Escolar Documentos
Profissional Documentos
Cultura Documentos
Contents
Examination again The rescheduled class Todays topics
Pseudo-science and hoaxes The philosophy of science Scientific reasoning
Deduction Induction Inference to best explanation Explanation Causality
Conclusion
2
Examination
Confirming ...
3 questions each relating to both parts of the course
From which you select 1
This class will be in 1467-515 Tirsdag den 10.04.2012 fra kl. 9 til kl. 12
According to Popper concepts like repression and sublimation could permit compatibility with any data the theory is unfalsifiable
6
Popper argued that by contrast Einsteins theory of gravitation makes very definite predictions
These can be tested against observations
Arthur Eddington organised an expedition to the island of Prncipe near Africa to observe the Solar eclipse of 29 May 1919 that provided one of the earliest confirmations of relativity During the eclipse, he took pictures of the stars in the region around the Sun According to the theory of general relativity, stars with light rays that passed near the Sun would appear to have been slightly shifted because their light had been curved by its gravitational field. This effect is noticeable only during eclipses, since otherwise the Suns brightness obscures the affected stars. Eddington showed that Newtonian gravitation could be interpreted to predict half the shift predicted by Einstein 7
Scientists frequently find empirical facts that dont exactly fit with the predictions of a theory
But rather than reject the theory, they attempt to save it By taking air resistance into by postulating some other factor that needs to be taken account we could explain into account
almost any empirical finding This is reminiscent Saussures In the case of Galileos falling objects experiment, if a of lead balloon postulation of proto-Indo-European and a feather were dropped at the same time, they would not reach the ground at the same time laryngeals to account for certain Note the obvious parallel to Poppers critique of Freud:
otherwise inexplicable facts about Greek morphology. Instead of rejecting his theory, he takes air resistance into account Subsequently vindicated by Hittite.
A more complex case observations of the orbit of Uranus showed some differences from the predictions of Newtonian physics
Independently in 1846 Adams and Leverrier suggested the existence of another planet that provided the gravitational force that would be responsible for the irregularity in the orbit of Uranus This planet (Neptune) was subsequently discovered almost exactly where predicted
9
Especially in the case of social sciences including linguistics Linguists theories are not always do on occasion critique one another on the easily falsifiable in practice basis of the unfalsifiablity of the others claims.
Even in physics means of measurement Extreme difficulties in applying this notion to the
Generally speaking I dont hold this critique in high regard: present falsifiability may be beyond
linguistic sign It is the less interesting claims that are more easily I believe this is the case for string falsified theory claims not mediated by the sign
It is an important part of the scientific endeavour to account for conflicts exceptional facts while retaining the theory, and working within it
More on this later
10
Popper attempted to characterise science in a Platonic way in terms of an essential feature that must be possessed
His approach to this task seems eminently reasonable:
Compare instances of pseudo-science with good science, and identify the differences
The point is he may have selected the wrong differences
One of the most famous hoaxes in the history of science was the Piltdown Man:
Bone fragments were presented as the fossilised remains of a previously unknown early human.
Parts of a skull and jawbone, said to have been collected in 1912 from a gravel pit at Piltdown, East Sussex, England
The Latin name Eoanthropus dawsoni ("Dawson's dawnman", after the collector Charles Dawson) was given to the specimen The significance of the specimen remained the subject of controversy until it was exposed in 1953 as a forgery
The lower jawbone of an orangutan that had been deliberately combined with the skull of a fully developed modern human
Marc D. Hauser (born 25 October 1959) is an American evolutionary biologist and a researcher in primate behavior and animal cognition who taught in the Psychology Department at Harvard University.
In August 2010, a committee of Harvard faculty found Hauser solely responsible for eight counts of unspecified scientific misconduct. On August 1, 2011 Hauser resigned his position at Harvard
Hwang Woo-suk (Korean: , born January 29, 1953) is a South Korean veterinarian and researcher, professor of theriogenology and biotechnology at Seoul National University (dismissed on March 20, 2006)
Became infamous for fabricating a series of experiments, which appeared in high-profile journals, in the field of stem cell research Until November 2005, he was considered one of the pioneering experts in the field, best known for two articles published in the journal Science in 2004 and 2005 where he reported to have succeeded in creating human embryonic stem cells by cloning. 15
Captain Cooks South Sea Island Vocabularies which contained, among other things:
Vocabularies from languages spoken in places where Cook never landed Wordlists that must have been gathered during circumstances of conflict with indigenes Error analysis:
The canonical vocabularies showed the normal quota of errors and inexplicable forms to be expected in such early lists, somewhat enhanced by Lanyon-Orgills ineptitude at identifying modern equivalents for the words. With the Lanyon vocabularies, by contrast, he scored 100 percent, and the resemblances between manuscript and modern language were uncannily close. The apparently inescapable conclusion was that the Lanyon lists were created not by contemporaries of Cook, but on the basis of later sources by a modern hand, using a linguistic equivalent of the distressing applied by makers of fake antiques (for example English-based spellings such as ee, oo would be substituted for modern i, u). 21
22
Wearing your scientists hat, these questions dont really concern you
But many scientists also doff their scientific hats, and think about these questions
Many prominent scientists like Descartes, Newton, Einstein thought about such questions, e.g.
How science should be carried out How much confidence to place in its methods Whether there are limits to scientific knowledge and praxis
The increasing specialisation of science has resulted in less interest in the philosophical questions by scientists
And also the polarisation of science and humanities (where philosophy belongs) of the modern university system
25
Scientific reasoning
We now discuss and evaluate some processes employed or claimed to be employed in scientific reasoning
Deduction Induction Inferences of best explanation Explanation
26
Deduction
This method of reasoning everyone accepts as valid
Study of deductive reasoning is a part of logic
Linguistic examples?
27
A. All languages have vowels and consonants B. Nyulnyul is a language C. Nyulnyul has vowels and consonants A. All languages have nouns B. Shua is a language C. Shua has nouns
This principle tells us that if the premises are true, so is the conclusion
The reasoning is valid and exciting eh? But the premises need not necessarily be
E.g. Not all languages have nouns (as a distinct part of speech)
28
29
Induction
Many people have suggested a naive inductivist account of science
That places induction as the central method of scientific reasoning
Examples of observational statements of course, some interpretative knowledge beyond simple observation is essential: these are not completely atheoretical statements
Gooniyandi is spoken in the town of Fitzroy Crossing Shua has a click sound in the word kick The last fluent speaker of Unggumi died in the last decade of the twentieth century
In principle, these statements can or could be established by observation the third one, with some qualifications:
It was observable at some point in time
These are singular statements; in science we are concerned not with singular statements, but with making generalisations
With statements that are universal in application:
Statements that apply to all phenomena of a particular kind
31
The naive inductivist view of science is that science is based on the principle of induction:
If a large number of Xs have been observed in a wide range of conditions, and if all these Xs without exception possess the property Y, then all Xs have the property Y
Linguistic example
We observe through collection of many instances of Agent NPs in Gooniyandi, in a range of different transitive clauses, with many different types of NP (pronominal, animate, inanimate, ...) each of which is marked by ngga We conclude that all Agent NPs are marked by the morpheme ngga
Seems reasonable But how certain can we be of our conclusion
Can we be any more certain than Russells turkey? Not really
36
In fact, once we stop observing elicited examples and look at narratives and everyday speech we find that there are exceptions
Agent NPs that are not marked by this morpheme
A difficulty is that what counts as a broad range of circumstances is not clear
Our elicitation aimed at gathering a broad range of data But in fact it was broad in the wrong ways There is no way we can determine what the relevant range of circumstances is
37
Some have suggested a way out is to introduce probabilities into the story
That the inductive inferences be evaluated in terms of probabilities
E.g. the likelihood that the our general statement is valid given the set of specific statements Possibly this might sound like a reasonable solution
Intuitively it is reasonable to expect that with increasing numbers of singular statements the probability will increase But the turkey example shows that this is not always valid And how could one calculate the probability?
38
Intuitively, if we have a good explanation the probability of the universal statement is increased
But to include this involves a problem for the nave inductivist story of the scientific process
It cant be a simple unidirectional process
39
Another difficulty with the proposals concerns the singular statements conflicting with the generalisations
In our Gooniyandi word structure example, we will eventually find words like mm, aa, mhm
But the obvious thing to do is not to immediately throw out the generalisation
Rather, we look at refining the notion of word
40
Philosophers of science do not generally regard the nave inductivist story as a plausible model of the scientific process
Or consider induction to have a central place in scientific reasoning
This is not to say that it has no place at all:
Presumably we use something like induction in everyday life
From observations we construct expectations as to what will happen a matter of survival
41
42
Clearly the premise does not logically entail the conclusion in these cases There are other possibilities, also consistent with the facts:
Although not a good explanation, there is a tiny chance that the similarities are a result of chance They might alternatively be the result of borrowing between the two languages
Observe that among the similarities in the lexicon there are a number of recurrent sound correspondences
Common origin together with regular sound change provides a better explanation than borrowing
46
47
This method of reasoning is common in linguistics Another example comes from my own descriptive work on Gooniyandi Here is a potted version of the story of the analysis of the pronoun system in the language
I observed early on that there were two 1st person nonsingular pronominals, ngidi and yaadi I concluded that the language made a distinction between inclusive and exclusive
Which is widespread in Australian languages Specifically,
Ngidi would be 1exclusive, dual or plural Yaadi would be 1 inclusive, dual or plural
Alan Rumsey drew exactly the same conclusion for Bunuba when he did his first fieldwork a couple of years earlier
48
These are not the only possibilities but we were content with them
For a while anyway
Within a few weeks of fieldwork however I found that people were just not using the 1st person pronouns as they should be if it was an inclusive-exclusive system
Observations conflicted with expectations Ngidi was used for not just 1&3 but also 1&2 Yaadi was never used for 1&2, only for 1&2&3 The observations were followed up with experiments i.e. getting speakers interpretations of invented forms: I asked speakers what does yaadi-yoorroo mean? Some rejected the form Some said it means we three!
49
I of course told Alan Rumsey (my supervisor then) about this problem
New to him, so he also tested things out in Bunuba
Finding exactly the same situation as in Gooniyandi
Yaadi means
1&2&3 1&2&3
50
In the mid 1990s Rumsey and myself both suggested explanations Mine was:
The system is an inclusive-exclusive one, but it operates on different principles than the standard one
What is included or excluded is not the hearer
Just apply logic to
It is hearers specifically the hearer plus others And you get the right
results
Ngidi excludes hearer plus others Yaadi includes hearer plus others
53
54
Explanations
One of the most important aims of science is to explain the world
Astronomers aim to explain why solar eclipses occur when they do Linguists aim to explain e.g.:
Why languages have the structures they do Why and how some languages are similar to one another How languages are used in speech Why a bilingual person choses to use one language rather than another in a given context
55
What is an explanation? Carl Hempel (1905-1997) suggested the covering law model for scientific explanation in 1950s
While a student in Gttingen he encountered David Hilbert and was impressed by his attempt to base all of mathematics on solid logical foundations derived from a limited number of axioms (Hilbert's Program)
56
Hempels suggestion:
The premises should entail the conclusion (logically a deductive argument) The premises should be true The premises should comprise at least one general universal statement at least one general law
Schematically:
General laws Particular facts Phenomenon to be explained
58
A linguistic example:
We want an explanation for why a bilingual speaker in the Austrian village Oberwart switched from Hungarian to German in an argument In the following discourse a mother is collecting her daughter, who has been looked after by the grandparents during the day. The girl has been misbehaving, and the grandfather sympathises with her.
59
The switch to German justifies the choice of child-rearing methods. This switch to German ends the argument between the mother and grandfather
60
We have there general laws (concerning language affiliations and choice) and particular facts (about the language use in the community)
We conclude that the switch to German in the conversation was motivated by the speaker wishing to distance herself from her interlocutor
She establishes herself as an Austrian, not an Oberwartian Austrian is dominant, the language of the powerful By implication she is the powerful one with the powerful argument So we can account for the effect of her language choice for her successfully winning the argument over her father
62
Similarly for the other linguistic example, optional ergative case marking in Gooniyandi though with some twists
A general statement I have proposed in recent work is given that a grammatical morpheme is optional:
Usage vs. non-usage always codes meaning relating to the interpersonal dimension
Specifically, this relates to the general dimension of joint attention
63
Prominence and backgrounding correlate with figure and ground but they are not identical notions
64
Flashing light off tells you that these guys are just cruising, looking for trouble: police function backgrounded
FL AS H IN G
hy
Use
System CATEGORY
What typeof ve hicle ? Car
FUNCTION
What function doe s theve hicles e rve ? Police vehicle
65
This means that there is an additional factor that must be taken into account:
Markedness
Non-use
No meaning [backgrounded]
No meaning [backgrounded]
Meaning [+backgrounded]
Meaning [+backgrounded]
67
In my papers on the topic, I show how some specific instances of non-use can be explained in this way:
Now we are explaining the specific statement not the general law (as above)
68
aa, ngidi garndiwangoorroo, garndiwirri ngidi yoowooloo-yoorroo, aa we many two we man-DU baraj-jirr++a-yi, thinga, Gooniyandi track-1excNOM+3sgACC+A-DU foot We all we two Aborigines tracked him on foot.
Here the non-use of the ERG on ngidi suggests pragmatically low agentivity of the speaker group.
Why, to what purpose would the speaker do this? It primarily concerns the construal of the narrative world as one in which:
The person being followed expended a lot of effort trying to find his way home, but failed and died. The followers easily tracked him, and found his body.
69
The next example is the only instance in which an inanimate Agent is not ergatively marked:
(1) thinga gilba-yirr++di-yi / gamba / yilij-jin++a foot find-1excNOM+3sgACC+DI-DU water rain-1excACC+3sgNOM+A garr garrwaroo / after afternoon We found his tracks, but it rained on us that afternoon. Gooniyandi
The reason: non-use of the ergative implicates low agentivity, low potency
Lack of effect on the Undergoer as the next sentence says, it was just a sprinkling of rain. NB not lack of volitionality
70
There are two important difficulties with Hempels story see Okasha 2002 First, what explains what? The structure of explanations is symmetric:
Flagpole example provides a clear illustration
We can measure the height of a tree from:
Length of its shadow Angle of the sun
We could use the height of the tree and angle of the sun to explain the length of the shadow In Hempels scheme, we could equally explain the height of the tree by the length of the shadow and the angle of the sun! 71
72
The causal criterion also explains why Johns taking the pill is not the explanation of his not being pregnant
Clearly the cause of this situation is his sex
74
Things are not always the same in linguistics it is not clear that causality ought to be invoked in all instances of explanation in our subject
In our language choice example, we explain
The choice of German to win the argument The winning of the argument by choice of German
My opinion is that the situation here is different to our flagpole and unpregnant male examples
There is a good reason why
The type of linguistic phenomenon we are trying to explain here concerns the linguistic sign
Recall Saussures model of the intrinsic relation between signifier and signified No causal relation here We can view the sign from either perspective, and thats what we are doing in our two alternative explanations of language choice and case-marker use
Of course, I do not suggest that causal explanations have no role at all in linguistics
Only that causality is not the ultimate arbiter in all instances
76
Conclusion
Well be talking more about the philosophy of science including linguistics later
What I try to do in this course is to encourage some of this self-awareness and reflexive thinking about our subject, linguistics
And to see it in the wider contexts of both
Its history of development Its ecology, the other disciplines that it interacts with
In short our interests are in the history and philosophy of the science of linguistics 77
78