a r X i v :a s t r o -p h /0303619v 1 28 M a r 2003
Astronomy &Astrophysics manuscript no.(will be inserted by hand later)
Simultaneous single-pulse observations of radio pulsars
III.The behaviour of circular polarization
A.Karastergiou 1,2,S.Johnston 2,and M.Kramer 3
1Max-Planck Institut f¨u r Radioastronomie,Auf dem H¨u gel 69,53121Bonn,Germany 2School of Physics,University of Sydney,NSW 2006,Australia
3
Jodrell Bank Observatory,University of Manchester,Maccles?eld,Chesire SK119DL,UK
the date of receipt and acceptance should be inserted later
Abstract.We investigate circular polarization in pulsar radio emission through simultaneous observations of PSR B1133+16at two frequencies.In particular,we investigate the association of the handedness of circular polarization with the orthogonal polarization mode phenomenon at two di?erent frequencies.We ?nd the association to be signi?cant across the pulse for PSR B1133+16,making a strong case for orthogonal polarization modes determining the observed circular polarization.The association however is not perfect and decreases with frequency.Based on these results and assuming emission occurs in superposed orthogonal polarization modes,we present a technique of mode decomposition based on single pulses.Average pro?les of the polarization of each mode can then be computed by adding the individual mode-separated single pulses.We show that decomposing single pulses produces di?erent average pro?les for the orthogonal polarization modes from decomposing average pro?les.Finally,we show sample single pulses and discuss the implications of the frequency dependence of the correlation of the circular polarization with the orthogonal polarization mode phenomenon.Key words.pulsars:PSR B1133+16–polarization
1.Introduction
Perhaps one of the most elusive qualities of the highly po-larized pulsar radio emission is its circular polarization.Although it is generally lower than its linearly polarized counterpart,it is amongst the highest observed in astro-physical objects.Circular polarization in pulsars has been attributed in the past to propagation e?ects on the ra-diation as it travels through the pulsar magnetosphere,geometrical e?ects or the emission mechanism itself.Cordes et al.(1978)were the ?rst to point out an as-sociation between the position angle (PA)of the linear polarization and the handedness of the circular polariza-tion.More speci?cally,90o jumps in the PA were observed to occur simultaneously with sense reversals in the circular polarization (V )in PSR B2020+28.The authors argued that this could be adequately explained by disjoint or su-perposed orthogonal polarization modes (OPMs).In the disjoint case,each of the OPMs is emitted independently and only individual modes are observed at a time.On the other hand,the superposed case supports the idea that the observed radiation at any instant is a mix of both modes,which are e?ectively emitted at the same time,despite each mode having its own properties.A number of authors argued for the identi?cation of OPMs as the natural polar-
ization modes in the pulsar magnetosphere (e.g.Barnard &Arons 1986,McKinnon 1997,von Hoensbroech et al.1998,Petrova 2001).Elliptically polarized natural modes could then provide the origin for the circular polarization as a propagation e?ect.
Stinebring et al.(1984a,b)studied the polarization behaviour of single pulses from a number of pulsars us-ing grey-scale density plots to investigate the ?uctuations of the PA,the linear polarization,L ,and V .They no-ticed that PA jumps often occur at the same pulse longi-tude at 430and 1404MHz and that the orthogonal po-larization mode (OPM)phenomenon was becoming more prevalent at higher frequencies.They also noticed that at the phase bins where OPM jumps occur,the degree of linear polarization was signi?cantly reduced,which led to the idea that,at any given instant,the radiation ob-served is the incoherent sum of OPMs.This has the ad-vantage of naturally reducing the linear polarization and attributing its ?uctuations to ?uctuations in the polariza-tion of the OPMs,which have a constant degree of lin-ear polarization.In contrast,if OPMs were emitted in-dependently,the emission mechanism would have to in-trinsically produce a largely ?uctuating degree of linear polarization.McKinnon &Stinebring (1998)realized that the assumption of incoherent addition of OPMs can be
2 A.Karastergiou et al.:Simultaneous single-pulse observations of radio pulsars
used to decompose an integrated pulse pro?le into the pro?les of each OPM.By accounting for the e?ects of in-strumental noise,this method can be used to reproduce the histograms of the distributions of the PA and L from Stinebring et al.(1984b).Mode decomposed average pro-?les of PSRs B0525+21and B2020+28were shown in McKinnon&Stinebring(2000)using their technique on the average pro?les of these pulsars.
In this paper,the association of circular polarization with the orthogonal polarization modes is examined in PSR B1133+16at1.41GHz and4.85GHz,having stud-ied the correlation of OPMs between these two frequen-cies in Karastergiou et al.(2002,hereafter Paper II).We make use of contingency tables to quantify this associa-tion at each frequency,which we compare to the?ndings of Cordes et al.(1978).We develop a method to decom-pose single pulses into the individual OPMs.The method is simple and only assumes that the observed Stokes I,Q, U,V are the incoherent sum of the100%polarized OPMs. We show the di?erent mode-decomposed pro?les of PSR B1133+16by following our method and that of McKinnon &Stinebring(2000)and discuss the di?erences.Finally, we present example single-pulses from our simultaneous data set to highlight the observed e?ects and we discuss possible interpretations.
2.The data
We use the same dataset on PSR B1133+16as in Paper II. It consists of4778pulses at4.85and1.41GHz observed with the E?elsberg and Jodrell Bank radio telescopes re-spectively.Full details of the observing,calibration and data reduction can be found in Paper II.The mean S/N of V detected in the single pulses is<<1.However,in a signi?cant number of cases,the signal in V exceeds the rms of the noise by many times at both frequencies.At 4.85GHz,where the?ux density of this pulsar is much less than1.41GHz,the sensitivity of the E?elsberg tele-scope largely compensates.This is evident from the S/N of V shown in Fig.1and favours our results to be in-trinsic to the pulsar emission rather than products of the instrumental noise.
2.1.Distributions of V and V/I
To investigate the single-pulse behaviour of circular polar-ization,the value of V is plotted against the correspond-ing pulse phase for every single pulse.Such
density plots reveal the pulse to pulse?uctuations of V.Fig.1shows two such plots for the single-pulse data of PSR B1133+16 observed simultaneously at4.85GHz and1.41GHz.The density plots reveal that the distributions of V at each pulse phase bin range over a few tens of rms units at both frequencies.
In the leading component,it has been shown that the OPMs are almost equally strong and dominate equally of-ten at both frequencies(Paper II).The distribution of V in this component shows an almost equally dense popula-Fig.1.The distribution of V and V/I in PSR B1133+16, at4.85GHz[top]and1.41GHz[bottom].Each plot has three horizontal panels.The top panel shows the total power(I)integrated pulse pro?le,the middle panel shows the distribution of?ux densities of V in units of the rms and the bottom panel shows the distribution of the frac-tion V/I.In the bottom panel,a threshold has been ap-plied to exclude the noise:only cases where the?ux den-sity of V is more than5×rms are shown.
tion of positive and negative values.On the other hand, the trailing component is almost always dominated by one OPM and the respective V distribution favours negative values.Under incoherent addition of OPMs,the observed V is the di?erence between the V of each OPM and there-fore the above observation is in accordance with a particu-lar handedness of V being associated with a certain OPM.
The seemingly bimodal distributions of V/I in the leading component in Fig.1are caused by the threshold applied to exclude the noise.Fig.2shows the speci?c dis-tributions of V and V/I together with the PA histogram of bin50.The bottom row corresponds to the1.41GHz data and the top row to the4.85GHz data.The thresh-
A.Karastergiou et al.:Simultaneous single-pulse observations of radio pulsars
3 Fig.2.The distribution of PA,V and V/I in bin50,near
the peak of the leading component,at4.85GHz[top]and
1.41GHz
[bottom].
Fig.3.The distribution of PA,V and V/I in bin65,near
the peak of the trailing component,at4.85GHz[top]and
1.41GHz[bottom].
old on V is not applied here,in order to demonstrate how
much of the distribution lies within the instrumental noise,
the rms of which is represented by a small horizontal bar
inside the V distribution.Only the pulses in which the
signal-to-noise(S/N)of the total power exceeds2and
the fractional linear polarization exceeds20%are consid-
ered,in order to exclude ambiguous PA values.Bin50
exhibits a clear bimodal PA distribution at both frequen-
cies and at the same time a mean V value very close to0.
On the other hand,Fig.3shows bin65from the trailing
pulse component,that presents a slightly di?erent picture.
The PA distributions at both frequencies are dominated
by one OPM(at4.85GHz this is not entirely clear due
to the small population of points that survive the thresh-
old)and the V distributions have a signi?cantly non-zero
negative mean.This picture is consistent with incoher-
ently superposed OPM,each mode favouring a particular
handedness of V:in bin50where both modes dominate
equally often,V is distributed around0,which is not the
case for bin65,where one mode essentially dominates and
the handedness of V associated with this mode gives mean
V its negative value.
Figs.2and3show a di?erence between the distribu-
tions in V and V/I.All four V/I histograms can be ad-
equately approximated by Gaussian distributions.In bin
50,the best?t has a mean of?2.5%and aσof19%at
4.85GHz and a mean of0.6%and aσof18%at1.41GHz.
In bin65on the other hand,the best?t has a mean of
?12%and aσof27%at4.85GHz and a mean of?13%
and aσof24%at1.41GHz.These broad distributions are
consistent with those seen in other pulsars(Stinebring et
al.1984a,b)and are largely dominated by instrumental
noise(McKinnon2002).
In summary,the density plots of the circular polariza-
tion in Fig.1and the individual distributions in Fig.2
and Fig.3,together with the OPM analysis presented in
Paper II,provide evidence that each OPM is associated
with a certain handedness of V.This association,how-
ever,is obviously not perfect,and we discuss this in the
following.
2.2.The V-OPM correlation
In Cordes et al.(1978),observations of PSR B2020+28at
404MHz were used to investigate a possible connection
between the handedness of V and the PA value.To this
goal,they plotted V versus PA in the single pulses they
observed.In their paper,they give an example of a par-
ticular pulse longitude where the distribution of PAs is
bimodal,in other words a phase bin where the dominant
OPM alternates from pulse to pulse.The plot of V versus
PA for this bin demonstrates a clear correlation between
the handedness of V and the OPM.
In Paper II,a detailed study of the frequency depen-
dence of the OPMs in PSR B1133+16demonstrated that,
at higher frequencies,the OPMs are seen to dominate al-
most equally often and the individual?uxes of the OPMs
become more equal to each other.Given this frequency
evolution,we investigate whether the V-OPM association
is also frequency dependent.Instead of V-PA plots,we
follow a di?erent route,namely to display the tips of the
polarization vectors of all the single pulses from speci?c
bins on the Poincar′e sphere.The coordinates of a given
point on the surface of the sphere are given by two angles,
the latitude(tan?1V/L)and longitude(2P A).The lati-
tude is a measure of the ellipticity of the polarization state:
values of0o correspond to linearly polarized states whereas
values of±90o correspond to purely circular states of op-
posite handedness.Fig.4shows the bimodal distributions
of PA in phase bin50,the modes being～90o apart in PA.
It also shows that the majority of points in each mode pre-
fer a certain handedness in V and therefore lie on either
4 A.Karastergiou et al.:Simultaneous single-pulse observations of radio pulsars
Fig.4.Hammer-Aito?projection of the Poincar′e sphere,where the polarization states of a phase bin50from the single pulses are depicted as the tip of the polarization vector on the Poincar′e sphere.The left panel corresponds to 1.41GHz and the right panel to4.84GHz.Bimodal PA distributions are evident in both plots.
4.85GHz
V+
6594361475
1.41GHz
V+
4423861563
Table 1.Contingency tables for bin50of the PSR
B1133+16data,showing the frequency of occurrence of
all possible combinations in the single pulses.The top ta-
ble corresponds to the4.85GHz and the bottom table to
the1.41GHz data.
side of the equator(V=0).Comparing Fig.4to Fig.6
Cordes et al.(1978),reveals that the association of V with
PA is not as clear in PSR B1133+16at these frequencies
as it was in PSR B2020+28at404MHz.A quantitative
test is therefore necessary.
In the scheme of OPM superposition,it is very impor-
tant to stress again that the observed PA is that of the
dominant mode,regardless of the?ux density di?erence of
the two modes.Therefore,by testing the PA vs.the sign
of V association,one is e?ectively testing the association
of the handedness of V with OPM.For each phase bin of
the pulse and for the total number of pulses,we quantize
the observed quantities according to the following3×2
possibilities:dominant mode1,dominant mode2or not
de?ned PA,and positive-V or negative-V.By composing
such contingency tables for each phase bin,it is possible
to quantify the association between the handedness of V
and the dominant OPM.An example for bin50at each
observed frequency is shown in Table1.The numbers have
been obtained by applying a two step process.Where the
S/N of L in the given bin is less than2.5,we are un-
able to compute the PA and therefore classify these as
“Unde?ned”in Table1.Then,we compare the observed
PA in each single pulse to the PA of the integrated pro?le
in this bin,to determine which mode is dominating at this
particular instant.From the numbers presented in Table
1,it is possible to compare the observed cases to the null
hypothesis,by using the formula
(Obs?Null)2
D=±
A.Karastergiou et al.:Simultaneous single-pulse observations of radio pulsars5 2hardly ever dominates in the trailing pulse component
and therefore the corresponding D values are all close to
zero.
3.OPM separation
Our statistical analysis of the dual frequency PSR B1133+16data has shown us that there is a frequency de-pendent correlation between the handedness of V and the dominant OPM.The contingency tables,however,show that there is a signi?cant population of cases where the handedness of V is opposite from the expected.In other words,the V-OPM association is not perfect.This is a very important result with serious implications on the technique of decomposing pulse pro?les into the individ-ual OPMs.The signi?cance lies in the fact that each OPM that dominates in a given single pulse may be observed instantaneously with either handedness of V.However, averaging the single pulses and decomposing the average pro?le is based on an assumption that is not met,namely that each OPM is clearly associated with a particular sign of V in all the single pulses.To demonstrate this,we detail the mathematics of mode-decomposition in single pulses and show the di?erence between the obtained pro?les by decomposing and then averaging,and averaging and then decomposing the PSR B1133+16data.
The orthogonality of the modes means that their po-larization states at a given instant are represented by an-tiparallel vectors on the Poincar′e sphere.By de?nition therefore the PAs can only take values which are exactly 90o apart.Incoherent superposition allows the addition of the individual Stokes parameters.The mathematical pro-cess is as follows:
The observed Stokes parameters can be written in terms of the OPMs
I=I1+I2
Q=Q1+Q2
U=U1+U2
V=V1+V2
L=L1+L2,
(2) where L i=
L2+V2,and considering that the antiparallel vectors must have opposite senses of circular polarization,the or-thogonal polarization states imply that,
L i
P
,(3) |V i|
P
.(4) Also,the OPMs are100%polarized,which means that d2L+d2V=1or d V=
1?d2L(I1?I2),
(6) which are su?cient to solve for I1and I2,so that
I1=0.5·(I+
L
d L
).
(7) and therefore,Eqs.(3)and(4)lead to
L1=d L·I1
L2=d L·I2(8) V1=d V·I1
V2=?d V·I2.(9)
To decompose Q and U,the PA is used.The PA of the strongest mode should be equal to the observed PA,and the PA of the weak mode should be orthogonal to this. Therefore,
U
Q1
=?U2Q
2
,(10) from which Q i and U i can be determined as
Q i=
L i
L·U.
(11)
The above method shows that there is a unique way to decompose the received Stokes parameters into the Stokes parameters of the individual orthogonal modes.In prac-tice,special care must be taken in the cases of very small L and V to avoid signi?cant errors.The equations also express the intrinsic assumption of this method that the modes are orthogonal,so the circular polarization sense is clearly opposite in each mode in a given single pulse. However,the sense of V that is associated with a partic-ular OPM may change from pulse to pulse.
In Stinebring&McKinnon(2000),the average pro-?les of PSRs B0525+21and B2020+28are decomposed into the OPMs,using the assumption of superposed OPM emission.We apply our method of single-pulse decomposi-tion and their method of average pro?le decomposition to PSR B1133+16to identify and interpret the di?erences. Figs.
6and7show the results of the two di?erent ap-proaches to OPM-decomposition at4.85GHz and1.41
6 A.Karastergiou et al.:Simultaneous single-pulse observations of radio
pulsars
Fig.6.The top row of plots shows the average pulse pro?le of PSR1133+16at4.85GHz,decomposed into the pro?les for Mode1(left)and Mode2(right).The bottom row is the result of the decomposition into the individual OPMs of each single pulse,which are then averaged.In each plot,the main panel shows the total power I(solid),the linear polarization L(dashed)and the circular polarization V(dashed-dotted).The secondary panel on top shows the PA.
GHz respectively.First of all,the averaged pro?les created
using the two methods are obviously di?erent.The average
circular polarization in the case of the single-pulse decom-position has a constant sense across the pro?le,which is
not true for the average pro?le decomposition.Also,it can
be seen that the average?ux density of V in each of the modes is considerably smaller in the case of single-pulse
decomposition.Another aspect of the observed di?erences between the pro?les is the linear polarization,which is
～100%in the decomposed average pro?les and signif-icantly less in the decomposed single-pulse pro?les.We claim that the reduced degree of polarization is due to the
spread in the PA distributions.We have found that a45o
spread in the PA distribution of a particular phase bin,re-sults in a～20%decrease in the mean linear polarization after averaging a few hundreds of single pulses.For this reason,it is impossible to obtain mode-separated average pro?les constructed from decomposed single pulses that exhibit100%polarization unless the PA distributions are delta-functions.4.Simultaneous pulses
The signi?cant correlation between V and OPM is in agreement with the initial result for PSR B2020+28 (Cordes et al.1978)and constitutes evidence that circu-lar polarization is closely linked to the OPM phenomenon. However,we have demonstrated that decomposing single pulses into the OPMs and averaging produces di?erent mode-separated pro?les from averaging single pulses?rst and then decomposing the resulting pro?le.This happens because the correlation of V with OPM,although signif-icant,is not perfect,and becomes worse with frequency. Obviously,the entire mechanism that determines the po-larization we observe is not well understood.
In Karastergiou et al.(2000,Paper I)and Paper II it was shown that single pulses of PSRs B1133+16and B0329+54observed simultaneously at two,widely spaced frequencies may often disagree in terms of the dominant OPM.Despite these frequent cases of disagreement,the cases of agreement were statistically more signi?cant and therefore revealed a high degree of correlation in OPMs be-tween the observed frequencies.The correlation of OPMs
A.Karastergiou et al.:Simultaneous single-pulse observations of radio pulsars
7 Fig.7.Same as Fig.6for1.41GHz.
between frequencies is facilitated by the fact that the PAs of the OPMs are,by de?nition,orthogonal.Therefore,it is possible to distinguish the prevailing OPM at both fre-quencies and count the cases of agreement and disagree-ment.Unfortunately,the observed V has wide unimodal distributions around a usually very small mean.An inves-tigation of the correlation of the handedness of V between the frequencies,is therefore dominated by the mean and the rms of the V distribution.As an example,the bins in the trailing pulse component of PSR B1133+16as seen in Fig.1,show broad V distributions of a clearly negative mean.A test of the association in handedness of V be-tween the frequencies in this component obviously shows a high correlation,forced by the underlying distributions themselves.As for the leading pulse component,we have demonstrated that each mode is clearly associated to a particular handedness in V at a given frequency.The high degree of the correlation of the OPMs between the fre-quencies therefore also results in a high correlation in the handedness of V.To summarize,we see a high correlation in the handedness of V between the observed frequencies, which is what we expect from our previous knowledge. This correlation,however,does not add signi?cantly to our interpretation of the observations.
Given the simultaneity of our data,we have the op-portunity to inspect single pulses as a whole observed at both frequencies.Two example single-pulses are provided in Figs.8and9.Fig.8shows pulse number1562from PSR B1133+16at4.85GHz(top panel)and1.41GHz (bottom panel).The leading pulse component shows the characteristic swing and change of handedness in V usu-ally associated with core components.The sense of the swing is opposite between the two frequencies although the PAs agree.They both show a PA jump in the same direction,although the jump happens slightly earlier at the lower frequency.In pulse2438in Fig.9,an OPM jump occurs in bin48and the PA jumps back again in bin52, at both frequencies.Both jumps correspond closely to the peaks in V.The shape of the V pro?le around these bins is also very similar between the two frequencies,but the handedness is opposite.The examples also help to extend the signi?cance of our results to timescales comparable to the pulsar component widths.
5.Summary
In this paper we have demonstrated a method for decom-posing single pulses into the OPMs,assuming incoherently superposed OPM emission.We support this method in
8 A.Karastergiou et al.:Simultaneous single-pulse observations of radio
pulsars
Fig.8.Single pulse number 1562from PSR B1133+16,at 4.85GHz (top)and 1.41GHz (bottom).The solid line rep-resents the total power,the dashed line the linear polariza-tion,and the dashed-dotted line the circular polarization.The PA jumps in the same direction at both frequencies,but the circular polarization follows a swing of opposite handedness between the frequencies.
light of our ?ndings for PSR B1133+16,where the hand-edness of the circular polarization has been observed not to be perfectly associated to the dominant OPM.We have shown that the correlation of V with OPM is worse at 4.85GHz than at 1.41GHz,which is consistent with the con-clusion of Paper II that propagation through the pulsar magnetosphere causes more severe e?ects at higher rather than lower frequencies.Finally,we show two examples of single pulses to demonstrate that our bin-by-bin analysis is also applicable to structures with timescales similar to pulsar components.
Acknowledgements.The authors would like to thank every-body at the E?elsberg and Lovell radio telescopes who helped in any way with the observations.AK would like to thank the Deutscher Akademischer Austausch Dienst for their ?nancial
support.
Fig.9.Single pulse number 2438from PSR B1133+16,at 4.85GHz (top)and 1.41GHz (bottom).The circular polarization disagrees in handedness between the frequen-cies.The V ?ux densities match very well,showing a peak at each phase bin of the PA jump.References
Barnard,J.J.&Arons,J.1986,ApJ,302,138
Cordes,J.M.,Rankin,J.M.,&Backer,D.C.1978,ApJ,223,
961
Karastergiou,A.,Kramer,M.,Johnston,S.,et al.2002,A&A,
391,247
Karastergiou, A.,von Hoensbroech, A.,Kramer,M.,et al.
2001,A&A,379,270
McKinnon,M.1997,ApJ,475,763
McKinnon,M.&Stinebring,D.1998,ApJ,502,883
McKinnon,M.M.&Stinebring,D.R.2000,ApJ,529,435Petrova,S.A.2001,A&A,378,883
Stinebring,D.R.,Cordes,J.M.,Rankin,J.M.,Weisberg,
J.M.,&Boriako?,V.1984a,ApJS,55,247
Stinebring,D.R.,Cordes,J.M.,Weisberg,J.M.,Rankin,
J.M.,&Boriako?,V.1984b,ApJS,55,279
von Hoensbroech,A.,Lesch,H.,&Kunzl,T.1998,A&A,336,
209
Simultaneous interpretation The simultaneous interpretation means translating the speaker’s speech almost simultaneously, which is used for international conferences, seminars and so on. Simultaneous interpretation makes a feature of professional and academic. Some people may think simultaneous interpretation is just a simple thing when you are proficient in the language and are quick-minded. However, if you can’t grasp the real meaning of the speech and just do simple conversion, you may fail. I think, to be a good simultaneous interpreter, you should have the following abilities. To begin with, memory ability is very important. Memory ability includes short-term memory and long-term memory. And short-term memory creates the probability of simultaneous interpretation. The interpreter should memory the sentences in the limited time, or the short-term memory will disappear quickly. But the interpreter needs long-term memory to get rid of simple conversion. And also, long-term memory can help interpreters grasp the spokesman’s meaning well and truly. Then, interpreters should have the ability to predict the sentences and understand the speech in advance. Although the interpreter speaks after the spokesman, they also can catch up with the speed of spokesman. The decisive factors of prediction ability are language skills and the scope of knowledge. Furthermore, strain capacity is necessary in simultaneous interpretation. The interpreter can’t avoid suffering unexpected difficulties. For example, when missing some words, the interpreter should complement the sentence according to the specific circumstance. When the prediction is opposite to the original sentences, the interpreter should not translate again but add an explanation in time. Finally, we talk about storage capacity. In order to insure the fluency of simultaneous interpretation, the interpreter should store one or more conception that spokesman refers to. Because of there are many differences between Chinese and English, interpreters usually need adjust the structure of sentences. As a result, the storage capacity is very important for interpreters. In conclusion, to make simultaneous interpretation accurate, clear and integrated, the interpreter should cultivate the four abilities above. But these abilities are the foundation of simultaneous interpretation. Also, how to practice is very important. There are two model of practicing simultaneous interpretation. The first is Jill Model, which is SI=L+M+P+C. It means simultaneous interpretation depends on listening and analysis, short-term memory effort, speech production and coordination. Another model is Xiamen University Model. It is more difficult than Jill Model. And I want to talk about it in detail. Xiamen University Model says the analysis of discourse and cross-cultural communication is important. Only when you understand the differences of culture, you can understand the meaning exactly. And you should have good comprehension in source language and the knowledge. Moreover, skill and professional standard play an important role. The interpreter should abide by the criteria of interpreting. The last
Bilingualism:Language and Cognition 7(3),2004,227–240C 2004Cambridge University Press DOI:10.1017/S1366728904001609227 Components of simultaneous interpreting:Comparing interpreting with shadowing and paraphrasing ? INGRID K.CHRISTOFFELS ANNETTE M.B.DE GROOT University of Amsterdam Simultaneous interpreting is a complex task where the interpreter is routinely involved in comprehending,translating and producing language at the same time.This study assessed two components that are likely to be major sources of complexity in SI:The simultaneity of comprehension and production,and transformation of the input.Furthermore,within the transformation component,we tried to separate reformulation from language-switching.W e compared repeating sentences (shadowing),reformulating sentences in the same language (paraphrasing),and translating sentences (interpreting)of auditorily presented sentences,in a simultaneous and a delayed condition.Output performance and ear–voice span suggest that both the simultaneity of comprehension and production and the transformation component affect performance but that especially the combination of these components results in a marked drop in performance.General lower recall following a simultaneous condition than after a delayed condition suggests that articulation of speech may interfere with memory in SI. Introduction In international political or corporate meetings simul-taneous interpreting plays an important role in mediating communication.In daily life,we may have encountered simultaneous interpretations of live broadcasted state-ments or interviews on television news channels,such as CNN,and may have been intrigued by this capacity to verbally transform online a message from one language,the source language,into another language,the target language.In simultaneous interpreting (SI)it is required that interpreters both listen and speak at the same time.In this regard it contrasts with so called consecutive interpreting,where the interpreter alternates between listening and speaking and only starts to translate after the speaker has ?nished speaking. SI is a cognitively demanding task because many processes need to take place at one and the same moment in time.The interpreter has to comprehend and store input segments in the source language,transform an earlier segment from source to target language,produce an even *The authors thank Bregje van Oel and Anniek Sturm-Faber for their invaluable research assistance,and Susanne Borgwaldt,Lourens Waldorp,Jos van Berkum and Diane Pecher for their helpful comments on earlier versions of this paper.This research was conducted while I.K.Christoffels was supported by a grant (575-21-011)from the Netherlands Organization for Scienti?c Research (NWO)foundation for Behavioral and Educational Sciences of this organization.Portions of this research were presented at the Twelfth Meeting of the European Society of Cognitive Psychology in Edinburgh,UK,in September 2001. Address for correspondence Ingrid Christoffels,Maastricht University,Faculty of Psychology,Department of Neurocognition,P .O.Box 616,6200MD Maastricht,The Netherlands E-mail :i.christoffels@psychology.unimaas.nl earlier segment in the target language,and cope with time pressure since SI is externally paced;the speaker,not the interpreter,determines the speaking rate (e.g.Gerver,1976;Lambert,1992;Padilla,Bajo,Ca?n as and Padilla,1995).That SI is intrinsically demanding is illustrated by the fact that even experienced professional interpreters sometimes make up to several mistakes per minute,and that they usually take shifts of only 20minutes maximum to prevent fatigue (Gile,1997). During SI,in contrast to normal language use,the interpreter must both comprehend another person’s speech and produce their own speech at the same time.Although interpreters appear to take advantage of pauses in the input (Goldman-Eisler,1972,1980;Barik,1973),a com-parison of the input to their spoken output suggests that almost 70%of the time they are actually talking while processing the input (Chernov,1994).The simultaneity of comprehension and production is perhaps the most salient characteristic of simultaneous interpreting,and it is likely to be one of the main reasons why it is such a cognitively demanding task. Many people would agree that SI is one of the most complex language processing skills,and hence,characterizing this ability may be regarded as an important objective of psycholinguistic research.It is therefore surprising that experimental research on SI is sparse (see Christoffels and De Groot,in press,for a review).On the one hand,we may want to understand the processes involved in SI to extend our understanding of bilingual language performance in general.On the other hand,studying SI may re?ne our models on language
3.Simultaneous-Move Games We now want to study the central question of game theory:how should a game be played.That is,what should we expect about the strategies that will be played in a game.We will start from the simplist types of games,those of simultaneous-move games in which all players move only once and at the same time.We will…rst study simutaneous games with complete information,and then study simutaneous games with incompete information,where each player’s payo¤s may be known only by the player. Dominant strategy and dominated strategy One quite convincing way of reaching a prediction in a game is the idea of domi-nance.Let’s start from a simple example. The Prisoner’s dilemma.Two individuals are arrested for allegedly commiting a crime.The district attorny(DA)does not have enough evidenc to convict them. The two suspects are put in two separate jail cells,and are told the following:if one confesses and another does not confess,then the confessed suspect will be rewarded with a light sentence of1year,while the unconfessed prisoner will be sentenced to 10years;if both confess,then both will be sentenced to5years;if neither confesses, then the DA will still have enough evidence to convict them for a less serious crime, and sentence each of them to jail for2years.The game is depicted below: Prisoner2 DC C DC-2,-2-10,-1 Prisoner1 C-1,-10-5,-5 Each player has two possible strategies:DC or C.What strategy will each player choose?Notice that strategy C is best for each player regardless of what the other 1
Sequential-Simultaneous Games Subgame perfect equilibrium is used to solve sequential games and Nash equilibrium simultaneous games.How are we going to solve games with both simultaneous and sequential moves? CROSSTALK Figure 6.1A Two-Stage Game Combining Sequential and Simultaneous Moves GLOBALDIALOG Don’t Invest CROSSTALK Don’t 0,00,14Invest 14,0?2,?2 Figure 6.2Stage One Investment Game (After Using Backward Induction)Suppose the investment decision is not simultaneous and GlobalDialog has already made the investment.Should CrossTalk invest? CROSSTALK 0,140,6 Figure 6.3Two-Stage Game When GlobalDialog Has Already Invested The following four examples further illustrate the importance of orders of moves.Di?erent orders of moves may have no impact on the equilib-rium (Prisoners’Dilemma),may result in a ?rst-mover advantage (Chicken),or a second-mover advantage (Matching Pennies),or a new equilibrium (Monetary-Fiscal Policy Game). 1
Simultaneous interpreting: A relevance-theoretic approach BRANCA VIANNA Abstract This paper is an attempt to discuss a specific mode of translation—simultaneous interpreting (SI)—within the framework of relevance theory (RT). It is based on Ernst-August Gutt’s account of translation, Translation and Relevance: Cognition and Context (2000). The introduction presents the notion of simultaneous interpreting as interpretive use of language, in which the speech produced by the interpreter achieves relevance by virtue of its interpretive resemblance with the source speech. Part one reflects on the differences between written translation and simultaneous interpreting. Part two considers the question of effort, a crucial relevance theoretic concept, in terms of the effort expended both by the interpreter and by her listeners. Part three discusses the role of the interpreter in the communicative act and the possible degrees of interpretive resemblance achieved by the interpreter’s rendering of the source speech. Pa rt four suggests a few relevance-theoretic explanations to certain types of errors in simultaneous interpreting. The paper comes to the conclusion that relevance theory is not only an excellent framework for simultaneous interpreting research but also a possible tool for improving interpreter-training methods. 1. Introduction According to Grice, one of the essential features of human communication is the expression and recognition of intentions (Grice 1989). Relevance theory takes this idea and develops it into an empirical theory of inferential communication, according to which a communicator provides evidence of his intention to convey a certain meaning, which is inferred by the audience on the basis of the evidence provided (Sperber & Wilson 1995). According to this approach, most communication between humans is ostensive-inferential communication. There must be ostension on the part of the speaker and the hearer must infer the meaning of what is being ostensively communicated. Simultaneous interpreting (SI) is a service practiced in professional conditions, in which interpreters in a sound-proof booth with headsets, control consoles and microphones, and a direct view on the meeting room, deliver versions of the discourse in di¤erent languages ‘on line,’ w ith a lag of a few seconds, alternating every 20–30 minutes or as speakers take turns on the conference floor. (Setton 1999: 1) This is a special form of ostensive human communication: there is a speaker on a podium, with a microphone, addressing an audience which is there expressly to hear him and learn as much as possible from what he will say. The audience cannot help but perceive the ostensive communicative behavior of the speaker and will therefore automatically make inferences and attribute meaning to what he is saying. The role of the interpreter is to facilitate this exchange when speaker and audience do not speak the same language. The interpreter perceives the ostensive communicative behavior of the speaker and must make her1 own informative intention clear to her hearers. She is not the primary addressee of the communication; nor is she the initiator
SIMULTANEOUS MODELING OF SPECTRUM,PITCH AND DURATION IN HMM-BASED SPEECH SYNTHESIS Takayoshi Yoshimura,Keiichi Tokuda,Takashi Masuko,Takao Kobayashi and Tadashi Kitamura Nagoya Institute of Technology,Gokiso,Shouwa-ku,Nagoya,466-8555Japan Tokyo Institute of Technology,Nagatsuta,Midori-ku,Yokohama,226-8502Japan E-mail:yossie,tokuda,kitamura@ics.nitech.ac.jp,masuko,tkobayas@ip.titech.ac.jp ABSTRACT In this paper,we describe an HMM-based speech syn-thesis system in which spectrum,pitch and state duration are modeled simultaneously in a uni?ed framework of H-MM.In the system,pitch and state duration are modeled by multi-space probability distribution HMMs and mul-ti-dimensional Gaussian distributions,respectively.The distributions for spectral parameter,pitch parameter and the state duration are clustered independently by using a decision-tree based context clustering technique.Synthetic speech is generated by using an speech parameter genera-tion algorithm from HMM and a mel-cepstrum based vocod-ing technique.Through informal listening tests,we have con?rmed that the proposed system successfully synthe-sizes natural-sounding speech which resembles the speaker in the training database. 1.INTRODUCTION Although most text-to-speech synthesis systems can syn-thesize speech with acceptable quality,they still cannot synthesize speech with various voice characteristics such as speaker individualities and emotions.To obtain vari-ous voice characteristics in text-to-speech synthesis systems based on the selection and concatenation of acoustical u-nits,a large amount of speech data is necessary.However, it is dif?cult to collect,segment,and store it.From these points of view,in order to construct speech synthesis system which can generate various voice characteristics,we have proposed an HMM-based speech synthesis system[1]. Several HMM-based concatenative speech synthesis sys-tems have also been proposed(e.g.,[2],[3]).Our system, however,differs from them in that our system generates syn-thetic speech from HMMs themselves by using a speech pa-rameter generation algorithm[4]and a mel-cepstrum based vocoding technique[5],[6].In the parameter generation algorithm,by the inclusion of dynamic coef?cients in the feature vector,the dynamic coef?cients of the speech pa-rameter sequence generated in synthesis are constrained to be realistic,as de?ned by the parameters of the HMM-s.By transforming HMM parameters appropriately,voice characteristics of synthetic speech can be changed since the system generates speech from the HMMs.In fact,we have shown that we can change voice characteristics of synthetic speech by applying a speaker adaptation technique[7]or a speaker interpolation technique[8].However,to change not only speaker individuality but also speaking style such as emotion expression,pitch and duration parameters have Continuous probability distribution (Spectrum part) Stream 1 Stream 2 (Pitch part) Multi-space probability distribution Multi-space probability distribution Multi-space probability distribution Figure1.Feature vector. to be appropriately transformed. In this paper,in order to apply speaker adaptation/ interpolation techniques to spectrum,pitch and state dura-tion,simultaneously,and to synthesize speech with various voice characteristics,we construct a speech synthesis sys-tem in which spectrum,pitch and state duration are modeled simultaneously in a uni?ed framework of HMM.In the sys-tem,pitch and state duration are modeled by multi-space probability distribution HMMs[9]and multi-dimensional Gaussian distributions[10],respectively.The feature vec-tor of HMMs used in the system consists of two streams, i.e.,the one for spectral parameter vector and the other for pitch parameter vector,and each phoneme HMM has its state duration densities.The distributions for spectral pa-rameter,pitch parameter and state duration are clustered independently by using a decision-tree based context clus-tering technique[11].In the context clustering,we take account of contextual factors which affect spectrum,pitch and duration such as phone identity factors,stress-related factors and locational factors. The rest of this paper is structured as follows.Section 2describes simultaneous modeling of spectrum,pitch and state duration.Section3describes the proposed text-to-speech synthesis system.Experimental results are present-ed in Section4,and concluding remarks and our plans for future work are presented in the?nal section. 2.SIMULTANEOUS MODELING 2.1.Spectrum and Pitch Model We use mel-cepstral coef?cients as spectral parameter.Se-quences of mel-cepstral coef?cient vector,which are ob-tained from speech database using a mel-cepstral analysis technique[5],are modeled by continuous density HMMs. The mel-cepstral analysis technique enables speech to be re-synthesized from the mel-cepstral coef?cients using the MLSA(Mel Log Spectrum Approximation)?lter[6].