Please cite as:
Levinson, S. C., Brown, P., Danzinger, E., De León, L., Haviland, J. B., Pederson, E., &
Senft, G. (1992). Man and Tree & Space Games. In S. C. Levinson (Ed.), Space stimuli kit
1.2 (pp. 7-14). Nijmegen: Max Planck Institute for Psycholinguistics.
doi:10.17617/2.2458804.
REGULATIONS ON USE
Stephen C. Levinson and Asifa Majid
This website and the materials herewith supplied have been developed by members of the
Language and Cognition Department of the Max Planck Institute for Psycholinguistics
(formerly the Cognitive Anthropology Research Group). In a number of cases materials were
designed in collaboration with staff from other MPI departments.
Proper citation and attribution
Any use of the materials should be acknowledged in publications, presentations and other
public materials. Entries have been developed by different individuals. Please cite authors as
indicated on the webpage and front page of the pdf entry. Use of associated stimuli should
also be cited by acknowledging the field manual entry. Intellectual property rights are hereby
asserted.
Creative Commons license
This material is provided under a Creative Commons Attribution-NonCommercialShareAlike license (CC BY-NC-SA 4.0). This means you are free to share (copy,
redistribute) the material in any medium or format, and you are free to adapt (remix,
transform, build upon) the material, under the following terms: you must give appropriate
credit in the form of a citation to the original material; you may not use the material for
commercial purposes; and if you adapt the material, you must distribute your contribution
under the same license as the original.
Background
The field manuals were originally intended as working documents for internal use only. They
were supplemented by verbal instructions and additional guidelines in many cases. If you
have questions about using the materials, or comments on the viability in various field
situations, feel free to get in touch with the authors.
Contact
Email us via library@mpi.nl
Max Planck Institute for Psycholinguistics
P.O. Box 310, 6500 AH, Nijmegen, The Netherlands
7
I. Photo- Photo Matching Games: Instructions to Investigator
Photo-photo matching games have the advantage of offering an external standard of
"mismatch", and of not requiring videotaping. Many of us have already tried them.
They can be used for exploratory or for explicitly probing work. They are not so good
for investigations into the limits of a domain of "space" itself, since of course the
contrasts about which people will talk have already been selected in preparing the
photographs.
Three supersets are provided, one exploratory (de Le_n's "Space Games") one
principled ("Men and Tree"), and one that you will make yourself. De Le_n's superset
consists of 32 photos in contrast sets of 3 to 6 photos each, to be combined into games
of about 12 photos each, as you wish. Men and Tree set consists of four pre-combined
game sets of twelve photos each.
Standardized instructions for running a photo-photo matching game appear in the
discussion of the Men and Tree superset below. You are of course free to run the
games in other kinds of situations and configurations as well. Try the players back to
back, in public, however you like. But please do also run at least three pairs of players
for the Men and Tree sets in the standard manner described.
A. De León's "Space Games" Superset. Natural Objects
De León's photos are recommended for those who are just beginning to explore spatial
reference in their field language. They encode spatial contrasts that are not obviously
lexicalizable in English, and they have the virtue of having been produced precisely
for initial exploration, before Lourdes knew much about what to expect from her
language either. We have chosen several contrasts that use natural rather than
manufactured objects, since so many of our other stimuli are manufactured objects.
B. Men and Tree Superset
The Men and Tree superset is the first attempt to run a formally comparative task
across all of our field sites. That's why we ask you to make it a high priority to run
this particular superset under the playing conditions outlined below and with at least
three separate pairs of players. If you have difficulty with photos in your field
situation, try running through at least the contrasts involving the Men and the Tree
themselves as Object-Object matching tasks (you need two farm animal sets for that).
The superset of Men and Tree pictures deals with location on the horizontal plane (4
directions) with both featured (man) and non- featured (tree, balls) objects. For alike
pairs of objects (featured with featured, unfeatured with unfeatured), the in- frontof/behind dimension is also explored.
Note that "Men and Tree" is a notation for the full set of 4 photo-photo matching
games which explore these questions. There is one training game in the set (game 1)
and there are distractor photos -- intended to make the game more fun to play -within games 2, 3 and 4 of the set. This means that all photos in the "Men and Tree"
set do not necessarily depict a man or a tree. It also means that there is one entire
game (game 1) of the "Men and Tree" set in which not a single man or a single tree
appears.
8
Two players are to be set side by side with a screen separating them so that they
cannot see each others' pictures. Orientation of players should be varied across (not
necessarily within) game-playing sessions so that the players face towards the
different points of the compass or other previously established locally significant
absolute points of orientation.
For comparative purposes, it's useful to make the effort to set up a game-playing
situation which is as private and undistracted as possible. Kibbitzing and coaching are
interesting from many points of view but they complicate the comparative project
enormously.
Record the instructions that the players receive. Players should be told, in their own
language, "This is a game with photographs. You each have the same set of pictures,
and the game is for one person to choose pictures one by one and to tell the other
person which picture s/he has chosen, WITHOUT LOOKING (just with language, try
not to point or gesture), so that the other person can pick out the one that matches
from their own set. You can talk back and forth as much as you want, for as long as it
takes you to make sure you have picked the matching photo. I'll show you how to do
it while you play the first game with this set".
Within the Men and Tree set, Game 1 is a training set and should be played first.
(Note that it does not have any photos which actually are of men or trees.) Even if you
have experienced players, run Game 1 first, with less emphasis on the training aspect.
After Game 1, the order in which Games 2, 3, and 4 are played should be varied.
When playing Game 1, keep in mind that the idea is to show them exactly how you
want them to play the later games. Use the same procedure you will use for the other
three games. Language data from game 1 may turn out to be interesting in itself, but
the main point is to make sure that people know how to play.
If you find that it is too difficult for people to play the 4 games in sequence (takes too
long etc.), it is possible to have each pair of players play fewer games. In this case,
players should play first game 1 (training game) and then either game 2, 3, or 4. When
you have played game 1 plus game 2 with one pair of players, game 1 plus game 3
with another, and game 1 plus game 4 with yet another, you will have one full set of
Men and Tree games played. Alternatively, you could play game 1 plus game 2 with a
pair of players on Monday, and then game 3 and/or game 4 with the same pair of
players on Tuesday. The crucial point is to be sure that everyone who plays game 2, 3,
or 4 has first played game 1 -- so that he or she has had some practice with the rules of
the game.
Within games 2 , 3 and 4, any picture that does not show either two men, a man with a
tree, or two balls is a DISTRACTOR and can be removed from the game if you
decide that having fewer photos per game will make the game easier/more
interesting/more fun for your informants.
Within the Men and Tree set, game 2 explores reference to horizontal relationships
between two unfeatured objects (the coloured balls) and between a featured and an
unfeatured object (the man and the tree). Game 3 explores horizontal relationships
between two featured objects (the men) where the two are oriented the same way (the
men are always looking in the same direction as one another). Game 4 explores
9
horizontal relationships between two featured objects (the two men) when they are
oriented in opposite directions.
For each game, shuffle the photos beforehand and lay them out right way up in front
of each player in a 3 x 4 grid (3 across, 4 down). The players are then free to choose
the order in which to pick and describe the photos. They should each make an ordered
pile, to one side, of the photos they select, as they select them. Have players put the
photos down right side up in their piles; if they place them upside down, they will see
the numbers on the backs. The pile then recapitulates the order in which the photos
were described, but backwards.
In the course of the game the investigator notes the number of each photo described,
from the numbers written on the front of the director's set, at the moment of
description. The game ends when all 12 photos have been described and a match
selected for each of them. At the end of each game the investigator makes a note of
the order of selection of the Matcher's pile. At the end of the game, and continuing to
record, the Matcher and the Director compare their piles and mismatches are
discussed. Before the end of any game, players can decide that they've made a
mistake, and go back and change the match to a previous photo (without of course
looking at each other's piles). When this happens, there is a risk that the ordering,
even of the agreed-upon matched photos in the piles, will get mixed up. This is one
place where you may have to be quite intrusive in noting down the old order as well
as the new one, in order to preserve your record. In general though, the rule of thumb
is to let the players themselves run the game and to be as little part of it as possible.
Video taping is not necessary, although it is an option. Audio taping is essential. Take
notes as the game is being played about any information that will not be on the tape.
It seems to make it easier to play a series of games if you allow the players to switch
roles as they go (eg. Director in Game 1 becomes Matcher in game 2). In addition,
you get more language for the effort if you have players double up roles, and play
each game once in each role (i.e. once as Director and once as Matcher). This may,
however, be boring or slow for the players. Make your own decision. If you do this,
keep in mind that the second time through a set of photos for a given player has a
different status from the first, and keep records accordingly. Doubling up is
interesting if you want to investigate reductions and special accomodations that
develop as people become expert at the game. If you do double up, you still count
only one pair (the first run-through for each player) for the comparative project, where
the same people are playing.
As well as playing the games, have an assistant explain the differences among the
photos to you in elicitation fashion, on some other occasion. If you use your
elicitation assistant as a player of the games, don't do the elicitation until after he or
she has played the games for you!
C. Each Field Worker Makes Their Own
Finally, think about making in the field your own superset of photo matching games
which will correct all the shortcomings you so easily see in this set of games, and
which will highlight distinctions relevant in your language. Think in terms of
developing a superset of photos which the rest of us can take to our own field sites
10
next time, so as to bring back comparative data which will demonstrate how unique
and interesting your speakers really are.
It may be useful to bring back any objects you used in making your series, to be used
again in making more polished photos back at your home base.
11
II. Photo-Object Matching: Instructions for Investigator
Matching Games using three-dimensional objects in real orientation are more openended than photo-photo games for exploratory elicitation. They make fewer
presumptions about the kind of contrasts that will be relevant in a given language and
may be more naturalistic and easier on the players, in having a less competitive feel.
For these, you must videotape. Ideally, film both participants. Set up the participants
as for photo-photo matching, and explain the game in similar terms (to match the
picture without looking, and by using language alone -- no gesture -- to explain which
parts to pick and how to arrange them or put them together).
The game is played one photo scene at a time. The Director is given a picture, and the
Matcher is given the exact set of objects necessary to re-create the picture. The
Director describes the scene while the Matcher re-creates it as precisely as possible
without looking at the photo. From one game-playing session to the next, try to vary
the order in which particular scenes are presented.
In one version of this game (preferred by those of us with most experience at this kind
of game), the Director is allowed to see what the Matcher is doing. He or she can
therefore monitor and correct the Matcher's choices (in this case, you will probably
have to be quite explicit about discouraging gesture). The game is over when the
participants agree that they have achieved a match. At this point, the Matcher can be
shown the original photo, and discussion or repair of the re- created scene can be
recorded.
In another version, the Director cannot see what the Matcher is doing. After the
participants agree that a match has been achieved, the photo may be shown to the
Matcher and the Matcher's creation to the Director. They discuss and repair the
created scene, on record.
It is also possible at the original point of agreement as to Match, to allow the Director
to see the Matcher's creation, while still keeping the photo secret from the Matcher.
Then the game can continue as in the first version.
Other variations allow kibbitzing and coaching. You can give an audience access to
the Director's photograph, or only to the Matcher's array. The latter is probably kinder,
given the stress the Matcher is already under... If you only use one set of objects in
this kind of game, please use the Farm animals. If you only use two sets, use the Farm
animals and the Tinker Toys (you can use the designs on the instructions that come
with the Tinker Toys for "photos" to match, as well as the photos you have in this kit).
Use the Wooden Man as lowest priority.
A few words on Tinkertoy constructions:
These are potentially very interesting, in that they involve constructions in three
dimensional space. This is an area in which your pilot work now will be helpful in
seasons to come. Try out some variations and let us know what works and what
doesn't. Go ahead and invent new constructions for your informants to reproduce,
either by making new model photographs, or by using the duplicated pieces in your
Tinkertoy set to allow Object-Object matching. (We have provided extra pieces in
addition to those you need for the tinkertoy constructions photographed in the kit.)
12
You have been given a set of Tinkertoy photos containing 7 photos. Most of the
photos are non-representational in that they do not depict a "real" object. Nonrepresentational constructions are more likely to make the director rely on "abstract"
descriptions which involve angles, planes, etc. Note that this is typically an unnatural
task and may be quite challenging for many speakers. The one representational
construction (a wagon) gives a clear goal in construction and speakers have already
available part terms for reference. This may make it easier for some people.
We recommend doing the photos in increasing order of difficulty -- stopping
whenever the participants seem frustrated or too bored. Therefore the following
sequence is recommended: 2, 6, 3, 15, 8, 4, 7.
This game tends to be more linguistically rewarding if you can conduct it "blind"
(with the director and matcher unable to see what each other is looking at and any
gestures). However that may prove too difficult for many subjects. If it seems too
difficult, you might try the first one or two photos not blind and then ask if they would
like to try it blind. If they then agree, it may seem like less of an imposition to do it
blind.
13
III. Object-Object Matching Games
This is the game to use for peoples who have difficulty interpreting photographs; it
also is useful for getting information on absolute orientation systems (if such exist) for
which any rotatable images like pictures are intrinsically confusing. Object-object
matching is also preferrable to photos where arrays are complicated enough that
photos can't render them well. (We have had particular difficulty, for example, in
rendering "in front/behind" in photographs).
Object-object games are also possibly more fun to play than those involving photos.
They can be used to reproduce the photo game tasks, as long as you have two sets of
the objects pictured in the photographs; they can also be used to explore any other set
of spatial concepts such as topological (in/on/at) concepts, left/right front/back
up/down alignments, body part systems, absolute orientation systems, and more.
Unlike the photo games, however, this kind of game must be videotaped, so that an
adequate record exists of the object arrays, their disposition in space, and the degree
of correspondence between Director and Matcher arrays.
A. Objects from Our Stimuli Kit
The toy animals, people, tinker toys, wooden men, and trees supplied in the stimulus
kit can be used to play the same game with the actual objects. This version of the
game should be performed if the photo version is not feasible, and the same
oppositions and variations should be conducted (especially for the Men and Tree
photo superset please!).
We have enough farm animal sets for each field worker to take two. We ran into
problems with Tinkertoys however. In the absence of an abundance of tinkertoys,
perhaps the following compromise will work. Give the director the photo and the
tinkertoys and have him/her make the construction and then disassemble it. Then give
the tinkertoys to the matcher and have the director begin the instructions from looking
at the photo. This method should prove especially useful for those constructions
which have "tricks" to them (e.g. Tinkertoys, PO.TT 4 and 7). The idea is the same as
for photo-photo games. The two players are set side by side, with a screen separating
them. Players should be told that they both have the same set of objects, and the game
is for the Director to describe the setup of objects in front of him, and for the Matcher
to set up his objects in exactly the same way without looking at the Director's. Two
sets of the real objects - farm animals/trees/fences from our stimulus kit - are used
instead of photos. One set is reserved for the Director, one for the Matcher, and the
Investigator (perhaps using a photo as a model) sets up a single scene (in randomized
order across sessions), in front of the Director. The Matcher's job is to match the
completed scene, using his set of objects, on the basis of the Director's description.
Players can have as much discussion as they like as they play. It should be noted that
this is a harder task than the photo description game version of it.
After the Matcher has achieved a match (by his own standard), the two arrays should
be compared and discussed by the participants. The game may be played in various
ways:
i. with the Director able to see and correct the Matcher's array as he goes along (but
not vice versa),
14
ii. with the Director unable to see the Matcher's, and correction only allowed (by
interactional negotiation) after the Matcher achieves what he thinks is an adequate
match. Of course, this makes the task much more difficult and potentially frustrating - don't try this with everyone! After the Director sees what the Matcher has done,
allow the Director to give further corrections until satisfied with the work.
iii. Another variant of this game involves setting all the stimuli arrays out in front of
the Director at once, for example by using a 3 x 3 grid drawn on the ground, and
setting objects in specific configurations within each square of the grid. While little of
interest may result from the matrix layout itself, this format has the advantage of
minimizing the investigator's presence (once the initial setup has been made, s/he can
vanish), and allows the Director's entire set of arrays to be photographed as a whole,
freeing the video camera to focus on the Matcher's array. Care should be taken that
audio monitoring is adequate for both players however - two mikes plugged into the
camera, one for each player, is best.
The game may be varied in all three versions by giving the Matcher only the objects
which s/he needs to match the Director's scene, or by providing extra objects as well,
which must then be eliminated through dicussion with the Director. The first option is
easier for the players. Care should be taken in general to vary absolute orientation of
the players across game-playing sessions, to ascertain whether or not absolute
reference points are being used in spatial descriptions.
Again, sociolinguistic factors (sex, age, schooling, etc.) can be varied if deemed
relevant, and a total of three sets of games for each type of player-pair would be the
ideal (three adult female pairs, three adult male pairs, three cross-sex adults pairs, etc.
etc.).
B. Object-object arrays using local objects.
In this variant real objects taken from the local milieu (leaves, sticks, stones,
corncobs, feathers, pots, beans, sand, water, etc.) are used instead of photographs. The
investigator collects two identical sets of such objects, one for each of the players,
who again sit side by side but visually screened off from each other. Some of the
objects in one set (the Director's) are set out in a particular constructed pattern by the
investigator in front of the Director, whose job is to describe it so that the Matcher can
create the same array with his own objects (preselected by the investigator to be
equivalent to those of the Director for each array). Here again, videotaping is very
much the ideal.