Abstract The Imitation Game (Alan M. Turing, 1950), now
A computer program capable of passing I.Q.tests
Pritika Sanghi(psan5@https://www.wendangku.net/doc/8a16202451.html,)
School of Computer Science and Software Engineering
Monash University,Clayton,VIC3800Australia
David L.Dowe
School of Computer Science and Software Engineering
Monash University,Clayton,VIC3800Australia
Abstract
The Imitation Game(Alan M.Turing,1950),now commonly known as the Turing Test(see,e.g.,Oppy and Dowe,https://www.wendangku.net/doc/8a16202451.html,/entries/turing-test,2003), was proposed as a way in which thinking or intelligence could be ascribed to any agent-including a computer program or machine-able to play the game.People routinely ascribe intelligence to humans and other animals by a variety of means,including those discussed by Turing.But when humans wish to specifically quantify intelligence,this is most commonly done by means of an intelligence quotient(I.Q.)or other aptitude test.Many such aptitude test questions fit in to Turing's (1950)framework of being able to be typewritten to a teleprinter communicating between two rooms-or,using modern technology still well within the spirit of Turing's game,being able to be typed as text into a World Wide Web(WWW)page applet.Sequences of such questions-such as an entire I.Q.test of them-may well form a strict subset of Turing imitation games,since they typically are independent of one another and do not take any advantage(or even account)of the contextual (conversational)framework of Turing's game.We present here a fairly elementary WWW-based computer program(shown in large part at https://www.wendangku.net/doc/8a16202451.html,.au/~psan5)which,on a variety of I.Q.tests,regularly obtains a score close to the purported average human score of100.One conclusion that some might make is to ascribe intelligence to the program. Another conclusion to make is that the reason that I.Q. test success can be automated with comparative ease is that administering an I.Q.test requires little intelligence-it requires comparatively little more than giving a list of questions with known answers.Turing's imitation game test requires greater intelligence to pass largely because of the flexibility it permits to an intelligent questioner-such as in the use of language and in taking into account the responses to previous questions before continuing the line of questioning.We also briefly consider a d m i n i s t r a t i o n o f t h e i m i t a t i o n g a m e“t e s t”v i a“d e t e c t i o n p r o g r a m s”a s a t e s t i n i t s o w n r i g h t(D o w e a n d H a j ek, 1998;CalTech Turing Tournament, https://www.wendangku.net/doc/8a16202451.html,,2003).All other things being equal,a more intelligent administrator can administer a more challenging test-and this notion can be continued recursively(Dowe and Hajek,1998).
1.Introduction
Alan Turing suggested the Imitation Game(Alan M. Turing,1950),now known as the Turing Test,in1950 as a way of ascribing thinking,or intelligence,to machines.It involves the interrogation by a judge of a human and a machine using teletype from behind a screen where the judge knows that one is a human and one is a machine,but the judge does not know a priori which is which.The test requires the machine to fool the judge into believing that it is human and that the human is the machine-and the Turing Test has certainly been frequently discussed and surveyed (Moor,2000;Saygin et al.,2000;Copeland,2000;V. Akman and P.Blackburn,2000;Oppy and Dowe, 2003).I.Q.tests are used to measure the level of intelligence of humans.If a computer is made to take an I.Q.test,it,too,can be given a score based on its performance.Given that such scores are used as a measure of human intelligence,it seems plausible that such a score might be used as an indication of the level of intelligence of the computer.
We present here a program which takes I.Q.tests in the form of questions typed in the text box of a simple webpage.The score can be calculated based on the number of questions it answers correctly.In cases where the question cannot(yet)be represented to the program(e.g.,picture questions,such as in Figure1) and there are n options,the program gets a score of1/n for the question.This is because the probability of getting the answer right is1/n and so the long run average score from guessing is(n-1)/n x0+1/n x1= 1/n.
More is said about I.Q.tests and their constituent questions in Section2,the program in Section3,and the program's performance in Section4;and in Section 5we discuss administration of the Turing Test as a test in its own right and the relevance of information-theoretic compression to inductive learning and intelligence.
Figure1:A picture question.
2.I.Q.Tests
Intelligence Quotient or I.Q.(A.C.E.,unknown;H.J. Eysenck,1988;Helenelund HB,2001; http://heim.ifi.uio.no/~davidra/IQ/,unknown;I.Q.Test Labs,2003;KHAN-UUL Institute,2001;Mensa,2000; Testedich.de,2002)is used as a measure of human intelligence.The first I.Q.test was conducted by a French scholar,Alfred Binet,around1906(KHAN-UUL Institute,2001;Enyclopedia Britannica,2000). The original purpose of the test was to identify slow learners in school.I.Q.is the ratio of mental age over chronological age.For example,consider a child of age 12.A mental age of10will suggest an I.Q.of10/12x 100=83;and a mental age of14will suggest an I.Q.of 14/12x100=117.An I.Q.of100implies that the mental age is same as the chronological age.
If a computer program can achieve a score between 90and110,it can arguably be said to have average intelligence(H.J.Eysenck,1988).The program should ideally not be test specific,as it would be rather abnormal to get a score of over120on one I.Q.test and under50on another test.
2.1Forms of frequently asked I.Q.questions Most I.Q.tests seem to have certain similar characteristics.An analysis of various I.Q.tests(such as those listed above)shows that the following types of questions are common:
1.Insert missing number
a.At end
b.In middle
2.Insert missing letter
a.At end
b.In middle
3.Insert suffix/prefix to complete two or more
words
https://www.wendangku.net/doc/8a16202451.html,plete matrix of numbers/characters(see
Figure2)
5.Questions involving directions
6.Questions involving comparison
7.Picture questions(see Figure1)
8.Pick the odd man out(word or picture)
9.Coding
Figure2:A matrix question
3.The Program
The program,which can be accessed from https://www.wendangku.net/doc/8a16202451.html,.au/~psan5,recognizes these characteristics(forms of frequently asked I.Q. questions)from Section2.1and tries to find the best-suited solution.It is written in Perl and is about960 lines of code.
A parser with a restricted vocabulary and a basic string search for keywords can be implemented to recognise the type of question.It can then be simple for the program to calculate the answer based on some pre-defined rules.
A trivial way of recognising the question would be to l o o k f o r p a t t e r n s s u c h a s“Wh a t i s t h e n e x t n u m b e r i n t h e s e q u e n c e”.A s m a l l c h a n g e i n t h e f o r m a t o f t h e question will cause the current version of the program to fall over.We can overcome this by looking for certain keywords(e.g.,number+sequence).A comprehensive list of keywords can be made for each type of question.If two questions have similar or identical keywords,extra keywords or patterns need to be included to differentiate between them.
A thorough list of possible keywords for a certain type of question can be made by analysing a large number of I.Q.tests.Even then there may be cases where it cannot identify the question or identifies it incorrectly.Provided the keyword list is made properly, this should be rare.
Once the type of question is established,it can be simple for the program to find the answer.An algorithm that does calculations and/or searches can easily be made that can find most of the answers to the questions.
A large number of questions that involve simple logic can be programmed.
We n o w d i s c u s s t h e p r o g r a m’s a n s w e r s t o I.Q. questions of frequently asked forms,such as those from Section2.1.
3.1Insert missing number or letter at end Consider the question–“I n s e r t t h e n e x t n u m b e r i n t h e sequence-12345”.‘N u m b e r’c o u l d b e r e p l a c e d b y
‘d i g i t’,a n d‘s e q u e n c e’c o u l d b e r e p l a c e d b y‘s e r i e s’.I t c o u l d a l s o b e p h r a s e d a s“Wh i c h n u m b e r f o l l o w s logically-12345”o r“Wh a t i s n e x t i n t h e s e q u e n c e-12345”.K e y w o r d s f o r t h i s t y p e o f q u e s t i o n c o u l d b e ((insert||what||which)&&(number||digit||sequence ||s e r i e s)),w h e r e‘&&’d e n o t e s‘a n d’a n d‘||’d e n o t e s ‘o r’.A n e x a m p l e o f p h r a s i n g that can be used for more than one type of question would be–“Wh a t i s n e x t i n the sequence-12345”,o r“Wh a t i s n e x t i n t h e sequence-A B C D E”.T h e y c a n b e d i f f e r e n t i a t e d b y the fact that one has a sequence of numbers and the other has letters.The keywords for number sequences may be extended to check that there are only digits(0-9)and separators in the sequence section.The keyword for character sequences will not be extended as they can contain numbers and characters in the sequence section. In the case of sequence questions,certain types of sequences(Arithmetic Progression,Geometric Progression,Fibonacci Series,Powers of a series,etc.) are used frequently in I.Q.tests-see,e.g.,(A.C.E., unknown;H.J.Eysenck,1988;Helenelund HB,2001; http://heim.ifi.uio.no/~davidra/IQ/,unknown;I.Q.Test Labs,2003;KHAN-UUL Institute,2001;Mensa,2000; Testedich.de,2002).If the given sequence is any of the types of sequence described above,it can be checked with ease.If it is,the next number/character can be calculated simply according to the properties of the sequence,although at least three numbers/letters are required to find a pattern.The program can solve this type of question from I.Q.tests most of the time.There will be cases when an answer cannot be found or the answer found is incorrect.Since it is rare for humans to get all answers correct,it is not considered a big problem.
3.2Insert missing number or letter in middle This type of question will be recognised using a technique similar to the one used in Section3.1.An e x a m p l e w o u l d b e‘I n s e r t t h e m i s s i n g n u m b e r:1020? 4050’.T h e p r o g r a m w i l l l o o k f o r a s p e c i a l c h a r a c t e r(x ||_||?)inside the sequence,where,again,‘||’d e n o t e s ‘o r’.T h e n u m b e r/s o r l e t t e r/s f o r t h a t p o s i t i o n/s c a n b e guessed using the properties for sequences mentioned in Section3.1.The entire sequence is then checked.If valid,the answer is found.This is not yet implemented in the program.
3.3Insert suffix/prefix to complete two or more words
This kind of question can again be recognised by keywords.For questions involving suffix and prefix (e.g.,What completes the first word and starts the second:wi..nt),keywords could be((suffix&&prefix) ||(c o m p l e t e&&w o r d)),w h e r e,‘||’d e n o t e s‘o r’a n d ‘&&’d e n o t e s‘a n d’.B r u t e f o r c e i s t h e n u s e d t o s e a r c h for a suffix for the prefix from the word list.Next it checks if that suffix is a prefix for the suffix. Sometimes,only a suffix is requested for more than one word(e.g.,Insert a word of size2that completes the words:ma,fa,chara(..)).The technique remains much the same.Instead of checking if the prefix is valid the next time it will check if it is a valid suffix for the rest of the words.A repeated prefix can also be found using the technique with slight modifications(e.g.,Insert the word of size2that completes the words:(..)de,ke,lt, trix).Most of this is implemented in the program. Currently,the program requires the question to be re-f o r m a t t e d f r o m‘w i..n t’(i t i s f o u n d o n I.Q.t e s t s i n t h i s o r s i m i l a r f o r m a t) t o ‘2-wi-n t’. T h i s c a n b e i m p l e m e n t e d b u t i t h a s n’t y e t b e e n d o n e i n t h e p r o g r a m. With the re-formatted input the program generally finds the solution.
3.4Complete matrix of numbers/characters
In order to complete the matrix,patterns(e.g.,for Figure2,a column is double the previous column) existing inside the matrix have to be identified.Once they are identified,the pattern can be applied to find the most appropriate value.Patterns can be found using the following techniques:
1.The sum of rows can be the same.
2.It could be in the form described in Figure3,
w h e r e‘o’r e p r e s e n t s a n a r i t h m e t i c f u n c t i o n
s u c h a s‘+’,‘-’,‘*’,‘/’,‘^’,e t c.
3.O p e r a t o r s(‘+’,‘-’,‘*’,‘/’,‘^’,e t c.a n d o n e
‘=’)a r e i n s e r t e d b e t w e e n c o l u m n s.I f t h e s a m e
combination is valid for each row,that is the
pattern.
4.Zig-zag through rows to find a pattern.
5.The matrix is transposed and above steps are
repeated.
6.Columns are shuffled and steps3,4are
repeated.
This is not yet implemented in the program
Figure3:Patterns in matrix questions
3.5Questions involving directions
Keywords like(left||right||east||west||north||south) can be used to identify questions involving directions. Directions can be represented quite easily
using
degrees.Assuming north to be0o and moving clockwise,east becomes90o.The points can be represented relative to the starting point.The distance between them can be calculated using methods such as Pythago r a s’s t h e o r e m.F o r e x a m p l e,‘J o e m o v e s t h r e e blocks east,takes a right turn and walks further four s t e p s.H o w f a r i s h e f r o m h i s o r i g i n a l p o s i t i o n?’F r o m t h e f i r s t p a r t o f t h e s e n t e n c e i t w i l l t a k e‘t h r e e’a n d ‘e a s t’a n d m a k e t h e n e w p o s i t i o n3u n i t s90o right of n o r t h.N e x t,‘r i g h t t u r n’a n d‘f o u r’w i l l b e t a k e n i n t o consideration,making the new position90o right,4 units away from previous position.The distance between points can be calculated using properties of a triangle.This is not yet implemented in the program. 3.6Questions involving comparison
First,all the elements being compared should be listed. Positions are given to elements relative to others based on initial sentences.The list is parsed for elements, which can be given position relative to other elements.
C o n s i d e r,e.g.,“A i s t a l l e r t h a n B,B i s t a l l e r t h a n C”.I n the second parse,A,B,and C will be given positions relative to one another(rather than merely the two initial separate comparisons not relating A to C).The two extremes and the complete,total ranking can then be found.This is not yet implemented in the program. 3.7Picture Questions
The aim of picture question is to check for pattern recognition.It will be hard for a program to solve picture questions(see,e.g.,Figure1).This is mainly because it is tough to represent them in a teletype environment.If the picture were defined symbolically, perhaps too much work would have been done in overly assisting the program.If the picture is described(e.g., the second element in the sequence is an unshaded square with a diagonal top right to bottom left),then parsing will be a challenge(see also Section5.1).This is not yet implemented in the program.
3.8Odd man out
Odd man out questions are commonly used in I.Q.tests (A.C.E.,unknown;H.J.Eysenck,1988;Helenelund HB,2001;http://heim.ifi.uio.no/~davidra/IQ/, unknown;I.Q.Test Labs,2003;KHAN-UUL Institute, 2001;Mensa,2000;Testedich.de,2002).They require one to differentiate names of countries,cities, vegetables,fruits,etc.To differentiate between objects, one probably needs to understand their concept.But, there may be many categories for objects(e.g.,for picture questions from Section3.7,a circle is in both the family of geometric shapes,and also geometric shapes with no edges).It could be a challenge for some time yet to make a computer program understand the concept behind different pictorial objects.However,odd man out questions not involving pictures may well be amenable before too long to a search analogous to a much more complicated version of that required in Section 3.3.This is not yet implemented in the program.
3.9Coding
These questions generally involve coding from alphabets to numbers or vice versa(e.g.,If KNOW is 20232432,what is CODE?).These questions are of the kind
(if&&(a-z||A-Z||0-9)+&&is&&(a-z||A-Z||0-9)+ &&what&&is&&(a-z||A-Z||0-9)+),w h e r e‘+’d e n o t e s o n e o r m o r e a n d,a s b e f o r e,‘||’d e n o t e s‘o r’a n d ‘&&’d e n o t e s‘a n d’.A r e l a t i o n i s f o u n d b e t w e e n ‘K N O W’a n d‘20232432’b a s e d o n A C S I I v a l u e s. The same relation is then used to find the code for ‘C O D E’.T h i s i s n o t y e t i m p l e m e n t e d i n t h e p r o g r a m.
3.10Other kinds of questions
An I.Q.test that steers away from the usual way of testing(certain types of questions can be expected most of the time)or a non-standard test will be cause for concern.Not being able to identify the questions,the program will fail the test.It is highly unlikely that a h u m a n c a n’t c o m p r e h e n d a n y o r m o s t o f t h e q u e s t i o n s on an I.Q.test.
4.Results–t h e p r o g r a m’s“I.Q.”
We present here the results of the program on various I.Q.tests.Most of the questions from the I.Q.tests were re-formatted before being entered to the program(see Section3).
Table1:I.Q.Scores on various tests.
Average
A.C.E.I.Q.Test108100
Eysenck Test1107.590-110
Eysenck Test2107.590-110
Eysenck Test310190-110
Eysenck Test4103.2590-110
Eysenck Test5107.590-110
Eysenck Test69590-110
Eysenck Test7112.590-110
Eysenck Test811090-110
I.Q.Test Labs5980-120
Testedich.de–The I.Q.
Test
84100
I.Q.Test from Norway60100
Average96.2792-108
As seen from Table1,the program scores high on some tests and low on others.A link can be seen between the score and the type of I.Q.test.The program
can attain a high score with ease on I.Q.tests which are more focussed on mathematics,pattern recognition, logical reasoning and computation.On the other hand, I.Q.tests that are based on general knowledge,language skills and understanding are a challenge to the program. The case is often the reverse for humans.Of course, human I.Q.tests are anthropomorphic(or “c h a u v i n i s t i c”),a n d a n i n t e l l igent non-English speaker, non-human earthling or extraterrestrial could be expected to struggle on an English-language I.Q.test.
5.Some thoughts and discussion
First,apart from relatively minor issues such as m e m o r y a n d r u n n i n g s p e e d,t h e“i n t e l l i g e n c e”o f a computer presumably depends almost solely on the software program it is running.That said,we now ask some additional rhetorical questions.
Are I.Q.tests really a measure of our intelligence? Will getting a higher score than a human mean that the computer program is more intelligent than that human? Without necessarily fully answering either or both of these two questions,let us take this discussion in two directions.In so doing,we shall consider two possible modifications to the Turing test.
5.1Administering the Turing test
Recalling Sections2.1and3,most I.Q.test questions fit s t r a i g h t f o r w a r d l y i n t o T u r i n g’s o r i g i n a l c o n v e r s a t i o n a l framework.With however much more work(to a human or possibly a non-human possibly poised with pen(cil)and paper),picture/diagram questions can probably also–by careful description–be brought w i t h i n T u r i n g’s c o n v e r s a t i o n a l f r a m e w o r k.S o,I.Q.t e s t q u e s t i o n s s e e m,b y a n d l a r g e,t o f i t n e a t l y i n t o T u r i n g’s conversational framework.Whereas the judge(or adm i n i s t r a t o r)o f T u r i n g’s i m i t a t i o n g a m e t e s t c a n l e a d the conversation in any number of directions given the conversation so far,I.Q.test questions asked largely or totally neglect the answers to previous questions.In other words,an I.Q.test requires less intelligence to administer than a Turing imitation game test–and this is essentially why it is less challenging and easier(for a computer program)to pass.This raises the issue that intelligence is required to administer a Turing test–and this ability could be used as a measure in a test for intelligence.Of course,we can continue this recursively (Dowe and Hajek,1997;Dowe and Hajek,1998).More explicitly,some intelligence is required to pass a Turing test(TT0).If we set up a new test,TT1,which is to correctly administer/judge TT0,then that presumably or seemingly requires more intelligence to pass.The test f o r“d e t e c t i o n p r o g r a m s” i n t h e C a l t e c h T u r i n g Tournament(CalTech,2003)is a case in point.And,of course,we can continue this recursively(Dowe and Hajek,1997;Dowe and Hajek,1998)and,e.g.,set up a new test,TT2,which is to correctly administer/judge/detect in TT1.(Passing the Turing Test,TT0,is analogous to writing a good academic paper.Passing TT(1)is analogous to being a good referee.Passing TT(2)is analogous to being a good member of a program committee or editorial board–one must be able to choose appropriate referees.) Continuing recursively by induction,given test TT(i), we can set up a new test,TT(i+1),which is to correctly administer/judge/detect in TT(i).Etc.While all of the a b o v e i s t r u e a n d T T(1),T T(2),…,T T(i),… a r e a l l interesting,salient and worthwhile directions in which to re-examine the Turing Test(TT0),it would also at least appear to follow by mathematical induction that e a c h o f T T(1),T T(2),…,T T(i),… c a n b e e x p r e s s e d–albeit seemingly in increasing order of difficulty–in T u r i n g’s o r i g i n a l c o n v e r s a t i o n a l f r a m e w o r k,T T0.
5.2Inductive learning,compression and MML The other direction in which we take this discussion is the observation–see(Dowe and Hajek,1998)and elsewhere–that traits which seem to be necessary for (human)intelligence and which certainly are assessed in human I.Q.tests include rote learning(and memory), deductive learning(e.g.,via modus ponens)and inductive learning.Inductive learning is perhaps the most important,significant and impressive of these. When asked for a list of great thinkers and minds,the primary reason for the inclusion of Newton,Darwin, Einstein and others is because of their inductive inferences-or inductive learning-of theories(gravity and laws of motion,evolution,relativity,etc.).
The relevance of compression to learning languages is discussed in(Wolff,1995).The Minimum Message Length(MML)theory of inductive inference(Wallace and Boulton,1968;Wallace and Freeman,1987; Wallace and Dowe,1999)states that the best theory,H, to infer from data,D,is that which minimises the (compressed)length of a two-part message transmitting H followed by D given H.MML is a quantitative form o f O c k h a m’s r a z o r(N e e d h a m a n d D o w e,2001), rewarding simple theories which fit the data well.The relationship between MML,Kolmogorov complexity (Kolmogorov,1965)and related(information-theoretic) works(e.g.,(Solomonoff,1964))is thoroughly discussed in(Wallace and Dowe,1999).MML is relevant to inductively learning all range of models–not just languages–from data.
Given the relevance of two-part MML compression to inductive learning,Dowe and Hajek have argued (Dowe and Hajek,1997;Dowe and Hajek,1998)that a n“a d d i t i o n a l r e q u i r e m e n t o n T u r i n g’s t e s t i s t o i n s i s t that the agent being subjected to the Turing test not only pass the test but also have a concise,compressed representat i o n o f t h e s u b j e c t d o m a i n”.I n d e p e n d e n t l y
but quite relatedly,Hernandez-Orallo and Minaya-C o l l a d o(1998)a l s o p r o p o s e t h a t“i n t e l l i g e n c e i s t h e a b i l i t y o f e x p l a n a t o r y c o m p r e s s i o n”.T h e y t h e n g o o n t o propose a variation of the I.Q.test based on Kolmogorov complexity.
6.Conclusion
While Section5considers two interesting possible modifications to the Turing(imitation game)test,the bulk of the paper concerns the performance of a comparatively small computer program(approximately 960lines of Perl code)on human I.Q.tests.The program obtains very close to the purported human average score both on a variety of I.Q.tests and on average.Although some human pre-processing admittedly takes place with some forms of questions,it should be emphasised(see Section3)that both viable improvements in the parser(s)and an increase in the number of question forms attempted should enhance the p r o g r a m’s s c o r e w h i l e r e d u c i n g o r e v e n e l i m i n a t i n g human pre-processing requirements.In addition and at the very least,even the current preliminary version of the program could assist and augment the score of a human able to parse the questions.
7.Acknowledgments
The second author expresses his life-long gratitude, encouragement,best wishes and more to a life-long special friend.
8.References
A.C.E.(unknown).Official A.C.E.I.Q.Test, Association of Christian Era(A.C.E.), https://www.wendangku.net/doc/8a16202451.html,/.
V.Akman and P.Blackburn(2000).Editorial:Alan Turing and Artificial Intelligence.Journal of Logic, Language&Information(vol.9,no.4,pp391-395). CalTech Turing Tournament(2003).Turing Tournament at California Institute of Technology (CALTECH),https://www.wendangku.net/doc/8a16202451.html,.
B.J.Copeland(2000).The Turing Test,Minds and Machines(vol.10,no.4,pp519-539).
D.L.Dowe and A.R.Hajek(1997).A computational extension to the Turing Test,Technical Report #97/322,Department of Computer Science,Monash University,Clayton3168,Australia.
D.L.Dowe and A.R.Hajek(1998).A non-behavioural, computational extension to the Turing Test, Proceedings of the International Conference on Computational Intelligence&Multimedia Applications(ICCIMA'98)(pp101-106),World Scientific publishers(ISBN981-02-3352-3),Feb. 1998,Gippsland,Australia.
Enyclopedia Britannica(2000).Britannica Encyclopedia2000Deluxe Edition CD-ROM.H.J.Eysenck(1962).“K n o w y o u r o w n I.Q”,Penguin, U.K.
Helenelund HB(2001).I.Q.Tests, https://www.wendangku.net/doc/8a16202451.html,/iq-tests.html.
J.Hernandez-Orallo and N.Minaya-Collado(1998),A Formal Definition of Intelligence Based on an Intensional Variant of Algorithmic Complexity, Proceedings of International Symposium of Enginee r i n g o f I n t e l l i g e n t S y s t e m s(E I S’98)(pp146-163),Tenerife,Spain.
http://heim.ifi.uio.no/~davidra/IQ/(unknown).I.Q. Test,Norway.
I.Q.Test Labs(2003).I.Q.Test, https://www.wendangku.net/doc/8a16202451.html,/quizzes/quiz1/index.htm. KHAN-UUL Institute(2001),IQ center,http://khan-uul.mn/Eng/IQ_centre.html#4.
A.N.Kolmogorov(1965).Three approaches to the quantitative definition of information,Problems in Information Transmission(vol.1,no.1,pp1-7). Mensa(2000).Mensa Workout, https://www.wendangku.net/doc/8a16202451.html,/workout.html.
J.H.Moor(2000).Alan Turing(1912-1954),Minds and Machines(vol.10,no.4,p461).
S.L.Needham and D.L.Dowe(2001).Message Length a s a n E f f e c t i v e O c k h a m’s R a z o r i n D e c i s i o n T r e e Induction,Proc.8th International Workshop on Artificial Intelligence and Statistics(AI+STATS 2001)(pp253-260),Key West,Florida,U.S.A. Oppy,G.R.and D.L.Dowe(2003).Stanford Encyclopedia of Philosophy (https://www.wendangku.net/doc/8a16202451.html,)entry on the Turing Test (https://www.wendangku.net/doc/8a16202451.html,/entries/turing-test),Thu.10 Apr.2003.
A.P.Saygin,I.Cicekli and V.Akman(2000).Turing Test:50Years Later,Minds and Machines(vol.10, no.4,pp463-518).
R.J.Solomonoff(1964).A Formal Theory of Inductive Inference I and II,Information and Control,7(pp1-22&224-254).
Testedich.de(2002).The I.Q.Test, https://www.wendangku.net/doc/8a16202451.html,/tests/iqtest.php3.
Alan M.Turing(1950).Computing Machinery and Intelligence,Mind,Vol59,(pp433-460);also at https://www.wendangku.net/doc/8a16202451.html,/Prizef/TuringArticle.html.
C.S.Wallace and
D.M.Boulton(1968).An information measure for classification,Computer Journal(vol. 11,pp185-194).
C.S.Wallace and
D.L.Dowe(1999).Minimum Message Length and Kolmogorov Complexity, Computer Journal(vol.42,no.4,pp270-283).
C.S.Wallace and P.R.Freeman(1987).Estimation and inference by compact coding,Journal of the Royal Statistical Society(Series B)(vol.49,pp240-252).
J.G.Wolff(1995).Learning and reasoning as information compression by multiple alignment, unification and search,In A.Gammerman(ed.), Computational Learning and Probabilistic Reasoning (pp223-236),Wiley,New York.
全国翻译价格
全国翻译价格 关于全国各地区翻译价格我们根据客户的不同需求和具体情况，提供多种等级和特色的翻译服务，供客户选择：（注：以下报价均为参考价格，精确报价将根据稿件内容的难度、技术处理的复杂程度和时限要求的缓急而定。
品质控制 坚持高端定位是外语通翻译的核心要素，追求高品质翻译需要译员具备深厚的语言功底和专业背景知 识，更需要严格的质量控制体系来管理这一过程： 外语通六阶梯质量控制体系 第一阶梯：译文评估承接 分析稿件性质、用途要求、商务背景、专业术语、数量和交稿时间等，确定是否有100%的把握承接， 否则坚决放弃，以免因质量或交稿时间耽误客户和影响品牌形象。 第二阶梯：专业译员翻译 专业背景的译员只专注于一个行业领域的精准翻译，项目经理根据译文评估，从外语通全球译员库中 分析挑选多名此行业的专业译员成立项目组，统一专业术语和标准，协同翻译。 第三阶梯：翻译质量监控 项目经理监控翻译进展，每日集中疑难词汇，请签约专家释疑。每日抽查译文质量，及时解决译文质 量问题。 第四阶梯：译文校对排版 汇总所有译文，查错补漏，进一步统一术语，按原文进行排版，形成完整初稿。 第五阶梯：专家译审修改 专家译审对翻译初稿进行翻译准确性审核，确保译稿忠于原文，专业词汇纯正地道。 六阶梯：外籍母语润色第 在华外籍翻译（外译中稿件由中文功底深厚的编辑）对译稿的语法、词汇进行修正和润色，确保译稿 纯正、地道，达到母语品质。 外语通翻译严格执行《ISO译文质量体系》，《翻译质量国家标准GB/T 19682-2005》： 译文质量标准Ⅲ类通用笔译Ⅱ类专业笔译Ⅰ类高级笔译译文用途内容概要、参考资料一般文件和材料正式文件、法律文书、出版物错漏译率小于5‰小于2‰0‰ 译员经验3年以上5年以上8年以上 译员学历硕士以上硕士以上硕士以上 行业背景常识业内资深 海外背景无/短期中期长期 译文校对有有有 专家译审无有有 母语润色无无有 译文排版简单排版详细排版出版级别
翻译公司收费标准
翻译公司收费标准 1.客户需要翻译的目标语言的普遍性和稀缺性可能导致非常 不同的费用。英语比较普遍，需求大，市场专业的英语翻译人 才也很多，翻译公司无论是从降价到抢占市场，还是成本核算 来考虑，英语收费都比较合理和透明。 其他诸如法语、德语、日语、俄语排在第二梯队，翻译公司收 费标准一般都是200-280元，视稿件专业度和数量略有调整； 意大利，西班牙，越南，泰文等东南亚语种已经接近稀有语 种了，翻译报价至少300元千字起。 2.根据翻译项目类型 常见的翻译方法主要包括翻译翻译、同声传译、本地翻译、口译翻译等，翻译项目自然是不同的收费。 3.根据翻译项目时长 这一时期的持续时间主要是指项目长度：同声传译、会议翻译、商务洽谈、双语主持人、口译、护送翻译、展览翻译，当然，视频翻译、音频翻译按时间计算的时间和会议类型是一个重要因素，是翻译时间决定翻译价格的一个重要因素。 4.根据翻译项目字数
翻译项目的字数是影响收费的重要因素之一，翻译字数主要对于笔译而言，例如：文件翻译、图书翻译、资料翻译、画册翻译等等，这些文件资料的字数决定了项目的翻译价格和翻译收费标准。 5.根据翻译项目语种 主流语种：英语、日语、韩语等和小语种：阿拉伯语、希腊语、印尼语等的翻译收费标准区别。我们知道：“物以稀为贵，”所以小语种的翻译报价会比主流语种收费要高的。 6.根据翻译项目难易程度 对于翻译公司来说，翻译费在很大程度上取决于翻译的难度程度，不同的行业术语不同，难度不同; 专业翻译公司将根据翻译人员的翻译水平、专业知识、翻译经验等方式来评价自己的翻译团队，高层次的翻译人员当然都是高收费; 如通用翻译、精细翻译、出版层次等不同类型的翻译报价不同，稿件的行业领域、材料难度、选择翻译类型等都是决定翻译公司收费标准的因素。
英文合同翻译价格 英文合同翻译需要多少钱
英文合同翻译价格英文合同翻译需要多少钱 在企业的经营过程中，有时候可能会涉及到翻译这个问题，但是一般的小企业并没有专门的人去做这件事情，大部分都是外包。那么对于企业来讲，翻译一份英文合同需要多少钱呢？作为浙江省最大的翻译公司，以琳翻译就在这里为大家解读一下。 一般来讲，翻译这项服务都是以字数来计价的，市场上的一般的价格是50-80元/千字，这是一个基本的价格。但是不同的公司的专业性质不一样的话，所给出的价格也是不一样的。对于公司的衡量标准来讲，影响价格的因素主要有：公司的资历、翻译人员的专业性、翻译文件的种类、难度等。所以，如果你需要去找翻译公司去服务，那么就需要考虑这些方面的东西。而对于合同这种文件，对于公司来讲是十分重要的，所以也需要去找专业的公司去进行翻译，如果是找一个资质不够的公司或者团队，那么就可能产生一些意想不到的问题，从而影响到公司的最终利益。 下面，我们来看看以琳翻译给出的翻译的价格。 从上面的价格可以看出，以琳翻译给出的价格是高于一般市场上的价格的，最低级别的翻译是160元/千字，然后分为A、B、C三级。C级译稿为普通中籍译员+中籍译员审核，满足客户对译文的普通要求。这是对于一般的合同而言的，但是如果是部分专业性质较强或者要求比较高的译文的话，那么可以选择更高级别的翻译，当然价格还是相对比较高的。 那么以琳翻译的资质是怎么样呢？我们再来看一下。 杭州以琳翻译有限公司是浙江省最大的实体翻译公司、中国翻译协会单位会员、美国翻译协会会员、全国翻译专业硕士研究生教育实习基地、西博会指定合作伙伴、以琳杭州翻译公司翻译团队成员均具有五年以上专业翻译、项目管理经验，绝大部分成员具有十年以上行业翻译经验。翻译服务涵盖英语、法语、韩语、日语、德语、俄语、西班牙语、葡萄牙语、
专业英文翻译中文收费标准
精诚英语翻译报价50-80元千字（市场价格100左右 精诚英语翻译工作室是由众多英语方面精英组成的翻译团队,一直致力于为广大中小企业和个人提供专业低价中英文翻译服务。价格是我们永远的优势！！！！最低价格支付宝担保交易，让你省钱又放心接受试译！！自信源于专业可以百度搜索精诚英语翻译找到我们 选择我们的理由：可以百度搜索精诚英语翻译找到我们 1.保证价格最低，团队网络化运作，无需经营成本，可以通过低价让利于客户。（有些客户看到这么低的价格还不敢相信，但是对于我们来说是完全可以接受的。） 2.保证准时、保密、准确 3.接受淘宝交易，让您没有任何担忧。 4.长期翻译经验，保证质量让您满意。 龚如心遗产案虽然告一段落，「遗产」二字仍然成为近日香港的焦点。新春期间，民政事务局局长曾德成表示，政府将展开全港非物质文化遗产首期普查，希望市民为遗产清单提出建议。 「遗产」是「资产」？ 近五、六年间，香港对保护本地小区和文化传统的意识高涨，现在政府带头要列一个「非物质文化遗产」清单，理应是很受欢迎之举。不过普查尚未展开，就引来学者争议，其中单是「非物质文化遗产」这个译名，就引起不少误会。 「非物质文化遗产」的原文是intangible cultural heritage（英文）或patrimoine culturel immateriel （法文），是联合国在1997年以尊重多元文化为大原则而提出的概念，并由联合国教科文组织制定「保护非物质文化遗产公约」，2006年生效。 「非物质文化遗产」是中国大陆的翻译，香港有学者不约而同就「非物质」和「遗产」二字提出质疑。香港城市大学中国文化中心主任郑培凯早在2005年就大声疾呼译名不妥。他认为原文heritage/patrimoine的意义是「传承」而非资产，不容易引发出财产的概念。而现在约定俗成译作「遗产」，容易令人觉得祖宗留下的东西，是可以变卖和投资的生财工具，与联合国提出的文化传承精神背道而驰。郑教授认识，正确的译名是「非物质文化承继」或「非实物文化传承」。另一位民俗学研究者陈云进一步指出，intangible「乃触摸不到的事，无形无相之事」，应用「精神价值」代之，「非物质」有消灭了精神之嫌，所以中国人应堂堂正正将之翻译为「无形文化传承」。 姗姗来迟的「遗产」 不论是「非物质」还是「无形」，「遗产」还是「承传」，即使公约成员国中国曲译甚至错译，香港特区政府还是只能照单全收。而且，随之而来的不止是字面的斟酌，而是「遗产」的搜寻和管理问题。 中国自2005年起，就开始非物质文化遗产（由于这个名称已约定俗成，故下文仍沿用之，并简称为「非遗」）普查，并陆续列出清单。香港也在翌年提出编制非遗清单，但却延至去年才聘专家普查，估计最快要2012年才完成。 非遗普查尚未展开，在国家文化部的再三邀请（或是说催促？）下，去年九月，香港终于申请将长洲太平清醮、大澳端午游涌、大坑舞火龙和香港潮人盂兰胜会列为第三批国家级非遗，预计今年六月有结果。 其实香港已错过了2006和2008年首批和第二批的申报机会，所以，至目前为止，在中国的文化版图上，香港是唯一没有任何有形和无形「遗产」的主要城市／特区，就连比邻的澳门也凭神像雕刻工艺获得2008年国家级非遗之「奖项」。 有人说非遗不过是人有我有，纯粹锦上添花；也有人说，中国在维护主权和领土完整的概念下，又怎能在文化层面少了香港一席？香港能够「出产」一个非遗，中国在全球的文化图谱中就多一个筹码。 姑勿论背后原因为何，由于「保护非物质文化遗产公约」也适用于香港，香港特区政府就有责任找出和保护濒危失传、与社会关系密切及具香港独特性的文化传统。现在起步虽迟，但为时未晚。 「遗产」的管理问题 不过，既然政府要展开普查，另一个问题来了。民间传统应该是属于民间的，并由民间自行发展，还是属于官方，由政府承担保护与管理？ 据政府委聘负责首期普查的香港科技大学华南研究中心主任廖迪生表示，政府至今仍未有任何政策配合或承诺给予全面的保护，所以，即使清单出炉，有些遗产仍有可能难逃「破产」的命运。他强调制作非遗清单只是第一步，更重要的是如何保护这些项目。 再问民政事务局，曾德成局长除了曾向立法会议员表示，制定清单是向国家文化部申请列为国家级非遗的第一步，进而再向联合国教科文组织提出申报为世界非物质文化遗产，他所提出的，就是以遗产作招徕吸引外地游客，「以提升香港作为旅游目的地的吸引力」。 这才是令人担心的地方。「非遗」这个金漆招牌在中国许多地方都有点石成金之效。戴上这个冠冕，民俗文化很容易沦为生财工具、游客的消费品，连婚嫁仪式也可用来表演，完全违背了保护非遗的原意。曾德成之言，是否意味着香港也要跟着祖国一起走上同一条路？
英文翻译价格
英文翻译价格 根据以英文作为母语的人数计算，英文是最多国家使用的官方语言，英语也是世界上最广泛的第二语言，也是欧盟，最多国际组织和英联邦国家的官方语言之一。但仅拥有世界第二位的母语使用者，少于标准汉语。上两个世纪英国和美国在文化、经济、军事、政治和科学上的领先地位使得英语成为一种国际语言。如今，许多国际场合都使用英语做为沟通媒介。英语也是与电脑联系最密切的语言，大多数编程语言都与英语有联系，而且随着网络的使用，使英文的使用更普及。英语是联合国的工作语言之一。 为了方便大家了解英文翻译价格，小编在目前汇集最多翻译团队的高校译云上面获得了不同翻译精英团队所展示的价格。 暨南大学翻译中心：中英---普稿---150---千字英中---普稿---250---千字 武汉理工大学-外国语学院MTI翻译中心 ：中英互译中英130-150 英中100-130 华中科技大学-翻译研究中心 ：中英互译中英120-150 英中100-120 湖南科技大学MTI中心：中英---普稿---150---千字 上海师大外国语学院翻译中心： 中英---普稿---200元---千字英语普通文本译成汉语---120元---千字西南大学翻译中心：中英---普稿----300---千字英中---普稿---200---千字 上海理工大学MTI翻译中心：中英---普稿---100---千字 南京财经大学外国语学院翻译研究中心：中英---普稿---100---千字 一般英文翻译价格是是在100—300元每千字，根据译员质量、翻译内容、需要的时间等都会有一定的波动，所以以上价格供大家参考，具体的可以准备好稿件了去问，这样会更加准确一些。
翻译服务收费标准
翻译服务收费标准 一、笔译人民币元/千字中文( 加急加收30% —70% ，专业加收50% ) 语种中译外外译中外译外 英语170 140 面议 日语170 140 韩语190 160 德语220 180 俄语220 180 法语220 180 意大利语280 250 西班牙语280 250 葡萄牙语290 260 阿拉伯语350 320 越南语430 400 荷兰语510 460 波兰语380-480 360-40 塞尔维亚语370-470 420-530 泰国语260-380 280-520 老挝语320-420 370-480 印度语320-420 370-480 希腊语370-470 420-530 哈萨克语280-380 300-410 瑞典语300-400 340-450 丹麦语320-420 370-470 印度尼西亚语330-450 350-460 蒙古语300-400 350-460 1、字数计算：以中文版稿件在Windows word文档显示的字符数（不计空格）为基准。也即包含了标点符号，因其为理解语义的必需。 2、图表计算：图表按每个A4页面，按页酌情计收排版费用。 3、外文互译：按照中文换算，即每个拉丁单词乘以二等于相应的中文字数。 4、日翻译量：正常翻译量3000-5000字/日/人，超过正常翻译量按专业难易受20%加急费． 5、付款方式：按预算总价的20%收取定金，按译后准确字数计总价并交稿付款。 6、注意事项：出差在原价格上增加20%，客户负责翻译的交通、食宿和安全费用。 二、口译价格： (1) 交传报价（元/人/天，加小时按100-150元/小时加收费用）类型英语德、日、法、俄、韩小语种 一般活动700 800 1500 商务活动500-1200 500-1500 800-3000 中小型会议1200-3000 1500-3000 2500-3000 大型会议1200-4000 2500-6000 4000-9000 (2) 同传报价（元/人/天） 类别中-英互译日、韩、德、俄、法、韩-中互译小语种-中互译 商务会议5000-8000 6000-10000 8000-10000 中小型会议5500-8000 7000-12000 8000-12000 大型国际会议6000-9000 8000-12000 12000-16000
浙江杭州小语种翻译公司报价
加快实施“走出去”战略是适应全球化新形势和我国发展新变化，培育参与和引领国际合作和竞争新优势的重要举措。据预测，今后五年，我国对外投资规模总量超过5000亿美元，并且年增速在10%以上。浙江是外贸大省，境外经贸合作区建设走在全国前列，境外投资一直保持高速增长，《浙江省利用外资和境外投资“十三五”规划》也作出了明确部署。跨境法商论坛拟计划在杭州举办，将促进浙企与马来西亚、新加坡相关机构的直接对话，帮助浙江企业进一步拓展“一带一路”市场，寻找国际合作渠道，促进有效投资。 各式各样大手笔的贸易投资合作，也预示着新的发展机会。对从事语言服务工作的人士无疑是好消息。据中青在线记者报道，在之前“一带一路”国际合作高峰论坛会议中，现场的同传耳机里，18种工作语言分别是：汉语、英语、法语、俄语、西班牙语、柬埔寨语、捷克语、匈牙利语、印度尼西亚语、哈萨克语、老挝语、蒙古语、波兰
语、塞尔维亚语、土耳其语、越南语、日语和韩语。每个座位上都放有一张列举了这些语言的名单，除了标注与同传耳机频道对应的16种语言以外，同传耳机中的第17、第18频道分别是日语和韩语。 这无疑是翻译从业者的一剂强心针，不少小语种从业者和学习者表示，政府、学校、企业等对于小语种的关注度，现在越来越高了。更日常的商务沟通、贸易合作等等势必越来越密集和频繁，毫无疑问，国内的翻译人才供不应求，高质量的翻译人群，已经成为最抢手的人才。 随着翻译需求的不断加大，国内的翻译产业开始蓬勃发展。从2012年到2015年，我国翻译专业硕士(MTI)学位授予点也由2007年的15所大学增加至2016年的215所大学，国内翻译公司数量已达到72，485家。咱们先了解一下市场小语种的翻译报价参考： *以上报价为人民币基准价，不含税金，仅供参考，具体报价根据需译资料数量、领域、难易等具体确定！
中译英翻译报价单中译英翻译价格
精诚翻译公司五周年五折优惠中5年经验！先翻译后付费学生客户送50元优惠券，可以搜索精诚翻译公司者50元翻译找到我们 The book of“Chinese Language" aims at improving students' Chinese accomplishment and emphasizes on humanity, instead of literature or humanism. The book mainly consists of literary works, and also involves philosophy, history, art, science and technology, so as to interest students of different majors and help broaden their vision. 1Ancient literature 2 Modern literature 3 English-Chinese translation In the textbook of Public English, one third of contents have been updated, therefore, the contents become more interesting, comprehensive and practical. The attached vocabularies enable students to master the law of learning vocabularies, and the actual use of English sentences help improve their ability of using English. "Fundamentals of Management" is a subject which discusses basic theories, principles, management functions and general method of management activities, it also combines theories with practice closely. Through studying this course, students should be familiar with basic concept of management, relevant knowledge system and thoughts by various management school; understanding basic management theories and relevant principles in depth, as well as planning, decision-making, organization, leadership, motivation, control, communication and other basic management functions and methods. "The History of Chinese Culture" The book' author is an contemporary Chinese culture expert who has published many cultural works. This book focuses on evolution of academic thought and culture, at the same time, it also consider behavior, system,
山东青岛小语种翻译公司报价
“一带一路”倡议提出五年来，青岛市紧紧围绕“新亚欧大陆桥经济走廊主要节点城市”和“海上合作战略支点”功能定位，积极融入大局，深入贯彻国家政策，在经贸合作、外事外宣、金融保障、海洋科技合作等领域着力打造对外开放新高地。对“一带一路”沿线国家累计投资69.8亿美元，承包工程完成营业额130.6亿美元。全市银行机构累计为“一带一路”沿线国家和地区有投资项目的84家大型企业授信860.3亿元，贷款余额445.1亿元，支持3000余家企业与“一带一路”沿线61个国家开展贸易投资与建设合作。市级领导出访沿线国家达50余批次；接待国外重要来宾、团组来访530余批次，友城总数达到70个，遍及全球37个国家。 各式各样大手笔的贸易投资合作，也预示着新的发展机会。对从事语言服务工作的人士无疑是好消息。据中青在线记者报道，在之前“一带一路”国际合作高峰论坛会议中，现场的同传耳机里，18种工作语言分别是：汉语、
英语、法语、俄语、西班牙语、柬埔寨语、捷克语、匈牙利语、印度尼西亚语、哈萨克语、老挝语、蒙古语、波兰语、塞尔维亚语、土耳其语、越南语、日语和韩语。每个座位上都放有一张列举了这些语言的名单，除了标注与同传耳机频道对应的16种语言以外，同传耳机中的第17、第18频道分别是日语和韩语。 这无疑是翻译从业者的一剂强心针，不少小语种从业者和学习者表示，政府、学校、企业等对于小语种的关注度，现在越来越高了。更日常的商务沟通、贸易合作等等势必越来越密集和频繁，毫无疑问，国内的翻译人才供不应求，高质量的翻译人群，已经成为最抢手的人才。 随着翻译需求的不断加大，国内的翻译产业开始蓬勃发展。从2012年到2015年，我国翻译专业硕士(MTI)学位授予点也由2007年的15所大学增加至2016年的215所大学，国内翻译公司数量已达到72，485家。咱们先了解一下市场小语种的翻译报价参考： *以上报价为人民币基准价，不含税金，仅供参考，具体报价根据需译资料数量、领域、难易等具体确定！ 尽管翻译人才输出与翻译机构成立都发展迅速，但真正能满足企业发展需求的高质量翻译机构却并不好找。一方面是因为从事翻译行业的，不少都是翻译新手，没有行业经验和过硬实力，另一方面，也是因为翻译机构资质参差不齐，没有统一的规范，翻译价格和翻译质量也受到影响。怎么选择靠谱的翻译机构就成为企业的困惑点。
小语种翻译现状
小语种翻译现状 目前很多大城市会出现一种奇怪的现象，当小众语种国家人代表在我国开会发言演讲时，台下的大家只能眼巴巴的望着他却不知所云，居然没人能翻译，这种现象引起了各相关部门的广泛关注。翻译达人表示说，我国很多二三线城市同声传译人才缺乏，翻译机构规模档次不够。面对二三线城市建设步伐的加快以及国际性会议的增多，高端翻译市场亟待提升档次。 国际会议尴尬 西安高新区工作的张先生讲述到，某次会议当天，来自亚洲的一些国家的部门代表轮流发言，参与会议的也都是来自周边国家的企业以及政府代表。每人配备同声传译设备，并且有同传人员现场翻译，一切看上去井井有条。可当轮到格鲁吉亚一位代表发言时，同传人员却“撤退”了。说着格鲁吉亚语的这位代表只好自顾自说了半天，下面参会人员看上去一片茫然。 人才严重匮乏 目前大多二三线城市的小语种翻译人员都是从北京、上海、广州这些大城市请过来的，并且价格不菲！同声传译人才对翻译水平的要求最高，不仅要具备良好的口语功底，还要对当地的文化有所了解，一般都需要有国外生活几年的经历才能胜任。大多二三线城中外语翻译人才还仅仅停留在英语方面，小语种翻译人才严重匮乏。而能提供大多小语种翻译的，一般只有像翻译达人这样的翻译平台。 小语种翻译专业开设少 导致小语种翻译人才稀缺的因素有很多，其中有一个因素就是开设小语种专业的高校少。比如小语种罗马尼亚语，全国仅有1所大学开设了这个专业，即北京外国语大学。还有上文提到的格鲁吉亚语，目前在我国还没有一所大学开设了
这个专业。曾经北大给俄语专业的学生开设了这个课程，请的格鲁吉亚的外教讲授的，但是因为外教的变动，这课程现在也没有了。因为开设小语种的学校少，学习该语种的人自然也少。随之就是翻译的价格水涨船高。可以说，翻译人才越难找，收费也就越贵。 翻译市场亟待壮大 人才的匮乏会明显的限制整个翻译市场的壮大与发展，国际性会议的大幅度增多，更导致了小语种人才严重稀缺。而懂得小语种的翻译人才，目前正在成为各大公司热捧的对象。甚至因为人才稀缺，企业在招聘标准上也低出不少，如某些公司招聘负责商务谈判及日常翻译工作的俄语翻译，已经打出“应届毕业生也可”的条件。所以，目前我国的小语种翻译市场亟待扩大，也需要更多的小语种翻译人才。