path: root/kemmer-handout.tex

\documentclass[9pt,a4paper,twocolumn]{extarticle}

\usepackage[top=2cm]{geometry}
\usepackage[british]{babel}
\usepackage{stfloats}

\usepackage{handouts}

\usepackage[hidelinks]{hyperref}
\usepackage{tikz}
\usetikzlibrary{arrows,fit,calc,shapes,positioning}
\usepackage[font=small]{caption}
\usepackage{subcaption}

\title{\Large Handout of ``The Middle Voice''\footnote{Suzanne Kemmer (1993).}}
\author{Camil Staps}

\newcommand{\RM}[0]{\textsc{rm}}

\begin{document}

\maketitle

\subsection*{Reflexive and Associated Middle Situation Types}
\summary{
	\subsubsection*{The direct reflexive}
	According to Faltz (1977), the archetypal semantically reflexive context is a simple two-participant clause
		where one participant is Agent or Experiencer and the other a Patient, while both refer to the same entity\pagenr{42}.
	This is the semantic prototype, since whenever a language marks coreference in other situtations,
		it also does in this situation\pagenr{43}.
	This reflexive is called the \term{direct reflexive}.
	It is defined by
		\term{coreference}, narrowed down by
		\term{scope}\note{only simple clauses} and
		\term{thematic roles}\note{Agent/Experiencer and Patient (Faltz)/Stimulus(Kemmer)}.

	A \term{reflexive marker} (\RM) is
		``a productive grammatical device that is used obligatorily
		to mark direct reflexive contexts in at least the third person''\pagenr{47}.
	All reflexive-marking languages use {\RM}s in the direct reflexive,
		which is a second indicator that this is the semantic reflexive prototype.

	The direct reflexive is a special case of a \term{two-participant event}.
	The prototypical two-participant event has a human Agent act volitionally,
		exerting physical force on an inanimate definite Patient,
		which is directly affected by that event\pagenr{50}.
	\parnote{This definition can be easily extended to include mental state- and perception-verbs,
		using Experiencer/Stimulus terminology and/or the Initiator/Endpoint macroroles\pagenr{51}.}

	\subsubsection*{Body action middles}
	\term{Grooming actions} often have different markings than reflexive verbs,
		so they are not a subset of the reflexive situation type\pagenr{54}.
	\term{Changes in body posture}\note{stand (up); sit (down); lie (down)}
		often appear as bare intransitives and therefore cannot be seen as direct reflexives either\pagenr{55}.
	Also \term{nontranslational motion} verbs\note{turn, twist, bend (e.g. one's head)}
	are often expressed using middle forms\pagenr{56},
		as are verbs of \term{translational motion}\note{fly, flee, run, etc.}\pagenr{57}.

	All these verbs may be marked differently from reflexive and two-participant verbs because coreference is expected.
	The actions of e.g. self-washing are also different from being washed or washing something else\pagenr{60}.

	Reflexive markers may be added when more distinguishability of the participants is required\pagenr{66}.
	This indicates that the {\RM} may have this function in a reflexive context as well.
	\parnote{And indeed, even though the participants in a reflexive context are coreferent, some degree of individuation is maintained.}

	\subsubsection*{Event schemas}
	We can graphically depict the results from the previous sections in \autoref{fig:refmid-events}.

	\begin{figure}[h]
		\begin{subfigure}[b]{.5\linewidth}
			\centering
			\begin{tikzpicture}[node distance=1cm,every node/.style={circle,draw},scale=0.9,>=stealth]
				\node (A) {};
				\node[right of=A] (B) {};
				\draw[->] (A) -- (B);
				\path (A) edge[bend left,dotted] (B);
			\end{tikzpicture}
			\caption{Reflexive}
		\end{subfigure}%
		\begin{subfigure}[b]{.5\linewidth}
			\centering
			\begin{tikzpicture}[node distance=2cm,every node/.style={circle,draw},scale=0.9,>=stealth]
				\node (A) {};
				\path (A) edge[loop left,looseness=5,->,in=315,out=45] (A);
			\end{tikzpicture}
			\caption{Body action middle}
		\end{subfigure}
		\caption{Event schemas\label{fig:refmid-events}\pagenr{71}}
	\end{figure}

	The relationship between two-participant, reflexive, middle and one-participant can be summarised as in \autoref{fig:middle-transitivity-line} below.
	Two-participant events have two distinct participants filling two semantic roles in an asymmetric relation.
	Reflexive events behave in the same way, but have one entity fulfilling both roles.
	For reflexive events, there is a distinction between the agent and the patient, while for middle events, there is not.
	This absence of differentiation causes it to lie further away from the two-participant event type\pagenr{72}.

	\begin{figure*}[t]
		\centering
		\begin{tikzpicture}[node distance=8em,scale=0.9]
			\node             (a) {Two-participant};
			\node[right of=a] (b) {Reflexive};
			\node[right of=b] (c) {Middle};
			\node[right of=c] (d) {One-participant};

			\node[xshift=-5em,yshift=-1em] at (a) (plus) {+};
			\node[xshift=5em, yshift=-1em] at (d) (min)  {-};

			\draw[<->] (plus) -- (min)
				node[below,midway] {Degree of distinguishability of participants};
		\end{tikzpicture}
		\caption{Distinguishability of participants\label{fig:middle-transitivity-line}\pagenr{73}}
	\end{figure*}

	\subsubsection*{Indirect situation types}
	Indirect reflexives occur when
		(a) a verb has three participants (Agent, Patient and Recipient or Beneficiary),
		(b) the Agent is coreferent with the Recipient/Beneficiary and
		(c) the Agent and the Recipient/Beneficiary are normally distinct\pagenr{74}.
	This is depicted in \autoref{fig:events-indref}.

	The indirect middle comprises actions that one \emph{normally} or \emph{necessarily} performs for one's own benefit
		(which is different from an indirect middle in that it violates property (c) above)\pagenr{78}.
	See \autoref{fig:events-indmid}.

	\begin{figure}[h]
		\begin{subfigure}[b]{.5\linewidth}
			\centering
			\begin{tikzpicture}[node distance=1cm,every node/.style={circle,draw},scale=0.9,>=stealth]
				\node (A) {};
				\node[right of=A] (B) {};
				\node[right of=B] (C) {};
				\draw[->] (A) -- (B);
				\draw[->,dashed] (B) -- (C);
				\path (A) edge[bend left,dotted] (C);
			\end{tikzpicture}
			\caption{Ind. reflexive\pagenr{77}\label{fig:events-indref}}
		\end{subfigure}%
		\begin{subfigure}[b]{.5\linewidth}
			\centering
			\begin{tikzpicture}[node distance=1cm,every node/.style={circle,draw},scale=0.9,>=stealth,->]
				\node (A) {};
				\node[right of=A] (B) {};
				\path (A) edge[bend left] (B);
				\path (B) edge[bend left,dashed] (A);
			\end{tikzpicture}
			\caption{Ind. middle\pagenr{81}\label{fig:events-indmid}}
		\end{subfigure}
		\caption{Event schemas for indirect situations}
	\end{figure}

	\subsubsection*{Logophoric situation types}
	It is possible that multiple events are combined%
		\note{e.g. ``Margaret says that Geoffrey is silly.''}\pagenr{82}.
	The case discussed here is when the participant in the main event is the \term{Mental Source} of the dependent event.
	In this case, there is coreference of Mental Source and the subject of the dependent event,
		yet the main event is neither direct nor indirect reflexive
		since it violates the thematic aspect\note{see \emph{The direct reflexive}; \term{thematic roles} above}\pagenr{86}.

	\term{Object control},
		where the referent of the object of the main clause is the initiator participant of the dependent event%
		\note{e.g. ``I forced him to go''},
		is commonly found with manipulative verbs in many languages\pagenr{88}.

	Whether the object of the main clause is marked using a middle or reflexive marker
		seems to depend on the \term{relative distinguishability of the participants}%
		\note{cf. body action middles and \autoref{fig:middle-transitivity-line} above}\pagenr{92}.
}

\subsection*{Related Semantic Domains}
\summary{
	\subsubsection*{Reciprocal situation types}
	The prototypical reciprocal context is a simple event frame expressing a two-participant event with two relations
		where each participant is the Initiator in one and the Endpoint in the other (see \autoref{fig:event-reciprocal})\pagenr{97}.

	\begin{figure}[h]
		\centering
		\begin{tikzpicture}[node distance=1cm,every node/.style={circle,draw},scale=0.9,>=stealth]
			\node (A) {};
			\node[right of=A] (B) {};
			\node[below of=A,yshift=3mm] (C) {};
			\node[right of=C] (D) {};
			\draw[->] (A) -- (D);
			\draw[->] (C) -- (B);
			\path (A) edge[bend left,dotted] (B);
			\path (C) edge[bend right,dotted] (D);
		\end{tikzpicture}
		\caption{Prototypical event schema for reciprocal contexts\label{fig:event-reciprocal}\pagenr{97}}
	\end{figure}

	The reciprocal situation type is similar to the reflexive situation type
		in that all participants are both Initiator and Endpoint\pagenr{98}.
	It is also similar to collective situations%
		\note{e.g. ``The guests left.''},
		but adds that all participants are Endpoints\pagenr{99}.
	Lastly, it compares to chaining situations%
		\note{e.g. ``The graduates followed each other up onto the platform.''}\pagenr{101}.

	Naturally reciprocal events%
		\note{that are semantically necessarily or usually reciprocal}
		are different from prototypical reciprocal events in the
		\term{relative distinguishability of events}\pagenr{112}:
	for naturally reciprocal events%
		\note{John and Mary kissed},
		the events are simultaneous,
		whereas for prototypical reciprocal events%
		\note{John and Mary kissed each other},
		they are not.

	Distinguishability of events and of participants (see under Logophoric situation types) are comparable\pagenr{121}.
	The encompassing property is \term{relative elaboration of events}.
	When markers are more prominently present, the events are more elaborate.

	\subsubsection*{The cognition middle}
	Simple mental events can be depicted as in \autoref{fig:events-mental}.
	\autoref{fig:events-mental-expstim} shows a two-participant event%
		\note{e.g. ``I pity him.''},
		where the Experiencer directs his attention on the Stimulus
		and the Stimulus brings about the mental event in the mind of the Experiencer.
	\autoref{fig:events-mental-exp} shows a one-participant event%
		\note{e.g. ``The people scared.''}\pagenr{128}.

	\begin{figure}[h]
		\begin{subfigure}[b]{.5\linewidth}
			\centering
			\begin{tikzpicture}[node distance=1cm,every node/.style={circle,draw},scale=0.9,>=stealth,->]
				\node (A) {};
				\node[right of=A] (B) {};
				\path (A) edge[bend left,dashed] (B);
				\path (B) edge[bend left,dashed] (A);
			\end{tikzpicture}
			\caption{Experiencer--Stimulus\label{fig:events-mental-expstim}}
		\end{subfigure}%
		\begin{subfigure}[b]{.5\linewidth}
			\centering
			\begin{tikzpicture}[node distance=2cm,every node/.style={circle,draw},scale=0.9,>=stealth]
				\node (A) {};
				\path (A) edge[loop left,dashed,looseness=5,->,in=315,out=45] (A);
			\end{tikzpicture}
			\caption{Experiencer only\label{fig:events-mental-exp}}
		\end{subfigure}
		\caption{Cognitive event schemas\pagenr{128}\label{fig:events-mental}}
	\end{figure}

	The affectedness of the Initiator is an inherent part of a mental event, like with middle types.
	They also share the low degree of distinguishability of participants\pagenr{129}.

	Verbs of emotion, emotive speech\note{e.g. `complain'} and emotive vocalisations\note{e.g. `laugh'}
		are similar to mental events and share the high degree of affectedness of the Experiencer,
		although they lack volitionality\pagenr{130}.

	Cognition verbs\note{e.g. `interpret', `consider'} may also appear with middle markers.
	\parnote{They seem related to emotion verbs:
		in all but one of the languages considered,
		when cognition verbs could appear with middle markers, emotion verbs could as well,
		and vice versa.}

	Perception events can be split up in three categories:
		passive Experiencer-based\note{e.g. ``I smell garlic''};
		active Experiencer-based\note{e.g. ``I smelled the meat to see if it was still good''};
		Stimulus-based\note{e.g. ``Garlic smells good''}\pagenr{136}.
	They appear less with middle markers,
		possibly because the Experiencer is less affected than with other cognition events.

	Cognition types are most commonly complex%
		\note{e.g. ``I think that Thatcher will be ousted''}.
	They may be represented as in \autoref{fig:events-complex-cognition}.

	\begin{figure}[h]
		\centering
		\begin{tikzpicture}[node distance=1cm,every node/.style={circle,draw},scale=0.9,>=stealth,->]
			\node (A) {};
			\node[right of=A,xshift=-2mm] (B) {};
			\draw (A) -- (B);

			\node[rectangle,inner sep=2mm,fit=(A) (B)] (D) {};
			\node[left of=D,xshift=-8mm] (E) {};
			\draw[dashed] (E) -- (D);
			\path[dashed,draw] (D) -- ($(D)-(0,.8)$) -- ($(E)-(0,.8)$) -- (E);
		\end{tikzpicture}
		\caption{Complex cognitive event schema\pagenr{139}\label{fig:events-complex-cognition}}
	\end{figure}

	\subsubsection*{Other related situation types}
	Middle markers are also used for \term{spontaneous events}\note{e.g. `grow', `be born'}\pagenr{144}.
	They are different from other middle situation types in the complete lack of volitional initiation by the Patient\pagenr{146}.

	In some languages, middle markers can be used to express situations
		where an external causer is understood to exist, but de-emphasised for non-specificity or relative unimportance%
		\note{e.g. ``This book reads well'', which implies a reader}\pagenr{147}.
	As with spontaneous events, the focus is on the affected entity.
	This use can be called \term{passive middle}\pagenr{149}.
	\parnote{Further distinctions can be made, but are irrelevant to Kemmer's study.}
}

\subsection*{Hypotheses and Predictions}
\summary{
	Based on results from the previous sections, the reflexive-middle domain can be plotted as in \autoref{fig:domain-map}.
	Prototype categories have been encircled.
	Plain arcs indicate semantic similarity (dotted arcs are discussed below)\pagenr{202}.

	\begin{figure*}[b]
		\centering
		\begin{tikzpicture}[scale=0.9,every node/.style={align=center,text width=2cm,scale=0.9},every path/.style={draw}]
			\node[text width=15mm,draw,ellipse] (dir-refl) {Direct Reflexive};
			\node[text width=28mm,above=4cm of dir-refl,draw,ellipse] (two-p-evts) {Two-Participant};
			\path (dir-refl) -- (two-p-evts);

			\node[above=5mm of dir-refl,xshift=-15mm] (indir-refl) {Indirect Reflexive};
			\node[above=13mm of indir-refl,xshift=-20mm] (indir-mid) {Indirect Middle};
			\path (dir-refl) -- (indir-refl) -- (indir-mid);

			\node[above=5mm of dir-refl,xshift=15mm] (logo-refl) {Logophoric Reflexive};
			\node[above=8mm of logo-refl,xshift=15mm] (logo-mid) {Logophoric Middle};
			\path (dir-refl) -- (logo-refl) -- (logo-mid);

			\node[left=of dir-refl] (recip) {Reciprocal};
			\node[left=of recip] (nat-recip) {Natural Reciprocal};
			\path (dir-refl) -- (recip) -- (nat-recip);

			\node[right=of dir-refl,yshift=5mm] (pass-mid) {Passive Middle};
			\node[right=of pass-mid,draw,ellipse,yshift=-5mm,text width=15mm] (passive) {\sc Passive};
			\node[below=5mm of pass-mid] (emo-mid) {Emotion Middle};
			\node[below=15mm of emo-mid,xshift=-5mm] (cog-mid) {Cognition Middle};
			\path (dir-refl) -- (pass-mid) -- (passive);
			\path (dir-refl) -- (emo-mid) -- (pass-mid);
			\path (dir-refl) -- (cog-mid) -- (emo-mid);

			\node[below=of dir-refl,text width=28mm] (nontrans-mot) {Non-Translational Motion};
			\node[below=of nontrans-mot] (posture-change) {Change In Body Posture};
			\node[below=of posture-change] (trans-mot) {Translational Motion};
			\node[below=of trans-mot,draw,ellipse,text width=25mm] (one-p-evts) {One-Participant};
			\path (dir-refl) -- (nontrans-mot) -- (posture-change) -- (trans-mot) -- (one-p-evts);

			\node[left=5mm of nontrans-mot,yshift=-5mm] (grooming) {Grooming};
			\path (dir-refl) -- (grooming) -- (one-p-evts.north west);

			\node[right=of cog-mid,yshift=-5mm] (spont) {Spontaneous Action or Process};
			\path (dir-refl) -- (spont);
			\path (one-p-evts.north east) -- (spont) -- (passive);

			\node[above=2mm of two-p-evts] {\sc Active};
			\node[below=2mm of one-p-evts] {\sc Active};

			\path[dotted,line width=2pt] (indir-mid) -- (logo-mid) -- (pass-mid) -- (emo-mid) --
				(spont) -- (cog-mid) -- ($(posture-change)!0.7!(trans-mot)$) --
				(grooming) -- (nat-recip) -- (indir-mid);
		\end{tikzpicture}
		\caption{Semantic relations among middle and other situation types\pagenr{211}\label{fig:domain-map}}
	\end{figure*}

	Reflexive types share coreference of Initiator and Endpoint\pagenr{204}.
	Each middle type is linked to the situation type(s) with which it has the greatest semantic affinity.
	The passive is midway on the transitivity spectrum,
	since it has two participants, but the event is treated as having only one salient entity\pagenr{205}.

	We have seen that relative elaboration of events is the key property
		by which reflexive and middle events are distinguished.
	Middle morphology, then, is an expressive strategy for variations in the conceptual structuring of events\pagenr{209}.

	The dotted arcs in \autoref{fig:domain-map} indicate the properties that middle types share (low elaboration of events)\pagenr{211}.
}

\end{document}