It’s Not Over When It’s Over: Semantics of English Prepositions Andrea Tyler

It’s Not Over When It’s Over:
A Cognitive Linguistic Analysis of the
Semantics of English Prepositions
Andrea Tyler
Vyvyan Evans, Yiyoung Kim and Dasha
Shakhova made invaluable contributions to
this project.
CL emphasizes a number of basic tenets
that allow us to systematically investigate
areas of language that previous models
ignore altogether or characterize as
 And it turns out that many of these
“unexplorable” or “uninteresting” areas
pose particularly difficult challenges to
second language learners.
The Problem
Traditional accounts have represented the
semantics of English prepositions as highly
arbitrary; the various uses are presented as
unorganized lists of meanings that have
accidentally come to be associated with a
particular preposition.
By and large, these are the accounts that
ELT texts and grammars are based on.
a. The flag is hanging at half-mast over the capitol
building: located higher than
b. Arlington Cemetery is over the river from the White
House: on the other side
c. The negotiations are over: completed or finished
d. The ship took the troops over to the Gulf region:
transfer of an entity from one location to another
e. The woman placed a gas mask over her face:
f. There are over 300,000 soldiers deployed in the
Middle East : more than
g. The President chose intervention over negotiation
h. The picture of mom holding an apple pie is
over/above the mantel.
The Alternative
CL gives us an alternative. The multiple meanings
associated with English prepositions can be
represented as being systematically related within
a motivated semantic polysemy network. These
meanings are organized around a central, abstract,
main meaning.
Some of the central concepts from CL which we
have drawn on in our analysis of the semantics of
English prepositions are:
Contextualized Nature of Language
In naturally occurring language use, lexical
items always occur in context.
 The exact understanding of a lexical item is
always influenced by the context in which it
 Work in pragmatics shows us that some
inferencing is always involved in the
interpretation of utterances.
Communicative Nature of Language and
Contextualized Interpretation of Lexical
 Lexical item is initially used to indicate an
established meaning
 A speaker would only use the lexical item to mean
something different from the established meaning
if they believed the listener had a reasonable
chance of understanding the new meaning
 This understanding presumably would come from
inferences arising from the contexutalized use of
the lexical item
 Repetition --> independent distinct senses
General Cognitive Processes
Instead of assuming that there is a separate language
module which operates under its own principles, Cognitive
Linguists ask how much of language can we explain
through general cognitive processes.
Expect to find key principles or facets of general cognition
also occurring in language
EX. Basic aspects of perceptual system will be reflected in
Humans are unique
Many ways in which we are unique.
 Only species to develop cloth making and by extension to
wear clothing
 Only species to develop the ability to develop elaborate
means of deception that doesn’t have anything to do with
stimulus in the immediate environment
 Only species to develop trade
 Also only species to develop language.
Wouldn’t want to say we have special modules or genes for
cloth making, trading, etc. These are accounted for by our
general cognitive abilities. Perhaps language should be
analyzed in the same way.
Humans do have a unique neurological and
physical architecture
(1) Cognitively:
 Incredibly good at classifying, drawing generalizations
over several specific instances (which are not exact in
color, shape, size, and other physical properties)
 Inferring on the basis of very little evidence
 Remembering past experiences and relating them to
similar, ongoing experiences. Build up complicated,
patterned, systematic meaning structures SCHEMA. Use
that structure to organize new information
 We do not have mental telepathy
Humans do have a unique neurological and
physical architecture
(2) Physically:
 Stand on our hind legs
 Cannot resist gravity (in the way a
humming bird can)
 Have an asymmetrical front/back
orientation with our most important
perceptual organs located in our faces
Humans live in a particular physicalspatial environment.
The ways we perceive and interact with that
environment have important consequences
for our conceptualizations, which are in turn
reflected in language.
All these elements shape our cognition.
Embodied Experience:
Concepts Are Not Propositional in
 Cognitive linguists argue that 1) human
conceptual structure is crucially shaped by our
human perceptions of and interactions with the
real world and 2) language is a reflection of
human cognitive structure. *
 How perceptions of the real world are represented
in memory is unlikely to be in terms of bundles of
linguistic features.
 Bird [+feather, +wings, +sits on a nest, etc.]
 Conceptualized spatial relations coded by
prepositions are not likely to be represented
conceptually by semantic features.
Experiential Correlation
Humans regularly observe the recurrent co-occurrence of
two distinct phenomena. With repeated exposures, the two
distinct but co-occuring phenomena become closely
associated in memory such that we conceptualize and talk
about one in terms of the other.
For example, an increase in quantity is associated with an
increase in elevation. Thus, we can use language about
vertical elevation to describe an increase in quantity.
MORE IS UP (Grady, 1997, 2001; Lakoff & Johnson, 1999)
The Central Meaning of a Preposition
Denotes a Spatial Relationship
 Prepositions
code conceptual spatial
relations between two entities,
 Central Sense- Spatial relationship
 Prompt for spatial scene
Real World Force Dynamics
As a default, speakers assume that all
elements in a conceptual spatial scene
are subject to real-world force
dynamics, such as the assumption that
objects are subject to gravity (Talmy,
1988; 2000).
Ways of Viewing a Scene
Every spatial scene is conceptualized
from a particular vantage point. The
conceptualizer represents the default
vantage point and is usually off-stage.
However, the same scene can be
viewed from different vantage points.
Shifts in vantage points can give rise to
new inferences, which in turn give rise
to new sense. (Langacker, 1987)
A Cognitive Analysis of over
Central representation of over
We use diagrams to avoid prepositional definitions and because prepositions in
their central senses represent spatial relations between two objects. We make
no claims about the psychological validity of these diagrams.
In the sentence, The picture is over the mantel the
picture represents the TR and the mantel represents the
Extended Senses: the A-B-C trajectory
(1). In the real world we often encounter TRs which are in
motion. Sometimes moving TRs encounter some
impediment (LM) to forward motion and in order to
continue their forward motion at some point move to a
position that is higher than the impediment. English
speakers use the preposition OVER in describing this
motion, as in a sentence such as: The horse jumped over
the hurdle.
(2). Only at point B, the TR, the horse, is located higher
than the LM, the hurdle. The verb jump implies point A,
but nothing in the utterance explicitly refers to point C. We
infer point C because of what we know about horses and
hurdles and jumping, including knowledge of gravity and
Thus, we rely on our knowledge of force dynamics to
establish a full interpretation of this sentence.
(3). Point C in the ABC trajectory has become associated
with a number of consequences.
These associations have, in turn, given rise to several
distinct non-spatial meanings associated with over.
Once a distinct meaning, such as COMPLETION,
becomes associated with point C in the ABC trajectory, the
word over can be used to denote this meaning even when
the original spatial configuration of the TR being located
higher than the LM is no longer involved.
The On-the-Other-Side Sense
When you move from point A to point C, you end up on
the other side of the LM from where you start: Thus the
meaning ON THE OTHER SIDE, as illustrated in the
Arlington Cemetery is over the river from the White
The Completion Sense
When the TR comes down at point C, that portion of the
movement, or the particular action being considered, is
completed or finished. Thus the meaning of completion as
in the sentence: The negotiations are over.
The Transfer Sense
When an entity moves from point A to point C, the entity
has been transferred from point A to point B. Thus the
meaning of TRANSFER, as in the sentence The ship
carried the troops over to the Gulf region.
Extended Senses: The UP Relation
The More Sense
An essential aspect of the spatial relationship
denoted by over is that the TR is in a vertical
relationship vis-a-vis the LM and that the TR is in
an UP relationship to the LM.
 We now have over 300,000 troops in the middle
The Preference Sense
English speakers identify being in an UP
position as generally positive.
Kids usually prefer cake over broccoli.
The picture is above/over the desk.
 The lamp is above/over the table.
She hung the backpack over the back
of the chair.
She hung the backpack above the back
of the chair.
The Key Question: Is such an analysis useful for
second language learners?
Hypothesis: A cognitive-based model of the
semantics of English prepositions, which
represents the many meanings associated with a
preposition as a motivated polysemy network, has
the potential to provide a useful, organized rubric
for L2 learners.
A. A clever theoretical description of language is not a
silver bullet for second language learning. Having a richer,
more accurate description of the language should represent
a valuable advance for second language learning BUT
B. Have to attend to principles from work in psychology
and the field of SLA, i.e. the importance of noticing,
interestingness, role of pushed input, and following Ortega
and Norris (2000) explicit instruction followed by
communicative tasks
C. A major challenge is to find ways to make the
appropriate theoretical notions precise yet accessible to
second language learners and teachers.
Current project in the preliminary pilot stage
Two intact adult ESL classes at the
intermediate level. Mixture of L1
backgrounds. Overall pretest scores were
similar, with the Cognitive group scoring
slightly lower than the Control group.
 Control = 13 (7)
 Cognitive = 14
Current project in the preliminary pilot stage
 Instruction:
Over instruction and task(s) (1 hour)
To/For/At instruction and task(s) (1 hour)
 SPRING BREAK (not by design)
 Post Test
The picture of my mother is over the piano.
The doorknob is over the keyhole.
The lamp is over the table.
ABC Trajectory
On the Other Side
The table cloth is over the table.
 The paper is over the hole in the wall.
 The board is over the hole in the ceiling
Task: Matching Headlines with Short Paragraphs
Beauty over brains?
The movie “Legally Blonde” is a typical romantic comedy that says
there's no relationship between appearance and intelligence. The
message certainly is worth supporting: A young woman learns that
having good looks is not as important as having integrity. Elle is the
prototypical California blonde coed: energetic, cute and popular. She's
sure that her snobbish East Coast boyfriend is going to propose, but he
has other ideas. He's going back East to enroll in Harvard Law School
to become a politician. As much fun as Elle has been, she doesn't fit
the image he envisions for himself. In short, she's not smart enough.
The worst is over
The head officer of Oracle, Jeff Henley Tuesday said, "We're optimistic.
We definitely think we are done with the bad times and things are going
to start getting better." In December, Henley said that they believed that
the company's business had bottomed in its recently completed second
Control group
Subjects were shown exactly the same
examples, but without the cognitive
 We also scrambled the order of presentation
of the different meanings so as not to
hypothesized semantic network.
 The explicit instruction for the control group
was approximately 10 minutes shorter. They
were given a second communicative task to
ensure that both groups received equal
Results (VERY tentative)
A. Clear limitations
 Groups not matched for size* – Control = 6 ;
Cognitive = 12
 Pretest – individual work; Exercise -- in pairs
and quite different in form than the pretest
 Best we can say is that the results are
*1 student in the Control group scored 7 on the pretest and on the
exercise; 2 students in the Cognitive group scored 7 on the
pretest and on the exercise. These students were eliminated
from this particular analysis.
B. Preliminary findings
Cognitive 4/7
Cognitive 11(92%)