A Translation Mining view on reference: method, data and analysis
Bert Le Bruyn CLS colloquium Nijmegen, 2024-6-20
Translation Mining
3
Translation Mining
> ° Time in Translation project (Henriëtte de Swart & Bert Le Bruyn | Utrecht | 2017-2022).
> Exploiting translations to map out and analyze cross-linguistic variation.
> Research method complementing questionnaires, experiments, monolingual corpus research, etc.
Source Text I
Source Text II (replication)
...
Source Text n+I (replication)
Experiments (triangulation)
> Methodological reflection (see, e.g., Le Bruyn et al. 2023; Le Bruyn et al. 2024).
4
Translation Mining
> Main methodological insights:
• Translation data are highly systematic, allowing for insightful analysis of general patterns of variation and of the individual datapoints that give rise to them;
• They can be used both from an exploratory and a hypothesis-driven perspective.
• They are particularly useful in the study of the division of labor between forms, within and across languages;
Beyond Time in Translation
6
Beyond Time in Translation
> Original focus on meso-variation in the use of have/be + past participle in Western European languages.
> Move to macro-variation:
• Chou Mo’s dissertation work (Mo 2022): study of Mandarin aspect through the Mandarin translation of L’Étranger.
• Jianan Liu’s dissertation work (ongoing): study of reference in Mandarin, including cross-linguistic joint work on a.o. Hindi and Russian with Olga Borik, Shravani Patil, Daria Seres, Hagay Schurr, Sadhwi Srinivas and me aka the HHRM group.
7
Beyond Time in Translation
Mandarin
Hindi
Russian
(Wright & Givón 1987; Li & Thompson 1989)
8
Beyond Time in Translation
Cheng & Sybesma (1999):
9
Beyond Time in Translation
> Basic questions:
> Is there an interaction between Mandarin bare nouns on the one hand and the numeral yi (‘one’)/the demonstrative on the other hand?
> If so, how should we account for this interaction?
> Roadmap:
> Basic data
> Indefiniteness in Mandarin
> Definiteness in Mandarin
> Wrap-up
Basic data
11
Basic data
> Translations to Mandarin.
> Manual (computer-assisted) extraction of all nominal expressions (including pronouns) from the English original (N=1210).
Chapter 1, Harry Potter and the Philosopher’s Stone
12
Basic data
(selection)
(n=52)
(n=140)
(n=90)
13
Basic data
indefinite singular
definite singular
indefinite plural
(selection)
Indefiniteness in Mandarin
15
Basic data
indefinite singular
16
Basic data
一 只 猫 在 看 地图 。
yī zhǐ māo zài kàn dìtú.
It was on the corner of the street that he noticed the first sign of something peculiar -
one cl cat asp see map
Dumbledore gave a great sniff as he took a golden watch from his pocket [...]
a cat reading a map
从 衣袋 里 掏 出 一 块 金 表
cóng yīdài lǐ tāo chū yī kuài jīn biǎo
from pocket in pull out one cl gold watch
17
Basic data
18
Basic data
The formal literature – to the best of my knowledge – does not predict variation:
> Chierchia (1998)
> Dayal (2004)
> Cheng & Sybesma (1999)
> Jiang (2020)
random variation?
19
Going cross-linguistic
20
Going cross-linguistic
21
Going cross-linguistic
22
Going cross-linguistic
Interim conclusion
> The numeral ‘one’ uses in Mandarin cannot be attributed to random variation. If they could, we would expect similar ‘random’ uses in other articleless languages.
23
Going cross-linguistic
24
Going cross-linguistic
25
Going cross-linguistic
(selection)
26
Going cross-linguistic
Interim conclusion
> The numeral ‘one’ uses in Mandarin cannot be attributed to random variation. If they could, we would expect similar random uses in other articleless languages like Russian.
> The numeral ‘one’ uses in Hindi/Marathi and Mandarin seem to form a scale, suggesting that at least some of the factors influencing the choice between bare nouns and ‘one’ N are constant.
27
Dayal (2004, 2011), Huang (2015), Luo (2022)
> Dayal (2004, 2011):
• Hindi has a general ban on singular indefinite bare nouns requiring the use of an indefinite determiner (like, e.g., the numeral ‘one’).
• Bare nouns that do occur in singular indefinite environments are (pseudo-)incorporated.
> Huang (2015) suggests that (pseudo-)incorporation might also exist in Mandarin and manifests itself – among others – in ‘well-established’ verb-noun combinations: kan shu (‘read book’), kai wanxiao (‘make joke’).
> Luo (2022) is the first to propose a formal semantic analysis of pseudo-incorporation in Mandarin and builds on Schwarz (2014) but implements it in a kind approach to bare nouns.
28
Dayal (2004, 2011), Huang (2015), Luo (2022)
29
Dayal (2004, 2011), Huang (2015), Luo (2022)
Hypothesis
The numeral ‘one’ functions as an indefinite article in Mandarin, blocking bare nouns from appearing with an indefinite interpretation in regular argument position. If bare nouns get an indefinite interpretation, they are pseudo-incorporated.
[Further differences between Hindi/Marathi and Mandarin should be attributed to independent factors.]
Challenge
Unlike Hindi (and other (pseudo-)incorporation languages like Hungarian), Mandarin has no form-based criteria to decide whether we’re dealing with (pseudo-) incorporation: it has rigid word order, no case markers and no generalized singular/plural distinction.
30
Dayal (2004, 2011), Huang (2015), Luo (2022)
31
Dayal (2004, 2011), Huang (2015), Luo (2022)
Rising to the challenge
For Dayal, Huang and Luo, a crucial property of (pseudo-)incorporation is that – as far as the object position is concerned – it is limited to ‘well-established’ verb-noun combinations.
If we can operationalize what it means to be a ‘well-established’ verb-noun combination, we can check whether the bare nouns we find in object position in singular indefinite contexts in Mandarin are part of this type of combinations, allowing for a meaning-based assessment of their (pseudo-)incorporation status.
32
Le Bruyn et al. (2016) and Liu et al. (2022)
> Le Bruyn et al. (2016) is cast in dynamic predicate logic and analyzes pseudo-incorporation of verb-noun combinations as involving a verb that targets an explicit or implicit argument of its object noun and prepares it for binding by its subject. Illustrative examples are have son (explicit relational argument) and read book (implicit qualia-based ‘user’ argument).
> In Liu et al. (2022), we argue that the way Le Bruyn et al. (2016) model pseudo-incorporation derives the well-establishedness constraint rather than stipulating it.
> In this talk, I use it as a way to decide whether verb-noun combinations qualify as allowing for pseudo-incorporation.
33
An extended (and more focused) dataset
> The basic data I have used up till now is limited to the first chapter of Harry Potter and the Philosopher’s Stone and includes translations of a N in all possible positions.
> We keep the object noun data from the first chapter but extend them with object noun data from the whole book (N=156).
> Our operationalization of our (pseudo-)incorporation check is limited to nouns in object position.
34
Annotation and analysis
> We annotated all the bare noun and ‘one’ N indefinite objects in their Mandarin translations for:
• incorporation: does the V-O combination it occurs in qualify as ‘well-established’?
• form: is the indefinite object rendered as a bare noun (N) or as a structure with a numeral (‘one’ N)?
> We rely on conditional inference trees (Tagliamonte & Baayen 2012) to assess whether incorporation is a good predictor for form.
35
Further annotations
> We further annotated all datapoints for a number of other characteristics that we assumed could influence form:
• non-referential: does the noun refer to an actual individual in the universe of the story? (yes/no)
• modification: is the noun modified or not? (yes/no)
• aspect: does the aspect marker le occur on the verb? (yes/no)
• NEG/DIS: see next slide (yes/no)
36
Further annotations: NEG/DIS
Most of them had never seen an owl even at nighttime .
‘ You must come and stay this summer , ’ said Ron , ‘ both of you - I 'll send you an owl. ’
他们 大多 甚至 夜里都 从未见过*猫头鹰*。
“你今年暑假一定要来我们 家里 玩 ， ” 罗恩 说 ， “ 你们俩都来 —— 我 会派*猫头鹰* 去 邀请 你们 的 。 ”
37
Some examples
一 只 猫 在 看 地图 。
yī zhǐ māo zài kàn dìtú.
It was on the corner of the street that he noticed the first sign of something peculiar -
one cl cat asp see map
Dumbledore gave a great sniff as he took a golden watch from his pocket [...]
a cat reading a map
从 衣袋 里 掏 出 一 块 金 表
cóng yīdài lǐ tāo chū yī kuài jīn biǎo
from pocket in pull out one cl gold watch
38
Some examples
Quirrell raised his hand to perform a deadly curse, but Harry, by instinct, reached up and grabbed Quirrell 's face -
A ghost wearing a ruff and tights had suddenly noticed the first-years .
A barn owl brought Neville a small package from his grandmother .
一个 穿 *轮状皱领紧身衣* 的幽灵突然发现了一年级新生 。
一只猫头鹰 从纳威的奶奶那里给他带来 了 *一 个* *小 包裹* 。
奇 洛举起 手 ，准备念*一个* *死咒* ， 可是 哈利出于 本能 ，猛地抬手抓向 奇洛的脸 ——
39
Results
40
The exception
It was on the corner of the street that he noticed the first sign of something peculiar - a cat reading a map. For a second, Mr Dursley didn't realise what he had seen - then he jerked his head around to look again. There was a tabby cat standing on the corner of Privet Drive, but there wasn't a map in sight.
但是 没有 看见 地图 。
dànshì méiyǒu kànjiàn dìtú .
but not see map
41
Conclusion
> Is there an interaction between bare nouns on the one hand and the numeral yi (‘one’) on the other hand?
> How should we account for this interaction?
The differences we find between Mandarin and Russian/Polish strongly suggest that there’s a genuine interaction between Mandarin bare nouns and the numeral yi.
The account I argue for is one in which the numeral functions as an article, blocking indefinite readings of bare nouns in regular argument position, and in which indefinite readings of bare nouns are only possible in (pseudo-)incorporation contexts.
Definiteness in Mandarin
43
Basic data
definite singular
44
Going cross-linguistic I
> Intuition from Jenks (2018): the contrast we find between bare nouns and demonstratives in Mandarin is the same as the one we find between weak (uniqueness) and strong (familiarity) definites in the German prepositional domain.
The priest is looking at his watch.
The priest is looking at his watch.
The priests are looking at their watches.
The priests are looking at their watches.
The priests are looking at their watches.
The man has a walking stick.
The man has a walking stick.
A man and a woman are walking in the park.
The man has a walking stick.
A man and a woman are walking in the park.
55
Going cross-linguistic I
56
Going cross-linguistic I
> Intuition from Jenks (2018): the contrast we find between bare nouns and demonstratives in Mandarin is the same as the one we find between weak (uniqueness) and strong (familiarity) definites in the German prepositional domain.
> Bremmers et al. (2022): contrastive study based on Harry Potter and the Philosopher’s Stone (full volume) where we compare German PPs with weak and strong definites to bare nouns and demonstratives in Mandarin.
57
Prediction
Results
German contr. def.
German uncontr. def.
Mandarin dem.
Mandarin BN
German contr. def.
German uncontr. def.
Mandarin dem.
Mandarin BN
10
3
45
34
Going cross-linguistic I
58
Going cross-linguistic I
59
Going cross-linguistic I
60
Going cross-linguistic I
61
Going cross-linguistic I
Interim conclusion
There are strong indications that Mandarin relies on bare nouns to convey weak (uniqueness) definiteness and that demonstratives are sometimes used for strong (familiarity) definiteness. However, there is no one-to-one mapping between demonstratives and strong definiteness given that bare nouns also regularly appear in familiarity contexts.
There’s an intuition that demonstratives are required for anaphoricity in ‘less tightly connected’ discourse (see also Simpson & Wu 2022 and Dayal & Jiang 2022).
62
Going cross-linguistic II
Mandarin
Hindi
Russian
63
Going cross-linguistic II
Mandarin
Hindi
Russian
17 anaphoric definites that take a demonstrative in at least one language:
14 in Mandarin
4 in Hindi
3 in Russian
64
Going cross-linguistic II
At half past eight, Mr Dursley picked up his briefcase, pecked Mrs Dursley on the cheek and tried to kiss Dudley goodbye but missed, because Dudley was now having a tantrum and throwing his cereal at the walls. ‘Little tyke,’ chortled Mr Dursley as he left the house. He got into his car and backed out of number four's drive.
It was on the corner of the street that he noticed the first sign of something peculiar - a cat reading a map. For a second, Mr Dursley didn't realise what he had seen - then he jerked his head around to look again. There was a tabby cat standing on the corner of Privet Drive, but there wasn't a map in sight. What could he have been thinking of? It must have been a trick of the light. Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive - no, looking at the sign; cats couldn't read maps or signs. Mr Dursley gave himself a little shake and put the cat out of his mind.
65
Going cross-linguistic II
At half past eight, Mr Dursley picked up his briefcase, pecked Mrs Dursley on the cheek and tried to kiss Dudley goodbye but missed, because Dudley was now having a tantrum and throwing his cereal at the walls. ‘Little tyke,’ chortled Mr Dursley as he left the house. He got into his car and backed out of number four's drive.
It was on the corner of the street that he noticed the first sign of something peculiar - a cat reading a map. For a second, Mr Dursley didn't realise what he had seen - then he jerked his head around to look again. There was a tabby cat standing on the corner of Privet Drive, but there wasn't a map in sight. What could he have been thinking of? It must have been a trick of the light. Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive - no, looking at the sign; cats couldn't read maps or signs. Mr Dursley gave himself a little shake and put the cat out of his mind.
66
Going cross-linguistic II
At half past eight, Mr Dursley picked up his briefcase, pecked Mrs Dursley on the cheek and tried to kiss Dudley goodbye but missed, because Dudley was now having a tantrum and throwing his cereal at the walls. ‘Little tyke,’ chortled Mr Dursley as he left the house. He got into his car and backed out of number four's drive.
It was on the corner of the street that he noticed the first sign of something peculiar - a cat reading a map. For a second, Mr Dursley didn't realise what he had seen - then he jerked his head around to look again. There was a tabby cat standing on the corner of Privet Drive, but there wasn't a map in sight. What could he have been thinking of? It must have been a trick of the light. Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive - no, looking at the sign; cats couldn't read maps or signs. Mr Dursley gave himself a little shake and put the cat out of his mind.
67
Going cross-linguistic II
At half past eight, Mr Dursley picked up his briefcase, pecked Mrs Dursley on the cheek and tried to kiss Dudley goodbye but missed, because Dudley was now having a tantrum and throwing his cereal at the walls. ‘Little tyke,’ chortled Mr Dursley as he left the house. He got into his car and backed out of number four's drive.
It was on the corner of the street that he noticed the first sign of something peculiar - a cat reading a map. For a second, Mr Dursley didn't realise what he had seen - then he jerked his head around to look again. There was a tabby cat standing on the corner of Privet Drive, but there wasn't a map in sight. What could he have been thinking of? It must have been a trick of the light. Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive - no, looking at the sign; cats couldn't read maps or signs. Mr Dursley gave himself a little shake and put the cat out of his mind.
68
Going cross-linguistic II
It was on the corner of the street [...] a cat [...] Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive [...] Mr Dursley gave himself a little shake and put the cat out of his mind.
It was on the corner of the street [...] a cat [...] Mr Dursley blinked and stared at CAT. CAT stared back.
As Mr Dursley drove around the corner and up the road, he watched DEM. CAT in his mirror. CAT was now reading the sign that said Privet Drive [...] Mr Dursley gave himself a little shake and put CAT out of his mind.
Mandarin translation
69
Going cross-linguistic II
It was on the corner of the street [...] a cat [...] Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive [...] Mr Dursley gave himself a little shake and put the cat out of his mind.
It was on the corner of the street [...] a cat [...] Mr Dursley blinked and stared at CAT. CAT stared back.
As Mr Dursley drove around the corner and up the road, he watched CAT in his mirror. CAT was now reading the sign that said Privet Drive [...] Mr Dursley gave himself a little shake and put CAT out of his mind.
Hindi translation
70
Going cross-linguistic II
It was on the corner of the street [...] a cat [...] Mr Dursley blinked and stared at the cat. It stared back.
As Mr Dursley drove around the corner and up the road, he watched the cat in his mirror. It was now reading the sign that said Privet Drive [...] Mr Dursley gave himself a little shake and put the cat out of his mind.
It was on the corner of the street [...] a cat [...] Mr Dursley blinked and stared at CAT. CAT stared back.
As Mr Dursley drove around the corner and up the road, he watched CAT in his mirror. CAT was now reading the sign that said Privet Drive [...] Mr Dursley gave himself a little shake and put CAT out of his mind.
Russian translation
71
Going cross-linguistic II
Interim conclusion
Our Mandarin data are fully in line with the intuition that demonstratives are preferred over bare nouns if the sentence in which the anaphor occurs is ‘less tightly connected’ to the sentence in which its antecedent occurs.
Mandarin furthermore turns out to occupy a special position in this respect, suggesting that Mandarin demonstratives interact with bare nouns in an article-like way.
72
Going cross-linguistic III
> With demonstratives doing double duty as definite articles, we need to determine the exact cut-off point between their definite article and demonstrative uses.
> To do so, we need to widen the contexts we look into from definites and their translations to definites/demonstratives and their translations.
> Demonstratives have two main uses: deictic and anaphoric ones. We focus here on the anaphoric ones. For definites, we also focus on their anaphoric uses.
> The data we use consist of the English anaphoric definites and demonstratives in Chapters 11, 12 and 13 that are translated to Mandarin as bare nouns or demonstratives.
73
Going cross-linguistic III
74
Going cross-linguistic III
‘You know what this means?’ he finished breathlessly. ‘He tried to get past that three-headed dog at Halloween! That’s where he was going when we saw him – he’s after whatever it’s guarding! And I’d bet my broomstick he let that troll in, to create a diversion.’
> English anaphoric demonstratives belong to the dialogue register, highlighting the shared experience of speaker and hearer.
75
Going cross-linguistic III
76
Going cross-linguistic III
77
Going cross-linguistic III
He had to clap his hands to his mouth to stop himself screaming. He whirled around. His heart was pounding far more furiously than when the book had screamed – for he had seen not only himself in the mirror, but a whole crowd of people standing right behind him.
> Mandarin anaphoric demonstratives also appear in the narrative register and are consequently not limited to shared experiences of the speaker and hearer.
> They are used to refer back to referents introduced in contexts that are less tightly connected to the current context.
78
Going cross-linguistic III
Harry felt in the pocket of his robes and pulled out a Chocolate Frog, the very last one from the box Hermione had given him for Christmas. He gave it to Neville, who looked as though he might cry.
‘You’re worth twelve of Malfoy,’ Harry said [...]
‘Thanks, Harry ... I think I’ll go to bed ... D’you want the card, you collect them, don’t you?’
As Neville walked away Harry looked at the Famous Wizard card.
‘Dumbledore again,’ he said.
79
Going cross-linguistic III
[...] He set off, drawing the Invisibility Cloak tight around him as he walked.
The library was pitch black and very eerie. Harry lit a lamp to see his way along the rows of books. The lamp looked as if it was floating along in mid-air [...]
80
Going cross-linguistic III
familiarity definites
uniqueness definites
deictic demonstratives
anaphoric demonstratives
basis = uniqueness in model/(con)text
basis = previous/current shared experience
anaphoric index part of semantics
anaphoric index part of presupposition
familiarity definites+
anaphoric index part of semantics
uniqueness through anaphoric index
81
Going cross-linguistic III
English, Hindi, Russian demonstratives
Mandarin demonstratives
Mandarin bare nouns
English definites, Hindi/Russian bare nouns
82
Conclusion
> Is there an interaction between bare nouns on the one hand and the demonstrative on the other hand?
The differences we find between Mandarin and Russian/Hindi and between Mandarin and English strongly suggest that there’s a genuine interaction between Mandarin bare nouns and the demonstrative.
83
Conclusion
> How should we account for this interaction?
The account I argue for is one in which the demonstrative functions as a definite article and is obligatory for anaphors that have been introduced in contexts that are ‘less tightly connected’ to the current context. These uses lack the experiential restriction we find in anaphoric uses of English demonstratives and their ‘definite-like’ uniqueness is guaranteed through the inclusion of an anaphoric index in their semantics.
Wrap-up
85
Data and analysis
> Is there an interaction between Mandarin bare nouns on the one hand and the numeral yi (‘one’)/the demonstrative on the other hand?
> If so, how should we account for this interaction?
Yes!
For indefiniteness: numeral yi functions as an indefinite article, blocking the use of bare nouns in regular argument position
For definiteness: the demonstrative functions as a definite article, blocking one particular definite use of bare nouns
86
Method
• Translation data are highly systematic, allowing for insightful analysis of general patterns of variation and of the individual datapoints that give rise to them;
• They can be used both from an exploratory and a hypothesis-driven perspective.
• They are particularly useful in the study of the division of labor between forms, within and across languages;
These insights extend from the meso-variation perspective of Time in Translation to the macro-variation we tackled in today’s talk.
Thank you for your attention!