INTRANSITIVE COPY PRONOUNS, MENTAL SPACES, AND THE ACCESS PRINCIPLE:
A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
by
KYLE A. HUBBARD

Thesis
Submitted in Partial Fulfillment of the Requirements for
the Degree of

MASTER OF LINGUISTICS
in the
FACULTY OF GRADUATE STUDIES
TRINITY WESTERN UNIVERSITY

August 2023
© Kyle A. Hubbard, 2023

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

i

Contents
1. Introduction ............................................................................................................................... 1
1.1 Intransitive Copy Pronouns .................................................................................................. 2
1.2 Defining the middle .............................................................................................................. 4
1.3 Mental spaces and the Access Principle ............................................................................... 8
1.4 Key terms and cognitive linguistics .................................................................................... 12
2. Characterizing ICPs from cognitive linguistics ................................................................... 15
2.1 Building a theory from [ + ICP ] middle verbal events ...................................................... 16
2.1.1 ICPs and their relationship to middle verbs ................................................................. 16
2.1.2 ICPs, mental space structure, and the transfer of FORCE.............................................. 20
2.1.3 ICPs and their relationship to other participants .......................................................... 25
2.2 Extending the theory to cover non-middle verbal events ................................................... 30
3. Analyzing the purported functions of ICPs .......................................................................... 36
3.1 Countering the purported Cognate Object (CO) function .................................................. 38
3.2 Explaining the purported Subject Point-of-View (SPOV) function ................................... 43
3.3 Establishing the Disjuncture Navigation (DN) function..................................................... 44
3.3.1 The DN function and semantic effects ......................................................................... 46
3.3.1.1 ICPs and totality .................................................................................................... 49
3.3.1.2 ICPs and mirativity ............................................................................................... 49
3.3.1.3 ICPs and inchoativeness ....................................................................................... 50
3.3.2 The DN function and discourse ................................................................................... 51
4. Conclusion ............................................................................................................................... 67
Abbreviations .............................................................................................................................. 70
Appendix ...................................................................................................................................... 71
References .................................................................................................................................... 88

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

1

1. Introduction
My intent in this thesis is to argue that the diverse functions of Intransitive Copy Pronouns
(ICPs) can be adequately explained with reference to normal cognitive processes. In particular, I
show that the Access Principle is the enabling element in a number of seemingly unrelated use
environments and functions associated with ICPs. I also show that ICPs can be used to help
navigate cognitive disruptions, which may account for what motivates their use. This all argues
against the common approach in the literature that treats ICPs as a special phenomenon with
divergent and often unexplainable functions and use-environments. Instead, I find it more
appropriate to analyze the ICP as a normal participant index used as a relatively straightforward
means of organizing participants and their conceptual world. In this thesis, I attempt to explain
ICPs using well-established cognitive linguistic principles, and in so doing hope to offer fairly
simple explanation for much of what makes ICPs seem strange and unusual.
In §§1.1-1.2, I introduce Intransitive Copy Pronouns (ICPs) and give my working
definition of “middles”. Doing so is especially necessary in light of the different definitions of
the middle voice. I then provide an introduction to mental spaces and the Access Principle in
§1.3, along with an introduction to some general cognitive linguistic principles. I also give
definitions for several terms I use in my analysis (§1.4). In §2, I build a characterization of the
ICP that draws from interspace relationships and the Access Principle. In particular, I look to the
way ICPs are used with middle verbs to show that [ + ICP ] constructions encompass a mental
space structure involving a single participant divided according to two different causal
perspectives. The fact that these constructions encompass different mental spaces, and the
existence of a close connection between the participants in those spaces, suggests the
applicability of the Access Principle. As I argue in the latter parts of §2.1, I believe the Access

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

2

Principle provides sufficient explanation for things like the apparent coreferentiality of ICPs and
their associated syntactic subject, without requiring a reworking of preestablished theory. I
follow this in §2.2 with an explanation of how the mental space structure of [ + ICP ]
constructions accounts for the use of ICPs with non-middle verbs (i.e., those not clearly fitting
the categories given by Kemmer (1993) ).
I apply my thesis to a variety of purported ICP functions in §3; here I leverage mental
space structure, the Access Principle, cognitive disjuncture, and the limits of working memory –
all elements of cognitive linguistics – to help explain why ICPs are used in certain environments,
and what motivates their use. This entails in part examining a purported functional connection
between ICPs and cognate objects in §3.1 – a connection I counter on the basis of their different
derivations (despite the existence of superficial similarity in their cognitive structure). I also
examine the purported link between ICPs and subject point-of-view (§3.2) – which I largely
affirm – and show how my thesis both clarifies the nature of the participants involved and builds
a theory-grounded explanatory framework. Finally, I use my thesis to help expand the known
functions of ICPs (§3.3). I do so by showing how the unique cognitive structure of [ + ICP ]
constructions allows them to be used for navigating instances of change in the cognitive
environment. In so doing, I explain some of the various (and seemingly unrelated) semantic
effects associated with ICP use. I also give evidence suggesting the existence of a previously
unreported ICP use related to participant tracking and mental activation of referents in discourse.
1.1 Intransitive Copy Pronouns
Intransitive Copy Pronouns (ICPs) are a phenomenon found in several languages of West Africa.
Though they differ slightly across languages, prototypical ICPs are post-verbal indexes (i.e.,
pronominal suffixes) reflecting or re-referencing the referent of the pre-verbal subject index. For

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

3

this reason, they are sometimes called recapitulative pronouns (Storch 2009: 123; see also Koops
& Bendor-Samuel 1974). In some languages, they constitute an entirely separate referent
paradigm, distinct from both subject and object indexes; in others, they closely resemble direct
object or possessive pronouns. ICPs frequently occur with semantically intransitive verbs; as
such, their use creates a transitive-like construction. The following examples are a sampling
from across the literature:

1

(1)

nà pòrò-no
1SG go_out.AUX1-ICP
‘I went out.’
(Kanakuru,1 West Chadic; Newman 1971: 190)

(2)

sèm bè-tá
sáy
man came-ICP AUX.FOC2
‘The man came.’
(Miya, West Chadic; Schuh 1989: 173)

(3)

tè
mùràn-rò
3SG.F died-ICP
‘She has died.’
(Widala,3 West Chadic; Jungraithmayr & Leger 1993: 169)

(4)

à-nzá-n-kà
3SG.M-run-ICP-PRF
‘He ran.’
(Gidar, Biu-Mandara; Frajzyngier 2008: 138; in Anderson 2011: 146)

Under the earlier name Dera.

2

Early work glossed this as a totality extension. I have glossed it as an auxiliary focus marker under the influence
of more recent research, especially Schuh (2017).
3

Also known as Kholok.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

4

Constructs similar to ICPs have been noted as far back as Welmers (1948) and Samarin (1967).
The term “Intransitive Copy Pronoun”, however, was not used till Newman (1971). He based
this term on the observation that ICPs in Kanakuru (West Chadic, Nigeria) are an obligatory
suffix occurring with intransitive verbs in certain tenses, which a) reflect subject information,
and b) render a transitive-like construction. This general analysis was extended by Frajzyngier
(1977), who noted that ICPs in Pero (West Chadic, Nigeria) occur only in non-stative
constructions without an object, and that they render an inchoative sense.
Some of these earliest observations about ICPs and how they function have since been
amended or greatly expanded. A significant factor in this has been the discovery of ICPs in
Benue-Congo languages. Some of these languages exist within a Sprachbund in Central Nigeria
alongside genetically unrelated West Chadic languages, which are well known for having ICPs.
This motivates the theory that ICPs spread as an areal feature (Wolff & Gerhardt 1977; Storch
2009; Storch, Atindogbé & Blench 2011; Hellwig 2011) within and around the Central Nigerian
plateau, though the existence of ICP-like structures in Narrow Grassfields languages of
Cameroon, such as Ngamambo (Asongwed 1980) and Bafanji (Nicolle In press), suggests an
even wider distribution. ICP research continues to develop, and an increasing number of
phenomena are now analyzed as having formal and functional overlap with “classical” ICPs;
among them both copy pronouns (Storch, Atindogbé & Blench 2011; Atindogbé & Chibaka
2011) and focus pronouns (Koops 2011; see also the same focus pronouns under other names in
Shimizu 1980; Koops & Bendor-Samuel 1974; Noss 1981; Storch 1999).
1.2 Defining the middle
In §2.1 below I use the relationship between ICPs and middle verbal events to build a case for
explaining ICPs in terms of cognitive linguistic principles, chief among them the Access

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

5

Principle. There are at least two major approaches to defining the term “middle”. The first has
its roots in descriptive analysis of specific languages. It is a form-function approach, as it
considers “middles” (properly: morphosyntactic markers of middle voice) to consist of a group
of similar forms that share a common function (Bakker 1994: 23). This approach is particularly
evident in classical grammars, especially for languages like Ancient Greek; the grammars of both
Kühner (1898: 100) and Thompson (1902: 305–306) identify similar morphosyntactic forms,
grouping them together on the basis of their shared function (i.e., that of indicating self-affecting
verbal action). This form-driven morphosyntactic category is then used as a prototype for
assessing other morphosyntactic elements. From this perspective, a given morphosyntactic
element is a middle if it a) is similar in form to the categorical prototypes, and b) matches the
function of the categorical prototypes. Thus, Dixon and Aikhenvald (2000: 11–12) can argue
that “middle” ought to be reserved for cases that mirror the Ancient Greek middle. In this view,
there is no typologically valid “middle” consisting of dissimilar forms and strategies (such as
passives, reciprocals, reflexives, etc.) used for expressing self-affectedness, but rather only those
constituent elements (passives, reciprocals, etc.) as distinct, unrelated phenomena. This is
fundamentally the approach adopted recently by Inglese (2021); though he looks at a wide range
of languages, and though his conclusions are likely valid within the framework of his aims, the
definition of “middle” he proposes ultimately depends on starting with forms and then assessing
whether those forms have the particular functions he’s looking for.
An alternative perspective is that adopted by Kemmer (1993; 1994), who owes much to
the observations of Lyons (1968). In Kemmer’s perspective, self-affecting action is a semantic
category that can be grouped together into a language-independent, functional category with a
variety of formal manifestations. This is a function-form approach, since it interprets the middle

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

6

as a function expressed through a variety of forms, rather than as a group of similar forms
sharing a common function; as such, it has been associated with discourse-pragmatic and
typological aims (Bakker 1994: 23). This perspective is reflected in the approaches of Keyser
and Roeper (1984), Fagan (1988), Hardy (1994), LaPolla (1996; 2010), Amberber (2010),
Martin (2010), Rice (2010), and Campbell (2017), along with many others. The verbs of the
middle share a common function, though they may be dissimilar in form. This function can be
described in terms of two semantic properties, which I adapt from Kemmer (1993: 238). The
first is that the initiator of these middle verbal events is also an endpoint or “affected entity” of
the same verb. The second is that the verbs themselves are characterized by a low degree of
elaboration of events, which is to say that they are characterized as having a low degree of
distinguishability of the participants and verbal sub-events that make up the overall action
(Kemmer 1994: 211).
Despite challenges to this perspective – brought most recently by Inglese (2021)4 – I
continue to use Kemmer and her broad functional/semantic middle as my starting point. First,
my aim is more typological than descriptive, in that I’m trying to understand the function of ICPs
and how they fit alongside other systems with related functions; thus, I’m less interested in
deciding whether the ICP fits a certain prototype (or network of prototypes) than I am in

Inglese takes issue with the Kemmer’s relative imprecision. He argues a) that the process Kemmer uses to
individuate situation types is overly controversial, and b) that the semantic category of middles she uses in
determining whether or not morphosyntactic elements count as “middles” lacks an “explicit and operationalizable
characterization” (Inglese 2021: 493). He proposes instead a narrower definition of “middle marker” that accounts
for these two criticisms and draws from a wider and more diverse dataset. Yet while I acknowledge the grounds for
these criticisms, they don’t apply to my thesis. That’s because I am not arguing that ICPs are middle markers;
instead, I’m arguing that ICPs are a means of expressing a particular verbal participant’s relationship with the verbal
action. ICPs often occur with “middle” verbs, but they are theoretically acceptable even with other verb types, as I
show in later sections. ICP use is governed by normal cognitive processes relating to participant structure and
attention management, and ultimately has little overlap with Inglese’s work.
4

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

7

deciding whether the ICP functions similarly to both the prototypes and a range of other systems
that have similar function.5 Second, the function-form view of the middle has been widely
adopted in ICP research. Scholars as early as Jungraithmayr (1970) have explicitly analyzed
ICPs as middle constructs, despite dissimilarities between them and the Ancient Greek middle.
Frajzyngier (1989) connected ICPs with the reflexive in Romance and Slavic languages on the
basis of their “common semantic conditions” (Frajzyngier 1989: 119). This constitutes an
implicit function-form approach to typifying the ICP that later researchers, such as Leger and
Zoch (2011: 41), made explicit. Schuh (2005: 22) noted that ICPs affect valency in a way that
makes them “strikingly like middle reflexive pronouns”. Storch, et al. (2011: 7) link ICPs to
similar self-affective (i.e., functionally “middle”) constructs in English. Leger and Zoch (2011:
37) explicitly invoke Kemmer’s “middle” to explain ICPs, and note their use to mark selfinitiated and self-affecting verbs. My adoption of the function-form approach to middles fits
well with these and other observations made over several decades.
I do this with certain misgivings, however. My thesis supports the idea that languages
can approach participant organization with a great deal of fluidity. This makes sorting verbs into
neat categories of inherently “middle” and inherently “non-middle” very difficult, despite the
existence of criteria made for that purpose (i.e., Kemmer 1993: 16–20).6 We tend to categorize
“middle verbs” the same way we categorize “transitive verbs”, or “intransitive verbs”, which is
to say, we treat them as discrete points when in reality, they exist on a continuum whose

5

This typological aim combines with a general cognitive orientation to produce the typological-cognitive
perspective at the core of my thesis. For explicit discussion and adoption of this perspective, see Kibrik (2011: §1.7
and throughout).
6

I discuss some of these difficulties in §2.2.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

8

boundaries are language-specific and mutable. Nevertheless, ICP research has used the language
of “middles” for such a long time that it is difficult to enter the conversation without adopting
that same language. So when I talk about “middle verbs” in this thesis, that should be
understood as shorthand for “verbs that are often construed as involving the initiator as an
affected entity”. This follows the Kemmerian sense of middles and fits in well with other ICP
research. It also proves helpful on a practical level: it is fairly simple with these sorts of verbs to
show the mental space structure being profiled by the ICP, which also makes it easy to see the
cognitive relationships that allow ICPs to be used in all their various ways. By the end of §2.2, it
should be clear that ICPs do more than simply mark middle verbs, which constitutes gentle
pushback on some recent trends in ICP research (see Leger & Zoch 2011: 42, among others).
1.3 Mental spaces and the Access Principle
The Access Principle is a key tenet of Cognitive Linguistics that describes the relationship
between entities inhabiting different mental spaces. Before I expand on this, it is necessary to
explain mental spaces. Mental spaces are conceptual groupings of information built up during
discourse (Fauconnier 1985: 16). For the utterance “If I were a rich man…”, two mental spaces
are set up; one contains the speaker (i.e., the referent “I”, here labeled as “SELF”), and the other
contains the second referent that the first is being compared against (i.e., “a rich man”). This is
evident in the following example (note that R represents the space corresponding to the speaker’s
reality, while C represents the counterfactual reality introduced linguistically by the combination
of “if” and use of the past tense):

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

9

(5)
a1

a1: SELF

R

a2

a2: A RICH MAN

C

Note that the two participants a1 and a2 are linked, though they constitute entities within separate
conceptual groupings (that is, within the separate but linked mental spaces R and C). Each
mental space – and there may be several in existence at any given stage of discourse – is built
from a base space, the starting point of a discourse-oriented network of conceptual groupings
(Fauconnier 1985; 2007).
I return now to the Access Principle. The Access Principle states that “an expression that
names or describes an element in one mental space can be used to access a counterpart of that
element in another mental space” (Fauconnier 1997: 41). Accessing an element’s counterpart in
this way also grants access to any information already associated with that counterpart. And
since it relies on the same type of cognitive connection, this associated information can be
updated with new information as well. In essence, given two linked referents – a base referent
existing in the base space and an other element in an other space – we have two possible options
for adding information relevant to the base element and its space: the direct way and the Accessmediated way. The direct way is to explicitly express the base referent as it exists in the base

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

10

space, adding information relevant to the base space. The Access-mediated way is to explicitly
express the linked other element that exists in the other mental space, while still adding
information that is relevant to the base space. With the Access-mediated way, though each of
the linked referents exists in separate mental spaces, information related to a participant in one
mental space is added by accessing that participant from within the connected – but distinct –
mental space inhabited by the other element. This is illustrated in the following example, which
I derive from Evans and Green (2006: 375–378):

(6) a.

James Bond is a top British spy.

b. In the war, he was an officer in the Royal Navy.

c.

James Bond served on HMS Espionage.

a1

a1: NAME: JAMES BOND
PROPERTY: TOP BRITISH SPY

a2

a2: HE
PROPERTY: OFFICER
IN THE ROYAL NAVY

WAR
SERVING ON WARSHIPS frame
<OFFICER, ON SHIP>

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

11

The sentence in (6a) above sets up a mental space containing the referent James Bond, and
attributes a property to that referent (i.e., that he is a “top British spy”). I follow common usage
in leaving this space unlabeled, since there is no explicit information provided in building the
space inhabited by James Bond (such as “in Ian Fleming’s novels”, or “in today’s Britain”).
Sentence (6b) introduces a second referent, He, along with the explicit space builder “in the
war”; note that because of this space builder I’ve labeled the space containing He the WAR space.
James Bond and He, which are inferred as referring to the same “real” individual, are labelled as
a set of linked referents, a1 and a2, which are mapped to each other. The property of He that was
given in (6b) – i.e., that He was an “officer in the Royal Navy” – is added to the WAR space’s
frame information. Doing so arguably7 also prompts for the second set of WAR space frame
information that concerns the relationship between ships – a known element of the Royal Navy –
and those who serve aboard them. This frame information helps shape how the referent HMS
Espionage in (6c) is understood. Though this is a complex example, what is essential is that it
shows the Access Principle at work: information relevant to the WAR space (i.e., the identity of
the ship) which helps further describe the WAR space’s participant (i.e., He) is being added from
within the base space, by explicitly referring to James Bond. This works because referents a1 and
a2 are mapped, and because the Access Principle allows for the flow of information between
mental spaces through linked referents.

I say arguably because the exact point at which frame information is added isn’t always clear; this is similar to the
reality that lets mental spaces be built both explicitly and implicitly. It may be the case that the warship frame
information comes into existence when the Royal Navy is mentioned; it may also be the case that mention of the
Royal Navy prepares the hearer to interpret new information in a way that might fit any of a variety of frames
deemed pertinent to the Royal Navy property, in which case the frame only becomes manifest when information that
might fill it is explicitly mentioned. I make no claim either way, as it doesn’t affect the resulting interspace access.
7

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

12

When an entity in a higher space accesses a mapped entity in a lower, subordinate space,
we may call it downward linking, or downlink. The same process done from lower to higher
may be called uplink. No matter the direction of access, the referent that is explicitly mentioned
in order to access its mapped counterpart is known as the trigger; the element that is so accessed
is known as the target. In the above example, the trigger (a1) is accessing the target (a2) via
downlink, so the connection between the two referents is shown as a directional arrow. Another
way to describe this access process would be to say that the target is accessed from the
perspective of the trigger. “Perspective” here refers to the use of the trigger as the funneling
entity pushing information toward a linked space. I normally avoid talking about the Access
Principle in terms of perspective-orientation, as it enters murky territory with regards to how we
understand the relationship between participants and things like volition and intentionality. With
that said, however, I find it important to acknowledge the language of “perspective”, since it
plays a role in explaining some of the purported functions of ICPs that I discuss in later sections
(see especially the Subject Point-of-View (SPOV) function in §3.2).
1.4 Key terms and cognitive linguistics
Throughout this thesis, I use the term base participant or base referent to refer to the “real”
participant – i.e., the physical “subject”, as an actual, extant entity. This relates to the idea of
schematization, which is a less granularly detailed abstraction of a given element (chicken may
be a particular element; a schematized representation of this would be foul or even bird – note
that there may be multiple levels of schematization). I owe the term base to a number of sources,
among them Langacker (1987), whose base designates the foundation or scope of a predication
(Langacker 1987: 183). This base is the general context information providing the background
for understanding a more-prominent profile element (Langacker 1987: 118–119). The base and

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

13

profile together constitute the full picture of an expression’s semantic value (Langacker 1987:
183). “Base participant” also alludes to the related idea of a base space, which as I noted above
is the starting place of a discourse-oriented network of mental spaces. This base space is the root
mental space that helps frame the discourse, and it and its contents are accessible to entities
contained in other subordinate mental spaces. This accessibility is mediated by the Access
Principle. By speaking of a base participant or base referent I acknowledge the concept of
cognitive linking that undergirds both Langacker (1987) and Fauconnier (1985; 2007), along
with the idea that there is a “home space” – the base – from which other entities draw meaning,
and which themselves are also affected by their subordinate entities.8
I offer a fuller explanation of the participant structure of middle situation types in §2.1
below; in short, however, there is reason to believe that middle verbal events are often expressed
as involving two participants, one more agentive and the other less agentive. The more-agentive
participant is typically given titles like initiator (instigator, etc.), while the less-agentive is
typically called an endpoint (recipient, etc.). Initiator and endpoint are especially common
terms, and they help make the directionality of force transfer in a middle verbal event apparent.
However, I find they risk over-specifying the often-imprecise thematic role of the participants in
question. Furthermore, thematic roles are not static; rather, they may be modified by changing
how the verbal event is construed. Doing so ultimately affects how we interpret the level of
volition of the event’s participants, which plays a role in how the event is interpreted.9 Because

This focus on cognitive links and interdependency seems to be shared by others: see, for example, the “base
scenario” of Tuggy (2007: 105).
8

9

See especially §3.3.1 on the connection between ICPs and mirativity.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

14

of this, I find it useful to highlight the directionality of force transfer in a way that avoids terms
like initiator and endpoint that might point to more specific thematic roles.
There is little consensus concerning the nature, number, and hierarchical organization of
thematic roles. Numerous contradictory and mutually exclusive systems have been proposed
(see inter alia Fillmore 1968; Foley & Van Valin 1984; Givón 1984; Bresnan & Kanerva 1989;
Jackendoff 1990; Dowty 1991; Van Valin 1993; Goldberg 1995; Davis & Koenig 1998;
Culicover & Jackendoff 2005; Van Valin 2005). This makes assigning thematic roles to
participants in [ + ICP ] constructions difficult. Even assuming we can confidently determine an
appropriate hierarchical organization between thematic roles, it is unclear exactly how the
thematic roles in that hierarchy are affected by changes in event construal. This problem, which
has also been noted by Croft (1998; 2012) and Dowty (1991), becomes especially clear in cases
where there is a mismatch between a participant’s semantic role and the role they are construed
as having in a given utterance; an example of this would be cases where a semantic experiencer
is construed as an agent. Ultimately, this leads to confusion in how to identify participants in
middle events.
I sidestep these challenges by adopting a relative mapping approach, following Dowty
(1991), Levin and Rappaport Hovav (2005), and using markings derived from Foley (2007).
With this approach, I use [ + A ] to mark the more-agentive participant, and [ – A ] for the lessagentive participant. These markings relate to how verbal participants are perspectivized: either
as actor ( [ + A ] ) or as undergoer ( [ – A ] ).10 For transitive expressions in which both

Actor/undergoer – along with related terms like P(roto-)Agent/P-Patient, or the agonist/antagonist of Talmy
(1988) – are often given as discrete points in a globally valid hierarchical organization of roles. Evidence that such
roles are globally valid, or that any hierarchy exists beyond the local level, is mixed, and I prefer to avoid terms that
10

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

15

arguments are core arguments of the verb, the more-agentive participant will be expressed as the
[ + A ] participant and the less-agentive participant will be expressed as the [ – A ] participant.
Sorting participants into [ ± A ] poles makes it clear which participant is construed as the main
instigator while also indicating the direction of force transfer.11
2. Characterizing ICPs from cognitive linguistics
In this section, I show how ideas from cognitive linguistics – especially mental spaces and the
Access Principle – can be used to account for ICPs in a variety of environments. These same
principles, along with consideration of other cognitive categories like working memory and
attention management, explain what would otherwise seem to be completely unrelated functional
uses of the ICP. To establish the validity of describing ICPs in terms of mental spaces and the
Access Principle, I look first to middle verbal events, which are a common environment for
ICPs. In particular, I explore the participant structure of such events and argue that they
represent a splitting of the unary base participant into separate, force-differentiated mental
instantiations. I show how the relationship between these mental instantiations (i.e., the different
participants profiled in [ + ICP ] middle verbal events) is mediated by the Access Principle. I
then apply the same theoretical model to non-middle verbs, with the goal of validating the use of
cognitive linguistic principles to describe the form, and function, and use-environment of ICPs.

might be confused with that perspective where possible. As I argue above, relative agentiveness marking highlights
force transfer without making claims about particular roles.
Note that though in some cases “±”indicates the presence or absence of a given characteristic, here I use it in the
same way we mark electrical charge; for both electrical currents and transitivity relationships, FORCE flows from the
positive [ + ] pole to the negative [ – ] pole. Unlike an electrical current, both poles of the transitivity relationship
might be characterized by a high degree of agentiveness; nevertheless, one of those participants may be more
agentive (or may be construed as such), in which case FORCE flows conceptually from that pole toward the other.
11

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

16

2.1 Building a theory from [ + ICP ] middle verbal events
In this section, I use available literature to affirm the observation that ICPs are strongly (though
not exclusively) associated with middle verbs. This is important because it validates using
[ + ICP ] middle verbs as the starting point for my ICP analysis. In the following paragraphs, I
outline arguments from the literature establishing the association between ICPs and middle
verbs. I follow this by arguing that middle verbal events can be validly interpreted as
semantically transitive, regardless of whether or not they are coupled with an explicitly
expressed second participant. Given the nature of middle events, I also demonstrate from this
point that the two participants need not exist in a “real” sense; rather, they only need to exist as
mental instantiations of the base participant. From there, I do two things: first, I show how the
ICP represents a [ – A ] instantiation of the base participant, existing in a separate mental space
from the [ + A] instantiation; second, I show how the Access Principle mediates the relationship
between the ICP and its base referent. Finally, I return briefly to consider the way “real”
participants can be split into force-differentiated participants even in cases where there is
no obvious candidate for a semantic force-receiving participant. This entails examining
instances of ICPs used with non-middle verbs, and establishing why doing so is theoretically
valid, even if the motivation for doing so is still conditioned by grammatical or pragmatic
considerations.
2.1.1 ICPs and their relationship to middle verbs
In the literature, ICPs frequently occur with middle verbs of the following types, drawn from
Kemmer (1993); note that I group these into four functional classes of my own making:

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
Table 1 Middle situation typesa
Class

Situation type

17

Examples/comments

Motional

Translational motion
Non-translational motion
Changes in body posture

climb up; go/leave; walk/stroll; fly
stretch one’s body; turn; bow
sit down; kneel; lie down; rise/get up

Affective

Indirect/self-benefactive event
Naturally reciprocal event
Grooming/body care
Spontaneous event

acquire; lay claim to; ask/request; take for oneself
meet; embrace; wrestle; converse/agree
wash; get dressed; shave
germinate/sprout; come to a stop; become

Expressive/cognitive

Emotive speech action
Other speech actions
Cognition
Emotional reaction

complain; lament
confess; admit one’s guilt; be boastful; boast
think; consider; ponder/meditate; believe
become frightened; be angry; grieve/mourn

Referential

Logophoric reference

Marks participants in a dependent clause that are
coreferential with subject of main verb
Marks subject as corresponding to patientive
participant of unmarked root verb

Passive/impersonal/facilitative
a

All examples taken from Kemmer (1993: 16–20)

Kemmer’s classifications are more granular than required for my purposes; in particular, the
distinction she makes between “emotive speech actions” and “other speech actions” and between
emotional reactions and cognition seems unclear. I suspect that the ostensible differences
between them are simply variant outworkings of a single underlying principle. This explains
why I group the different middle situation types together into the four functional classes given
above. These functional classes are intended to guide my interpretation of the role played by a
given middle situation type in communication.
My work in this section is indebted to Leger and Zoch (2011), who examined data from
several decades of ICP research, and found examples of ICPs with at least six of Kemmer’s
thirteen middle situation types, covering the motional, affective, and expressive/cognitive
classes. In a quick survey of other works, I find evidence of three additional situation types.
These are: cognition – forget, in Storch and Coly (2014: 6); non-translational motion – turn, in

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

18

Schuh (1998: 178); and other speech acts – whisper, in Schuh (1998: 181). This leaves only four
[ + ICP ] middle situation types absent from the literature: emotive speech actions, naturally
reciprocal events, and the two members of the referential class. I have not found any examples
of emotive speech actions in the literature, but I have no reason to suspect that they would be
incompatible with ICPs in principle.12 In any case, I find that ICPs are well-attested in the
expressive/cognitive class. Naturally reciprocal events are absent from the datasets I have
available, but I find this appropriate given how I analyze ICPs in the following sections.13 This
leaves only the referential situation types. The passive/impersonal/facilitative (PIF) middle uses
a middle marker14 to support an O argument being expressed as an So argument (Kemmer 1993:
20). This is different from what we typically observe with ICPs. When [ + ICP ] constructions
have two expressed participants, the ICP – the [ – A ] index – is encoded as the O argument.
When only a single participant is expressed, the ICP is removed and the participant that would
have been encoded as the [ + A ] index of the [ + ICP ] construction is treated as the Sa argument
of what is now a [ – ICP ] construction. In other words, the normal counterpart to a twoparticipant construal with the ICP as the O argument is a one-participant construal with no ICP at
all, and with only the [ + A ] index of the [ + ICP ] construction given explicit expression as the

12

Both emotive speech actions and emotional reactions can be interpreted as overwhelming semi-volitional mental
state changes. There is room for future research to explore this more fully.
Naturally reciprocal events already have a second “real” participant, completely separate from the agentive
participant. The agentive participant instigating the naturally reciprocal event is also the patient of the same action,
initiated by the other participant; nevertheless, their affectedness comes from an action instigated by a completely
separate real-life participant. This is distinct from what I believe happens with ICPs, which could explain the lack of
[ + ICP ] naturally reciprocal events in the literature. An alternative explanation may be that some [ + ICP ]
languages employ a separate reciprocal marker in such contexts, effectively preempting use of the ICP.
13

14

Defined here as a language-specific morphosyntactic marker associated with the expression of semantically
related middle situation types (Kemmer 1993: 15).

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

19

Sa argument. The PIF middle is defined in such a way that it only applies to cases where the O
argument becomes an So argument, which is generally not the case for ICPs.15 Since the PIF
middle doesn’t account for the argument indexation we normally see with ICPs, I am unsurprised
to see no evidence of it in the ICP datasets I have available.
This leaves only logophoric reference middles. This situation type constitutes a form of
extra-clausal reference. For these middle situation types, a verb in the subordinate clause is
marked with a specific “Kemmerian” middle marker, which refers to an explicit participant
reference appearing with a separate verb in the matrix clause. The logophoric reference middle,
as Kemmer defines it, requires the existence of a middle-marked predicate. In such situations,
the middle marker indicates that the verbal participant is coreferential with the participant
expressed for a different verb in another clause. I consider this a type of extrinsic definition,
since the middle marker is only fully defined with reference to a participant given syntactic
expression elsewhere. Based on my reading of Kemmer’s examples, the middle marker in these
situation types does not constitute a full participant reference in its own right. Furthermore, I see
no evidence in the literature that ICPs are defined by arguments given explicitly in a separate
clause. Schuh (2005: 22), however, notes that ICPs seem to do this with arguments given in the
same clause. What links both perspectives is that the element in question is defined extrinsically
– that is, it is defined with reference to information not available from the element itself. While I
argue later in this thesis that the locus of this defining information isn’t the other syntactic

I hedge with “generally” because there is limited evidence that the ICP might be able to appear in the [ – A ]
syntactic position without a [ + A ] argument. This is seen in clause 56a of the Appendix. If this is a legitimate case
of ICPs appearing without an explicit [ + A ] participant, then that could arguably be an example of the ICP used in a
way consistent with the PIF middle. With that said, this may be conditioned by other processes, and as such bears
further research. For now, the validity of expressing the ICP without a corresponding [ + A ] argument – along with
any bearings that might have on demonstrating use of the ICP with the PIF middle – must remain unresolved.
15

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

20

participant, but rather the “real” participant to which they both refer, the fact that both Schuh’s
perspective and Kemmer’s logophoric reference middle depend on a similar cognitive
mechanism suggests functional similarity between the two. This similarity, considered together
with the other examples of ICPs occurring with Kemmer’s middle situation types I have given
above, support using middle verbs as a starting point for my analysis.
2.1.2 ICPs, mental space structure, and the transfer of FORCE
In the previous section, I examined the literature and available data to make a case for
associating ICPs with middle verbal events. Though this is not the only environment ICPs are
found in, it is a helpful one. This is because there are certain characteristics of middle verbal
events that make my larger theoretical position more readily demonstrable. In this section, I
examine the various ways of expressing the core arguments of middle verbs. Doing this lets me
draw attention to the nature of self-affecting action and how that ought to inform our analysis of
the participant structure of middle verbs. In particular, my reasoning reflects others in the
literature in arguing that middle verbs are semantically transitive. Though I offer a more
nuanced perspective on this in §2.1.3 and §2.2, arguing outward from the assumed transitivity of
middle verbs proves to be helpful in building a theoretical foundation for analyzing less
straightforward cases. In further support of this, I present a theoretical argument consisting of
diagrams modelling the transfer of the property of FORCE. I then apply this model to both a
prototypical transitive and a middle verb, showing that each can be validly interpreted as
encompassing semantic transitivity. For middle verbal events construed in this way, I argue that
the second participant is simply a mental instantiation of the base referent (in particular, the
[ – A ] instantiation). When this participant is tacit or implicit, it’s because the speaker has
decided to not profile it; in cases where it is explicit, it’s because the speaker has decided to

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

21

profile it. There are a number of reasons a speaker might decide to profile this participant, some
of which I discuss in §3.
There is significant variation between languages in how the participants of middle verbs
are expressed. This is evident in the following examples from French and English:

(7)

a. il
se
lave
3SG.M REFL wash
‘He washes himself.’ [syntactic object is present, referencing the semantic object]

b. he
washes
3SG.M wash
‘He washes.’ [no syntactic object is present, but still has a semantic object]

These examples illustrate two approaches to expressing middle semantics, outlined by Kemmer
(1993: 58–59): in (7a), the French verb laver is treated as an ordinary two-participant verb
alongside special additional marking to indicate that it is self-affecting; in (7b), the English verb
wash is interpreted as inherently self-affecting in the absence of explicit participant profiling to
the contrary (i.e., He washes vs. He washes the dog). Kemmer extends this observation to argue
that certain verbs, especially those dealing with grooming and changes in body posture, naturally
pattern together with either (7a) or (7b) depending on how strongly they imply self-affecting
participation. What I deem important is that, in either case, a second, force-receiving participant
is assumed to exist, whether it is explicitly profiled or not.
These two participants exist in a transitive relationship. Transitivity involves the transfer
of force from one participant to another, embodying what García-Miguel calls “an asymmetric
energetic interaction between two participants” (García-Miguel 2007: 764). This definition is
similar to ones adopted by Kemmer (1993), Rice (1987), Hopper and Thompson (1980), and

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

22

others. Such “asymmetric energetic interactions” are readily apparent when they involve two
distinct “real” participants – i.e., a perfectly individuated agentive participant acting on (and
modifying) a perfectly individuated patientive participant (García-Miguel 2007: 764).
Nevertheless, force transfer between participants can exist even with the mentally instantiated
participants of middle verbal events. In English, this is easiest to see with verbs like wash or
especially shave; absent explicit marking to the contrary, the act of shaving is typically
understood as being carried out on the self. This is because there is a strong sense that a) shaving
is something one does to one’s own benefit (i.e., not as directed against another, except in
exceptional circumstances), and b) the action of shaving requires that a thing of some sort –
some part of the body (e.g., the face) – be shaved. For the event of shaving, the person
performing the action of shaving is also the person upon which the action of shaving is being
carried out. Thus, the “real” flesh-and-blood participant is understood in terms of two roles: both
as an exerter of force and as a receiver of force, simultaneously. Both this [ + A ] version of the
“real” participant and the [ – A ] version are mere mental instantiations of the base referent –
there aren’t suddenly two physical human referents involved. And though in the case of he
shaves the force-receiving version of the “real” participant isn’t explicitly profiled, that
participant exists regardless, as a function of how we understand the action of shaving to be
carried out. This relates to what we observe with verbs like go, which are frequently expressed
with ICPs; here, the base participant is construed as being split into a [ + A ] version of the self
and a [ – A ] version, the self-as-body. This conceptual splitting of the “real” participant into two
force-differentiated mental instantiations is the perspective implicitly adopted by Kemmer
(1993: 58), when she analyzes the verb going as involving volitional action performed by a
force-initiating participant that is carried out upon the self (interpreted as a force-receiving

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

23

body). Apart from Kemmer’s validation of the split-self perspective, I also have reason to
believe it is the operating perspective at work in a number of West African languages, where the
self-affecting nature of middle events is marked through the use of the clitic word “body”,
expressed together with an attached participant index.16
The validity of interpreting middle verbal events as semantically transitive is made
especially clear when diagrammed. In the diagrams below, I show how, given two participants
interacting in the context of a verbal event, a conceptual property called FORCE is transferred
from one participant to the other. This transfer of FORCE represents the completion of the verbal
action as applied against, upon, toward, or to the benefit of the “receiving” participant. This
transfer is shown in the abstract in example (8) below, with FORCE moving from the [ + A ]
participant (where it is located before the event occurs) to the [ – A ] participant (where it is
located after the event occurs). This is consistent with the progression of the verbal event from
its ante-state to its post-state. Example (9) show this model applied to a prototypical, twoparticipant transitive event. Example (10) does the same, but for a middle verb of the affective
class:

16

Two examples of this from Bole (West Chadic, Nigeria); note 1) the cliticization of the lexical word body to
indicate self-affect, and 2) the repetition in (b) of the index -mú, which is consistent with its different referential uses
(i.e., as a subject index affixed directly to the verb and as an index in the cliticized body+index “ICP” construction):
(ii)

a.

b.

pet-é=jì:-tò
go_out-FUT=body-3SG.F
‘She will go out.’ (Bole, West Chadic; Lukas 1971: 12; in Frajzyngier 1977: 79)
’yòr-á:-mú=jì:-mú
stand-PL-1PL=body-1PL
“(Let’s) stand!” (Bole, West Chadic; Ryan 2019: 140)

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(8)

Model representation:

Base
event

Base
event

[–A]

[ + A ]FORCE

Post-occurrence

‘The man kicked the dog.’

kick

kick

DOG

MANFORCE

DOGFORCE

MAN

Pre-occurrence

(10)

[ – A ]FORCE

[+A]

Pre-occurrence

(9)

24

Post-occurrence

‘The man shaved.’
shave

MANFORCE
Pre-occurrence

shave

[–A]

MAN

[ – A]FORCE

Post-occurrence

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

25

The transfer of FORCE diagrammed in (8) above is a property of transitive semantics. As such, it
applies equally to situations like (9), where the [ – A ] and [ + A ] participants are completely
different, and in situations like (10), where there is no “real” [ – A ] participant to express, but
which, as I’ve argued above, may be validly interpreted as involving such a participant. While
middle verbal events like that in (10) may not have a second “real” participant existing alongside
the base participant, the existence of the transferred FORCE element (as attached to the conceptual
[ – A ] entity) gives theoretical license for a speaker to express this transferred FORCE. This is
done by explicitly expressing some sort of “container” or entity to which that FORCE can be
understood as attached. Incidentally, this may also help explain why the ICP is optional in many
[ + ICP ] languages; for verbs that are strongly understood to affect the base participant as both
force-exerter and force-receiver (as is the case with verbs like wash and shave in English), the
existence of the base participant as an affected entity may be so assumed (i.e., so present and
cognitively accessible) that explicitly profiling the transferred FORCE with an ICP may be
processed as a sort of “doubling up” on that information. In cases like this, highlighting
transferred FORCE by explicitly profiling it – when it is already assumed that the base participant
is involved as an entity receiving that FORCE, even in the absence of explicit profiling – would
very likely be interpreted as emphasis.17
2.1.3 ICPs and their relationship to other participants
In this section, I use the mental space structure of [ + ICP ] constructions to show how the
Access Principle accounts for not only the relationship between the ICP and its base referent, but
also between the ICP and its associated [ + A ] participant. I have already built my case for

17

This would likely produce a range of effects, some of which I discuss in §3.3.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

26

interpreting the base participant in middle verbal events as being conceptually split into two
separate mental instantiations, representing the [ + A ] and the [ – A ] participant. In middle
verbal events, these two semantic participants exist in an asymmetric energy transfer
relationship. Even though there is only a single “real” participant to which both mental
instantiations are mapped, the nature of middle semantics points to an underlying transfer of
FORCE.

I argue that this relationship constitutes semantic transitivity, and gives validation to

explicitly profiling each of the two mental instantiations, even though they are mapped to a
common base participant. I examine these relationships in this section, and in so doing show
why the [ + A ] index and ICP are indexed in the same syntactic positions occupied by the
syntactic subject and direct object (respectively). I also engage with the idea that ICPs are
coreferential to the syntactic subject, and argue instead that the relationship between the two
participants is more appropriately explained as common coreference to their shared base
participant.
First, I address the relationship between ICPs and other objects. This observation has
been made explicitly by Storch et al. (2011: 6), who link ICPs to the object pronoun on the basis
of a) their inherent force-receiving nature, and b) their common syntactic expression. In
addressing this, I look especially to Croft (2012), who argues that the key determiner of whether
entities with particular participant roles are realized as subject or object depends on where they
exist relative to one another in the causal chain (Croft 2012: 4). In other words, for a given
verbal event construal, the participant construed as directing FORCE (i.e., the one closer to the
beginning of that verb’s causal chain, the [ + A ] argument) is realized as the syntactic subject,
and the participant construed as receiving FORCE (i.e., the one nearer the end of the causal chain,
the [ – A ] argument) is realized as the syntactic object. Croft’s argument supports my earlier

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

27

appeal to FORCE transfer and the structure of verbal events in transitive construal; the participant
associated with ante-state FORCE, near the beginning of the verb’s causal chain, is realized as
syntactic subject, while the participant associated with post-state FORCE, near the end of the
verb’s causal chain, is realized as syntactic object. Thus, for middle verbal events where the ICP
is explicitly expressed, we expect (and observe) that this participant will be indexed postverbally, a syntactic position common to force-receiving participants in [ + ICP ] languages.
The second point I address is my claim that the Access Principle offers the best
explanation for why the [ + A ] and [ – A ] participant in [ + ICP ] constructions match each other
in person, number, and gender. The existence of shared participant information between ICP and
[ + A ] index has been noted countless times in ICP research, at least as far back as Newman
(1971); it’s ultimately the reason for the “copy” part of the name Intransitive Copy Pronoun. Yet
I find that there are certain challenges to this perspective. The most important of these come out
of Van Hoek’s (2007) work on the acceptability criteria of coreference. The argument I derive
from her work runs (briefly) as follows: for two referents to be accepted as valid, appropriate
forms of coreference, one must be in the dominion of the other – that is, one must be interpreted
from the perspective of the other, as when a pronoun is used in place of a cognitively accessible
full nominal. The most important element in determining whether or not one referent is in the
dominion of the other is prominence, which is a measure of cognitive availability and relative
salience. Van Hoek’s definition of dominion assumes that the dominion-exercising referent is
offstage, but prominent, or highly accessible (Van Hoek 2007: 898). This introduces two related
challenges: first, since both the ICP and the [ + A ] index are explicitly profiled, neither is clearly
offstage to begin with; second, since both the ICP and the [ + A ] index are explicitly profiled,
neither is clearly more salient than the other; thus neither is clearly greater in prominence. This

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

28

is backed up by Langacker’s (1987: 234) observation that standard measures of determining
participant prominence, such as trajector/landmark alignment, tend to be less applicable to verbs
that fit the expressive/cognitive class. In other words, the kinds of verbs that are consistently
associated with ICPs don’t have a clear way of measuring prominence, which means (following
Van Hoek’s acceptability criteria) they don’t have clear criteria for measuring dominion.
The lack of a clear dominion relationship between the ICP and the [ + A ] index suggests
to me that the common characterization of ICPs and [ + A ] indexes being coreferential to each
other is questionable, if not outright untenable. A ready alternative to this analysis is provided
by framing the ICP/[ + A ] relationship in light of the Access Principle. This alternative
argument has the benefit of accounting for the evident coreference between the two participants
(i.e., the relationship evinced by the existence of shared participant information) but without
positing a coreference relationship that would be difficult to support from theory. From the
perspective of the Access Principle, [ + ICP ] constructions consist of a base participant being
construed as both an affected and non-affected affecting entity. Each of these two participants
repeats information about the base participant, and each adds additional linguistic information
(e.g., constituent order) used to mark the role that the mental instantiation of the base participant
is construed as having. Given a verbal event with a single base participant: if that verbal event is
construed as having a single participant (see (11) below), information about the base participant
is expressed in the mental instantiation of that participant, along with information that encodes
its role. If that same verbal event is construed as having two participants (as in (12) below), then
information about the base participant is expressed in both mental instantiations, along with
information that feeds into how each participant’s distinct role is encoded:

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(11)

‘I went (out).’

a1
< 1; SG >

a1: Base Participant

a2
< 1; SG; +A >

a2: Reflected: < 1; SG >
Added: < [ + A ] >

(12) nà
pòrò-no
1SG go_out.AUX1-ICP
‘I went out.’
(Kanakuru, West Chadic; Newman 1971: 190)

a1
< 1; SG >

a2
< 1; SG; + A >

a2: Reflected: < 1; SG >
Added: < [ + A ] >

a1: Base Participant

a3
< 1; SG; – A >

a3: Reflected: < 1; SG >
Added: < [ – A ] >

29

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

30

As another advantage to this analysis, the base participant is both highly accessible and
completely offstage, making it a valid choice for the dominion-exercising referent. And since
both the ICP and the [ + A ] index are accessing and reflecting information about the base
participant along separate referential pathways, there is no longer any conflict caused by both
profiled participants being equally prominent.
As both of the force-differentiated participants in middle verbal events are coreferential
with the base participant, [ + ICP ] constructions are arguably a type of distributed participant
reference. This is because in such constructions, the single base participant is being given
expression in two syntactic positions, corresponding to the two force-differentiated mental
instantiations of that base participant. Distributed participant reference is not without support,
having been discussed by Haspelmath (2013), Ryan (2019), and Van Hoek (2007). Haspelmath
(2013) argues that free person forms (i.e., full NPs and non-clitic pronouns) regularly occur
alongside argument indexes that encode the same participant information. Ryan (2019: 139–
140) discusses ICPs and distributed reference (as “multiple exponence”), noting that both the
subject index and the index in the ICP construction don’t amount to “vacuous repetition”, but
rather have different uses; in the terms I have used in this thesis, the first appearance of the index
accesses the base participant as the [ + A ] participant, while the index that makes up the ICP
construction accesses the base participant as the [ – A ] participant.
2.2 Extending the theory to cover non-middle verbal events
In the preceding section, I examined [ + ICP ] constructions expressing middle semantics to
provide evidence for a particular mental space structure. From this mental space structure, I built
a model of inter-participant relationships, mediated by the Access Principle, that I find accurately
accounts for the apparent coreference of the ICP and its associated syntactic subject. This was

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

31

done with reference to middle verbs (as I defined them in §1.2). But ICPs are associated with a
range of different verbs, including situation types that are only marginally “middle”. The
analysis I have presented above should theoretically be valid even with these types of verbal
events that fall outside the middle core. In this section, I test my analysis against these types of
verbs in order to demonstrate that ICPs can theoretically occur with any type of verb – regardless
of whether or not it would typically be considered a middle verb – as long as it is construed as
affecting the base participant as both a force-exerting and a force-receiving entity. Establishing
this obviates the need to determine whether and how a given verb fits Kemmer’s middle situation
types. In such cases, the existence of the ICP is prima facie evidence of self-affect, which is the
defining feature of middle verbs. This nullifies any expectation in our analysis that ICPs should
only be able to occur with a specific group of well-defined verbs, and makes analyzing ICP use
with any verb significantly easier: if an ICP exists, it is because the verbal event is interpreted as
being self-affecting on some level.
Although many middle verbal events naturally suggest the existence of a second
participant, as I demonstrated in previous sections, this second participant is essentially just a
container used to profile post-state FORCE. Middle verbs may have a strong candidate to fill the
[ – A ] syntactic position and allow for the profiling of post-state FORCE; however, there is
nothing prohibiting non-middle verbs from doing the same thing. This is ultimately what we
observe. Languages are free to adopt any of a variety of strategies for profiling post-state FORCE.
One such strategy is reflected in (13) below, which uses a reflexive participant reference as a
container or vehicle, allowing the transferred FORCE to be explicitly profiled. I argue that ICPs
have a similar function, though they are dissimilar from the particular reflexive vehicle used in
(13) in key respects. Another strategy adopted by some languages to express transferred FORCE

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

32

is to use a dummy object, as in (14). These dummy objects differ significantly from ICPs, a
point I discuss further in §3.1 below:

(13)

þykk-ja=sk
think-INF=REFL;ACC
‘Think.’
(Icelandic; Kemmer 1993: 19)

(14)

I thought a thought.

I find (13) in particular paints a clear picture of the ICP as a means of profiling post-state FORCE.
This is because the =sk middle marker seems to have developed diachronically from a standalone
logophoric reflexive pronoun (Kemmer 1993: 182); in other words, both the =sk middle marker
in (13) and the ICP profile post-state FORCE by expressing the self-as-affected-entity.
In (14) I see evidence that even prototypically one-participant events can be expressed as twoparticipant events. This supports Storch et al.’s (2011: 7) observation relating example (14) to
other, similar strategies for avoiding intransitive syntax. The most important thing I see in both
(13) and (14) is that a second, force-receiving entity – the [ – A ] participant – is chosen and
given explicit expression in order to profile post-state FORCE. In (13), this force-receiving entity
is the self-as-affected-entity; in (14), the force-receiving entity is the “thought”, which is a sort of
dummy object, and which follows the =sk in (13) in functioning as a vehicle for profiling where
the FORCE has gone to. This same idea is, perhaps, even more evident in (15), where the ICP and
the possessive index are expressed identically:

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(15)

33

mbɔ́
á-ɽɔ́ y
à-kpíɽìwáy=↓mbɔ́
à-zí=↓mbɔ́
02.PRO 02.SC-pick 06A-small.hoe=02.POSS 02.SC-go=ICP
‘They picked their small hoes; they went.’
(Dũya, Benue-Congo; Marggrander 2018: 183)

Here, the possessive index and the ICP both indicate that the preceding element is applied toward
or against the participant indicated by the suffix (i.e., the possessive index/ICP); in other words,
FORCE

is transferred conceptually from the preceding element, with the participant indicated by

the suffix as the endpoint of that transferred FORCE. With the ICP in particular, the participant
against which this FORCE is directed is the [ – A ] instantiation of the undifferentiated base
participant (the same base participant it shares with the pre-verbal [ + A ] index). It is this [ – A ]
index that acts as a container allowing post-state FORCE to be profiled.
Thus, on a theoretical level, a [ – A ] mental instantiation of the base participant is a valid
candidate for profiling the location of the verb’s post-state FORCE. Thus, as long as it is
considered a licit option within the language, having an ICP appear with non-middle verbs is
totally acceptable from a theoretical standpoint. This is important because it offers an
explanation for [ + ICP ] constructions involving non-middle verbs (i.e., verbs not associated
with Kemmer’s middle types). I give some examples of possible non-middle verbs below:18

(16)

bònò bòkké=jìːnì19
house burn:SBJV=ICP
‘The house burned (down).’
(Bole, West Chadic; Schuh & Gimba 2001: 10)

18

Among these are candidates I have found in previously unpublished discourse data from Ishɛ (Benue-Congo,
Nigeria). Note the [clause number] for these latter examples, allowing cross-reference against the appendix.
Technically, this should be two separate morphemes, =jìː and -nì, glossed as =body and – 3SG.M, respectively. I
have followed Schuh (2017: 285) in glossing them as a single entity so as to not obscure the fact that it functions as
a single unit, and is employed in a way that mirrors ICPs in other languages.
19

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(17)

múm
ì-sí-fás=ṹ
ú-gbɛ́tà=↓ŋó
1SG.PRO 1SG-NEG.FUT-pay=ICP 03-salary=2SG.POSS
‘I will not pay your salary.’
(Dũya, Benue-Congo; Marggrander 2018: 181)

(18)

à-zhéé
à-shí=↓ké
06A-tiredness 06A.SC-NEG.COP=ICP
‘There is no tiredness. (I am fine.)’
(Dũya, Benue-Congo; Marggrander 2018: 181)

34

(19) à
na-ɬən
sáy
3SG20 be_done-ICP AUX.FOC
‘They are ripe.’
(Miya, West Chadic; Schuh 2017: 284)

(20)

máa faarà-ɬən
à
what happen-ICP INT
‘What happened?’
(Miya, West Chadic; Schuh 1998: 199)

(21)

fa
ta-səna-fə-uw
2SG NEG-night-ICP-NEG
‘Don’t spend the night.’
(Miya, West Chadic; Schuh 1998: 91)

(22)

amma
inte
nahã
e-jɛ-yir
however 1PL.PRO PERF.NEG 1PL-see-1PL.ICP
‘However, we have never seen the place.’ [48]

e-nɛŋ-e
CL-place-DEF

This is used like English “one” or French “on”, and can appear (as it does in this example) together with the plural
ICP -ɬən to refer to an impersonal or conceptual subject.
20

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(23)

ko

k-a-yari-bɔɔ
HAB-3PL-graze:Xeven CL.PL:X21 AM CL.PL-cow NEG 3PL.ICP22
‘Even the Fulani don't graze their cattle there.’ [69a]

(24) ni
LOC

agɛ

e

i-nak

35

hã

a-teu

a

a-koi-she
CL-forest-

hã

CL-center

AM

DEF

NEG

k-a-kaupi-bɔɔ
HAB-3PL-farm:X3PL.ICP

i-nak
CL.PLcow

i-ram
CL-farming

ni

hum ba
LOC there NEG
‘At the center of that forest till now, they don't usually farm there.’ [66]

As I explained in §1.2, defining the line between middle and non-middle verbs is messy. “Ripe”,
“happened”, and “burned (down)”, in particular, could perhaps be understood as spontaneous
events, which would make them members of the Affective class of middles. Kemmer (1993:
142) explicitly includes “ripen” as an example of spontaneous middle events, though it’s unclear
if that would apply the same way to a verb with the potentially more expansive meaning of
“be_done”. “Graze” and “farm” could arguably be self-benefactives (also part of the Affective
class), depending on how the -X morpheme is understood. At a minimum, neither “see” nor
“spend the night” – nor especially “pay” and the NEG.COP of (18) – seem to fit any of the middle
classes particularly well, and as such are probably the strongest candidates for non-middle ICP

Here and elsewhere, I use “X” to represent possible morphosyntactic elements of as-yet undetermined function.
This also leaves the door open to the element in question being a simple phonological alteration of the preceding
element, or any of a number of alternative analyses.
21

22

Any candidate for the role of the -i in yari (or kaupi in (24) ) is speculative. It may be related to the 3 SG negative
perfective paradigm, which is expressed using the negative hã, a pre-verbal third-person index a-, and a post-verbal
suffix -i. This is far from conclusive, and it raises its own questions. The parallel with (23) argues for interpreting
post-verbal -bɔɔ as an ICP, in which case this is a good candidate for being a non-middle verb with an ICP.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

36

use. What is interesting is that even these verbs can be understood as being self-benefactive with
only a little bit of imagination. This illustrates the problem with talking about verbs in categories
like “middle” and “non-middle”: verbs aren’t universally “middle verbs” or “non-middle verbs”,
just like they’re not universally transitive or intransitive. The only requirement for a verbal event
to be interpreted as self-affecting is that it be understood as self-affecting. This seems like a
tautology, but it is actually the key understanding that decouples ICPs (as a cross-linguistic
phenomenon) from the expectation that they should only occur with certain verbs. The
assumption of self-affect can be encoded lexically – such as with “shave” in English; it can also
be encoded syntactically, such as through use of an explicit force-receiving syntactic participant.
Cross-linguistically, some kinds of verbs are so commonly understood as self-affecting that they
emerge as a pattern in typological analysis: i.e., changes in body posture, non-translational
motion, etc., all of which are considered “middle verbs”. This is part of what contributes to the
idea that ICPs are mostly associated with certain verbs. Yet regardless of whether or not a given
verb fits these common “middle” patterns, the mere fact that it is construed syntactically as selfaffecting means that it is a verb of self-affecting action, just like the core “middle verbs”. As
long as there is sufficient motivation for the verb to be construed as self-affecting, and as long as
doing so would not be considered illicit within the language, then essentially any verb can
theoretically be expressed with a syntactic participant indicating self-affect.23
3. Analyzing the purported functions of ICPs

Note that this doesn’t universally license ICP use with all verbs. Each language has its own codified standards
related to ICPs and verbs. It does, however, mean that ICPs are not inherently limited to certain verbs. ICPs
indicate a particular relationship the base participant has with the verbal action, which is not limited to certain verbs.
Thus, ICP use with “non-middle” verbs should not be considered impossible, or even necessarily surprising.
23

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

37

In the previous section, I showed how the ICP can theoretically be used with verbs falling
outside the traditional core of Kemmer’s middle situation types. This makes it clear that
although the ICP is associated with middleness (in the sense of it explicating self-affecting
action), it is not particularly useful to think of it as being associated with a narrow range of
acceptable verbs. In some ways, this moves the foundation of ICP research, especially as given
in Leger and Zoch (Leger & Zoch 2011: 36–42). I find this shift actually brings stability, rather
than instability. ICP research is filled with widely varying uses, environments, and effects; for
many of these, the connection to the middle seems tenuous and indistinct. But the result of the
theoretical framework built in this thesis is that we no longer have to explain why a speaker is
using the ICP with verbs nearer or farther from the core of Kemmer’s middle types. In other
words, we no longer have to explain why the ICP is judged as producing such unusual effects in
some cases but not in others (if it is, in fact, just a middle marker); instead, we can just accept
that the speaker is construing the verbal action as applying to the initiator of the action, making
them an affected entity. This means the only question left to be answered is what the speaker
intends to communicate by using the ICP.24 The ICP builds or explicates a specific mental space
structure, and that mental space structure is used to a variety of ends. In the following sections, I
interact with the literature to show how my analysis of ICPs counters (§3.1), explains (§3.2), and
extends (§3.3) some of the ICP’s purported functions. In §3.1, I describe the superficial

24

In reality, there are other questions. I've shown in this section that ICPs can theoretically exist in a range of
environments. Yet in practice, they don't. I suspect that language-specific ICP use is a product of their diachronic
spread, and may be constrained by other forms in the language having similar functions. One possible universal
constraint is that they can't cooccur with a profiled O argument - see Schuh and Gimba (2001) on Bole (West
Chadic, Nigeria). This may be related to the prohibition on using ICPs with 'sneeze', or 'cough'; my guess is that in
such cases, the resultant "product" (i.e., the sneeze itself, or the cough itself) is so strongly accessible on a cognitive
level that it is the obvious candidate for profiling post-state FORCE, and using another, weaker option would be
judged incongruous and therefore unacceptable (for alternative explanation, see the footnote on "sneeze" in §3.1).

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

38

similarity between ICPs and cognate objects. I demonstrate that though these depend on a
similar cognitive relationship, there are important structural differences between the two that
suggest they should be carefully distinguished. In §3.2, I examine a recent trend in the literature
arguing that ICPs encode Subject Point-of-View (SPOV). In particular, I show how the mental
space and participant structure at the heart of my thesis also adequately explains how ICPs
encode SPOV. This also entails creating a tighter definition of the “subject” highlighted by
SPOV. In §3.3, I extend the Disjuncture-Navigation (DN) function of ICPs by organizing it into
two broad classes and introducing a potentially new discourse use that has not yet been discussed
in the literature. Doing this all has two results: first, it pares down the number of functions that
are associated with ICPs; second, it shows that the cognitive relationships that make ICPs work
are able to account for challenging uses – i.e., ICPs used with non-middle verbs and ICPs used to
capture vastly different functions. Ultimately, this contributes to ICP research by moving the
conversation toward a single root explanation that accounts for the wide variety we see in ICP
use (though there will undoubtedly still be a need for research explaining why this root
explanation manifests a certain way in specific cases).
3.1 Countering the purported Cognate Object (CO) function
Researchers have long pointed to the object-like nature of ICPs. This was given particular
emphasis by Storch et al. (2011), who note in a summary of previous ICP diagnostics25 that
“ICPs tend to formally resemble the object pronoun”, and that “this relates to cognate object
constructions, where intransitivity is avoided through the use of a dummy object” (Storch,
Atindogbé & Blench 2011: 6) Although more recent works touching on ICPs have emphasized

25

E.g., Newman (1971), Wolff and Gerhardt (1977), Frajzyngier (1977), Jungraithmayr and Leger (1993), and
Tuller (1997).

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

39

the relationship between ICP and subject,26 rather than ICP and object, Storch et al. remains the
most recent comprehensive diagnostic. My analysis supports some of Storch et al.’s
conclusions; in particular, I agree that there are similar cognitive mechanisms at work with both
cognate objects and ICPs. With that said, my analysis points to differences in the participant
structure of cognate objects and ICPs that I believe calls for the two to be carefully distinguished.
This results in a theory-tested means of bridging the gap between Storch et al.’s Cognate Object
(CO)-function perspective and the alternative SPOV-function perspective.
First, I give a brief explanation of cognate objects. Cognate objects can be analyzed as a
specific type of object nominalization, wherein the O argument of a transitive verb is a noun
derived from the verb itself (Dixon 2005: 329). Dixon categorizes object nominalizations as
either matching the form of their corresponding verb (25a) or else constituting a derived form of
the verb (25b):

(25) a. spray a nauseous
VERB

spray
OBJECT NOM [same form]

b. bear
a heavy burden
VERB
OBJECT NOM [derived]

As shown in (26) below, the cognate objects of [ + ICP ] languages tend to resemble Dixon’s
“derived form” object nominalizations:

26

I discuss this more in §3.2.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(26)

40

dáp
dábe
ku-∅
3SG-AOR hit
hit
‘S/he hit a hit [i.e., she gave a hit].’
(Hone, Benue-Congo; Storch 2009: 129)

If we don’t examine the mental space and participant structure of [ + ICP ] constructions, the
most straightforward analysis of ICPs is that they’re mere syntactic placeholders, just like
cognate objects. Ostensible evidence for this is that ICPs largely occur with “intransitive” verbs
– that is, verbs with no clear candidate for a second participant – where they act as a syntactic
participant in parallel to direct objects. Verbs in [ + ICP ] languages tend to be neutral with
regard to transitivity, 27 which means they can be used both as a verb with a clear second
participant and as a verb with no clear second participant. This lends credence to the idea that
ICPs exist only to give verbs-as-intransitives some syntactic participant to act against, though
they have no “real” participant they’re connected to. Thus, the thinking goes, ICPs and cognate
objects are both syntactic realities without a clear semantic reality backing them up, used
principally to avoid syntactic one-participant constructions (Storch, Atindogbé & Blench 2011:
7–8).
With that said, I think the mental space structure of both [ + ICP ] constructions and
cognate object constructions paints a slightly different picture. As I show below, ICPs and
cognate objects differ in that while ICPs are a force-receiving instantiation of the base
participant, cognate objects are a mental instantiation of the verbal action itself.28 From this
perspective, though both participants are vehicles for profiling post-state FORCE, cognate objects

27

For more on this, see Schuh (2017), referencing Hoffman (1963) and Jungraithmayr (1970).

28

This is supported by similar observations made by Schuh (1998) and especially Storch (2009).

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

41

represent EFFECT29 (by highlighting the thing done to the participant) while ICPs represent
AFFECT30 (by

highlighting the affectedness of the participant). The following examples show the

different mental space structure of cognate object constructions and [ + ICP ] constructions,
respectively:

(27)

dáp
dábe
ku-∅
3SG-AOR hit
hit
‘S/he hit a hit.’ [i.e., S/he gave a hit] (Hone, Benue-Congo; Storch 2009: 129)

a1
< PARTICIPANT >
< UNDIFFERENTIATED >

a2: ku
< PARTICIPANT >
<EFFECTOR >

a1: Base Participant; a2: Derived from a1

29

I.e., the action that was accomplished.

30

I.e., the participant’s response to that action.

b1: dáp
< ACTION >

b2: dábe
< EFFECT >
< FORCE >
b1: Verbal action; b2: Derived from b1

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

42

(28) nà
pòrò-no
1SG go_out.AUX1-ICP
‘I went out.’ (Kanakuru, West Chadic; Newman 1971: 190)

a1
< PARTICIPANT >
< UNDIFFERENTIATED >
a1: Base Participant

a2: nà
< PARTICIPANT >
< EFFECTOR >

a2: Derived from a1

a3: -no
< PARTICIPANT >
< AFFECT >
< FORCE >

a3: Derived from a1

Note that in both examples, the base participant, undifferentiated as to force-dynamic roles, is
mapped to a [ + A ] participant representing the EFFECTOR. And both examples have a [ – A ]
participant being used as a vehicle to profile post-state FORCE. The difference between the two is
that in (27), this second participant isn’t derived from the base participant (as with (28) ), but
rather from the conceptual entity ACTION. This means that although cognate objects work like
ICPs – that is, they receive FORCE – only [ + ICP ] constructions have a [ + A ] and [ – A ]
participant that are mapped to the same entity. This is important because it directly constrains
how we understand the relationship between ICPs and cognate objects. While the Storch et al.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

43

diagnostic correctly points out that the two phenomena share a cognitive mechanism, they are
ultimately manifestations of distinct mental space structures, with different derivational chains.31
3.2 Explaining the purported Subject Point-of-View (SPOV) function
In the previous section, I showed that cognate objects and ICPs differ substantially, despite
relying on a similar cognitive mechanism. I now examine the recent view that ICPs encode
SPOV. This perspective has been advanced especially by Frajzyngier (see Frajzyngier & Shay
2012: 294; Frajzyngier 2012: 585; Frajzyngier 2011: 54; 57–58; see also Storch 2011: 90–93 and
Ch. 6 of Frajzyngier, Johnston & Edwards 2005). SPOV is a type of subject focus, which I
follow Van Hoek (2007: 900) in defining as use of the “subject” for the reference point or locus
of empathy from whose perspective the verbal event is interpreted.32 This “subject” is rarely
defined; ostensibly, it refers to the syntactic subject. That said, I note that this “subject” need not
be a controlling participant (Frajzyngier 2012; Frajzyngier & Shay 2012; see also the same
perspective in Mous 2012), which opens the possibility that it could be either the [ + A ] or [ – A ]
participant that is in focus. I find it improbable that the ICP – itself a distinct syntactic
participant – causes attention to be directed toward the syntactic subject, a different syntactic

Consider also environment: cognate objects appear with “product” verbs like cough or sneeze (Amha 2009: 372–
373; Levin 1993: 95; Dixon 2005: 330), but ICPs don’t (Schuh 1998: 182; 2017: 283; Leger & Zoch 2011: 37–38).
This may be because a sneeze, for example, is readily interpreted as a punctiliar event with an exceptionally close
link between its existence and the verbal event that caused it. Furthermore, a sneeze seems to exist as a conceptual
reality even apart from conjuring imagery of an associated causal chain. This would make a sneeze something more
concrete than it being merely the reality marking the endpoint of a verbal action; this in contrast to a “hit” which
completes the “hit” causal chain but which seems to lack such a strong sense of a discrete and concrete “thing” (i.e.,
a “hit” as a stand-alone reality) having been brought into sudden existence. This would mean that sneeze-type verbs
have a relatively prominent candidate for EFFECT, which would potentially override using the relatively lessprominent AFFECT candidate.
31

32

I argue that this process works via manipulation of focal attention; for reflections of this, see Schuh (2005: 21) and
Grubic et al. (2010: 4).

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

44

participant, with which it has no clear, direct cognitive or attentional connection.33 In any case, it
seems unnecessary, as the subject-focus noted by Frajzyngier can be accounted for by my thesis:
the “subject” whose perspective is encoded by the ICP is the unary base participant. This
“subject”, which is undifferentiated as to force dynamics, is mapped to the ICP via the Access
Principle. The ICP serves as the focal element of the [ + ICP ] construction and highlights by
extension the base participant to which it is linked. This means that the ICP does, in fact, encode
SPOV, but does so by exploiting the normal cognitive mechanism linking the ICP and the base
participant, and not by shifting attention to the [ + A ] participant as expressed in the syntactic
subject.
3.3 Establishing the Disjuncture Navigation (DN) function
In the previous two sections, I brought the cognitive framework of [ + ICP ] constructions to bear
against two functions: first, a purported CO-function (which my thesis counters), and second, an
SPOV-function, which my thesis explains from the perspective of cognition and attention. In
this section, I argue that the cognitive framework I present in my thesis suggests that ICPs may
have a role in navigating disruptions in the cognitive environment – what I call cognitive
disjunctures (hence, the Disjuncture Navigation function). In support of this, I offer five
conclusions about verbal event construal, derived from Croft (2012):

If anything, the O argument – and not the A argument (i.e., the [ + A ] index) is the most obvious candidate for
trajector, in that it tends (weakly) to receive focal attention. Additionally, I also see little theoretical support for
thinking that explicitly profiling AFFECT would make the EFFECTOR more prominent (as would be the case if using
the force-receiving ICP did, in fact, highlight the force-directive [ + A ] index).
33

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

45

1) Construal isn't fixed, and any of a number of different construals can be chosen (Croft
2012: 14);
2) construal choice is, in part, motivated by the causal relationship between
participants (Croft 2012: 27);
3) certain construals can perform functions extending well beyond argument encoding,
such as the building of mental spaces and the tracking of referents
(Croft 2012: 20);
4) construal choice is (weakly) constrained by the nature of the verbal event and (more
strongly) constrained by the conventions of the language – i.e., whether or not
certain construals are deemed licit, or are deemed a departure from the norm
(Croft 2012: 29);
5) ultimately, construal choice is motivated by the various goals of the speaker, among
them discourse goals (Croft 2012: 18; 29).

The key takeaway from these five conclusions is that speakers choose a construal in service of
various goals, and within the context of various constraints.
In §3.3.1, I draw from these conclusions to argue that the various semantic effects
associated with ICPs are the result of interpretive assessments of the choice to use [ + ICP ]
construals rather than an alternative. These assessments can hinge on either the presumed
motivation for using ICPs or else the cognitive structure evinced by use of the ICP. In §3.3.2, I
again draw from the five conclusions above – along with previously unpublished discourse data
– to argue that ICPs are being used primarily to maintain participant activation across points of
cognitive disjuncture. Since the argument in §3.3.1 and the argument in §3.3.2 have a prominent

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

46

element of the disruption of the cognitive environment, I argue that ICP use in both cases
constitutes a means of navigating disjuncture.
Since there is so much cross-linguistic variation in ICPs, the assessments I make in
§§3.3.1 and 3.3.2 should be understood as preliminary results, to be tested in-depth by future
research. In particular, there is a need to analyze how ICPs produce semantic effects within the
unique linguistic context of each [ + ICP ] language (extending my work in §3.3.1), as well as to
study the DN function against more extensive (and more varied) discourse data (extending my
work in §3.3.2). Despite these limitations, my analysis in these sections helps demonstrate the
real-world application of my thesis. First, it helps answer yet-unanswered questions, such as
how ICPs can produce such a wide range of seemingly unrelated semantic effects. Second, it
helps make predictions about other uses for the ICP that have not yet been explored in the
literature, such as the use of ICPs in discourse.
3.3.1 The DN function and semantic effects
Interpretive assessments play a key role in determining how ICPs are understood. These
assessments are built on recognizing the choice to use a [ + ICP ] construal over alternatives
(related to conclusion one above), and then processing that choice and assigning it an
explanatory interpretation (related to conclusion five). Of particular note is that this applies even
in cases where ICP use is deemed “obligatory” – i.e., in cases, where there seems to be no free
choice that could result in an interpretive assessment. Though this may sound controversial, I
find it fits not only the theoretical analysis I’ve built in this thesis, but also the literature itself. A
cursory exploration of the literature produces no clear examples of obligatory ICP-use that also
produces semantic effects (i.e., totality, mirativity, etc.). In a number of prominent cases where
ICPs are associated with semantic effects, the ICPs themselves are explicitly deemed “optional”

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

47

(Newman 2003: 620) or “not… syntactically obligatory” (Schuh & Gimba 2001: 10). So there is
reason to believe that semantic effects are only present in cases where there is some motivation
(i.e., conclusion five) to choose a [ + ICP ] construction over a valid alternative.
Furthermore, it isn’t even evident that “obligatory use” is a legitimate category when
disconnected from semantic effects. Early work on Pero (West Chadic, Nigeria) and Kanakuru
(West Chadic, Nigeria) argued that ICPs are required with all verbs that are not inherently stative
(Pero; Frajzyngier 1977: 76), or with all verbs used intransitively (Kanakuru; Frajzyngier 1977:
83). These same perspectives have been repeated multiple times, including as recently as Schuh
(2017: 283). This sounds like obligatory use, but I question whether that’s an accurate
assessment: discussing Pero, Schuh notes that ICPs are obligatory not with all non-statives, but
rather only with those non-statives being used with an inchoative sense. This means that ICPs:
a) are not used in cases where expressing the semantic effect (i.e. inchoativeness) isn’t desired;
and b) are used in cases where expressing the semantic effect is desired.34 With Kanakuru, the
“obligatory use” covers only perfective verbs. Perfective verbs in particular seem to be easily
compatible with ICPs, in that their use carries an inherent understanding of transferred FORCE,
and thus completed action, as I’ve argued previously. Thus here also, ICPs are: a) obligatory in
cases where the semantic effect of “completeness” or “fully in-effect” is desired; and b) excluded
in cases where that semantic effect is not desired. Examples like this abound in the literature,
and future research could fruitfully re-examine each such case of ostensible “obligatory use”,
reframing them as cases where the ICP is being used because of its ability to communicate a
mental space structure that lends itself to a range of identifiable interpretive assessments.

34

It is no coincidence here that both inchoativeness and middleness are often expressed with ICPs: Croft (1994:
102) argues that both encode the fact that the initiator of the verbal event is also the affected entity.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

48

What these observations ultimately argue for is that ICPs are able to capture semantic
effects because the choice to use the ICP communicates something about the mental space (and
causal) structure of the verbal event (i.e., conclusions one and two above), which therefore drives
an interpretive assessment (i.e., conclusions three and five) – likely even in cases where the ICP
is deemed obligatory. The mental space structure communicated in this way constitutes a change
in the cognitive environment: granularly, because it introduces a participant whose existence
must then be interpreted; broadly, because it constitutes a motivated (or marked) construal
choice. Using the ICP asserts an updated cognitive picture. It bridges the pre-assertion cognitive
environment and the post-assertion cognitive environment. Thus, the ICP functions to: a)
communicate a change in the cognitive environment; b) help build an updated cognitive
environment; and c) help speaker and hearer navigate the disjuncture. The updated cognitive
picture and the difference between it and the old picture are interpreted, which may (among other
things) result in an assessment that certain semantic effects are present.
With this in mind, the remainder of this section examines a few common semantic effects
associated with ICPs and explains how they can be arrived at as interpretive assessments of the
DN function of ICPs. This serves two purposes. First, it helps to establish the validity of my
proposed DN function. Second, it shows how my thesis contributes to making sense of the
variation seen in ICPs and their associated semantic effects. Since – as I argue – the underlying
cognitive process of update—signal—interpretation35 is unchanging, understanding the semantic
effects of ICPs no longer requires that the semantic effects be linked to one another (i.e., as

35

This simplifies a process that actually looks more like: a) the cognitive picture is assessed by the speaker as in
need of an update; b) this is signaled; c) the signal is received by the hearer; d) the interpretation is made; and e) the
updated cognitive picture is activated (with the old picture rapidly decaying from working memory). The cognitive
framework enabling this is essentially the same as the one enabling the DN function in discourse (see §3.3.2).

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

49

though to show that they are not simply random). Rather, all that is necessary is to show that
each can be reasonably arrived at as the end result of the process. I do this briefly below, for a
selection of semantic effects commonly associated with ICPs. A possible avenue of future
research would be to cover the remaining semantic effects noted in the literature. This would not
only result in a clearer picture of the relationship between cognition and semantic effects, it
would also serve as part of the groundwork for an extensive comparative study of [ + ICP ]
languages looking for patterns in which semantic effects are produced and how they are
distributed.
3.3.1.1 ICPs and totality
The “totality” function of ICPs is arguably the most straightforward case, so I begin here.
Totality is an expression of the completeness, all-encompassing-ness, or thoroughness of an
action. It expresses finality, definitiveness, and irreversibility (Schuh & Gimba 2001: 10), and is
ultimately a declaration of the thorough, complete application of the verbal action (Newman
2000: 648). ICPs have been connected to totality at least as far back as Schuh (1972: 28).
Explained simply, the ICP signals totality because it naturally suggests the complete selfinvolvement of the participant. This is a result of its encoding of the participant from two
distinct perspectives, as both the [ + A ] and the [ – A ] participant. Using the ICP to highlight the
force-receiving nature of the participant stresses the action as being applied to or against a
participant, and thus highlights (by extension) the fact that the action is understood as a fait
accompli.
3.3.1.2 ICPs and mirativity
A similar case may be made for instances where the ICP is used to express mirativity.
Mirativity, following Aikhenvald (2004: 20), is concerned with “unexpected information”, and

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

50

includes expressions with “overtones of surprise.” Mirativity has been associated with various
ICP and ICP-like constructions by Storch (2009: 134; 2011: 96–98), Hellwig (2011: 74–77), and
Koops (2011: 107; 110–112), among others. The simplest explanation of why the ICP can
express mirativity is that it likely disrupts what the interlocutors expect about the relationship
between the participant and the verbal event being profiled. In other words, if it is assumed that
a simple verbal action done by the participant will involve only that participant in a [ + A ]
capacity, then the inclusion of the same participant in a [ – A ] capacity will be interpreted as
unexpected. Storch (2011: 97) gives an interesting example of “mirativity forms” – namely,
ICPs – used to describe the effects of spirit possession in Northern Jukun (Benue-Congo,
Nigeria). Actions undertaken by a possessed individual are expressed as involving that
individual both as a [ + A ] and a [ – A ] participant. In such cases, it is ostensibly deemed
surprising or unexpected that the possessed individual should “do” something while also being
carried along in the “doing” as a non-volitional participant, acted upon (and thus compelled) by
the possessing spirit.
3.3.1.3 ICPs and inchoativeness
Inchoativeness, which is related to the incipience of a verbal event, seems harder to explain. It
has been connected to ICPs by Frajzyngier (1977: 80–82), among others,36 though there is no
immediately apparent link between the ICP and the effect. This stands in marked contrast to the
relatively straightforward ICP/totality link. Totality strongly implies that the verbal action is
understood as fully taking effect. This is because the existence of the [ – A ] participant implies

36

Inchoative semantics have also been connected to middles (Arce-Arenales, Axelrod & Fox 1994: 17–18; Croft
1994: 102). This is noteworthy in light of the close connection between the mental space structure of [ + ICP ]
constructions and the mental space structure of middle verbal events.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

51

the transfer of FORCE to its post-state location, meaning there is an implicit (notional) temporal
shift from “before the action occurs” to “after the action occurs”. Thus, the action can be
understood on a conceptual level to be “complete”, even when it’s not expressed as such
aspectually (e.g., by use of perfective aspect). Mirativity can be understood on much the same
grounds; an action interpreted as surprising or unexpected is still applied to or against the [ – A ]
participant, which would, as with totality, imply its conceptual completion. Nevertheless, the
ICP is still used to express inchoativeness in various languages, which is prima facie evidence of
its licitness. A potential explanation for the relationship between the ICP and inchoative
semantics is that the ICP introduces an explicit participant into the cognitive environment,
causing the verbal event associated with that participant to be interpreted as cognitively new.
Having a cognitively new event suddenly manifest itself could be highlighting the shift from the
verbal event’s non-existence to its resultant existence, which would thus be interpreted as its
incipience. This would relate inchoativeness to punctual semantics, in that both are connected to
the verb analyzed with regard to a single, fixed point in time.37
3.3.2 The DN function and discourse
In §3.3.1, I argued that the semantic effects produced by ICPs can be explained by two factors:
1) the interpretive assessment of marked choice to use ICPs over alternative construals; and 2)
the interpretive assessment of the mental space structure expressed by use of the ICP. The ICP is
associated with cognitive disjuncture, and its use to navigate these disjunctures may result in any
of a variety of interpretive assessments, chief among which are a range of seemingly unrelated
semantic effects. Despite being seemingly random, these semantic effects are – as I argued –

This is supported by Leger and Zoch on Frajzyngier, who note the latter’s connection of punctuality with both
inchoative semantics and posture change (Leger & Zoch 2011: 36).
37

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

52

produced as a result of the ICP’s DN function. In this section, I argue that the cognitive structure
of [ + ICP ] constructions may allow ICPs to be used as aids to discourse navigation. To do this,
I explore the relationship between attentional focus, Working Memory, and cognitive
disjuncture. I then examine discourse data from Ishɛ (Benue-Congo, Nigeria) and show how ICP
use seems to be concentrated around changes in the referent that is receiving attentional focus.
This means that the ICP is another option in a range of referent-expression choices, one that
plays an important role in helping to navigate changes in the cognitive environment shared by
interlocutors.
I have explained in previous sections that cognitive disjunctures are points of change in
the cognitive environment. This may constitute a change in how the participant and their role is
to be interpreted (as with semantic effects); it may constitute a change in the context in which
participants exist or in which actions are undertaken. In all such cases, however, the cognitive
disjuncture is effectively making changes to the contents of Working Memory (WM). Working
Memory is a “small and quickly updated storage of information” (Kibrik 1999: 30) that plays an
important role in guiding decisions on referential choice (Kibrik 2000: 73). Referents are added
to WM via attentional focus and – unless they are reactivated – eventually pass out of WM via
decay (Kibrik 1999: 49). Working Memory is essentially a running clock; when an entity
receives attentional focus, it is entered (or entered anew) into WM, where it immediately begins
to decay and eventually pass out of WM, unless it is reasserted in WM. Elements that are
receiving focal attention are said to be active; those that are not receiving attentional focus (but
which are still present in WM) are said to be semi-active (Chafe 1987: 22).38 When focal

Chafe (1987: 22) calls this state of semi-activation “accessible”. I offer an alternative: all information in WM is
accessible. The closer something is temporally to having been activated (i.e., to having received attentional focus),
the more accessible it is; the further it is from having been activated (and thus, the closer it is to complete decay and
38

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

53

attention is directed toward a given element, it is activated and becomes highly accessible. Since
only one element can receive focal attention, all other elements in WM are necessarily in semiactivation. These elements continue their progress toward decay, with those closest to the
complete decay (i.e., those that are only marginally accessible) passing out of WM first, followed
by those that are relatively higher in accessibility. In a typical example of cognitive disjuncture
in discourse, attentional focus is redirected from one element to another; thus, the activation of
the attention-losing element is preempted by the activation of the attention-gaining element. The
attention-losing element is still present in WM, but its accessibility begins to decrease, leading to
marginal accessibility and eventual decay from WM.
Cognitive disjuncture in discourse takes many forms. Some of these are part of the
discourse’s global structure – the large-scale chunks into which discourse is segmented (Kibrik
2011: 14). An example of this type of cognitive disjuncture is the thematic unit boundary, which
signals the line separating one thematically connected group from another (Dooley & Levinsohn
2001: 18; §7). Others are more closely connected to the discourse’s local structure – i.e., smallscale chunks, such as clusters of prosodically linked clauses (sometimes called Elementary
Discourse Units (EDUs) ) (Kibrik 2011: 14). An example of this latter type of cognitive
disjuncture is the Subject Context change. Subject Context (SC) changes are those points in the
discourse where the participant receiving focal attention changes. This is typically understood to
occur between clauses. The four possible SCs are: S1, for subjects that are identical to that of the
previous clause or sentence; S2, for subjects that were the addressee of speech reported in the
previous sentence (in a closed conversation); S3, for subjects involved in the previous sentence

removal from WM), the less accessible it is. Thus, in addition to the activated/semi-activated distinction, we have
two poles for sorting the elements of WM: “highly accessible” and “marginally accessible”.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

54

in a non-subject role (other than in a closed conversation); and S4, for subject changes other than
those covered by S2-S3 (Dooley & Levinsohn 2001: 65).
Discourse-level cognitive disjuncture is often navigated through the use of extra encoding
material. Participant reference strategies range from zero anaphora (i.e., no explicit reference) –
low encoding – to a full NP – high encoding (Givón 1983: 18).39 Thematic unit boundaries are
often marked with relatively high levels of encoding – things like full NPs (Dooley & Levinsohn
2001: 20; see also Dooley and Levinsohn 2001: §§16-18). This is supported by the observation
that navigating changes in the cognitive environment requires increased cognitive processing
(Dooley & Levinsohn 2001: 57). It is unclear whether the extra encoding material is the cause of
the increased processing, or whether it simply accompanies it. Given that cognition drives
language – and indeed all human experience – so strongly, I suspect that the use of extra
encoding material may actually induce cognitive disjuncture, or at least prepare the way for it.
Testing this is outside the scope of this thesis. In any case, whether we assume that extra
encoding of participants is done in order to effect cognitive disjuncture or as a result of it, it
seems clear that the two are closely connected. In the case of thematic unit boundaries, using
extra encoding (i.e., a full NP) activates that referent in WM. This changes the cognitive
environment by allowing the previously active referent to be deactivated, at which point it moves
through successively lower stages of accessibility,40 before ultimately decaying completely and
disappearing from WM. Although the thematic unit boundary itself constitutes a cognitive
disjuncture, the referential change does as well. This referential change might potentially be

39

Free Person Forms (FPFs – i.e., non-bound pronouns) fall somewhere in the middle.

40

In reality, a continuum of accessibility without discrete points.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

55

additive, simply supporting another manifestation of the cognitive disjuncture associated with the
thematic unit, or it might be independent, and thus the only manifestation of that cognitive
disjuncture. I make no claim either way. The referential change constitutes a local-structure
manifestation of cognitive disjuncture that supports the existence of a global-structure
disjuncture (i.e., the thematic unit boundary). This is true even if it’s the only local-structure
manifestation being expressed. Understanding this is important, because it helps to establish the
idea that a relatively higher level of encoding material constitutes a cognitive disjuncture in its
own right. Again, it is not particularly important at this point to determine which came first (i.e.,
the cognitive disjuncture or the encoding of it); what matters is seeing that use of extra encoding
material updates the contents of WM, which is self-evidently a change in the cognitive
environment.
Using extra encoding material in participant reference is, as I have just argued, a type of
local-structure cognitive disjuncture. This is evident not only because of how we understand the
workings of WM, but also because of the well-attested connection between extra encoding and
thematic unit boundaries. I argue that ICPs may also be considered a type of extra encoding
material. The rules on ICP use vary significantly between [ + ICP ] languages. With that said,
data across the literature supports the idea that using the ICP is not the default level of encoding
material (except, depending on the language, in certain verbal TAMs, in negation, etc.). Based
on the data I have available, the default level of encoding in Ishɛ is the [ + A ] index. So when
the ICP is used in Ishɛ, it constitutes use of extra encoding material.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

56

I believe that at least some use of the ICP may be explained by understanding it as extra
encoding material being used to help navigate cognitive disjuncture.41 Specifically, I see
evidence supporting the idea that ICPs are being used near SC changes. While SC changes
themselves are frequently marked by even higher amounts of encoding material (especially full
NPs), ICP use seems to cluster near SC changes, as I show below. This suggests to me that the
ICP, while not constituting enough extra encoding material to be interpreted as an SC change
itself, nevertheless constitutes an increase in encoding material over the default. This would
serve to reassert the linked participant in WM, allowing it to be made (temporarily) more
accessible. The nearby SC change would assert a different referent, thereby deactivating the
participant expressed with the ICP. Nevertheless, this newly deactivated referent would still be
relatively highly accessible. Since even activated referents still continue their progress toward
decay, a participant that is preempted in WM by a new referent (as would happen at an SC
change) would already have become less accessible over time. By reasserting that participant
mildly through the use of the ICP, that participant is given a boost in its accessibility that helps it
to stay present and relatively highly accessible in WM despite an intervening referent being
activated at the SC change.
On a mechanical level, the ICP is able to do this because of the way it affects focus. The
ICP directs attentional focus to the base participant via the Access Principle (in particular,
uplink). Directing attentional focus to the base participant activates it and resets its decay timer

“Navigating” and “inducing” cognitive disjuncture are essentially matters of perspective. I suspect that the
speaker encodes disjuncture through the ICP, along with other means. These same signals of disjuncture are
received by the hearer and are then compiled and interpreted as signaling disjuncture on the part of the speaker, at
which point they actually induce disruption of the hearer’s cognitive environment in order to coordinate their shared
cognitive environment. I have no space to explore this further; in any case, what matters is that the ICP is being
used to aid in this process of navigating what it means to have a signaled change in the cognitive environment.
41

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

57

to zero. It makes no difference what the nature of the base participant’s relationship with WM
has been before this point. It may have already been present in WM; it may have decayed out of
WM; it may have never been in WM at all. As soon as it is asserted as existing in WM (i.e., by
having attentional focus directed toward it) it is handled as a new cognitive reality, becoming
activated and having its decay timer set to zero. In cases where a referent does already exist in
WM, using the ICP to manipulate attention and reassert the referent in WM is a valid means of
staving off decay. This keeps elements both present in WM and also highly accessible, which
may be desirable for any of a number of reasons. This may be of particular importance when
another referent is being introduced; since asserting a different referent in WM would cause the
base participant to become less accessible, it may be desirable to reactivate the base participant
in order to maintain its relatively high level of accessibility, even if it’s not otherwise close to
complete decay.42
Some reflection of this may be visible in data from Ishɛ, which I provide in full in the
appendix. Although the discourse sample I use is relatively small, the distribution of ICPs in it is
consistent with the theoretical position I have built. Below, I present a table that tracks how
ICPs pattern relative to SC changes. I also note the distribution of free person forms (FPFs) –
i.e., standalone pronouns – and full NPs, as these are referential strategies that use a relatively
high amount of encoding material.43 While it is possible that some of the ICP use shown in the

42

This might be related to the idea of privileged referents, in which case keeping a participant highly accessible
through the use of an ICP could be a type of local-VIP strategy. More data would help make this clear.
43

I break slightly from precedent (see especially Haspelmath 2013) by separating full NPs from FPFs. Though both
full NPs and free-standing pronouns are free (i.e., not attached to another syntactic entity) full NPs constitute selfdefining participant references and as such – unlike free-standing pronouns – don’t require the presence of a referent
in WM to give them definition. Because of the role WM plays in guiding use of various person forms, I find it
appropriate to treat full NPs and free-standing pronouns separately.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

58

table could be due to the desire to capture semantic effects, I have no evidence to that effect in
any of the glosses, translations, or additional comments provided to me; in any case, it’s unlikely
that all of them would be explained by semantic effects. At least some (if not all) of the variation
in ICP use in this data suggests some other motive. I believe the best explanation for this
variation lies in assuming that the ICP is being used to maintain the referent’s presence and high
accessibility in WM. I discuss specific examples below. First, the data, with SC changes
highlighted in bold:

Table 2 Activation strategies (ICPs vs FPFs) and their associated SCsa
Clause
SC
ICP
FPF
Full NP
Clause
1
2a
2b
3a
3b
4a
4b
4c
5a
5a
5a
5b
6
6
7
8
8
8
8
8
9a
9b
10a
10b
10b
10b
(10c)
10d
10e
a
b

S4
S1
S1
S1
S1
S1
S1
S1
S3
S4
=
S1
S1
=b
=
S4
S1
S4
S1
S4
S1
=
=
S4
S4

X
X
X
X

X

X
X

X
X
X

X

10e
10f
10g
11
12
13
14
15
16a
16b
17
17
18
19
20
21a
21b
22
23a
23b
24
25
25
26
27
28a
28b
(28c)
28d

SC
S1
S1
S1
(S3)
S1
S1
S1
S4
S1
S1
S1
=
S1
S4
S4
S4
S3
(S4)
S1
S4
S1
S1
S1
S1
S4
S1
S4

ICP

FPF

Full NP

X
X

X
X
X
X

X
X
X
X
X

X

X
(X)
X

X

Entries with surrounding parentheses denote tentative assessments.
‘=’ denotes separate verbs in a shared SC environment that form a distinct, prosodically linked set.

X

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
Table 2 (Cont.) Activation strategies (ICPs vs FPFs) and their associated SCs
Clause
SC
ICP
FPF
Full NP
Clause
SC
29a
29b
29c
29c
29d
29e
30
31
32
33a
33b
34
35
36
37
38
(39a)
39b
40a
40b
40b
41a
41b
42a
42b
42c
42d
42d
43a
43a
43b
44
44
45
45
46a
46b
(46c)
(46d)
46e
47a
47a
47b
48
49a
49b
49c
50a
50b
50c
c

S4
S4
S4
S1
S4
S4
S4
S4
S4
S1
S4
S4
S4
S4
S4
(S3)
S1
S1
=
S4
S1
S1
S1
S4
S4
S1
S4
=
S3
S4
=
S1
=
S4
S4
(S4)
S4
=
S4
S4
S4
S1
S1
S4
S4
S4

X
X
X

X

(X)c
X
X
X
X
X

X
X
X
X
X
X

X

X

X
X

X

X

51
52a
52b
53
54
54
55a
55b
56a
56b
56c
57
58a
58b
58c
59a
(59b)
59c
60a
60b
61a
61b
61c
61c
62a
62b
62c
62d
(62e)
62f
63a
63b
63c
63d
63e
63f
63g
64
65
65
66
67a
67b
68
69a
69b
70
70
71a
71b

An alternative interpretation would have all of 33a as the NP for 33b.

S1
S4
S1
S4
S4
S1
S1
S3
(S4)
S1
S1
S4
S4
S4
S1
S4
(S4)
S1
S4
S1
(S1)
(S1)
S4
=
S4
S1
S1
S1
S1
S1
S1
S1
S1
S1
S1
S1
S4
S1
=
S4
S1
S4
S4
S4
S1
S4
=
S4
S3

59

ICP

FPF

Full NP

X
(X)

X

X
X
X
X
X

X
X

X
X

X
X

X
X
(X)
X
X
X

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
Table 2 (Cont.) Activation strategies (ICPs vs FPFs) and their associated SCs
Clause
SC
ICP
FPF
Full NP
Clause
SC
71c
72a
72b
72c
(72d)
73
74a
74b
74c

S1
S4
(S1)
(S4)
S4
S1
S1
S4
S4

(X)
X
X
X
(X)

74c
75a
75b
75c
76a
(76b)
77
78

S1
S1
S1
S4
S4
(S1)
(S4)

60

ICP

FPF

Full NP

X
X
X

In general, ICPs tend to occur near SC changes, with full NPs occurring at SC changes. FPFs
(which may lie between ICPs and full NPs in terms of encoding material) are difficult to pattern,
but they may be slightly more likely to occur near SC changes than at SC changes. Some FPF
use may be conditioned by the particular verbs used in discourse, or else speaker judgements as
to the accessibility of a referent – the idea being that referents that are particularly low in
accessibility may require an FPF over an ICP. Exploring this would require more data, which
falls outside the scope of this thesis.
The dataset I reference above is likely too small to count as definitive evidence.
Nevertheless, it provides compelling initial support for the theoretical arguments I have
developed to this point. Note in particular the examples below, representing clauses [63a]
through [65]:

(29)

(30)

a-kama
3PLcontinue

a-wur-shi
CL-come_outPL.NOMR1

i-shɔuhɔ

a-wur-shi
CL-come_out3PL-continue PL.NOMR1

i-shɔuhɔ
CLlikeness

a-kama

CL-likeness

[63a]

[63b]

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(31)

(32)

(33)

(34)

(35)

a-nyiŋ
3PLbe_one

[63c]

a-nyiŋ
3PLbe_one

[63d]

a-ba
3PLuntil come

[63e]

a-wur-sha
3PL-come_outX

[63f]

har

a-maa-bɔɔ
3PL-finish-3PL.ICP [63g]
‘They kept coming out like that; they kept coming out like that one by one till they
were all finished coming out.’

(36) a-koi
ha-no
o-cu
ha
a-mɛɛk-kɔ
e-nyaŋ
CL-forest
DEM-DIST CL-time DEM 3SG-turn-3SG.ICP
CL-ruin
‘That forest [i.e., where they lived] has now turned to ruins.’
(37)

61

a-mɛɛk
a-sho
e-nɛŋ
e
i-ram
3SG-turn 3SG-be CL-place AM CL-farming
‘It turned into a place of farming.’

[64]

[65]

Examples (29) through (34) have the same SC, and are not marked by increased encoding
material (i.e., they have no ICP, FPF, or Full NPs). This is particularly interesting, since the verb
in (31) is used with an ICP elsewhere – notably in clause [62d]:44

The verb in (34) doesn’t appear with an ICP in the Ishɛ data. However, it fits the Motional class of middle verbs,
and likely could be expressed with an ICP, much like the verb cikpi (‘come_down’), which has an ICP in [16a].
44

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(38)

a-ba-bɔɔ

ni

o-ku-ŋwe

3PL-come-3PL.ICP

LOC

CL-edge-DEF

62

[62d]

Example (35) does have an ICP,45 and it is immediately followed by an SC change in (36)
(which also has an ICP). Example (37) keeps the same SC as (36), and is expressed without an
ICP, even though the verb is the same. This seems consistent with what we might expect if we
assume that the ICP is being used to keep participants present and highly accessible in WM.
When there is a pending SC change and the previously activated participant is proceeding toward
preemption, the ICP is used to briefly reactivate it, thereby keeping it present and highly
accessible in WM across the intervening activation of a different referent.
This connection between ICP use and cognitive disjunctures is supported not just by the
SC change, but by how the lines are delivered: (33) through (35) form a group of prosodically
linked clauses, offset from surrounding groups by distinct caesura.46 This constitutes a distinct
local structure. Examples (33) and (34), which have verbs that could be expressed with ICPs, are
in the middle of a group of cognitively related clauses, and are therefore less likely to be
expressed with ICPs (assuming the DN function being explored in this section). Example (35)
does have an ICP, but it is situated at the local-structure boundary, offset from the next group by
caesura (and possibly intonational contour). These offset-establishing features are associated

45

This is the only time this verb is used in the discourse data. However, since ICPs are optional in Ishɛ, I believe its
use in clause [63g] is likely to be conditioned by its proximity to the SC change in [64]. ICP use here may also be
reinforced by the fact that “finish” suggests fully applied action (i.e., totality).
Example (35) is delivered with a slight lowering of terminal pitch. This produces a “continuation” intonational
contour – common throughout the discourse – that may signal the persistence of a larger cohesive group across
points of cognitive disjuncture. Analyzing all the language’s intonational contours is outside the scope of this thesis.
46

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

63

with extra cognitive processing (Swerts & Geluykens 1994: 34–35), which is a hallmark of
cognitive disjuncture. The absence of ICPs within a prosodic group and the presence of ICPs at
the boundaries of prosodic groups, immediately before SC changes, suggests that ICP use is
motivated – in at least some cases – by the need to maintain the referent’s presence and relatively
high accessibility in WM across the activation of another referent.
Another possible example of the ICP being used to navigate cognitive disjuncture in
discourse comes in clauses [68] through [70]:

(39) a-nya-he
3PL-refuse-

u-neke

a-kau-pi
CL-farmingAPL
CL-person NOMR1
‘They refuse to let anyone farm there;’

e

i-ram

AM

CL-farm

[68]

(40) ko
agɛ
e
i-nak
hã
k-a-yari-bɔɔ
even CL.PL:X AM CL.PL-cow NEG HAB-3PL-graze:X-3PL.ICP

i-nak
CL.PL-cow

[69a]

(41) a-jẽẽ-bɔɔ
hum ba
3PL-go-3PL.ICP there NEG [69b]
‘even the Fulani don't graze their cattle there.’
(42) domin
a-bin
a
a-tɛtɛ
because CL.PL-thing AM CL.PL-elder

sh-a-kõ
BM-3PL-be_old

nik
incɛŋ
up_to now

a-na
a-sho ni
hum
3PL-continue CL-be LOC there [70]
‘(This is) because the ancient things of the elders are still there.’

In these examples, cognitive disjuncture exists between (39) and (40) – i.e., clauses [68] and
[69a]. This is evident not only in the SC change between the two clauses, but also in the caesural

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

64

offset between the two.47 This cognitive disjuncture is navigated with help of the ICP. The ICP
is also present in (41), which immediately precedes another SC change in (42). Here again, the
existence of a cognitive disjuncture between (41) and (42) is evinced by the SC change and by
distinct caesura that offsets the one clause from the other.48 Examples (40) and (41) are both part
of the same local structure, and both have the same SC. Although it is possible that the ICP in
(41) is influenced by the ICP in (40) (given that the two form a single sentence), it is arguably
more likely that the ICP in (41) is influenced by the SC change immediately following it, in (42).
Partial support for this may potentially be found in clauses [63f] and [63g], which constitute
another [ – ICP ] / [ + ICP ] clause pair within a single sentence:49

(43)

a. a-wur-sha
3PL-come_outX
b.

[63f]

a-maa-bɔɔ
3PL-finish3PL.ICP
[63g]
‘(They kept coming out like that; they kept coming out like that one by one till)
they were all finished coming out.’

47

Also present: another possible example of “continuation contour”.

48

I believe (40) may also exhibit contour signaling cognitive disjuncture – though not the same contour as in (39).

49

Alternatively, it may be that the ICPs in (40) and (41) express mirativity, while the ICP in (43b) (i.e., [63g] ) may
express totality (as suggested by “finish”). With that said, totality in [63g] would not explain [ + ICP ] “turn” in
[64]. ICP use with “turn” is optional, and “turn” is actually [ – ICP ] in [65]. This suggests that use of the ICP in
[63g] and [64] is due to the SC change (and also that lack of the ICP in [65] is due to the lack of SC change). This
favors interpreting the ICP in (40) and (41) as also motivated by SC change, even if there is also a mirative reading.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

65

In any case, there is a definite cognitive disjuncture following (41), evident in the SC change at
(42). This change is expressed using a full NP, which is consistent with the general pattern in the
data.
In this data, FPFs and ICPs may arguably be in free variation. Consider what happens
when FPFs and ICPs occur in close proximity: examples (44a) and the first part of (44b) use the
same verb (‘go’) – a verb known to be expressed with ICPs – but in this case, (44a) has an ICP
and (44b) has an FPF. The same is true in (45a) and (45b), as well as in [16b] (prosodically
linked with [16a] ) and [17].

(44)

a.

b.

(45)

a-jẽẽ-bɔɔ
i-lau-kpe
e
imbɔɔ
3PL-go-3PL.ICP
CL-hunting-DEF
AM
3PL.PRO
‘They went to their hunting on the Nkojo side.’ [7]

u-dẽ
CL-side

in-kojok
CL-Nkojo

imbɔɔ
k-a-jẽẽ
a-jẽẽ
a-dur
i-jok
i-rɛ
3PL.PRO HAB-3PL-go
3PL-go 3PL-catch CL-guinea_fowl
3PL-be_many
‘They typically went out and caught many (guinea fowl here in the forest of
Nkojo).’ [8]

a.

a-jẽẽ-bɔɔ
i-lau
u-dẽ
o
o-bẽ
a
a-shɛ
3PL-go-3PL.ICP
CL-hunting
CL-side
AM
CL-land
AM
CL-Ashɛ
‘(They were hunters, and they came from Kagoro and) went hunting on Ashe
lands.’ [4c]

b.

imbɔɔ
3PL.PRO
na

bik
COND

a-jẽẽ
3PL-go

k-a-jẽẽ
HAB-3PL-go

a-tũ-ã
3PL-stay-X

ni
LOC

i-gõõ
CL-hill

a-jei
CL-Ajei
‘(When they went, they usually stayed a short time at the Hill of Ajei;) that’s the
one that is in Katugal.’ [5a]
AM

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(46)

a.

a-cikpi-i
3SG-come_down-3SG.ICP

ni
LOC

i-gõõ-i
CL-hill-DEF

[16a]

b.

a-jẽẽ-i
u-dẽ
a
a-tõr
3SG-go-3SG.ICP CL-side AM CL-Katugal
‘He came down the hill and he went to the Katugal side.’ [16b]

c.

iye
3SG.PRO

g-a-jẽẽ
X-3SG-go

a-dok
3SG-found

e-nɛŋ
CL-place

66

ni
LOC

o-shĩ-ĩ50
CL-underneath-DEF

e

i-gõõ-i
CL-hill-DEF
‘He went again and found a place below the hill.’ [17]
AM

Assume for the moment that for each of these clause pairs there is a) strong motivation to assert
the referent in WM, and b) strong motivation to not express subsequent (identical) verbs with the
same SC using an ICP in both places. This would explain why the first instance uses the ICP and
the second uses the FPF: they both assert the referent in WM, but they are different enough in
form that they can be used to avoid repetition. With that said, prosodic evidence suggests there
may be a local structure boundary between all these clauses. This would argue against free
variation, and would suggest rather that the ICP is still being used to navigate a change to the
contents of WM (as with S2-S4 type SC changes), only in this case, the newly activated referent
in the subsequent clause is actually the same referent as in the [ + ICP ] clause. This newly
asserted referent is expressed using the FPF. It is worth noting that the final clause in each of the
examples above includes non-Main Event lines; this is evident from use of the HAB marker. This

50

This seems to be a different word in the audio. I have maintained it as it was given in the original transcription.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

67

would potentially mean that the FPF is the preferred choice for reasserting already-activated
referents in events that fall outside the Main Event line, with ICPs preferred elsewhere. I suspect
that using the FPF in these cases may ultimately be due to the fact that FPFs reassert referents in
WM more strongly than ICPs, but less strongly than full NPs. In any case, since the ICP is
acting the same in S2-S4 cases (where it is typically followed by a full NP) as in these particular
S1 cases (where it is followed by an FPF), this might also indicate that the true measure of an SC
change is not simply whether the participant is different, but whether it is treated as different:
i.e., by being given expression with a different (and higher) amount of encoding material than
would otherwise be expected. More data would likely clarify this relationship.51
4. Conclusion
My intent in this thesis has been to show that ICPs – with all their various manifestations,
environments, and constraints – are unified by a relatively simple cognitive structure. This
structure is best understood in terms either taken directly from or else influenced by cognitive
linguistics. It is ultimately this structure that accounts for the language-to-language variations
we see in the ICP. In order to demonstrate this, I first argued that ICPs are a forcereceiving mental instantiation of single “real” entity, which I defined as the base participant.
This force-receiving participant (i.e., [ – A ] ) is distinct from the force-exerting (i.e., [ + A ] )
participant, though both are directly linked to the same base participant. I showed how middle
verbs (as defined) lend themselves particularly easily to expression with both a [ – A ] and a
[ + A ] participant. While acknowledging that ICPs often occur with middle verbs, I nevertheless
demonstrated on theoretical grounds that the ICP is licensed with “non-middle” verbs. This was

51

A good starting point for future research analyzing the DN function could be the tone-marked discourse found in
Marggrander (2018), which comes from Dũya (Benue-Congo, Nigeria), a language closely related to Ishɛ.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

68

supported by a number of potential examples from the literature, though further research needs to
be done to validate these.
Starting from this theoretical perspective, I then analyzed various functions of the ICP.
Some of these were taken from the literature; others were new (or at least newly categorized)
functions of my own making. I leveraged my thesis’ mental space structure to show how ICPs
are distinct from cognate objects. I then used this same structure to further validate the
relationship between ICPs and Subject Point-of-View (SPOV). This is particularly important
because although the SPOV function has proven popular in recent works, it hasn’t yet been given
a particularly thorough treatment in the literature. My hope is that the theoretical foundation I
have built in my thesis will open the door to further research in this direction. Finally, I explored
the various ways the ICP is used to help navigate cognitive disjunctures. Although other
researchers have seen many of these same functions, it seems there hasn’t been a significant
effort to this point to determine whether or how those functions are related. As I have shown in
this thesis, I believe these various functions, different as they seem, are all outworkings of the
same cognitive structure that helps explain ICPs more generally.
As I see it, my thesis is a significant departure from the historical trend. If the main
question of ICP research to this point has been “what are ICPs?” or “what do ICPs do?”, then my
question here has been “how do ICPs work?”. This entails, to some extent, also answering the
“what” questions. I believe ICPs are a force-receiving participant index, a [ – A ] mental
instantiation of the base participant, to which the “subject index” (i.e., the [ + A ] index) is also
linked. I believe ICPs help encode SPOV (as I’ve defined it here). I also believe that ICPs allow
speakers to manipulate how they understand a participant’s relationship to the verbal action that
same participant has performed. And since this understanding is ultimately received by the

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

69

hearer, I believe the ICP is a type of “out of band signaling”52 that interacts with the
interlocutors’ shared cognitive environment, even as its various manifestations are shaped by
other cognitive and linguistic information. Though further research is required to completely
answer the “what” questions, it should be clear by now that these cannot be answered apart from
addressing the “how” of ICPs. I believe my thesis stands as a useful push in this direction. With
further research built on more complete data – especially natural discourse data – I believe it will
be possible to move the discussion of ICPs forward significantly, and to provide theorygrounded, testable answers to help make sense of ICPs used in unexpected places and ways.

52

I.e., an exchange of information about a message sent and received separately from the message itself. In RF
communications, this occurs on a different channel; in cognitive linguistics, this occurs on a distinct cognitive
pathway.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
Abbreviations
02

Class 02

FUT

Future

03

Class 03

HAB

Habitual

06A

Class 06a

ICP

Intransitive Copy Pronoun

1

First person

INF

Infinitive

2

Second person

INT

Interrogative particle

3

Third person

INTS

Intensifier

ACC

Accusative

LOC

Locative

ADD

Additive

M

Masculine

AM

Associative Marker

NDEF

Indefinite

AOR

Aorist

NEG

Negative

APL

Applicative

NOMR

Nominalizer

ASC

Associative

NOMR1(/2)

Nominalizer 1 (or 2)

AUX

Auxiliary

PERF

Perfect

AUX1

Auxiliary 1

PL

Plural

BM

Boundary Marker

POSS

Possessive

CL

Classifier

PRF

Perfect

CLM

Class marker53

PRO

Pronoun

COM

Comitative

PROX

Proximal

COND

Conditional

PURP

Purposive

COP

Copula

QM

Quotative Marker

DEF

Definite

REFL

Reflexive

DEM

Demonstrative

REL

Relative marker

DIST

Distal

SBJV

Subjunctive

DM

Discourse marker

SC

Subject Concord

F

Feminine

SG

Singular

FOC

Focus

X

Unknown

53

Tentative assessment; CLM also appears as CLM.ADD and CLM.PURP. An alternative is “Clause Marker”.

70

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

71

Appendix
The following table represents my best assessment of the clause breaks in the Ishɛ discourse
data. This data consists of approximately six minutes of discourse (comprising a total of 78
lines) recorded by Mr. Arams Zachariah of SIL Nigeria on 11 June, 2018. It has been shared
under a Creative Commons Attribution-ShareAlike 4.0 license. The data was originally
transcribed by one or more of David Rowbory, Moses Sabo, and Kathleen Spence. With the
exception of a few relatively minor emendations (and a fairly extensive restructuring of the
data’s clausal structure), I have kept the gloss as close as I could to the original that was provided
to me. It also means that the data does not indicate tone, as it was provided to me without tone
marking. Ishɛ is a tonal language and it is possible that information pertinent to understanding
ICPs could be carried by tone. Analyzing the data for tone is beyond the scope of this thesis;
with that said, there seems to be significant patterns of sentence- or clause-level intonational
contour that factor into my analysis of the DN function of ICPs (see §3.3.2). Furthermore, as the
original analysis was a work-in-progress, there are some irregularities in the gloss consistent with
the ongoing process of morphosyntactic evaluation. Throughout this table, an “X” represents a
morphosyntactic element of unknown function; this represents either the assessment of the
original glossing team or my own assessment, based on questions I have concerning the text. In
light of my analysis of ICPs, and in light of the theoretical framework I use throughout this
thesis, I have some thoughts on the function of the several “X” morphosyntactic elements. With
that said, a full analysis of the grammar of Ishɛ lies beyond the scope of this work. Though the
focus of this appendix is on clauses, the included free translations are indexed to whole lines, and
are provided either after individual clauses or after the last clause in a lettered sub-group.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(1)

u-tarihi
o
u-kojok
mi
CL-history AM CL-Nkojo DEM.PROX
‘The history of Nkojo here is like this:’

(2a)

ɛti

72

nɛ
u-sho
i- shɔu
here 3SG-be CL- likeness

ha-ne
DEM-PROX

QM

(2b)

ki

sh-e
esha
INTS BM-3SG X:be_long_ago
‘It was long, long ago.’

(3a)

e-ner
CL.PL-person

e
AM

i-hwɛi bom
CL-two be_there

(3b)

a-sho
ni
hum
3PL-be LOC there
‘There were two people.’

(4a)

imbɔɔ
a-sho
3PL.PRO 3PL-be

a-lau
CL.PL-hunter

(4b)

k-a-wur
X-3PL-come_out

ni
LOC

u-kagoro
CL-Kagoro

(4c)

a-jẽẽ-bɔɔ
i-lau
u-dẽ
o
o-bẽ
a
a-shɛ
3PL-go-3PL.ICP CL-hunting CL-side AM CL-land AM CL-Ashɛ
‘They were hunters, and they came from Kagoro and went hunting on Ashe lands.’

(5a)

imbɔɔ
bik
3PL.PRO COND

a-jẽẽ
k-a-jẽẽ
3PL-go HAB-3PL-go

a-tũ-ã
3PL-stay-X

ni
LOC

i-gõõ
CL-hill

na
AM

a-jei
CL-Ajei

(5b)

i-gõõ
nyom
a-tõr
no
CL-hill be_there CL-Katugal DIST
‘When they went, they usually stayed a short time at the Hill of Ajei; that’s the one that
is in Katugal.’

(6)

ni

a-aak
a-daak
hum
CLM 3PL-be_then 3PL-descend there
‘Then they descended there at that time.’

o-cu
CL-time

ha
DEM

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(7)

(8)

a-jẽẽ-bɔɔ
i-lau-kpe
e
imbɔɔ
u-dẽ
3PL-go-3PL.ICP CL-hunting-DEF AM
3PL.PRO CL-side
‘They went to their hunting on the Nkojo side.’
imbɔɔ
k-a-jẽẽ
3PL.PRO HAB-3PL-go

a-jẽẽ
3PL-go

a-dur
3PL-catch

73
in-kojok
CL-Nkojo

i-jok
CL-guinea_fowl

i-rɛ
3PL-be_many

i-rɛ
ni
iyɔɔ
ni
a-koi
a
a-kojok
3PL-be_many LOC
DEM.PRO LOC
CL-forest
AM
CL-Nkojo
‘They typically went out and caught many guinea fowl here in the forest of Nkojo.’
(9a)

o-cu
CL-time

(9b)

ni

o-jẽẽ
3SG-go

o-ba
3SG-come
‘Time went on.’
DM

(10a)

u-nyiŋ
CL-one

ni
LOC

a-hɛɛ
3SG-say
(10b)

(10c)

ime
1SG.PRO
ni
LOC

(10d)

e
REL

(10e)

sh-u
BM-3SG
shi
FUT

e
AM

imbɔɔ
u-ŋwɛ
3PL.PRO CL-child

kõ-i
be_old-NOMR1

sh-u
BM-3SG

QM

in-jẽẽ
in-jẽẽ in-nɔɔ
ni
1SG-go 1SG-go 1SG-sleep LOC

i-jok
CL.PL-guinea_fowl

ni

ni
CLM.ADD

in-nɛhe
1SG-avoid

shi
FUT

a-koi
CL-forest

k-i-sho
HAB-3PL-be

in-gui
1SG-return

in-hwa
i-hãrã
1SG-drink CL-suffering

ciki-i
be_young-NOMR1

ɛti

e-nɛŋ
CL-place

CLM.PURP

(10f)

o-shɔ
CL-inside

e
AM

e-wurta-ko
CL-come-NOMR

ha-no
DEM-DIST

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(10g)

ankaŋ

in-jẽẽ
NDEF
1SG-go
‘The first of them, the younger one, said to the older one, “I will go and settle in the
forest where the guinea fowl are typically found, so that I may avoid repeating the
suffering of coming out here whenever I go.’
54

(11)

iye
a-shok
hum
o-te
ha
3SG.PRO 3SG-get_up there CL-time DEM
‘He got up to go there at that time.’

(12)

a-jẽẽ-i
3SG-go-3SG.ICP
‘He went.’

(13)

jẽẽ

(14)

ni

e

CLM

AM

(15)

(16a)

(16b)

a-tɛɛr
a-teu
a55 a-koi
a
a-kojok
DM 3SG-clear CL-center AM CL-forest
AM CL-Nkojo
‘He cleared the center of the forest of Nkojo.’
a-jin
a-ɛi-le
3SG-build CL-house-DEF
‘He built his house there.’

iye
3SG.PRO

u-ner
e
i-hwɛi-le
a-shok
CL-person
AM CL-two-DEF 3SG-get_up
‘The second person himself got up.’
a-cikpi-i
3SG-come_down-3SG.ICP

ni
LOC

e
AM

ni
LOC

hum
there

iye
3SG.PRO

i-gõõ-i
CL-hill-DEF

a-jẽẽ-i
u-dẽ
a
a-tõr
3SG-go-3SG.ICP CL-side AM CL-Katugal
‘He came down the hill and he went to the Katugal side.’

54

From Hausa.

55

I hear an intervening sound in the audio that wasn’t transcribed. I have left this as it was given to me.

74

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(17)

iye
3SG.PRO

g-a-jẽẽ
a-dok
X-3SG-go 3SG-found

e-nɛŋ
CL-place

ni
LOC

75

o-shĩ-ĩ56
CL-underneath-DEF

e
AM

i-gõõ-i
CL-hill-DEF
‘He went again and found a place below the hill.’
(18)

a-jin
a-ɛi-le
3SG-build CL-house-DEF
‘He built his house there.’

e
AM

iye
o-teu-kpe
AM 3SG.PRO
CL-village-DEF
‘His village is called Katugal.’

iye
3SG.PRO

hum
there

(19)

e

(20)

u-yɔɔ
nɔko sh-a-jẽẽ-i
ni
a-koi
X-DEM.PRO other BM-3SG-go-3SG.ICP LOC CL-forest
‘This other one went to that forest.’

(21a)

e
AM

iye
3SG.PRO

o-teu-kpe
CL-village-DEF

o-gõõ-ni
3SG-be_called-X

o-gõõ-ni
3SG-be_called-X

a-tõr
CL-Katugal

ha-no
DEM-DIST

a-koi
CL-forest

e
AM

i-jok
CL-guinea_fowl

(21b)

hɛre
i-jok
k-i-sho
ni
a-koi-she
because CL-guinea_fowl HAB-3SG-be LOC CL-forest-DEF
‘His village is called Akoi e Ijok [Forest of the Guinea Fowl], because the guinea fowl
are usually in the forest.’

(22)

iyɔ

(23a)

iye
3SG.PRO

(23b)

ɛti

i-tuk
PRO.DEM 3SG-put
‘This one puts (the name) [i.e., this one names it].’
a-tar-ke
3SG-choose-APL

e-cok-e
CL-name-DEF

a-koi
e
i-jok
CL-forest
AM
CL-guinea_fowl
‘He chooses the name Akoi e Ijok.’
QM

56

This seems to be a different word in the audio. I have maintained it as it was given in the original transcription.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(24)

imbɔɔ
a-kojok
a-shok
3PL.PRO CL.PL-Nkojo 3PL-get_up
‘The Nkojo's, they got up.’

(25)

a-kama
a-na
a-ŋu-i
3PL-continue 3PL-continue CL-live-NOMR1
‘They still keep living in that forest.’

ni

(26)

a-kau
i-ram-ŋwe
e
imbɔɔ
ni
3PL-farm CL-farm-DEF AM 3PL.PRO LOC
‘They farmed their farms there;’

hum
there

(27)

ni

LOC

a-kpea ko
u-ŋwe
e
imbɔɔ
ni
3PL-do NDEF57 CL-what AM 3PL.PRO LOC
‘they did whatever they had to do there.’
CLM

(28a)

bik
COND

u-jẽẽ
ni
2SG-go LOC

(28b)

sh-u-jɛ
FUT-2SG-see

(28c)

e-nɛŋ
CL-place

(28d)

e

a-koi-she
CL-forest-DEF

e-nɛŋ
CL-place

57

hum
there

incɛŋ
now

imbɔɔ
shi k-a-sher
i-yɛi-le
e
imbɔɔ
3PL.PRO BM HAB-3PL-sit CL-discussion-DEF AM 3PL.PRO
‘If you go to that forest even up till now, you'll see a certain place, a place where they
usually sit for their discussion.’
bik
COND

(29b)

nik
up_to

ha-no
DEM-DIST

e-com
CL-certain

REL

(29a)

a-koi
CL-forest

76

a-sho
3PL-be

u-ba
3SG-come

From Hausa.

ni
LOC

u-bin
CL-thing
o-teu-kpe
CL-village-DEF

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(29c)

a-bik
3PL-COND

a-bera
3PL-want

(29d)

akwai
i-tɛi-le
there_is CL.PL-stone-DEF

(29e)

e

77

a-dam
CL-announcement

imbɔɔ
shi k-a-sher
ni
hum
3PL.PRO BM HAB-3PL-sit LOC there
‘If something [i.e., some issue] comes to the village and they want an announcement,
then there are the stones where they usually sit (to do that).’
REL

(30)

nik
incɛŋ ime
sh-in-na
a-yei
up_to now 1SG.PRO BM-1SG-continue CL.PL-discussion
‘Up till now, as I continue my discussion here:’

(31)

i-tɛi
ha-no
i-nɔɔ
ni
CL.PL-stone DEM-DIST 3SG-lie LOC
‘Those stones sill lie there.’

(32)

ni

(33a)

PROX

hum
there

o-shɔ
e
i-tɛi
ha-no
ke-dik58 e-sher-ti
LOC CL-inside AM
CL.PL-stone
DEM-DIST 1PL-just CL-sit-NOMR1
‘Among those stones, it's not just, "Sit (wherever)".’
ko

be

NDEF59

ASC.PL

i-mar-ke
CL-clan-DEF
(33b)

nɛ

e
AM

i-nyaa
CL-who

ni
LOC

a-ɛi-le
CL-house-DEF

e
AM

imbɔɔ
3PL.PRO

ni
LOC

imbɔɔ
3PL.PRO

a-sho
ni
o-tɛi-le
e
imbɔɔ
3PL-be LOC CL-stone-DEF AM 3PL.PRO
‘Whoever's in their house, in their clan, is at their (own) stone’

58

This is clearly preceded by another word, possibly hã (NEG) used as part of the negation paradigm.

59

From Hausa.

ba
NEG

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(34)

a-hãmbazo
a-sho
ni
o-tɛi-le
CL-Hãmbazo
3PL-be LOC CL-stone-DEF
‘The Ahãmbazo are at their stone.’

e
AM

78

imbɔɔ
3PL.PRO

(35)

a-nɔr
e
in-bini a-sho
ni
o-tɛi-le
e
imbɔɔ
CL.PL-south AM CL-Bini 3PL-be
LOC CL-stone-DEF AM 3PL.PRO
‘The Southerners of the Inbini (clan) are at theirs.’

(36)

a-nɔr
e
i-jim
a-sho
ni
o-tɛi-le
CL.PL-south AM CL-Jim 3PL-be LOC CL-stone-DEF
‘The Southerners of the Ijim (clan) are at theirs.’

(37)

(38)

e
AM

a-nɔr
e
i-pada
a-sho
ni
o-tɛi-le
CL.PL-south AM CL-Pada
3PL-be LOC CL-stone-DEF
‘The Southerners of the Pada (clan) are at theirs.’
a-gar
a
a-nkpaŋ
CL.PL-side AM CL.PL-Nyankpa
‘The Nyankpa side came later.’

(39a)

e imbɔɔ
u-tarihi-n
X 3PL.PRO CL-history-AM60

(39b)

ɛti

imbɔɔ
3PL.PRO

e
AM

imbɔɔ
sh-a-ba
3PL.PRO BM-3PL-come

ni
LOC

imbɔɔ
3PL.PRO

o-bur
CL-later

imbɔɔ
a-sho
a
a-nkpaŋ
3PL.PRO 3PL-be AM CL.PL-Nyankpa
‘Their history says they are of the Nyankpa.’
QM

(40a)

imbɔɔ
bik
3PL.PRO COND

(40b)

imbɔɔ
k-a-ba
a-sher e
imbɔɔ
ni
3PL.PRO HAB-3PL-come 3PL-sit AM 3PL.PRO LOC
‘When they come, they usually sit at the entrance.’

(41a)

a-kojok
a-nɛnɛ
CL.PL-Nkojo 3PL-take

60

From Hausa.

a-ba
3PL-come

imbɔɔ
ni
3PL.PRO LOC

o-cu
CL-time

ha
DEF

a-dɛi
CL.PL-entrance

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(41b)

imbɔɔ
a-sho
a-ner
e
i-jɛu
3PL.PRO 3PL-be CL-people AM CL-watching
‘The Nkojo's take them at that time and serve as the watchmen.’

(42a)

imbɔɔ
k-a-dẽẽ
3PL.PRO HAB-3PL-stand

(42b)

ni
CLM.ADD

(42c)

bik
COND

ni
LOC

a-dɛi
CL.PL-entrance

a-jɛ
3PL-see

a-sho
3PL-be

u-bin
CL-thing

(42d)

sh-u-biwi
u-ba
FUT-3SG-be_bad 3SG-come
‘They usually stand at the entrances in order to watch and see if anything bad might
be coming.’

(43a)

imbɔɔ
a-aak
3PL.PRO 3PL-be_then

a-hɛɛ-he
3PL-say-APL

a-neke
CL-people

(43b)

sh-a-sho
ni
o-shɔ-i
BM-3PL-be LOC CL-inside-DEF
‘They then tell their people that they're with.’

(44)

k-a-kama
a-kpe-a
HAB-3PL-continue 3PL-do-X
‘They keep on doing that;’

inom
like_that

(45)

k-a-kama
a-kpe-a
HAB-3PL-continue 3PL-do-X
‘they keep on doing that.’

inom
like_that

(46a)

akwai
e-nɛŋ
there_is CL-place

(46b)

ke-shi
1PL-BM

ke-hɛɛ
1PL-say

e-com
CL-certain

hum
there

nik
up_to

e
REL

incɛŋ
now

imbɔɔ
3PL.PRO

79

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(46c)

ɛti
QM

o-tai
o
u-ngɛlɛ
CL-cave AM CL-Ungɛle

(46d)

e-nɛŋ
CL-place

(46e)

e

inte
shi k-e-jẽẽ
e-wa
wa
a-wa-a
1PL.PRO BM HAB-1PL-go 1PL-bathe bathe CL-bathe-NOMR2
‘There is a certain place there up till that now that we call the Cave of Ungɛlɛ, a place
where we usually go do our bathing.’
REL

(47a)

nik
up_to

(47b)

ɛti

(48)

amma
inte
nahã
e-jɛ-yir
however 1PL.PRO PERF.NEG 1PL-see-1PL.ICP
‘However, we have never seen the place.’

(49a)

bik

incɛŋ sh-a-kõ
now BM-3PL-be_old

a-hɛɛ-wa
3PL-say-APL

inte
1PL.PRO

e-nɛŋ
e-com
e-sho
ni
hum
QM CL-place CL-certain
3SG-be LOC there
‘Even now, the elders tell us that another place is there [i.e., possibly that the Cave of
Ungɛlɛ is actually the location of another important place].’

COND

a-shei
3PL-be_not

imbɔɔ
3PL.PRO

(49b)

sh-a-kõ
BM-3PL-be_old

(49c)

a-hwɛŋ
e-nɛŋ-e
3PL-know CL-place-DEF
‘Only the elders know the place.’

(50a) ni
LOC

(50b)

80

o-cu
CL-time

i-ba
3SG-come

ha

bik

DEM

COND

u-kojok
CL-Nkojo

a-sho
3PL-be

i-wa
CL-war

e-nɛŋ-e
CL-place-DEF

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(50c) ko cak e
imbɔɔ
k-a-tõrõõ
u-tɔk-e
X all
AM
3PL.PRO HAB-3PL-line_up CL-together-DEF
‘At that time, if war were to come to the Nkojo's, they would all line up together.’
(51)

a-nyi-bɔɔ
hum
3PL-enter-3PL.ICP there
‘They entered there.’

(52a) a-ner
CL.PL-person

e
AM

i-wa
CL-war

ha-no
DEM-DIST

bik

a-ba
3PL-come

COND

(52b)

shina
a-doka
imbɔɔ
ba
FUT:NEG 3PL-find:X 3PL.PRO NEG
‘If any warriors were to come, they wouldn't find them;’

(53)

ni

g-a-ci-he
o-hwei ha-no
CLM
X-3PL-cover-APL CL-hole DEM-DIST
‘they would cover that hole using a millstone.’

(54)

sh-a-cɛi
k-a-dẽẽ
BM-3PL-be_female HAB-3PL-stand
‘The women usually stand outside;’

(55a)

ni
CLM

a-hɛ-he
3PL-say-APL

(55b)

e

(56a)

kɔtɔ inom
live like_that

(56b)

a-kɔtɔ
3PL-live

ni
COM

o-kɔk
CL-millstone

i-jɛu-kpe
CL-outside-DEF

a-neke
CL-people

imbɔɔ
sh-a-sho
ni
o-shɔ-i
REL 3PL.PRO BM-3PL-be LOC CL-inside-DEF
‘they tell the people who are inside (if something bad is coming).’

inom
like_that

(56c) a-kɔtɔ
inom
3PL-live like_that
‘(They) live like that; they live like that; they live like that.’

81

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(57) tɔ
i-ram-ŋwe
e
imbɔɔ
ni
o-cu
well CL-farming-DEF AM 3PL.PRO LOC CL-time
‘Well, at that time they farmed corn.’
(58a) kita a-ba
later 3PL-come
(58b) o-yer
CL-issue

o
AM

ni
LOC

ha
DEM

82

i-sho
a-gutamai
3SG-be CL-maize

o-bur
CL-later

u-citta
CL-ginger

o-ba
3SG-come

(58c) o-nyi
3SG-enter
‘Then later the deal with the ginger came on the scene.’
(59a) a-kojok
CL.PL-Nkojo
e
AM

a-kau
3PL-farm

u-citta
CL-ginger

ni
LOC

a-koi
ha-no
CL-forest DEM-DIST

shɔu-kpe
likeness-DEF

i-shɛɛ
CL-nonsense

(59b) tun ko
o-cu-i
61
until NDEF
CL-time-DEF
imbɔɔ62 k-a-nee
e-ɔrɔ
u-citta
ni
i-cɔk
REL 3PL.PRO HAB-3PL-continue
CL-scrape CL-ginger COM
CL-spoon
‘The Nkojo's farmed ginger like crazy in that forest, till they'd be sitting there scraping
ginger with a spoon.’

(59c) e

(60a) inte
nɛ
ke-shok
1PL.PRO PROX 1PL-get_up
(60b)

ke-ɔrɔ
u-citta
ha
ni
i-cɔk
1PL-scrape CL-ginger DEM COM CL-spoon
‘We got up and scraped ginger with a spoon.’

61

From Hausa.

62

Although this seems to be exceptionally reduced in the audio, I have kept the original gloss as it was given to me.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(61a)

kpe-a
do-X

inom
like_that

(61b) kpe-a
do-X

inom
like_that

(61c) har a-shu
a-ba
a-tan
until CL.PL-eye 3PL-come 3PL-open
‘They did that; they did that till their eyes came to be opened.’
(62a) imbɔɔ
a-shai
3PL.PRO 3PL-begin

a-wur-ti
CL-come_out-NOMR1

(62b) a-nyiŋ
3PLbe_one
(62c) a-nyiŋ
3PLbe_one
(62d) a-ba-bɔɔ
3PL-come-3PL.ICP
(62e) ni
LOC

ni
LOC

o-ku-ŋwe
CL-edge-DEF

e-nɛŋ
e
CL-place REL

(62f) imbɔɔ
a-kojok
sh-a-nɔ
inceŋ
3PL.PRO CL.PL-Nkojo BM-3PL-settle now
‘They started coming out one by one, coming to the edge of the place where the
Nkojo's are now settled.’
(63a) a-kama
a-wur-shi
3PL-continue CL-come_out-PL.NOMR1
(63b) a-kama
3PL-continue

a-wur-shi
CL-come_out-PL.NOMR1

i-shɔuhɔ
CL-likeness
i-shɔuhɔ
CL-likeness

83

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

84

(63c) a-nyiŋ
3PL-be_one
(63d) a-nyiŋ
3PL-be_one
(63e) har a-ba
until 3PL-come
(63f) a-wur-sha
3PL-come_out-X
(63g) a-maa-bɔɔ
3PL-finish-3PL.ICP
‘They kept coming out like that; they kept coming out like that one by one till they
were all finished coming out.’
(64)

a-koi
ha-no
o-cu
ha
a-mɛɛk-kɔ
e-nyaŋ
CL-forest DEM-DIST CL-time
DEM
3SG-turn-3SG.ICP CL-ruin
‘That forest [i.e., where they lived] has now turned to ruins.’

(65)

a-mɛɛk
a-sho
e-nɛŋ
e
i-ram
3SG-turn 3SG-be CL-place AM CL-farming
‘It turned into a place of farming.’

(66) ni
LOC

a-teu
CL-center

a
AM

a-koi-she
CL-forest-DEF

nik
up_to

incɛŋ hã
now NEG

k-a-kaupi-bɔɔ
HAB-3PL-farm:X-3PL.ICP

i-ram
ni
hum ba
CL-farming LOC there NEG
‘At the center of that forest till now, they don't usually farm there.’
(67a) k-a-hɛɛ
X-3PL-say
(67b) ɛti

a-koi
e
i-naŋbishi
kom
no
ni
hum
QM CL-forest AM CL-masquerade be_there DIST LOC there
‘They say, “That's where the Forest of the Masquerade is.”’

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(68) a-nya-he
3PL-refuse-APL
(69a)

ko
even

u-neke
CL-person

85

a-kau-pi
e
i-ram
CL-farming-NOMR1 AM CL-farm

agɛ
e
i-nak
CL.PL:X AM CL.PL-cow

hã
NEG

k-a-yari-bɔɔ
HAB-3PL-graze:X-3PL.ICP

i-nak
CL.PL-cow

(69b) a-jẽẽ-bɔɔ
hum ba
3PL-go-3PL.ICP there NEG
‘even the Fulani don't graze their cattle there.’
(70)

domin
a-bin
because CL.PL-thing

a
AM

a-tɛtɛ
sh-a-kõ
CL.PL-elder BM-3PL-be_old

nik
up_to

incɛŋ
now

a-na
a-sho ni
hum
3PL-continue CL-be LOC there
‘(This is) because the ancient things of the elders are still there.’
(71a) iyɔɔ
DEM.PRO

i-tuk
3SG-put

a-kojok
ni
CL.PL-Nkojo COM

a-tõr
CL.PL-Katugal

(71b) a-sho a-ndɔk
3PL-be CL.PL-friend
(71c) a-g-a-sho
a-ŋwɛ-bɔɔ
3PL-X-3PL-be CL.PL-brother-3PL.ICP
‘This all makes the Nkojo's and the Katugal's friends; it makes them brothers.’
(72a) i-naŋgbishi
CL-masquerade
(72b) bik
COND

bik63
COND

sh-a-tuk
FUT-3SG-put

e-wur-ta
3SG-come_out-X

ni
LOC

a-tõr
CL-Katugal

incɛŋ
now

i-tasha64
CL-Itasha

63

I believe I hear bik a-sho (COND 3PL-be), though I have left the gloss as it was given to me.

64

This appears to be a different word in the audio. I have left the gloss in its original form, however.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(72c) i-naŋgbishi
CL-masquerade

a
AM

a-kojok
CL.PL-Nkojo

bik
COND

a-nahã
3SG-not_yet

86

a-jẽẽ-i
CL-go-NOMR1

(72d) tɔ
a-kama
a-jɛu-pi
e
imbɔɔ
kom65
well 3PL-continue CL-watch-NOMR1 AM 3PL.PRO certain
‘If the masquerade goes on among the Katugal, if they put on the Itasha memorial,
and if the Nkojok's have not yet come out, then they keep watching for them there;’

(73)

sai

a-kojok
a-jẽẽ
DM CL.PL-Nkojo 3PL-go
‘(to see) if the Nkojo's go.’

(74a) bik

shi

COND

g-a-kpea
X-3PL-do:X

FUT

(74b) i-naŋgbishi
CL-masquerade

a
AM

ni
LOC

in-kojok
CL-Nkojo

a-tõr
bik
a-nahã
CL.PL-Katugal COND 3SG-not_yet

a-jẽẽ-i
CL-go-NOMR1

u-kojok
CL-Nkojo
(74c) sai
DM

a-jɛu
3PL-watch

(74d) sai

a-ba
DM 3PL-come
‘If the Nkojok's intend do the masquerade again, and the masquerade of the Katugal's
has not yet come out to Nkojo, (then) they watch (till) they come.

(75a) kita a-na
a-kama
later 3PL-continue CL-continue\NOMR

65

ume

e

DEM.PROX

AM

e-sha
CL-long_ago

a
AM

a-yitik
CL-remember\NOMR

o
AM

u-gai
CL-tree

ha-no
DEM-DIST

This is unclear; it is present (though unglossed) in the transcript, but it’s unclear if it exists in the associated audio.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS
(75b) hɛ
e
because AM

87

imbɔɔ
sh-a-wur
3PL.PRO BM-3PL-come_out

(75c) e-nɛŋ
e-nyiŋ
i-hwɛi ne
e
imbɔɔ
CL-place
3SG-be_one CL-two PROX AM 3PL.PRO
‘Since then, they've kept the memory of that tree from long ago, because the two
of them came out from the same place.’

(76a) u-ner
CL-person

a-tuk
3SG-put

e

in-kojok

ni

u-ner

a

AM

CL-Nkojo

COM

CL-person

AM

a-tõr
CLKatugal

iteŋgashe
CL-Lord

i-hwɛi~hwɛi
CL-two~two

(76b) i-nyom
e
imbɔɔ
no
CL-certain AM
3PL.PRO DIST
‘(For) the man of Nkojok and the man of Katugal: may the Lord make them both
as one.’
(77)

i-taka
a-le-i
e
i-hĩĩdẽ
3SG-add CL-eat-NOMR1 AM CL-success
‘(May the Lord) add success (to them).’

(78) o-sho
i-shɔuhɔ
3SG-be CL-likeness
‘Amen [lit., “may it be like that].’

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

88

References
Amberber, Mengistu. 2000. Valency-changing and valency-encoding devices in Amharic. In R.
M. W. Dixon & Alexandra Y. Aikhenvald (eds.), Changing valency: Case studies in
transitivity, 312–332. Cambridge: Cambridge University Press.
Anderson, Gregory D. S. 2011. Auxiliary verb constructions in the languages of Africa. Studies
in African Linguistics 40(1 & 2). 1–258.
Asongwed, Tah. 1980. Sentence negation in Ngamambo. In Larry M. Hyman, Jan Voorhoeve &
Luc Bouquiaux (eds.), L’Expansion Bantoue: Actes du colloque international du CNRS,
Viviers (France), 4-16 Avril 1977 (Colloques Internationaux Du Centre National de La
Recherche Scientifique 9), 555–561. Paris: SELAF.
Atindogbé, Gratien & Evelyn Chibaka. 2011. Pronoun serialization in Kamtok: The syntaxsemantic interface. In Anne Storch, Gratien Atindogbé & Roger Blench (eds.), Copy
pronouns: Case studies from African languages (Kay Williamson Educational
Foundation. African Languages Monographs 3), 193–212. Köln: Rüdiger Köppe Verlag.
Bakker, Egbert. 1994. Voice, aspect and Aktionsart: Middle and passive in Ancient Greek. In
Barbara A. Fox & Paul J. Hopper (eds.), Voice: Form and function (Typological Studies
in Language 27), 23–37. Amsterdam: John Benjamins Publishing Company.
Bresnan, Joan & Jonni Kanerva. 1989. Locative inversion in Chicheŵa: A case study of
factorization in grammar. Linguistic Inquiry 20(1). 1–50.
Campbell, Akua. 2017. A grammar of Gã. Houston, TX: Rice University Ph.D.
Chafe, Wallace. 1987. Cognitive constraints on information flow. In Russell S. Tomlin (ed.),
Coherence and grounding in discourse: Outcome of a symposium, Eugene, Oregon, June
1984 (Typological Studies in Language 11), 21–51. Amsterdam: John Benjamins
Publishing Company.
Croft, William. 1994. Voice: Beyond control and affectedness. In Barbara A. Fox & Paul J.
Hopper (eds.), Voice: Form and function (Typological Studies in Language 27), 89–117.
Amsterdam: John Benjamins Publishing Company.
Croft, William. 1998. The structure of events and the structure of language. In Michael
Tomasello (ed.), The new psychology of language: Cognitive and functional approaches
to language, 67–92. Mahwah, NJ: Lawrence Erlbaum Associates.
Croft, William. 2012. Verbs: Aspect and causal structure (Oxford Linguistics). Oxford: Oxford
University Press.
Culicover, Peter W. & Ray Jackendoff. 2005. Simpler syntax (Oxford Linguistics). Oxford:
Oxford University Press.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

89

Davis, Anthony R. & Jean-Pierre Koenig. 2000. Linking as constraints on word classes in a
hierarchical lexicon. Language. Linguistic Society of America 76(1). 56–91.
Dixon, R. M. W. & Alexandra Y. Aikhenvald (eds.). 2000. Changing valency: Case studies in
transitivity. Cambridge: Cambridge University Press.
Dooley, Robert A & Stephen H Levinsohn. 2001. Analyzing discourse: A manual of basic
concepts. Dallas, TX: SIL International.
Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67(3). 547–619.
Evans, Vyvyan & Melanie Green. 2006. Cognitive linguistics: An introduction. Edinburgh:
Edinburgh University Press.
Fagan, Sarah M. B. 1988. The English middle. Linguistic Inquiry 19(2). 181–203.
Fauconnier, Gilles. 1985. Mental spaces: Aspects of meaning construction in natural language.
Cambridge, MA: The MIT Press.
Fauconnier, Gilles. 1997. Mappings in thought and language. Cambridge: Cambridge University
Press.
Fauconnier, Gilles. 2007. Mental spaces. In Dirk Geeraerts & Hubert Cuyckens (eds.), The
Oxford handbook of cognitive linguistics, 351–376. Oxford: Oxford University Press.
Fillmore, Charles. 1968. The case for case. In Emmon Bach & Robert Harms (eds.), Universals
in linguistic theory, 21–119. New York: Holt, Rinehart, and Winston.
Foley, William. 2007. A typology of information packaging in the clause. In Timothy Shopen
(ed.), Language typology and syntactic description, vol. 1, 362–446. 2nd edn.
Cambridge: Cambridge University Press.
Foley, William A. & Robert D. Van Valin. 1984. Functional syntax and universal grammar
(Cambridge Studies in Linguistics 38). Cambridge: Cambridge University Press.
Frajzyngier, Zygmunt. 1977. On the intransitive copy pronouns in Chadic. In Papers from the
eighth conference on African linguistics (Studies in African Linguistics Supplement 7),
73–84. Los Angeles: Department of Linguistics, UCLA.
Frajzyngier, Zygmunt. 1989. A grammar of Pero. Berlin: D. Reimer.
Frajzyngier, Zygmunt. 2008. A grammar of Gidar (Schriften Zur Afrikanistik). Vol. 13.
Frankfurt am Main: Peter Lang.
Frajzyngier, Zygmunt. 2011. Point of view of subject: Possessive subject pronouns in Wandala.
In Anne Storch, Gratien Atindogbé & Roger Blench (eds.), Copy pronouns: Case studies

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

90

from African languages (Kay Williamson Educational Foundation. African Languages
Monographs 3), 47–64. Köln: Rüdiger Köppe Verlag.
Frajzyngier, Zygmunt. 2012. Typological outline of the Afroasiatic phylum. In Zygmunt
Frajzyngier & Erin Shay (eds.), The Afroasiatic languages (Cambridge Language
Surveys), 505–624. Cambridge: Cambridge University Press.
Frajzyngier, Zygmunt, Eric Johnston & Adrian C. Edwards. 2005. A grammar of Mina (Mouton
Grammar Library 36). Berlin: Mouton de Gruyter.
Frajzyngier, Zygmunt & Erin Shay. 2012. Chadic. In Zygmunt Frajzyngier & Erin Shay (eds.),
The Afroasiatic languages (Cambridge Language Surveys), 246–341. Cambridge:
Cambridge University Press.
García-Miguel, José. 2007. Clause structure and transitivity. In Dirk Geeraerts & Hubert
Cuyckens (eds.), The Oxford handbook of cognitive linguistics, 753–781. Oxford: Oxford
University Press.
Givón, Talmy. 1983. Topic continuity in discourse: An introduction. In Talmy Givón (ed.),
Topic continuity in discourse: A quantitative cross-language study (Typological Studies
in Language 3), 1–41. Amsterdam: John Benjamins Publishing Company.
Givón, Talmy. 1984. Syntax: A functional-typological introduction. Vol. 1. Amsterdam: John
Benjamins Publishing Company.
Goldberg, Adele E. 1995. Constructions: A Construction Grammar approach to argument
structure (Cognitive Theory of Language and Culture). Chicago: University of Chicago
Press.
Grubic, Mira, Suzanne Genzel & Frank Kuegler. 2010. Linguistic fieldnotes I: Information
structure in different African languages. Universität Potsdam.
Hardy, Donald. 1994. Middle voice in Creek. International Journal of American Linguistics
60(1). 39–68.
Haspelmath, Martin. 2013. Argument indexing: A conceptual framework for the syntactic status
of bound person forms. In Dik Bakker & Martin Haspelmath (eds.), Languages across
boundaries, 197–226. Berlin: De Gruyter.
Hellwig, Birgit. 2011. The semantics of copy pronouns in Goemai. In Anne Storch, Gratien
Atindogbé & Roger M. Blench (eds.), Copy pronouns: Case studies from African
languages (Kay Williamson Educational Foundation. African Languages Monographs 3),
65–79. Köln: Rüdiger Köppe Verlag.
Hoffman, Carl. 1963. A grammar of the Margi language. Oxford: Oxford University Press.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

91

Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language
56(2). 251–299.
Inglese, Guglielmo. 2021. Towards a typology of middle voice systems. Linguistic Typology
26(3). 489–531.
Jackendoff, Ray S. 1990. Semantic structures. Cambridge, MA: The MIT Press.
Jungraithmayr, Herrmann. 1970. Die Ron-Sprachen: Tschadohamitische Studien in
Nordnigerien; Eine morphologische Studie (Afrikanistische Forschungen 3). Glückstadt:
J. J. Augustin.
Jungraithmayr, Herrmann & Rudolph Leger. 1993. The Benue-Gongola-Chad Basin: Zone of
ethnic and linguistic compression. In G Nagel (ed.), Berichte des
Sondervorschungsbereichs 268, vol. 2, 161–172. Frankfurt am Main: Goethe-Universität.
Kemmer, Suzanne. 1993. The middle voice (Typological Studies in Language 23). Amsterdam:
John Benjamins Publishing Company.
Kemmer, Suzanne. 1994. Middle voice, transitivity, and the elaboration of events. In Barbara A.
Fox & Paul J. Hopper (eds.), Voice: Form and function (Typological Studies in Language
27), 179–230. Amsterdam: John Benjamins Publishing Company.
Keyser, Samuel Jay & Thomas Roeper. 1984. On the middle and ergative constructions in
English. Linguistic Inquiry. The MIT Press 15(3). 381–416.
Kibrik, Andrej A. 1999. Reference and working memory: Cognitive inferences from discourse
observations. In Karen Van Hoek, Andrej Kibrik & Leo Noordman (eds.), Discourse
studies in cognitive linguistics: Selected papers from the fifth International Cognitive
Linguistics Conference (Amsterdam Studies in Theory and History of Linguistic Science
4), vol. 176, 29–52. Amsterdam: John Benjamins Publishing Company.
Kibrik, Andrej A. 2000. A cognitive calculative approach towards discourse anaphora. In P
Baker (ed.), Proceedings of the Discourse Anaphora on Reference Resolution Conference
(DAARC 2000) (Technical Papers), vol. 12, 72–83. Lancaster: Lancaster University
Centre for Computer Corpus Research on Language.
Kibrik, Andrej A. 2011. Reference in Discourse (Oxford Studies in Typology and Linguistic
Theory). Oxford: Oxford University Press.
Koops, Robert. 2011. Focus pronouns in Kuteb. In Anne Storch, Gratien Atindogbé & Roger
Blench (eds.), Copy pronouns: Case studies from African languages (Kay Williamson
Educational Foundation. African Languages Monographs 3), 103–117. Köln: Rüdiger
Köppe Verlag.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

92

Koops, Robert & John Bendor-Samuel. 1974. The recapitulating pronouns in Kuteb. Journal of
West African Languages 9(1). 5–16.
Kühner, Raphael. 1898. Ausführliche Grammatik der Griechischen Sprache ii: Satzlehre. (Ed.)
Bernhard Gerth. 3rd edn. Vol. 1. Hannover and Leipzig: Hahnsche Buchhandlung.
Langacker, Ronald W. 1987. Foundations of cognitive grammar. Vol. 1. Stanford, California:
Stanford University.
LaPolla, Randy. 1996. Middle voice marking in Tibeto-Burman languages. In Pan-Asian
Linguistics: Proceedings of the Fourth International Symposium on Languages and
Linguistics, vol. 5, 1940–1954. Thailand: Mahidol University.
LaPolla, Randy. 2000. Valency-changing derivations in Dulong/Rawang. In R. M. W. Dixon &
Alexandra Y. Aikhenvald (eds.), Changing valency: Case studies in transitivity, 282–
311. Cambridge: Cambridge University Press.
Leger, Rudolph & Ulrike Zoch. 2011. Intransitive Copy Pronouns in Chadic. In Anne Storch,
Gratien Atindogbé & Roger Blench (eds.), Copy pronouns: Case studies from African
languages (Kay Williamson Educational Foundation. African Languages Monographs 3).
Köln: Rüdiger Köppe Verlag.
Levin, Beth & Malka Rappaport Hovav. 2005. Argument realization (Research Surveys in
Linguistics). Cambridge: Cambridge University Press.
Lukas, Johannes. 1971. Über das erweiterte Verb im Bolanci (Nordnigerien). Journal of African
Languages 10(1). 1–14.
Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University
Press.
Marggrander, Anna. 2018. A grammar of Dũya: A Plateau language of central Nigeria
(Westafrikanische Studien 39). Köln: Rüdiger Köppe Verlag.
Martin, Jack. 2000. Creek voice: Beyond valency. In R. M. W. Dixon & Alexandra Y.
Aikhenvald (eds.), Changing valency: Case studies in transitivity, 375–403. Cambridge:
Cambridge University Press.
Mous, Maarten. 2012. Cushitic. In Zygmunt Frajzyngier & Erin Shay (eds.), The Afroasiatic
languages (Cambridge Language Surveys), 246–341. Cambridge: Cambridge University
Press.
Newman, Paul. 1971. Transitive and intransitive in Chadic languages. In Afrikanische Sprachen
und Kulturen – Ein Querschnitt (Hamburger Beiträge Zur Afrika-Kunde 14), 188–200.
Hamburg: Deutsches Institut für Afrika-Forschung.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

93

Newman, Paul. 2000. The Hausa language: An encyclopedic reference grammar. New Haven:
Yale University Press.
Newman, Paul. 2003. Hausa and the Chadic Languages. In Bernard Comrie (ed.), The World’s
Major Languages, 705–724. 0 edn. New York: Routledge.
https://doi.org/10.4324/9780203214961-35.
Nicolle, Steve. In press. Narrative discourse. In Lutz Marten, Nancy Kula, Jochen Zeller & Ellen
Hurst (eds.), The Oxford guide to the Bantu languages. Oxford: Oxford University Press.
Noss, Philip A. 1981. Gbaya, phonologie et grammaire: Dialecte Yaayuwee. Meiganga: Eglise
Évangélique Luthérienne du Cameroun, Centre de Traduction Gbaya.
Rice, Keren. 2000. Voice and valency in the Athapaskan family. In R. M. W. Dixon &
Alexandra Y. Aikhenvald (eds.), Changing valency: Case studies in transitivity, 173–
235. Cambridge: Cambridge University Press.
Rice, Sally. 1987. Towards a cognitive model of transitivity. Unpublished.
Ryan, Kevin. 2019. Bole suffix doubling as morphotactic extension. In Margit Bowler, Philip
Duncan, Travis Major & Harold Torrence (eds.), Schuhschrift: Papers in honor of
Russell Schuh, 139–160. Los Angeles: University of California - eScholarship
Publishing.
Samarin, William. 1967. Field linguistics: A guide to linguistic field work. New York: Holt,
Rinehart, and Winston.
Schuh, Russel. 1989. Gender and number in Miya. In Zygmunt Frajzyngier (ed.), Current
progress in Chadic linguistics: Proceedings of the International Symposium on Chadic
Linguistics, Boulder, Colorado, 1-2 May, 1987 (Amsterdam Studies in the Theory and
History of Linguistic Science IV: Current Issues in Linguistic Theory), vol. 62, 171–181.
Amsterdam: John Benjamins Publishing Company.
Schuh, Russel. 1998. A grammar of Miya (University of California Publications in Linguistics
130). Berkeley: University of California Press.
Schuh, Russel. 2005. The totality extension and focus in West Chadic. International Berlin Focus
Conference.
Schuh, Russell G. 1972. Aspects of Ngizim syntax. Los Angeles: UCLA Ph.D.
Schuh, Russell G. 2017. A Chadic cornucopia. (Ed.) Paul Newman. Oakland: California Digital
Library.
Schuh, Russell Galen & Alhaji Maina Gimba. 2001. Substantive and anaphoric “thing” in Bole,
with remarks on Hausa “abu/abin.” UCLA Working Papers in Linguistics 6. 90–122.

ICPS: A CHARACTERIZATION FROM COGNITIVE LINGUISTICS

94

Shimizu, Kiyoshi. 1980. A Jukun grammar (Beiträge zur Afrikanistik 9). Wien: Institut für
Afrikanistik und Ägyptologie der Universität Wien.
Storch, Anne. 1999. Das Hone und seine Stellung im Zentral-Jukunoid (Westafrikanische
Studien 20). Köln: Rüdiger Köppe Verlag.
Storch, Anne. 2009. Hone. In Gerrit Dimmendaal (ed.), Coding participant marking:
Construction types in twelve African languages (Studies in Language Companion Series
110), 123–140. Amsterdam: John Benjamins Publishing Company.
Storch, Anne. 2011. Copy pronouns and pragmatics in Northern Jukun. In Anne Storch, Gratien
Atindogbé & Roger Blench (eds.), Copy pronouns: Case studies from African languages
(Kay Williamson Educational Foundation. African Languages Monographs 3), 139–162.
Köln: Rüdiger Köppe Verlag.
Storch, Anne, Gratien Atindogbé & Roger Blench (eds.). 2011. Copy pronouns: Case studies
from African languages (Kay Williamson Educational Foundation. African Languages
Monographs 3). Köln: Rüdiger Köppe Verlag.
Swerts, Marc & Ronald Geluykens. 1994. Prosody as a marker of information flow in spoken
discourse. Language and Speech 37(1). 21–43.
Talmy, Leonard. 1988. Force dynamics in language and cognition. Cognitive Science 12(1). 49–
100.
Thompson, John. 1902. A Greek grammar: Accidence and syntax for schools and colleges.
London: John Murray.
Tuggy, David. 2007. Schematicity. In Dirk Geeraerts & Hubert Cuyckens (eds.), The Oxford
handbook of cognitive linguistics, 82–116. Oxford: Oxford University Press.
Van Hoek, Karen. 2007. Pronominal anaphora. In Dirk Geeraerts & Hubert Cuyckens (eds.), The
Oxford handbook of cognitive linguistics, 890–915. Oxford: Oxford University Press.
Van Valin, Robert D. (ed.). 1993. Advances in Role and Reference Grammar (Amsterdam
Studies in the Theory and History of Linguistic Science Series 4, Current Issues in
Linguistic Theory 82). Amsterdam: John Benjamins Publishing Company.
Van Valin, Robert D. 2005. Exploring the syntax-semantics interface. Cambridge: Cambridge
University Press.
Welmers, W.E. 1948. The phonology and morphology of Kuteb. Sudan United Mission, ms.
Wolff, H. Ekkerhard & Ludwig Gerhardt. 1977. Interferenzen zwischen Benue-Kongo und
Tschad-Sprachen. Zeitschrift der Deutschen Morgenländischen Gesellschaft (Suppl.
3(2)). 1518–1543.