Assignment 1 (1) The following is a list of features which could cue for the categories indicated.  Compare the relative cue validity of  each feature for the two categories given, briefly explaining  why the feature might be more or less valid.   Feature        category Eye        pig, animal Hoof        horse, animal Siren        car, police car Cue validity has been defined as the conditional probability that an object will be in category X, given that it has some feature Y.   To estimate this probability one could count instances of occurrences of feature Y and note how often X occurs with Y. P(cat|cue) = P (cat & cue) / P (cue) The problem with this formulation of cue validity is that it doesn’t give us what we intuitively want cue validity to be which is predictiveness of feature for category. As a result, I received answers of various sorts for this question which were right based on either the formulaic definition of cue validity, or the intuitive idea of cue validity as a measure of predictiveness.   Because of the confusion surrounding this question, I decided to ignore it for grading purposes.    To see how the calculations actually work out for these, click on links below: cue validity of eye for pig or animal cue validity of hoof for horse or animal cue validity of siren for car or police car (2) Argue for the characterization of each of the following lexical items as ambiguous, polysemous or vague:  (more than one answer possible) Whiskers Shoe Pen Lexical items may refer to more than one referent (or pattern/assembly of features) and the relationship between these referents can differ in terms of the degree of feature overlap and the functional or experiential importance of the contrast (that is, if we have no particular need to distinguish a male fish from a female fish, the term ‘fish’ which refers to both is adequate for most people’s needs) A term can be called vague if it has referents that share most of the same features and differ in ways that are generally not functionally significant.  (ex.  ‘fish’ can be male or female;  ‘aunt’ can be mother’s or father’s sister) A term can be called polysemous if its referents are semantically related but differ in more significant ways.  (Ex. Paint can refer to painting a wall or creating a work of art –while there is significant overlap in these actions they are also functionally quite distinct, i.e.  we do not consider a painted wall a work of art, we do not display it in a museum, etc.  nor do we (normally) cover the interiors and exteriors of our houses with artwork) A term can be called ambiguous if its referents are semantically unrelated. For this question your answer should have included the referents for each term that you were considering and what kind of relationship held between them that determined whether the term was vague, polysemous or ambiguous.  (ex. Pen refers to different types of writing utensils –ball point pen, quill pen, etc. , but it also refers to an enclosure)  Thus for pen, you might have said that term is vague (or polysemous) because it refers to different types of functionally equivalent (or nearly equivalent) items.  You might also have said that pen is ambiguous because it refers to two semantically unrelated items, a writing utensil and an enclosure. (3) The use of ‘ear’ in ‘an ear of corn’ is etymologically not related to the term for the auditory organ ‘ear’  Ear 1:  auditory organ Etymology:    Middle English ere, from Old English Eare; akin to Old High German Ora ear, Latin auris, Greek ous Ear 2: the fruiting spike of a cereal (such as corn) including both the seeds and protective structures Etymology:    Middle English er, from Old English Ear; akin to Old High German ahir ear, Old English ecg edge Yet a folk etymology exists in which they seem related semantically.  Propose a mechanism by which this folk etymology was made.   Are these two meanings of ear related by ambiguity, vagueness or polysemy?  A folk etymology is a link made between lexical items on the basis of apparent (but historically not actual) form/meaning similarity.  For ear, a meaning link between the ear on the head and an ear of corn (historically distinct origins) seems likely or possible based on the similarity of shape and position between ears of corn and ears on heads.  A person making such a folk etymology would regard (although probably not consciously) the term ‘ear’ as polysemous since there is an apparent semantic link but substantial functional difference. (4) a.    What is meant by a basic level category? b.    How has the basic level been determined experimentally? Basic level categories in the abstract, are a means of classifying feature sets in an optimally useful way, keeping together items at a level that is neither too general (so that distinct things are lumped together) nor too specific (so that similarities between like items are ignored)  We might say that the basic level category is the category at which the tension between category homogeneity and category differentiation is optimally resolved.   Such a level seems to have psychological primacy in that it is the  level at which people are able to imagine a single shape, or motor routine for the category.  It is also the level which people are able to easily list features for. (5) Several researchers (Zwaan & Madden, Gibbs) have carried out various experiments to demonstrate that as language is being comprehended, experiential traces are being activated which affect how particular words and phrases are interpreted. Based on the kinds of results you’ve seen, and other things you’ve learned about category representation, make a prediction about what you would expect to happen in the following case: Subject sees a sentence, then a picture.   Task is to name the object in the picture.   (1)    John drove his family to Michigan. (2)    John drove his cattle to the market. (3)    John led his cattle to the market Picture of car is shown What sorts of constrasts in task performance do you expect with respect to which sentence is read.  Explain. Based on Z&M’s experiments demonstrating some kind of mental simulation during comprehension, it is likely that seeing sentence 1 would yield the fastest reaction time to a picture of a car given that the term ‘drive’ saliently evokes the image a car, an image which is reinforced by the rest of the sentence.  Sentence 2 should be intermediate because while the word ‘drive’ should still evoke an image of a car, that image should fade as the rest of the sentence is comprehended and it is understood that either the cattle were either being driven or led in the traditional sense of the word (with no vehicle/maybe horses) or perhaps driven in a truck or trailer.  Sentence 3 contains no language that would particularly evoke cars, and so response would be expected to be slowest. (6) Referring to Goldberg’s idea of constructions, explain how a sentence like John mooped Mary a glurgh.  is so readily comprehensible as an act of giving. The sentence is a ditransitive construction (NVNN) which by its early and frequent association with the word give has come to take on a schematic constructional meaning in which an object is transferred by one party to another.

Cue validity of ‘eye’ for ‘pig’ or for ‘animal’ Let x = population of animals in the world We know pigs are some fraction of the world’s animals, let’s just say 5% for example Then   0.05x = population of pigs in the world Let’s assume all animals have 2 eyes (not accurate, but close enough) And only animals have eyes Then  2x = population of eyes in the world To estimate probabilities, we can count actual occurrences of features and categories in a representative sample.  In this simplified illustrative example, our representative sample is the universe. P (animal|eye) = co-occurrence of animal & eye/ occurrence of eyes Since there are 2x eyes in the universe and every time we see one, we see an animal P (animal|eye) = 2x/2x = 1 Now for P(pig|eye) We still have 2x eyes in the world but now the only co-occur with pigs .05x * 2 times or 0.1x    so, P(pig|eye) = 0.1x / 2x = 0.05 Of course we don’t care about these actual numbers, just the fact that the cue validity of eye for animal is greater than the cue validity of eye for pig.  To understand it more intuitively we can think of it this way.  Every time we see an eye, we see an animal, but only some of the time when we see an eye are we seeing a pig. Cue validity of hoof for animal or for horse Let x = population of animals with hooves Assume only animals have hooves and every hoofed animal has 4 hoofs. Then  4x = population of hooves Assume 90% of hoofed animals are horses Every time a hoof occurs, there is an animal so, P (animal|hoof) =  4x/4x = 1 90% of those hooved animals are horses so 90% of the time a hoof is seen, a horse is seen P (horse| hoof) =  0.9 * 4x / 4x = 0.9 Because not all hoofed animals are horses the cue validity of hoof for animal is higher than the cue validity of hoof for horse, but, assuming most hoofed animals are horses, we can say these 2 validities are roughly equal. Cue validity of siren for car or for police car Let x = the population of police cars Assume all police cars have sirens, and all cars with sirens are police cars Given this, every time that a siren and car co-occur, a siren and a police car also co-occur But, assume other non-car objects (ambulances and fire trucks) also have sirens and let y = population of sirens that are not on police cars then (x+y) = population of all sirens P(police car| siren)  =  x / (x + y) And since we’re assuming that all co-occurrences of sirens and cars are instances of police cars P(car | siren) = x / (x + y) Bottom line:  cue validity of siren for police car is the same as cue validity of siren for car.