entailed by the verb. The subject also weakens indi-
viduation, because uno ‘one’ is an indefinite or impersonal pronoun. Example (8)
is the bridge context between the referential meaning of andar ‘to go’ and the dis-
course subjective meaning seen in (11) below. A bridge context, following Heine
(2002), means that a form or a construction may have two readings simultane-
ously, the conservative one and the innovative one; (8) is ambiguous between a
referential interpretation and a non-referential one.
(8) ¿qué cree, que uno anda aquí por su puro gusto?
what think.3sg that one go.3sg here for their just sake
‘What, do you think one is here for one’s health?’
(CORDE: 20th c., Mariano Azuela, Los de abajo)
Examples (9) and (10) below represent a further step in the weakening of the ref-
erential movement meaning of andar ‘to walk, to go’. The presence of a reflexive
clitic, se, in se anda makes the subject itself the goal of movement; thus, there is
not a locative argument. In other words, the locative argument is cancelled, and
because of that, the verb loses its movement meaning. In (10), the clitic contin-
ues being present, as in te anda and le anda, but there is no syntactic subject: the
locative goal has been replaced by an abstract goal, a human being. This is seen in
the pronoun a ti ‘to you’, understood from the context in (10a), or al niño ‘(to) the
child’ in (10b). The clitics, te and le, in (10) are still pronouns, which anchors their
reference to nominals.
Finally, the form ándale in (11) is the switch context (Heine 2002) that shows an
upgrading in the cline with an entrance into the periphery grammar: (11) only has
one interpretation, as a discourse marker. In (11) the verb shows syntactic rigidi-
fication, meaning it can only appear in third person singular anda, and the dative
clitic is an obligatory le; andar ‘to go’ is no longer a verb, but rather a discourse
marker having both an intersubjective (11a) and a subjective (11b) meaning. The
construction verb + dative clitic form a fixed expression, creating an autonomous
predication. This expression must appear alone, usually at the beginning or at the
end of the utterance, increasing in scope. Both the verb and the clitic lose their
relational capacity. The meaning of the verb no longer has to do with movement,
and thus it may co-occur with another verb of movement, vete ‘go’, as in (11a); this
co-occurrence means that the original etymological referential meaning of andar
‘to walk’ has been completely weakened. The upgrading in the cline is completed:
the construction leaves the sentence grammar domain and begins to function in
the periphery grammar.
(11) a. Y al ver a Buenaventura bien sport, le dio un tirón a su
and by see.inf to Buenaventura so sporty, dat give.3sg a tug to his
corbata y ordenó a uno de sus ayudantes: “¡ándale, vete
tie and ask.3sg to one of his assistants: walk.imper-to.it go.imper
por otra camisa y una chamarra!
for another shirt and a jacket”
‘And when he saw Buenaventura [looking] so sporty, he pulled on his tie
and yelled to one of his assistants, “Come on! Go get me another shirt
and a [leather] jacket!”’ (CREA: 21st c., Mexican newspaper)
The phrase “neither down nor up” describes changes that produce neither
stronger nor weaker grammatical ties and do not involve any movement between
language levels. This type of grammaticalization comes about without exiting the
sentence grammar. Usually, these processes maintain the same introductory form
or construction but change the distribution radically and the meaning along with
it.
These changes are neutral as to directionality. They can involve changes such
as the following: the spread of prepositions to new contexts and distributions
without changing their grammatical status as prepositions; the spread of some
conjunctions into new distributions without changing their grammatical status as
conjunctions; displacement from temporal to concessive, conditional, or causal
meanings; displacement from locative to temporal meanings; etc.
The change from temporal mientras ‘while’ > conditional mientras ‘if ’ is a par-
adigmatic example of this third type of directionality – or, more properly, non-
directionality. Examples (12)–(14) show the diachronic continuum of the entire
change. The process is a change by subjectification.
Latin temporal correlative conjunction-adverb dum ‘while’ + interim ‘in the
meantime’ is the etymon of Spanish mientras (12a). By a process of attraction in
Latin, interim began to be used next to dum. One of these clauses lost its intro-
ductory conjunction, meaning it was interpreted as the main clause. The clause
introduced by dum interim was subsequently interpreted as a subordinate tempo-
ral clause (12b). Latin particles underwent univerbation, giving rise to the Spanish
temporal conjunction mientras ‘while’ in (12c). The subordinate clause introduced
by mientras in (12c) is co-existential or simultaneous with the temporal line of the
predication denoted in the main clause. The indicative mood is proof that mien-
tras is factual in nature; it expresses something that is true in the world in which
the reference of the events is framed. In the temporal factual use of simultane-
ity, the subjects of both clauses are usually human – for example, los otros griegos
‘the other Greeks’ and estos quatro ‘these four [kings]’ in (12c) – and they are fre-
quently expressed in the third person.
(12) a. dum quoquetur… interim potabimus
while cook.3sg.pas meanwhile drink.1pl
‘While [the food] is being cooked, in the meantime we will drink.’
(Plautus, apud Lodge 1971: s.v. interim)
b. dum interim quoquetur, potabimus
while in the meantime cook.3sg.pas drink.1pl
‘While the food is being cooked, we will drink.’
(Late Latin, apud Corominas 1980–1991: s.v. mientras)
c. E entre tanto legaron ý el rrey Dohas e Ajax Thalamon
and meanwhile arrive.3pl there the King Dohad and Ajax Thalamon
e Agamenon e Menelao, e mientras los otros griegos lidiauan
and Agamenon and Menelao, and while the other Greeks fight.3pl
con los troyanos, tomaron estos quatro el puerto
with the Trojans, take.3pl these four the port
‘And in the meantime King Dohas and Ajax Thalamon and Agamemnon
and Menelaus arrived there, and while the other Greeks fought the Tro-
jans, these four assaulted the port.’
(CORDE: 13th c., Anonymous, Historia troyana)
Mientras in (13) introduces sentences which have the verb in the subjunctive
mood. The predication refers to a circumstance occurring later than the moment
of reference. The use of the subjunctive mood weakens or cancels the referential
reading of simultaneity, giving a non-factive reading, because the subject has not
carried out the action of the verb. The subjunctive mood also weakens the tem-
poral meaning of mientras. Examples in (13) are the critical or the bridging con-
text between the temporal etymological meaning of simultaneity in (12c) and the
innovative conditional meaning in (14) below.
(13) a. Un dia pensaron el gallo y el gato comerse la comida mientras
one day think.3pl the rooster and the cat to-eat the food while
estuviera en misa la beata
be.3sg at Mass the pious-woman
‘One day the rooster and the cat thought about eating the food while the
pious old woman was at Mass.’
(CORDE: 20th c., Anonymous, Cuentos extremeños)
Empirical evidence for round trip directionality is presented in (15).7 The example
in (15a) comes from Octavio de Toledo y Huerta (2002). Examples (15b) and (15c)
will be analyzed in the following sections in depth. The three examples provide
evidence for the hypothesis that round trip directionality forms a structural pat-
tern in language change.8
(15) a. (i) Full movement verb vaya: vaya a la escuela ‘go to school’ > (ii) Sub-
jective discourse marker of surprise / disgust ¡vaya!: ¡Vaya!, no me sabía
yo esas mañas ‘well [lit. go]! ‘I wasn’t aware of those bad customs’ > (iii)
Adjective vaya: vaya nochecita que pasé ‘what a night I had (without
sleeping)’
b. (i) Full ditransitive verb decir ‘to tell / to say’ + full subordinate comple-
tive sentence > (ii) Coalescence of verb + conjunction que > (iii) Eviden-
tial discourse marker of doubt dizque > (iv) Adjective dizque.
7. In Spanish there are no cases of the reverse round trip directionality PG > SG > PG. I do not
know why at this moment, but I advance a possible reason: the forms in PG are derived or sec-
ondary categories, in that PG is always coded with forms of SG, at least in Spanish, with the only
exception being primary interjections: oh, ay, uy. For that reason, it seems somewhat unlikely
that a form would move into PG, then after that become a SG form, and then after that to leave
its anchor status again to reach a new PG status.
8. There is no documentary evidence for interpretations other than that of a round direction-
ality. For instance, it might be possible to think about a retention or a latent niche of meaning
for an adjective, one that developed prior to or in parallel with the conventionalization of a dis-
course marker, but that possibility has never been attested.
Some aspects of this directionality must be commented upon. First, the result of
the round trip diachronic path is always an adjective. The reason for this seems
to be that the original source is a full verb; because of this, the new forms, both
in PG and in SG’, retain predicative force, like they are retaining a “memory” of
their original category. Adjectives in Spanish are nominal forms capable of having
predicative meaning. Second, the new adjective in SG’ has a fixed order: it comes
before the noun, which is not marked for quantifier or evidential order in Spanish,
but which is marked for general adjective order. This restriction in order, in my
opinion, results from the fact that the new forms are a derived category, undergo-
ing many changes in meaning and distribution during the long trajectory, before
reaching the category status of adjective.
4.4.1 Full (di)transitive verb decir ‘to tell’ / ‘to say’ > evidential discourse
marker of doubt dizque ‘supposedly’ > adjective dizque ‘supposed’
The particle dizque comes from the information verb decir ‘he/she/it tells, he/she/
it says’, plus the conjunction que ‘that’ which introduces a completive clause that
functions as the do of the main verb decir ‘to tell, to say’. This type of composite
clause is rooted in Latin and is documented in all stages of Spanish, as the exam-
ples in (16) show.
(16) a. Et por ende, amenazandol, dize que por la ssu mallat
and consequently threaten.ger-acus ay.3sg that for the his wickedness
es por derecho juyzio condenpnado
be.3sg by fair law condemned
‘And by that, threatening him, [he] says that because of his wickedness
he is condemned by fair law.’ (CORDE: 13th c., Alfonso X, Setenario)
b. y con juramento que de él se ha tomado dice que vio
and with oath that of him refl take.3sg say.3sg that see.3sg a
tanto, que cree que [la nave] iba lastrada dello
lot that believe.3sg that [the ship] be.3sg ballasted for-that
‘And by oath, he says that the ship was so filled with merchandise that it
was ballasted by that reason.’ (CORDE: 16th c., letter)
c. ¿Ostí dice que pescador es maleante; Ostí, chofer?
Ostí say.3sg that fisherman be.3sg bad-man Ostí driver
‘Does Osti say that the fisherman is a bad man?’
(CORDE: 20th c., José María Arguedas, El zorro de
arriba y el zorro de abajo)
Even from far back, as seen in (17), an abbreviated or eroded form is attested, diz
‘he/she/it tells, he/she/it says’. This form is followed by the conjunction que, either
with graphic separation as in (17a) or with coalescence of the two forms as in (17b),
with both constructions preserving the value of two clauses. Formally, examples
in (17) represent the source stage of the subjectification process by which this evi-
dential marker was recruited in Spanish.
(17) a. Et otrosí o diz que departiesen la luz de las tiniebras sse
and else or say.3sg that separate.3pl the light from the shadows pron
entiende que departieron
understand.3sg that separate.3pl
‘And else, where [God] says that the light is separated from the darkness,
it is understood that it occurred.’ (CORDE: 13th c., Alfonso X, Setenario)
b. En el cclxxxviiiº capitulo, que fue a los xxxv annos, dizque
in the 287 chapter that be.3sg at the 35 years say.3.sg-that
aviendo el rey don Ramiro contienda con el rey don Bermudo,…,
have.ger the king don Ramiro conflict with the king don Bermudo
adolecio en Leon e morio
fall.3sg-ill in Leon and die.3sg
‘In the 287 chapter, at the 35 years of his reign, [the book] says that when
King Ramiro fought King Bermudo … the first one fell ill in the city of
Leon and died.’ (CORDE: 14th c., Don Juan Manuel, Crónica abreviada)
In Examples (16) and (17) above, the verb dice/diz functions as a full verb, display-
ing a subject – either an animate subject, as in the case of ‘he’ in (16a) and (16b),
‘God’ in (17a), and Osti (a proper noun) in (16c); or an inanimate subject, as in the
case of ‘the book’ in (17b) – and a do clause. The verb preserves its etymological
referential meaning of information transfer in these examples.
Semantically, examples such as (18) must have been on the path to or an inter-
mediate stage in the progress toward the subjectification of dice + que as an evi-
dential marker. The verb is encoded in the impersonal form se dice ‘one tells, one
says’, which indicates that it is both possible that everyone might say it and that no
one is responsible for saying it.
(18) a. En Seçilia se dice que hay una sal que, echándola en
In Sicilia pron say.3sg that have.3sg a salt that throw.ger-acus in
el fuego, se desface é torna en agua.
the fire pron fade.3sg and turn.3sg into water
‘In Sicilian it is said that there is a kind of salt that becomes water,
putting it into fire.’
(CORDE: 15th c., Pero Díaz de Toledo, Diálogo e razonamiento
en la muerte del marqués de Santillana)
From the 16th to the 19th centuries, the meaning of the sequence diz que/dizque
vacillated amongst being a predication with two clauses, being an evidential
expression of uncertainty, and having both readings simultaneously; this can be
seen in (19). Such ambiguous expressions were very common in Spanish, and they
constitute the bridge context between the grammatical clause and the discourse
evidential marker; that is, the examples have two possible readings simultane-
ously, as a referential full verb and as an evidential non-verb form.
(19) a. Levantaronse çiertos indios en la provjnçia de Guaxaca,
stand up.3sg-pron some Indians in the providence of Oaxaca
diz que con acuerdo del dicho Pedro de Alvarado.
say.3sg that with accordance of-the mentioned Pedro of Alvarado
‘Some Indians stood up in the province of Oaxaca, (they) say-that (this
was) probably in accordance with the mentioned Peter of Alvarado.’
(CORDIAM: 16th c., Mexican letter)
b. que la tal Francisca Chicuagem le havía dado un azeite a
that the such Francisca Chicuaguem dat have.3sg given an oil to
Antonia. Dizque lo havía untado al otro su marido.
Antonia. say.3sg-that acus have.3sg put to-the other her husband
‘Francisca had given an oil to Antonia. (they) say-that (this was) appar-
ently to put on her husband.’
(CORDIAM: 17th c., Mexican legal document)
The particle que ‘that’ in (19) above introduces a clause that could function as do,
at least semantically; however, from a syntactic point of view, this clause is not
properly an object. It does not admit substitution by the neuter accusative clitic
lo ‘it’, a substitution that constitutes a classic test to identify a do in Spanish. In
addition, there is no identifiable subject of the verb diz in (19). This lack of sub-
ject paved the way for the acquisition of a discourse value of evidentiality. Via the
utterance introduced by diz que ‘say-that-supposedly’ in (19), the speaker-writer
communicates uncertainty about the event stated in the preceding sentence; they
are communicating that “everybody supposedly says that, but to me (the concep-
tualizer) there is no evidence.”
Throughout the 16th and 17th centuries, it is common to see dizque ‘suppos-
edly’ co-occuring with decir ‘to say’, as in (20). Those contexts are the proof that
the change from full ditransitive verb decir ‘to tell / to say’ + que ‘that’ to evidential
discourse marker is completed. These examples serve as proof that dizque ‘suppos-
edly’ has no more informational meaning; because of that, it can appear adjacent
to the full verb dice ‘says’, which is the etymon of the new evidential marker. Diz is
no longer a verb, and que is no longer a conjunction.
(20) a. y me levantó testimonio que dizque dixe yo quándo dizen
and acus raise.3sg accusation that supposedly say.1sg I when say.3pl
los artículos en la yglesia, que así como dizen el primero
the articles at the church that like-that like say.3pl the first-one
creer en un solo dios todopoderoso, que disque dezía yo
believe.inf in a one god almighty that supposedly say.1sg I
en lugar desto,… echo a dios en el fuego.
instead of-this put.1sg a god into the fire.
‘And he raised me a false accusation (he falsely accused me) that suppos-
edly I said that… instead that supposedly I said I put god into the fire.’
(CORDIAM: 16th c., Mexican legal document)
b. Esperen, aguarden, / que yo lo diré. / Porque, como dizque /
wait.imp hold on.imp that I acus tell.1sg because like supposedly
dice no sé quién, / ellas sólo saben / hilar y coser.
say.3sg neg know.3sg who they only know.3pl spin.inf and sew.inf
‘Wait, wait, I’ll tell it. Because, supposedly somebody says that…’
(CORDIAM: 17th c., Mexico, Sor Juana Inés de la Cruz, Villancicos)
Examples (21a)–(21c) show the last stage of the upgrading process; the form diz
‘say’ and the conjunction que ‘that’ undergo univerbation, forming dizque, and
the construction then becomes an unanalyzable form. The two formatives create
a new word, an evidential particle indicating the speaker’s uncertainty and doubt
with regards to the preceding discourse with a meaning close to ‘supposedly’.
Dizque often stands alone at the beginning or at the end of an utterance, isolated
from it by a pause (comma) and followed by a sustained suspensive intonation,
as seen in (21c). It is worth noting that the form dizque constitutes a lexical entry
in present-day Spanish dictionaries (see, for instance, Real Academia Española
& Asociación de Academias de la Lengua Española’s dictionary, 2013: s.v. dizque).
The directionality at this stage is an up in the cline, from sentence grammar to
periphery grammar.
Once the evidential discourse marker dizque ‘supposedly’ had come into being,
the form, which maintained the previous subjective meaning, came back to the
sentence grammar as an adjective, as examples in (22) show. The form dizque ‘sup-
posed’ began to operate as a full adjective: it can be found modifying nouns, the
nucleus of a full noun phrase, such as ley ‘law’ in (22a), imagen ‘look’ in (22b), and
comisionados ‘committee members’ in (22c); also, it appears in a distribution com-
mon for some adjectives in Spanish, following determiners such as la ‘the’ in (22a),
una ‘an’ in (22b), and sus ‘their’ in (22c). However, as a remnant of its previous life
as a verb and discourse marker, the new adjective has restrictions, occurring only
to the left of the nominal nucleus. The meaning of this new adjective is evalua-
tive, placing doubt on the referential features of the modified noun: e.g., dizque ley
implies “it is not a true law,” dizque imagen juvenil implies “it is not a true young
look,” and dizque comisionados implies “they are not true union commissioners.”
As many adjectives in Spanish, dizque may also modify other adjectives, as in the
case of dizque altruistas ‘supposedly altruistic’ in (23).
(22) a. los transeúntes viles, amparados por la dizque ley, solían correr
the passer-by vile protected by the supposed law use.3pl run.inf
tras el ladrón.
behind the thief.
‘Vile people, protected by the supposed-bad law, used to run…’
(CREA: 20th c., Fernando Vallejo, La virgen de los sicarios)
The directionality at the stage represented by (22) and (23) is a down in the cline,
from periphery grammar to sentence grammar; however, it takes the form of a
new grammatical category, adjective. This is different from the category of its ety-
mon, which was a verb + conjunction.
In summary, the directionality of the change as a whole is a round trip. First, it
exits from the sentence grammar, and then second, it is reinserted into it. In both
processes, it is involved in creation of new categories via reanalysis. The process is,
therefore, as follows: (i) Full (di)transitive verb decir ‘to tell / to say’ > (ii) Eviden-
tial discourse marker of doubt dizque ‘supposedly’ > (iii) Adjective dizque ‘sup-
posed’ ‘not true’.
4.4.2 Noun Phrase + relative sentence que + copulative sentence > Discourse
marker of evidentiality quesque > Adjective quesque
This change follows the same diachronic path as the previous one. Quesque dis-
plays a round trip diachronic path similar to that of dizque ‘supposedly, not
true’ and that of vaya ‘go’ in Example (15a) above. Examples (24)–(26) show the
diachronic continuum by which two sentences gradually evolve into a discourse
marker of evidentiality,9 with this discourse marker later reinserted into the sen-
tence grammar as an adjective.10
Example (24) shows the source construction of the new discourse particle
quesque ‘apparently’ and the new adjective quesque ‘not true, not totally’. The new
form comes from que + es + que, which are in fact two clauses; the first is a rela-
tive clause introduced by the relative pronoun-conjunction que ‘that, which’, hav-
ing a nominal antecedent fiestas ‘celebrations’. The first predications are fiestas del
Yahuayra, que … ‘celebrations of Yahuayra that … ’ in (24a) and mi primera peti-
ción que … ‘my first request that … ’ in (24b). The second clause is a copulative
clause; this can be seen inserted into the relative clause, as seen in es que en ella
pedían … ‘is that in it they prayed for … ’ in (24a) and es que declare los otros
vezinos que … ‘is that he tells the other neighbors that … ’ (24b). The utterance
introduced by the second que ‘that’, which follows the verb es ‘to be’, functions as
the subject of the copulative verb. In (24a), que en ella pedían … ‘that in it they
prayed for … ’ is the subject of es ‘is’ (24a); equally, in (24b) que declare los otros
vezinos … ‘that he tells the other neighbors that … ’ is the subject of es ‘is’. The two
predications, relative + copulative, inform or explain the reader what kind of fies-
tas ‘celebrations’ are, what kind of petición ‘request’ is.
(24) a. Y en este mes haçían las fiestas del Yahuayra, que
and in this month make.3pl the celebrations of-the Yahuayra which
es que en ella pedían al Haçedor que… produxiesen bien en
be.3sg that on it ask.3pl to-the Maker that… produce.3pl well on
aquel año y que fuese próspero.
that year and that be.3sg prosperous
‘In that month they made celebrations of Yahuayra, which is that they
asked for good harvest times.’ (CORDIAM: 17th c., Peruvian chronicle)
9. The diachronic attestation of quesque is harder to track than dizque in old texts, because
quesque is considered more colloquial than dizque; because of that, the documentation in old
Spanish texts is very scant.
10. For Aikhenvald (2004: Chapter 1), dizque is the only true evidential marker in Spanish; she
does not attest quesque. However, there is no doubt that quesque is another evidential marker,
one that is amply documented in many American Spanish dialects. This article shows in passing
empirical support for evidentiality.
Example (25) displays the new evidential marker quesque. The three forms que-es-
que undergo univerbation into quesque ‘apparently, supposedly’. The whole con-
struction constitutes an unanalyzable form: quesque is no longer the sum of two
sentences, there is not a copulative verb, and there is no longer a relative pro-
noun conjunction or a second conjunction. Rather, the three formatives create a
new word, an evidential particle indicating the speaker’s uncertainty and doubt
with regards to the preceding discourse, with a meaning close to “supposedly” or
“apparently” – essentially, “others could say that, but I doubt it.” At this stage, the
directionality is an up in the cline.
(25) El primer día que fuimos, quesque a posar, como dicen, fue, ora
the first day that go.1pl supposedly to pose.inf as say.3pl be.3sg now
verá usté, el lunes.
see.2sg you the Monday
‘The first day we went, supposedly to pose, as they say, it was on Monday.’
(CREA: 20th c., Eladia González, Quién como Dios)
Finally, once the evidential discourse marker quesque ‘apparently’ had come into
being, the form, which maintained the previous subjective meaning, came back to
the sentence grammar as an adjective; this is shown in (26). The form quesque ‘not
true, not actual’ works as a full adjective: it appears modifying nouns that serve
as the nuclei of full noun phrases, such as especialista ‘specialist’ in (26a), hombre
‘man’ in (26b), and siesta ‘nap’ in (26c); also, it appears in a distribution typical of
certain adjectives in Spanish, occurring after a determiner such as el ‘the’ in (26b),
un ‘an’ in (26a), and una ‘an’ in (26c). As a remnant of its previous verbal and dis-
course life, the new adjective is restricted in its placement: it can only occur to the
left of the nominal nucleus. The meaning of this new adjective is evaluative, plac-
ing doubt on the referential features of the modified noun, such as quesque espe-
cialista which carries the meaning of “not a true specialist,” quesque muy hombre
which means “not a true man,” and quesque siesta, “not an actual nap.” As with
The directionality at this stage is a down in the cline, from periphery grammar to
sentence grammar; however, in this stage, it takes the form of a new grammatical
category – adjective – that is very different from its two source clauses.
In summary, the directionality of this change is a round trip. First, the form
exits from the sentence grammar and enters into the periphery grammar; then,
second, it is reinserted into the sentence grammar. Both processes result in the
creation of new categories via reanalysis. The process is as follows: (i) Noun
Phrase + relative sentence are introduced by the pronoun que + copulative sen-
tence + conjunction que > (ii) Form becomes a discourse marker of evidentiality
quesque ‘apparently’ > (iii) Form becomes adjective quesque ‘not true, not actual’.
The proof that all forms and constructions found in (15) complete a round trip –
up and then down in the cline, SG > PG > SG’ – is that all of them exhibit para-
digmatization, which is suggestive of grammaticalization (Lehmann 1995 [1982]).
Paradigmatization is understood as the possibility of alternative choices with sim-
ilar meaning and distribution. The new adjectives coming from the periphery
grammar – vaya, dizque, and quesque – exhibit distribution like that of other
adjectives with similar meanings, forming a class with them, as (28)–(30) show.
(28) a. vaya cochazo ‘what a big car’
b. qué cochazo ‘what a big car’
c. menudo cochazo ‘what a big car’
(29) a. la dizque ley ‘the supposed law’
b. la supuesta ley ‘the supposed law’
c. la aparente ley ‘the supposed law’
(30) a. un quesque especialista ‘a supposed/false specialist’
b. un aparente especialista ‘a supposed/false specialist’
c. un falso especialista ‘a supposed/false specialist’
5. Conclusions
recreation of the grammar of a language, from the core relations to the peripheral
relations, and again from the periphery to the core.11
The changes analyzed here have also demonstrated some well-known aspects
of grammaticalization. Grammaticalization is a cumulative or stratified process:
the old or conservative form/meaning and the new or innovative form/meaning
may live together for centuries. Grammaticalization is the major factor responsi-
ble for the creation of new categories in language; additionally, it is asymmetrical
in direction, as even in the round trip cline it is irreversible and follows specific
paths or clines.
Finally, the article has shown that the distributions of forms and constructions
play major roles as conditioning factors in the direction and results of grammat-
icalization. Distribution is as important as meaning. Context and specific distrib-
ution – i.e. forms in specific construction – are the locus at which directionalities
and diachronic processes are shaped.
Acknowledgements
Preliminary versions of this article were read at the International Conference Syntax of the
World’s Languages VII, at Universidad Nacional Autónoma de México, August 2016, and at the
23rd International Conference on Historical Linguistics, at the University of Texas, August 2017. I
am indebted to colleagues and audience for criticisms and insightful comments; I am indebted
also to two anonymous reviewers, as their comments improved the analysis. Errors are mine.
Abbreviations
11. An issue outside the scope of this article is why all directionalities except directionality
A produce subjective, intersubjective, epistemic forms, or, in a broad sense, modalized forms.
Modality here is understood according to the traditional meaning in Romance linguistics, i.e.
the integration of the speaker’s point of view in the utterance.
References
Joseph, Brian D. & Richard Janda, eds. 2003. The Handbook of Historical Linguistics. London:
Blackwell. https://doi.org/10.1002/9780470756393
Kaltenböck, Gunther, Bernd Heine, & Tania Kuteva. 2011. On Thetical Grammar. Studies in
Language 35:4.848–893. https://doi.org/10.1075/sl.35.4.03kal
Kuryłowicz, Jerzy. 1965. The Evolution of Grammatical Categories. Diogenes 51.55–71.
https://doi.org/10.1177/039219216501305105
Lehmann, Christian. 1995 [1982]. Thoughts on Grammaticalization. München-Newcastle:
Lincom Europa.
Lindström, Thérese. 2004. The History of the Concept of Grammaticalization. Sheffield
University PhD dissertation.
Lodge, González. 1971 [1901–1933]. Lexicon Plautinum. Hildesheim: Georg Olms.
Luraghi, Silvia. 2005. Does a Theory of Language Change Need Unidirectionality? Logos and
Language: Journal of General Linguistics and Language Theory 6:2.9–19.
Meillet, Antoine. 1965 [1912]. L’évolution des formes grammaticales. Linguistique historique et
linguistique générale by Antoite Meillet, 130–148. Paris: Honoré Champion.
Newmeyer, Frederick J. 2001. Deconstructing Grammaticalization. Language Sciences
23:2.187–229.
Norde, Muriel. 2009. Degrammaticalization. Oxford/New York: Oxford University Press.
https://doi.org/10.1093/acprof:oso/9780199207923.001.0001
Norde, Muriel & Karin Beijering. 2014. Facing Interfaces: A Clustering Approach to
Grammaticalization and Related Changes. Folia Linguistica 48:2.385–424.
https://doi.org/10.1515/flin.2014.014
Norde, Muriel, Karin Beijering & Alexandra Lenz. 2013. Current Trends in Grammaticalization
Research. Language Sciences 36.1–6. https://doi.org/10.1016/j.langsci.2012.03.015
Octavio de Toledo y Huerta, Álvaro S. 2002. ¿Un viaje de ida y vuelta? La gramaticalización de
vaya como marcador y cuantificador. Anuari de Filologia 11–12.47–72.
Olmen, Daniël van, Hubert Cuyckens & Lobke Ghesquière, eds. 2017. Aspects of
Grammaticalization, (Inter)Subjectification and Directionality. Berlin: De Gruyter
Mouton.
Real Academia Española & Asociación de Academias de la Lengua Española. 2009. Nueva
gramática descriptiva de la lengua Española. Madrid: Espasa.
Real Academia Española & Asociación de Academias de la Lengua Española. 2013. Diccionario
de la lengua Española. Madrid: Espasa.
Torres Cacoullos, Rena. 2016. La gramaticalización. Enciclopedia de lingüística hispánica ed. by
J. Gutiérrez-Rexach, 504–514. London: Routledge.
Traugott, Elizabeth. 2001. Legitimate Counterexamples to Unidirectionality. Paper presented at
Freiburg University, Freiburg, October 17, 2001. Available at www.stanford.edu/traugott
Traugott, Elizabeth. 2003a. From Subjectification to Intersubjectification. Motives for Language
Change ed. by Raymond Hickey, 124–139. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511486937.009
Traugott, Elizabeth. 2003b. Constructions in Grammaticalization. In Brian D. Joseph &
Richard Janda, eds., 624–647.
Traugott, Elizabeth & Richard Dasher. 2002. Regularity in Semantic Change. Cambridge:
Cambridge University Press.
Traugott, Elizabeth & Graeme Trousdale. 2013. Constructionalization and Constructional
Changes. Amsterdam/Philadelphia: John Benjamins.
https://doi.org/10.1093/acprof:oso/9780199679898.001.0001