LET THE DATA SPEAK FOR THEMSELVES: A FORM-DRIVEN CROSS ...
LET THE DATA SPEAK FOR THEMSELVES: A FORM-DRIVEN CROSS-LINGUISTIC STUDY OF TENSE AND ASPECT
CHOU MO (C.MO@UU.NL)
15, JUNE, 2018
UIL OTS, UTRECHT UNIVERSITY Research Questions ? How are the different tense notions expressed in an aspect-prominent language like Mandarin? ? What are the similarities and differences between the temporal systems of tense-prominent and aspect-prominent languages?
Step 2: Translation Distance Calculation
Example No. French German Dutch Spanish English Mandarin
Mandarin Aspect Markers (Xiao & McEnery, 2004: 171, 240):
(I) Perfective Aspect Markers ? Le: actual marker, e.g. chi-le `eat-le'. (Verbal le vs. Sentential le) ? Guo: experiential marker, e.g. kan-guo `see-guo'. ? RVC: completive marker, structure `simple verb form + resultative complement (verb/adj.)', e.g. xie-hao `write-done', chi-wan `eat-finish'. The complement indicates the result of an action. ? Unmarked duplication: delimitative marker, simple verbal form is duplicated, e.g. yao-yao `shake-shake'. (II) Imperfective Aspect Markers ? Zhe: a progressive/state marker, e.g. baoliu-zhe `keep-zhe', and wo-zhe `hold-zhe'. ? Zai: a progressive marker, e.g. zai-jixu `zai-continue'. (III) Zero Form ? Aspectually unmarked verb: simple verb form, e.g. xunzhao `look for'. Perfective or Imperfective, dependent on context.
Methodology (computational methodology from van der Klis et al. (2017)) Parallel Corpus Data (French, German, Dutch, Spanish, English, Mandarin) Chapters 1-3 in French novel L'?tranger (app. 300 sentence examples) Chapters 1 & 17 in English novel Harry Potter and the Sorcerer's Stone (app. 1000 sentence examples) Step 1: Sentence Tense/Aspect Marker Annotation Example No. 24950 French: J' ai bu. (pass? compos?) German: Ich habe getrunken . (perfekt) Dutch: Ik dronk. (o.v.t.) Spanish: Beb?. (pret?rito indefinido) English: I drank. (simple past) Mandarin: (he-le `drink-le') (le) Tense/Aspect Tuple No. 24950 :
French Map
German Map
24950
pass? perfekt o.v.t. pret?rito simple
le
compos?
indefinido past
25851
pass? perfekt v.t.t. pret?rito simple
le
compos?
perfecto present
compuesto
Tense/Aspect
0
0
1
1
1
0
Variation
Total tense/aspect variation between 2 examples Translation distance =
Number of languages in comparison
Total variation: 3 (0+0+1+1+1+0); 6 languages are compared. The translation distance between Examples 24950 and 25851 is 3/6 (0.5).
Descriptive statistics are collected: Tense/aspect marker frequency & tuple frequency
Aspect Form unmarked
le rvc duplication zhe non-verb guo zai
Frequency 154
77 50 11 9 5 2 2
Percentage
49.7% 24.8% 16.1% 3.5% 2.9% 1.6% 0.6% 0.6%
Aspect Pt Pt Pre Pt
Form Nart1 Impf
Cont
unmarked 59 42 19 13
Le
31 12 3
0
rvc
20 4 0
0
duplication 10 0 0
0
zhe
6 40
4
non-verb 2 21 3
0
zai
0 00
4
Mandanrin Aspect Form Distribution in Camus dataset
Step 3: Multi-dimensional Maps (Smaller Translation Distance, Closer Dots )
Dutch Map
Spanish Map
TABUDAG 2018
`Time in Translation' group (Henri?tte de Swart, Bert Le Bruyn, Martijn van der Klis) See 'Time in Translation' Project at
Top 4 EU language Tuples in Camus Dataset Pt Nart 1 (Past narrative): Perfekt, simple past, pret?rito indefinido, pass? compos?, ovt Pt Impf (Past imperfective): Pr?teritum, simple past, pret?rito imperfecto, imparfait, ovt Pre (PRESENT): Pr?sens, simple present, presente, pr?sent, ott Pt Cont (Past continous) : pr?teritum, past continuous, pret?rito imperfecto, imparfait, ovt
Mandarin Aspect Form Distribution (among Top 4 EU language tuples in Camus Dataset) ? Unmarked: Pt Nart 1 > Pt Impf > PRESENT> Pt Cont ? Le: Pt Nart 1 > Pt Impf > PRESENT> Pt Cont=0 ? Rvc: Pt Nart 1 > Pt Impf > PRESENT= Pt Cont=0 ? Duplication: Pt Nart 1 > Pt Impf=PRESENT= Pt Cont=0 ? Zhe: Pt Nart 1 > Pt Impf > Pt Cont > PRESENT=0 ? Non-verb: Pt Impf > PRESENT > Pt Nart 1 > Pt Cont=0 ? Zai: Pt Cont> Pt Nart 1 = Pt Impf = PRESENT=0 * Guo: only 5 examples, co-occur with either PERFECT or PLUPERFECT tuples.
Subset theory for EU PERFECT (see MDS Maps) Perfect use in EnglishPerfect use in Spanish Perfect use in Dutch Perfect use in GermanPerfect use in French (de Swart, 2017)
Mandanrin Aspect Distributon in PERFECT vs. PAST Competition
German [+Narration]
PERFECT
PAST
Unmarked, le, rvc, zhe,
Unmarked (3rvc, 2non-verb)
duplication (guo, zai, non-verb)
Dutch [+Boundedness] Spanish [+Deictic adverbs]
English [+Classical Perfect]
Le, unmarked (guo,1zhe)
Le, unmarked (guo, 1non-verb,1rvc) Le, unmarked (guo, 3non-verb,1zhe)
Unmarked, le, rvc, zhe, duplication (zai, non-verb) Unmarked, le, rvc, zhe, duplication(zai, non-verb)
Unmarked, le, rvc, zhe, duplication (zai, non-verb)
English Map
Mandarin Map
Aspect Marker
unmarked
rvc
le1 (verbal le)
le12 (mix le)
Pt Nart 2 Count Percentage 142 46.41%
63 20.59% 41 13.40%
22 7.19%
PRESENT
Pt Impf
Pt Nart 3
Pt Cont
Count Percentage Count Percentage Count Percentage Count Percentage
62
66.67% 52 55.91% 16 57.14% 13 52.00%
2
2.15% 7
7.53%
1
3.57%
1
4.00%
2
2.15%
3
3.23%
7 25.00% 5
20.00%
11
11.83% 4
4.30%
2
7.14%
0
0.00%
IMPERATIVE
PLUPERFECT
FUTURE
Count Percentage Count Percentage Count Percentage
5
45.45% 0
0.00%
0
0.00%
4
36.36% 3 33.33%
0
0.00%
0
0.00%
1 11.11%
0
0.00%
0
0.00%
0
0.00%
0
0.00%
Duplication 14 4.58%
1
1.08%
0
0.00%
0
0.00%
0
0.00%
1
9.09%
0
0.00%
0
0.00%
zhe
14 4.58%
1
1.08%
4
4.30%
0
0.00%
5
20.00%
0
0.00%
0
0.00%
0
0.00%
non-verb 9
2.94%
11
11.83% 22 23.66% 2
7.14%
0
0.00%
1
9.09%
0
0.00%
0
0.00%
zai
1
0.33%
2
2.15%
0
0.00%
0
0.00%
0
0.00%
0
0.00%
0
0.00%
0
0.00%
Aux
0
0.00%
1
1.08%
1
1.08%
0
0.00%
1
4.00%
0
0.00%
0
0.00%
4 100.00%
guo
0
0.00%
0
0.00%
0
0.00%
0
0.00%
0
0.00%
0
0.00%
5 55.56%
0
0.00%
Sum
306 100.00%
93
100.00% 93 100.00% 28 100.00% 25 100.00% 11 100.00% 5 100.00% 4 100.00%
Top 8 EU language tuples in Harry Potter Dataset
Conclusions from 2 Datasets
Pt Nart 2: Pr?teritum, simple past, pret?rito indefinido, pass? simple, ovt
? EU languages distinguish PAST/PRESENT/FUTURE ,
PRESENT: Pr?sens, simple present, presente, pr?sent, ott
Mandarin distinguishes FUTURE/NON-FUTURE (Aux can
Pt Impf: Pr?teritum, simple past, pret?rito imperfecto, imparfait, ovt
mark FUTURE) , but not PAST/PRESENT.
Pt Nart 3: Perfekt, simple past, pret?rito indefinido, pass? compos?, vtt
? In EU languages, (almost) no variation in PLUPERFECT,
Pt Cont: Pr?teritum, past continuous, pret?rito imperfecto, imparfait, ovt
PRESENT and FUTURE, but competition in PAST and
IMPERATIVE: Imperativ, imperative, imperativo, imp?ratif, imperatief
PERFECT domains.
PLUPERFECT: Plusquamperfekt, past perfect, pret?rito pluscuamperfecto, plus-que-parfait, vvt ? Mandarin has rich forms for PAST & PRESENT, but
FUTURE: Futur I, simple future, futuro pr?ximo, futur proche, ott
restricted forms for PLUPERFECT & FUTURE, both of which
Mandarin Aspect Form Distribution (among Top 8 EU language tuples in Harry Potter Dataset) ? Unamrked: Pt Nart 2 > PRESENT> Pt Impf >Pt Nart 3>Pt Cont >IMPERATIVE ? Rvc: Pt Nart 2 > Pt Impf > IMPERATIVE>PLUPERFECT> PRESENT>Pt Nart 3=Pt Cont ? Le1 (verbal le): Pt Nart 2> Pt Nart 3 > Pt Cont >Pt Impf > PRESENT> PLUPERFECT ? Le12 (mix le): Pt Nart 2 > PRESENT > Pt Impf > Pt Nart 3 (>Pt Cont=0) ? Non-verb: Pt Impf > PRESENT> Pt Nart 2> Pt Nart 3 >IMPERATIVE (>Pt Cont=0) ? Aux: FUTURE > PRESENT= Pt Impf = Pt Cont ? Zhe: Pt Nart 2 > Pt Cont >Pt Impf > PRESENT ? Duplication: Pt Nart 2> PRESENT=IMPERATIVE ? Zai: PRESENT>Pt Nart 2 ? Guo: PLUPERFECT
must be marked in Mandarin. ? Iceberg in Mandarin aspect research: unmarked (50% data, Pt Nart > Pt Impf > Pt Cont), rvc (Pt Nart > PRESENT) . ? Verbal le: Pt Nart > Pt Impf > PRESENT, but also in Pt Cont. (why?) Sentential le: Pt Nart > PRESENT > Pt Impf ? Duplication: Pt Nart (V-le-V), PRESENT, IMPERATIVE (V-V) ? Zhe: Pt Nart > Pt Impf > PRESENT ? Zai: few in Narration, more imperfective than zhe ? Guo: reside within European PERFECT & PLUPERFECT ? Non-verb: prefer strong imperfective contexts
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
- let the data speak for themselves a form driven cross
- rule 1 0 1 terminology rule approved by the supreme court
- using parsed corpora to compare the evolution of word
- chapter iv finding and discussion
- serbian an essential grammar
- problems of english grammar hm
- the sketch engine
- quantitative research dissertation chapters 4 and 5
- comparing the evolution of v2 in english and french
- a comparative analysis of the arabic and english verb
Related searches
- think for themselves synonym
- speak for someone else synonym
- bible verse let the children come
- let the children come unto me verse
- let the children come to me bible
- let the children come jesus
- write the chemical equation for a photosynthesis
- pandas check data type for a column
- where is the data analysis on excel
- simply the quest who let the gods out
- poetry for the aztec was considered as a very scared practice they used this li
- praise the lord let the earth lyrics