diff --git a/.gitignore b/.gitignore
index d7e9a0b..0095e5a 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,4 +1,3 @@
-*.conllu
 *.pyc
 *.ipynb
 
diff --git a/requirements.txt b/requirements.txt
index 98ea625..c70f2c9 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -3,6 +3,7 @@ build>=1.2.1
 flake8>=7.1.0
 jsonargparse[signatures]>=4.32.0
 lightning>=2.4.0
+parameterized>=0.9.0
 pytest>=7.4.4
 transformers>=4.44.0
 torch>=2.4.0
diff --git a/tests/testdata/datasets/README.md b/tests/testdata/datasets/README.md
new file mode 100644
index 0000000..5c1a228
--- /dev/null
+++ b/tests/testdata/datasets/README.md
@@ -0,0 +1,5 @@
+This directory contains small "toy" samples drawn from Universal Dependencies
+data for English, Greek, and Russian for `udtube_test.py`. The `_train.conllu`
+files are used to train and validate the model, and the `_expected.conllu`
+files are the result of applying the model to the training data.
+Each file contains ten sentences.
diff --git a/tests/testdata/datasets/el_expected.conllu b/tests/testdata/datasets/el_expected.conllu
new file mode 100644
index 0000000..0dc2f13
--- /dev/null
+++ b/tests/testdata/datasets/el_expected.conllu
@@ -0,0 +1,270 @@
+# newdoc id = gdt-20120309-elwikinews-5160
+# sent_id = gdt-20120309-elwikinews-5160-1
+# text = Η Μάντσεστερ Γιουνάιτεντ ηττήθηκε από την Ατλέτικο Μπιλμπάο με σκορ 2:3
+1	Η	ήισ	INTJ	ADP	Case=Acc|Gender=Neut|Number=Sing	2	det	_	_
+2	Μάντσεστερ	μάντσεσο	PUNCT	DET	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	4	nsubj	_	_
+3	Γιουνάιτεντ	γιουνάιτο	INTJ	ADP	Foreign=Yes	2	flat	_	_
+4	ηττήθηκε	ηττεθηκαίνω	[UNK]	_	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	0	root	_	_
+5	από	από]DAP[	PROPN	_	[PAD]	7	case	_	_
+6	την	ω	[UNK]	ADP	[PAD]	7	det	_	_
+7	Ατλέτικο	ατλύτι	INTJ	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	4	obl:agent	_	_
+8	Μπιλμπάο	μπιλμο	[UNK]	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	7	flat	_	_
+9	με	με]DAP[	PROPN	ADP	Case=Gen|Gender=Neut|Number=Plur	10	case	_	_
+10	σκορ	όεσκο	INTJ	DET	Foreign=Yes	4	obl	_	_
+11	2:3	όε2ο	PROPN	_	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	10	nmod	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-3
+# text = Χθες, η Μάντσεστερ Γιουνάιτεντ ηττήθηκε με σκορ 2:3 από την Ατλέτικο Μπιλμπάο, στα πλαίσια της φάσης των 16 του Γιουρόπα Λιγκ 2011-2012.
+1	Χθες	ήιθεσ	INTJ	ADP	Foreign=Yes	6	advmod	_	SpaceAfter=No
+2	,	,]DAP[	PART	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	1	punct	_	_
+3	η	ω	PART	ADP	[PAD]	4	det	_	_
+4	Μάντσεστερ	μάντσεσο	_	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	6	nsubj	_	_
+5	Γιουνάιτεντ	γιουνάιτο	INTJ	_	Foreign=Yes	4	flat	_	_
+6	ηττήθηκε	ηττήθηκε]DAP[	INTJ	NUM	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	0	root	_	_
+7	με	ώ	PROPN	NUM	[PAD]	8	case	_	_
+8	σκορ	όεσκο	INTJ	NUM	Case=Acc|Gender=Neut|Number=Sing	6	obl	_	_
+9	2:3	2	PROPN	_	Case=Acc|Gender=Neut|Number=Sing	8	nmod	_	_
+10	από	από]DAP[	PROPN	_	[PAD]	12	case	_	_
+11	την	ω	[UNK]	_	[PAD]	12	det	_	_
+12	Ατλέτικο	ατλέτιόσ	INTJ	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	6	obl:agent	_	_
+13	Μπιλμπάο	μπιλμο	PROPN	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	12	flat	_	SpaceAfter=No
+14	,	,]DAP[	PART	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	17	punct	_	_
+15-16	στα	ω	DET	_	[PAD]	_	_	_	_
+15	σ	ο	[UNK]	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	17	case	_	_
+16	τα	τα]DAP[	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	17	det	_	_
+17	πλαίσια	όλείσο	INTJ	_	[PAD]	6	obl	_	_
+18	της	τή	INTJ	DET	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	19	det	_	_
+19	φάσης	ω	ADV	ADJ	Case=Gen|Gender=Neut|Number=Plur	17	nmod	_	_
+20	των	ω	AUX	DET	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Pass	21	det	_	_
+21	16	16]DAP[	ADV	ADP	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Pass	19	nmod	_	_
+22	του	του]DAP[	[UNK]	DET	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	23	det	_	_
+23	Γιουρόπα	γιουρο	INTJ	ADP	Foreign=Yes	21	nmod	_	_
+24	Λιγκ	ι	INTJ	DET	Foreign=Yes	23	flat	_	_
+25	2011-2012	20ό1ε20ο	PROPN	[PAD]	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	23	nmod	_	SpaceAfter=No
+26	.	.]DAP[	INTJ	ADP	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	6	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-4
+# text = Το σκορ του αγώνα άνοιξε ο Γουέν Ρούνι στο 22ο λεπτό, ωστόσο οι φιλοξενούμενοι ισοφάρισαν με τον Λλορέντε στο 44'.
+1	Το	ήιτσ	ADP	ADP	Case=Acc|Gender=Neut|Number=Sing	2	det	_	_
+2	σκορ	όεσκο	INTJ	_	Foreign=Yes	5	obj	_	_
+3	του	όετο	ADP	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	4	det	_	_
+4	αγώνα	ω	PROPN	ADP	Case=Acc|Definite=Def|Gender=Fem|Number=Plur|PronType=Art	2	nmod	_	_
+5	άνοιξε	άνοιξε]DAP[	INTJ	_	[PAD]	0	root	_	_
+6	ο	ο]DAP[	PROPN	_	[PAD]	7	det	_	_
+7	Γουέν	όεουο	PROPN	ADP	Foreign=Yes	5	nsubj	_	_
+8	Ρούνι	όεούο	PROPN	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	7	flat	_	_
+9-10	στο	στο]DAP[	PROPN	_	[PAD]	_	_	_	_
+9	σ	ω	PRON	_	Foreign=Yes	12	case	_	_
+10	το	το]DAP[	PART	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	12	det	_	_
+11	22ο	22ο]DAP[	PROPN	_	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	12	amod	_	_
+12	λεπτό	λεπή	PROPN	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	5	obl	_	SpaceAfter=No
+13	,	,]DAP[	PART	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	17	punct	_	_
+14	ωστόσο	ωετόσαίνω	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	17	cc	_	_
+15	οι	ο	PART	_	Foreign=Yes	16	det	_	_
+16	φιλοξενούμενοι	φιλοξενόύεενο	PART	ADP	Foreign=Yes	17	nsubj	_	_
+17	ισοφάρισαν	ισοόάεισο	PROPN	_	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Act	5	conj	_	_
+18	με	ω	PROPN	_	[PAD]	20	case	_	_
+19	τον	ο	PROPN	ADP	NumType=Card	20	det	_	_
+20	Λλορέντε	λλοεένταίνω	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	17	obl	_	_
+21-22	στο	όεσο	PROPN	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	_	_	_	_
+21	σ	ω	PRON	_	Foreign=Yes	23	case	_	_
+22	το	όεο	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	23	det	_	_
+23	44'	όε4ο	SYM	_	Case=Nom|Gender=Neut|Number=Sing|Person=3|PronType=Rel	17	obl	_	SpaceAfter=No
+24	.	όεο	PROPN	ADP	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	5	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-5
+# text = Στο δεύτερο ημίχρονο, η Ατλέτικο πέτυχε δύο τέρματα με τους Όσκαρ ντε Μάρκος (71ο λεπτό) και Ικέρ Μουνιάιν (90ο λεπτό).
+1-2	Στο	ήιστσ	PART	ADP	[PAD]	_	_	_	_
+1	Σ	ω	PUNCT	DET	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	4	case	_	_
+2	το	ή	PUNCT	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	4	det	_	_
+3	δεύτερο	δεετεραίνω	PUNCT	ADP	Foreign=Yes	4	amod	_	_
+4	ημίχρονο	ημίχρο	ADV	ADP	Foreign=Yes	8	obl	_	SpaceAfter=No
+5	,	ή	PUNCT	ADP	Foreign=Yes	4	punct	_	_
+6	η	ω	DET	ADP	[PAD]	7	det	_	_
+7	Ατλέτικο	ατλέτο	_	ADP	Foreign=Yes	8	nsubj	_	_
+8	πέτυχε	όπετυο	ADJ	ADP	NumType=Card	0	root	_	_
+9	δύο	ώ	PROPN	ADP	NumType=Card	10	nummod	_	_
+10	τέρματα	όέεμαο	PROPN	ADP	NumType=Card	8	obj	_	_
+11	με	ω	PROPN	ADP	NumType=Card	13	case	_	_
+12	τους	τοή	DET	ADP	Foreign=Yes	13	det	_	_
+13	Όσκαρ	ύσι	SYM	ADP	Foreign=Yes	8	obl	_	_
+14	ντε	ω	PROPN	ADP	NumType=Card	13	flat	_	_
+15	Μάρκος	μάρκ	PROPN	_	Case=Gen|Gender=Masc|Number=Sing	13	flat	_	_
+16	(	εαίνω	SYM	_	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	18	punct	_	SpaceAfter=No
+17	71ο	όε7ο	PROPN	ADP	NumType=Card	18	amod	_	_
+18	λεπτό	λεπή	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	8	obl	_	SpaceAfter=No
+19	)	ομ)	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	18	punct	_	_
+20	και	όεκο	SYM	ADP	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	21	cc	_	_
+21	Ικέρ	όεικο	SYM	ADP	Foreign=Yes	8	conj	_	_
+22	Μουνιάιν	μόυειάο	SYM	_	NumType=Card	21	flat	_	_
+23	(	όεο	SYM	_	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	25	punct	_	SpaceAfter=No
+24	90ο	όε9ο	_	ADP	NumType=Card	25	amod	_	_
+25	λεπτό	λεπή	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	21	orphan	_	SpaceAfter=No
+26	)	όεο	PROPN	ADP	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	25	punct	_	SpaceAfter=No
+27	.	όεο	INTJ	ADP	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	8	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-6
+# text = Ωστόσο, ο Γουέν Ρούνι με πέναλτι μείωσε το σκορ για την Μάντσεστερ.
+1	Ωστόσο	ωετόσαίνω	DET	ADP	Foreign=Yes	8	cc	_	SpaceAfter=No
+2	,	ω	PART	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	1	punct	_	_
+3	ο	ω	PART	ADP	Foreign=Yes	4	det	_	_
+4	Γουέν	όεουο	PROPN	ADP	Foreign=Yes	8	nsubj	_	_
+5	Ρούνι	όεούο	PROPN	[PAD]	Foreign=Yes	4	flat	_	_
+6	με	όεο	INTJ	ADP	Case=Acc|Gender=Neut|Number=Sing	7	case	_	_
+7	πέναλτι	όέεαλο	INTJ	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	8	obl	_	_
+8	μείωσε	μείωσαίνω	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	0	root	_	_
+9	το	εταίνω	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	10	det	_	_
+10	σκορ	εσκοαίνω	PROPN	ADP	Foreign=Yes	8	obj	_	_
+11	για	ω	PROPN	ADP	[PAD]	13	case	_	_
+12	την	ετηαίνω	PROPN	ADP	Foreign=Yes	13	det	_	_
+13	Μάντσεστερ	μάντσεσο	_	ADP	Foreign=Yes	8	obl	_	SpaceAfter=No
+14	.	ω	PART	ADP	Foreign=Yes	8	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-7
+# text = Οι δύο αντίπαλοι θα ξανασυναντηθούν στις 15 Μαρτίου στο Στάδιο «Σαν Μαμές», με τους Ισπανούς να χρειάζονται νίκη και ισοπαλία και με τους κόκκινους διαβόλους να χρειάζονται νίκη με διαφορά δύο τερμάτων.
+1	Οι	ήιοσ	PART	ADP	[PAD]	3	det	_	_
+2	δύο	όεδο	PUNCT	[PAD]	Foreign=Yes	3	nummod	_	_
+3	αντίπαλοι	ανόίεαλο	PUNCT	ADP	Foreign=Yes	5	nsubj:pass	_	_
+4	θα	ήιθσ	PUNCT	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	5	aux	_	_
+5	ξανασυναντηθούν	ξανασυναότεθοο	AUX	_	Foreign=Yes	0	root	_	_
+6-7	στις	όεστο	SYM	ADP	[PAD]	_	_	_	_
+6	σ	ω	ADV	ADP	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	8	case	_	_
+7	τις	ω	SYM	ADP	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	8	det	_	_
+8	15	15]DAP[	ADV	ADP	Case=Nom|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	5	obl	_	_
+9	Μαρτίου	μαρτο	SYM	ADP	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	8	nmod	_	_
+10-11	στο	στο]DAP[	SYM	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	_	_	_	_
+10	σ	ω	[UNK]	PROPN	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	12	case	_	_
+11	το	το]DAP[	SYM	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	12	det	_	_
+12	Στάδιο	στάδ	PROPN	ADP	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Pass	5	obl	_	_
+13	«		SYM	ADP	[PAD]	14	punct	_	SpaceAfter=No
+14	Σαν	σ	SYM	ADP	Foreign=Yes	12	flat	_	_
+15	Μαμές	μαμ	SCONJ	ADP	Foreign=Yes	12	flat	_	SpaceAfter=No
+16	»	»]DAP[	PROPN	ADP	Foreign=Yes	15	punct	_	SpaceAfter=No
+17	,	,]DAP[	PART	ADP	_	20	punct	_	_
+18	με	εμαίνω	PROPN	ADP	[PAD]	20	case	_	_
+19	τους	τοή	PROPN	ADP	Foreign=Yes	20	det	_	_
+20	Ισπανούς	ιήπαιούσ	SYM	ADP	Foreign=Yes	5	obl	_	_
+21	να	εναίνω	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	22	aux	_	_
+22	χρειάζονται	χρειόζεντο	SYM	_	NumType=Card	20	acl	_	_
+23	νίκη	όενίο	PROPN	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	22	obj	_	_
+24	και	όεκο	PROPN	_	NumType=Card	25	cc	_	_
+25	ισοπαλία	ιόοεαλο	DET	ADP	Foreign=Yes	23	conj	_	_
+26	και	κή	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	30	cc	_	_
+27	με	εμαίνω	[UNK]	ADP	[PAD]	30	case	_	_
+28	τους	τοή	SYM	ADP	Foreign=Yes	30	det	_	_
+29	κόκκινους	κόκκινουσ]DAP[	SYM	ADP	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	30	amod	_	_
+30	διαβόλους	διαβόλουσ]DAP[	SYM	DET	Foreign=Yes	20	conj	_	_
+31	να	όεο	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	32	aux	_	_
+32	χρειάζονται	χρειόζεντο	SYM	_	NumType=Card	30	acl	_	_
+33	νίκη	όενίο	PROPN	ADP	NumType=Card	32	obj	_	_
+34	με	όεο	INTJ	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	35	case	_	_
+35	διαφορά	όιεφοο	SYM	ADP	[PAD]	33	nmod	_	_
+36	δύο	όεδο	SYM	[PAD]	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	37	nummod	_	_
+37	τερμάτων	τερμάτων]DAP[	PROPN	ADP	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	35	nmod	_	SpaceAfter=No
+38	.	.]DAP[	INTJ	ADP	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	5	punct	_	_
+
+# newdoc id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-1
+# text = Στον ποταμό Αχέροντα.
+1-2	Στον	στον_	PART	ADP	[PAD]	_	_	_	_
+1	Σ	ο	SYM	PROPN	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	3	case	_	_
+2	τον	τή	SYM	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	3	det	_	_
+3	ποταμό	ποτα	INTJ	ADP	_	0	root	_	_
+4	Αχέροντα	αχέροο	SYM	ADP	NumType=Card	3	flat	_	SpaceAfter=No
+5	.	ο	INTJ	ADP	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	3	punct	_	_
+
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-2
+# text = Ο ποταμός Αχέροντας αποτελούσε κατά τη μυθολογία το δρόμο μέσω του οποίου ο Άδης μετέφερε τις ψυχές στο βασίλειό του, στη λίμνη Αχερουσία.
+1	Ο	ο_	PART	ADP	[PAD]	2	det	_	_
+2	ποταμός	ποταο	INTJ	ADP	_	4	nsubj	_	_
+3	Αχέροντας	αχέροντή	PUNCT	ADP	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	2	flat	_	_
+4	αποτελούσε	αποόεεούο	SCONJ	_	_	0	root	_	_
+5	κατά	κατά]DAP[	ADJ	DET	[PAD]	7	case	_	_
+6	τη	τη]DAP[	INTJ	DET	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	7	det	_	_
+7	μυθολογία	μυθοεογίαίνω	PART	DET	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	4	obl	_	_
+8	το	το]DAP[	PART	_	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	9	det	_	_
+9	δρόμο	δρό	SYM	ADP	_	4	obj	_	_
+10	μέσω	όεμέο	SYM	_	[PAD]	12	case	_	_
+11	του	του]DAP[	SYM	ADP	[PAD]	12	det	_	_
+12	οποίου	όοεοίο	PART	ADP	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	15	obl	_	_
+13	ο	ύι	PART	ADP	[PAD]	14	det	_	_
+14	Άδης	άο	PUNCT	ADP	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	15	nsubj	_	_
+15	μετέφερε	μετέφεή	SYM	_	_	9	acl:relcl	_	_
+16	τις	τή	SYM	ADP	[PAD]	17	det	_	_
+17	ψυχές	ψυο	PART	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	15	obj	_	_
+18-19	στο	στο]DAP[	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	_	_	_	_
+18	σ	ω	[UNK]	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	20	case	_	_
+19	το	το]DAP[	PART	PROPN	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	20	det	_	_
+20	βασίλειό	βασίλο	INTJ	ADP	[PAD]	15	obl	_	_
+21	του	τή	SYM	DET	[PAD]	20	nmod	_	SpaceAfter=No
+22	,	ή	PART	ADP	[PAD]	25	punct	_	_
+23-24	στη	ω	SYM	ADV	[PAD]	_	_	_	_
+23	σ	ω	ADJ	_	Foreign=Yes	25	case	_	_
+24	τη	τη]DAP[	PART	PROPN	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	25	det	_	_
+25	λίμνη	λίμνη]DAP[	[UNK]	ADP	[PAD]	20	nmod	_	_
+26	Αχερουσία	αχερω	ADJ	ADP	Foreign=Yes	25	flat	_	SpaceAfter=No
+27	.	.]DAP[	PART	ADP	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	4	punct	_	_
+
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-3
+# text = Πηγάζει από τα ορεινά του Νομού Ιωαννίνων και έπειτα από διαδρομή 64 χιλιομέτρων εκβάλλει στο Ιόνιο Πέλαγος.
+1	Πηγάζει	πηω	INTJ	ADP	_	0	root	_	_
+2	από	ω	INTJ	PROPN	[PAD]	4	case	_	_
+3	τα	τα]DAP[	SYM	ADP	[PAD]	4	det	_	_
+4	ορεινά	ορει	SYM	ADP	[PAD]	1	obl	_	_
+5	του	τή	SYM	ADP	[PAD]	6	det	_	_
+6	Νομού	νοο	PROPN	ADP	NumType=Card	4	nmod	_	_
+7	Ιωαννίνων	ιωαννίνή	SYM	ADP	Foreign=Yes	6	nmod	_	_
+8	και	κ	INTJ	PROPN	[PAD]	14	cc	_	_
+9	έπειτα	έπώ	INTJ	ADP	[PAD]	14	advmod	_	_
+10	από	αή	INTJ	ADP	[PAD]	11	case	_	_
+11	διαδρομή	διαερομαίνω	INTJ	ADP	[PAD]	9	obl	_	_
+12	64		SYM	ADP	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	13	nummod	_	_
+13	χιλιομέτρων	χιλιόμετρο	PROPN	ADP	[PAD]	11	nmod	_	_
+14	εκβάλλει	εόβελλο	[UNK]	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	1	conj	_	_
+15-16	στο	στο]DAP[	INTJ	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	_	_	_	_
+15	σ	ο	[UNK]	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	18	case	_	_
+16	το	ο	PART	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	18	det	_	_
+17	Ιόνιο	ήιινισ	SYM	ADP	NumType=Card	18	amod	_	_
+18	Πέλαγος	πέλαγ	SYM	ADP	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	14	obl	_	SpaceAfter=No
+19	.	.]DAP[	INTJ	ADP	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	1	punct	_	_
+
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-4
+# text = Ο Αχέροντας διασχίζοντας την κοιλάδα που σχηματίζεται ανάμεσα στους ορεινούς όγκους της Δυτικής Ηπείρου διέρχεται από ένα στενό φαράγγι μεταξύ των βουνών Παραμυθίας και Σουλίου, το οποίο ονομάζεται "Στενά του Αχέροντα".
+1	Ο	ο_	PART	ADP	[PAD]	2	det	_	_
+2	Αχέροντας	αχέρονο	INTJ	_	Foreign=Yes	16	nsubj	_	_
+3	διασχίζοντας	διασχίζονο	INTJ	_	Foreign=Yes	16	advcl	_	_
+4	την	ύι	PART	PROPN	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	5	det	_	_
+5	κοιλάδα	κοιλάδα]DAP[	INTJ	ADP	[PAD]	3	obj	_	_
+6	που	πή	PART	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	7	nsubj:pass	_	_
+7	σχηματίζεται	σχηματίζετή	PROPN	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	5	acl:relcl	_	_
+8	ανάμεσα	ανάμεή	[UNK]	ADP	[PAD]	7	advmod	_	_
+9-10	στους	στοή	SYM	ADP	[PAD]	_	_	_	_
+9	σ	ο	[UNK]	_	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	12	case	_	_
+10	τους	τοή	SYM	PROPN	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	12	det	_	_
+11	ορεινούς	ορεινο	PROPN	NUM	Foreign=Yes	12	amod	_	_
+12	όγκους	όγκοή	ADP	ADP	[PAD]	8	obl	_	_
+13	της	τή	PROPN	ADP	[PAD]	15	det	_	_
+14	Δυτικής	δυτικήσ]DAP[	PROPN	ADP	Foreign=Yes	15	amod	_	_
+15	Ηπείρου	ηπείρου]DAP[	ADJ	ADP	Foreign=Yes	12	nmod	_	_
+16	διέρχεται	διέρχετή	INTJ	PROPN	_	0	root	_	_
+17	από	αή	INTJ	_	[PAD]	20	case	_	_
+18	ένα	ένα]DAP[	ADJ	ADP	[PAD]	20	det	_	_
+19	στενό	_	[UNK]	_	[PAD]	20	amod	_	_
+20	φαράγγι	φαράο	INTJ	PROPN	NumType=Card	16	obl	_	_
+21	μεταξύ	όμεταο	PROPN	_	[PAD]	23	case	_	_
+22	των	τή	SYM	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	23	det	_	_
+23	βουνών	όβευνο	SYM	ADP	NumType=Card	20	nmod	_	_
+24	Παραμυθίας	παραμυθο	SYM	_	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	23	nmod	_	_
+25	και	όεκο	SYM	PROPN	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	26	cc	_	_
+26	Σουλίου	σουλο	PUNCT	ADP	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	24	conj	_	SpaceAfter=No
+27	,	ύι	PART	ADP	[PAD]	30	punct	_	_
+28	το	το]DAP[	PART	_	[PAD]	29	det	_	_
+29	οποίο	ώ	PART	ADP	[PAD]	30	nsubj:pass	_	_
+30	ονομάζεται	ονομάύει	INTJ	_	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	20	acl:relcl	_	_
+31	"	εαίνω	INTJ	_	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Act	32	punct	_	SpaceAfter=No
+32	Στενά	στο	SYM	ADP	_	30	xcomp	_	_
+33	του	του]DAP[	SYM	PROPN	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	34	det	_	_
+34	Αχέροντα	αχέρονή	SYM	ADP	NumType=Card	32	nmod	_	SpaceAfter=No
+35	"	ή	SYM	_	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	32	punct	_	SpaceAfter=No
+36	.	.]DAP[	INTJ	ADP	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	16	punct	_	_
+
diff --git a/tests/testdata/datasets/el_train.conllu b/tests/testdata/datasets/el_train.conllu
new file mode 100644
index 0000000..49ece6c
--- /dev/null
+++ b/tests/testdata/datasets/el_train.conllu
@@ -0,0 +1,270 @@
+# newdoc id = gdt-20120309-elwikinews-5160
+# sent_id = gdt-20120309-elwikinews-5160-1
+# text = Η Μάντσεστερ Γιουνάιτεντ ηττήθηκε από την Ατλέτικο Μπιλμπάο με σκορ 2:3
+1	Η	ο	DET	DET	Case=Nom|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	2	det	_	_
+2	Μάντσεστερ	Μάντσεστερ	X	X	Foreign=Yes	4	nsubj	_	_
+3	Γιουνάιτεντ	Γιουνάιτεντ	X	X	Foreign=Yes	2	flat	_	_
+4	ηττήθηκε	ηττώμαι	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Pass	0	root	_	_
+5	από	από	ADP	ADP	_	7	case	_	_
+6	την	ο	DET	DET	Case=Acc|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	7	det	_	_
+7	Ατλέτικο	Ατλέτικο	X	X	Foreign=Yes	4	obl:agent	_	_
+8	Μπιλμπάο	Μπιλμπάο	X	X	Foreign=Yes	7	flat	_	_
+9	με	με	ADP	ADP	_	10	case	_	_
+10	σκορ	σκορ	X	X	Foreign=Yes	4	obl	_	_
+11	2:3	2:3	NUM	NUM	NumType=Card	10	nmod	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-3
+# text = Χθες, η Μάντσεστερ Γιουνάιτεντ ηττήθηκε με σκορ 2:3 από την Ατλέτικο Μπιλμπάο, στα πλαίσια της φάσης των 16 του Γιουρόπα Λιγκ 2011-2012.
+1	Χθες	χθες	ADV	ADV	_	6	advmod	_	SpaceAfter=No
+2	,	,	PUNCT	PUNCT	_	1	punct	_	_
+3	η	ο	DET	DET	Case=Nom|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	4	det	_	_
+4	Μάντσεστερ	Μάντσεστερ	X	X	Foreign=Yes	6	nsubj	_	_
+5	Γιουνάιτεντ	Γιουνάιτεντ	X	X	Foreign=Yes	4	flat	_	_
+6	ηττήθηκε	ηττώμαι	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Pass	0	root	_	_
+7	με	με	ADP	ADP	_	8	case	_	_
+8	σκορ	σκορ	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Plur	6	obl	_	_
+9	2:3	2:3	NUM	NUM	NumType=Card	8	nmod	_	_
+10	από	από	ADP	ADP	_	12	case	_	_
+11	την	ο	DET	DET	Case=Acc|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	12	det	_	_
+12	Ατλέτικο	Ατλέτικο	X	X	Foreign=Yes	6	obl:agent	_	_
+13	Μπιλμπάο	Μπιλμπάο	X	X	Foreign=Yes	12	flat	_	SpaceAfter=No
+14	,	,	PUNCT	PUNCT	_	17	punct	_	_
+15-16	στα	_	_	_	_	_	_	_	_
+15	σ	σε	ADP	ADP	_	17	case	_	_
+16	τα	ο	DET	DET	Case=Acc|Gender=Neut|Number=Plur	17	det	_	_
+17	πλαίσια	πλαίσιο	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Plur	6	obl	_	_
+18	της	ο	DET	DET	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	19	det	_	_
+19	φάσης	φάση	NOUN	NOUN	Case=Gen|Gender=Fem|Number=Sing	17	nmod	_	_
+20	των	ο	DET	DET	Case=Gen|Definite=Def|Gender=Fem|Number=Plur|PronType=Art	21	det	_	_
+21	16	16	NUM	NUM	NumType=Card	19	nmod	_	_
+22	του	ο	DET	DET	Case=Gen|Definite=Def|Gender=Neut|Number=Sing|PronType=Art	23	det	_	_
+23	Γιουρόπα	Γιουρόπα	X	X	Foreign=Yes	21	nmod	_	_
+24	Λιγκ	Λιγκ	X	X	Foreign=Yes	23	flat	_	_
+25	2011-2012	2011-2012	NUM	NUM	NumType=Card	23	nmod	_	SpaceAfter=No
+26	.	.	PUNCT	PUNCT	_	6	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-4
+# text = Το σκορ του αγώνα άνοιξε ο Γουέν Ρούνι στο 22ο λεπτό, ωστόσο οι φιλοξενούμενοι ισοφάρισαν με τον Λλορέντε στο 44'.
+1	Το	ο	DET	DET	Case=Acc|Definite=Def|Gender=Neut|Number=Sing|PronType=Art	2	det	_	_
+2	σκορ	σκορ	X	X	Foreign=Yes	5	obj	_	_
+3	του	ο	DET	DET	Case=Gen|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	4	det	_	_
+4	αγώνα	αγώνας	NOUN	NOUN	Case=Gen|Gender=Masc|Number=Sing	2	nmod	_	_
+5	άνοιξε	ανοίγω	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Act	0	root	_	_
+6	ο	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	7	det	_	_
+7	Γουέν	Γουέν	X	X	Foreign=Yes	5	nsubj	_	_
+8	Ρούνι	Ρούνι	X	X	Foreign=Yes	7	flat	_	_
+9-10	στο	_	_	_	_	_	_	_	_
+9	σ	σε	ADP	ADP	_	12	case	_	_
+10	το	ο	DET	DET	Case=Acc|Gender=Neut|Number=Sing	12	det	_	_
+11	22ο	22ος	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Sing|NumType=Ord	12	amod	_	_
+12	λεπτό	λεπτό	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	5	obl	_	SpaceAfter=No
+13	,	,	PUNCT	PUNCT	_	17	punct	_	_
+14	ωστόσο	ωστόσο	CCONJ	CCONJ	_	17	cc	_	_
+15	οι	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Plur|PronType=Art	16	det	_	_
+16	φιλοξενούμενοι	φιλοξενούμενος	ADJ	ADJ	Case=Nom|Gender=Masc|Number=Plur	17	nsubj	_	_
+17	ισοφάρισαν	ισοφαρίζω	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin|Voice=Act	5	conj	_	_
+18	με	με	ADP	ADP	_	20	case	_	_
+19	τον	ο	DET	DET	Case=Acc|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	20	det	_	_
+20	Λλορέντε	Λλορέντε	X	X	Foreign=Yes	17	obl	_	_
+21-22	στο	_	_	_	_	_	_	_	_
+21	σ	σε	ADP	ADP	_	23	case	_	_
+22	το	ο	DET	DET	Case=Acc|Gender=Neut|Number=Sing	23	det	_	_
+23	44'	44'	NUM	NUM	NumType=Card	17	obl	_	SpaceAfter=No
+24	.	.	PUNCT	PUNCT	_	5	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-5
+# text = Στο δεύτερο ημίχρονο, η Ατλέτικο πέτυχε δύο τέρματα με τους Όσκαρ ντε Μάρκος (71ο λεπτό) και Ικέρ Μουνιάιν (90ο λεπτό).
+1-2	Στο	_	_	_	_	_	_	_	_
+1	Σ	σε	ADP	ADP	_	4	case	_	_
+2	το	ο	DET	DET	Case=Acc|Gender=Neut|Number=Sing	4	det	_	_
+3	δεύτερο	δεύτερος	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Sing|NumType=Ord	4	amod	_	_
+4	ημίχρονο	ημίχρονο	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	8	obl	_	SpaceAfter=No
+5	,	,	PUNCT	PUNCT	_	4	punct	_	_
+6	η	ο	DET	DET	Case=Nom|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	7	det	_	_
+7	Ατλέτικο	Ατλέτικο	X	X	Foreign=Yes	8	nsubj	_	_
+8	πέτυχε	πετυχαίνω	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Act	0	root	_	_
+9	δύο	δύο	NUM	NUM	Case=Acc|Gender=Neut|Number=Plur|NumType=Card	10	nummod	_	_
+10	τέρματα	τέρμα	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Plur	8	obj	_	_
+11	με	με	ADP	ADP	_	13	case	_	_
+12	τους	ο	DET	DET	Case=Acc|Definite=Def|Gender=Masc|Number=Plur|PronType=Art	13	det	_	_
+13	Όσκαρ	Όσκαρ	X	X	Foreign=Yes	8	obl	_	_
+14	ντε	ντε	X	X	Foreign=Yes	13	flat	_	_
+15	Μάρκος	Μάρκος	X	X	Foreign=Yes	13	flat	_	_
+16	(	(	PUNCT	PUNCT	_	18	punct	_	SpaceAfter=No
+17	71ο	71ος	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Sing|NumType=Ord	18	amod	_	_
+18	λεπτό	λεπτό	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	8	obl	_	SpaceAfter=No
+19	)	)	PUNCT	PUNCT	_	18	punct	_	_
+20	και	και	CCONJ	CCONJ	_	21	cc	_	_
+21	Ικέρ	Ικέρ	X	X	Foreign=Yes	8	conj	_	_
+22	Μουνιάιν	Μουνιάιν	X	X	Foreign=Yes	21	flat	_	_
+23	(	(	PUNCT	PUNCT	_	25	punct	_	SpaceAfter=No
+24	90ο	90ος	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Sing|NumType=Ord	25	amod	_	_
+25	λεπτό	λεπτό	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	21	orphan	_	SpaceAfter=No
+26	)	)	PUNCT	PUNCT	_	25	punct	_	SpaceAfter=No
+27	.	.	PUNCT	PUNCT	_	8	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-6
+# text = Ωστόσο, ο Γουέν Ρούνι με πέναλτι μείωσε το σκορ για την Μάντσεστερ.
+1	Ωστόσο	ωστόσο	CCONJ	CCONJ	_	8	cc	_	SpaceAfter=No
+2	,	,	PUNCT	PUNCT	_	1	punct	_	_
+3	ο	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	4	det	_	_
+4	Γουέν	Γουέν	X	X	Foreign=Yes	8	nsubj	_	_
+5	Ρούνι	Ρούνι	X	X	Foreign=Yes	4	flat	_	_
+6	με	με	ADP	ADP	_	7	case	_	_
+7	πέναλτι	πέναλτι	X	X	Foreign=Yes	8	obl	_	_
+8	μείωσε	μειώνω	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Act	0	root	_	_
+9	το	ο	DET	DET	Case=Acc|Definite=Def|Gender=Neut|Number=Sing|PronType=Art	10	det	_	_
+10	σκορ	σκορ	X	X	Foreign=Yes	8	obj	_	_
+11	για	για	ADP	ADP	_	13	case	_	_
+12	την	ο	DET	DET	Case=Acc|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	13	det	_	_
+13	Μάντσεστερ	Μάντσεστερ	X	X	Foreign=Yes	8	obl	_	SpaceAfter=No
+14	.	.	PUNCT	PUNCT	_	8	punct	_	_
+
+# sent_id = gdt-20120309-elwikinews-5160-7
+# text = Οι δύο αντίπαλοι θα ξανασυναντηθούν στις 15 Μαρτίου στο Στάδιο «Σαν Μαμές», με τους Ισπανούς να χρειάζονται νίκη και ισοπαλία και με τους κόκκινους διαβόλους να χρειάζονται νίκη με διαφορά δύο τερμάτων.
+1	Οι	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Plur|PronType=Art	3	det	_	_
+2	δύο	δύο	NUM	NUM	Case=Nom|Gender=Masc|Number=Plur|NumType=Card	3	nummod	_	_
+3	αντίπαλοι	αντίπαλος	ADJ	ADJ	Case=Nom|Gender=Masc|Number=Plur	5	nsubj:pass	_	_
+4	θα	θα	AUX	AUX	_	5	aux	_	_
+5	ξανασυναντηθούν	ξανασυναντώ	VERB	VERB	Aspect=Perf|Mood=Ind|Number=Plur|Person=3|VerbForm=Fin|Voice=Pass	0	root	_	_
+6-7	στις	_	_	_	_	_	_	_	_
+6	σ	σε	ADP	ADP	_	8	case	_	_
+7	τις	ο	DET	DET	Case=Acc|Gender=Fem|Number=Plur	8	det	_	_
+8	15	15	NUM	NUM	NumType=Card	5	obl	_	_
+9	Μαρτίου	Μάρτιος	PROPN	PROPN	Case=Gen|Gender=Masc|Number=Sing	8	nmod	_	_
+10-11	στο	_	_	_	_	_	_	_	_
+10	σ	σε	ADP	ADP	_	12	case	_	_
+11	το	ο	DET	DET	Case=Acc|Gender=Neut|Number=Sing	12	det	_	_
+12	Στάδιο	στάδιο	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	5	obl	_	_
+13	«	«	PUNCT	PUNCT	_	14	punct	_	SpaceAfter=No
+14	Σαν	Σαν	X	X	Foreign=Yes	12	flat	_	_
+15	Μαμές	Μαμές	X	X	Foreign=Yes	12	flat	_	SpaceAfter=No
+16	»	»	PUNCT	PUNCT	_	15	punct	_	SpaceAfter=No
+17	,	,	PUNCT	PUNCT	_	20	punct	_	_
+18	με	με	ADP	ADP	_	20	case	_	_
+19	τους	ο	DET	DET	Case=Acc|Definite=Def|Gender=Masc|Number=Plur|PronType=Art	20	det	_	_
+20	Ισπανούς	Ισπανός	PROPN	PROPN	Case=Acc|Gender=Masc|Number=Plur	5	obl	_	_
+21	να	να	AUX	AUX	_	22	aux	_	_
+22	χρειάζονται	χρειάζομαι	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Plur|Person=3|Tense=Pres|VerbForm=Fin|Voice=Pass	20	acl	_	_
+23	νίκη	νίκη	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	22	obj	_	_
+24	και	και	CCONJ	CCONJ	_	25	cc	_	_
+25	ισοπαλία	ισοπαλία	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	23	conj	_	_
+26	και	και	CCONJ	CCONJ	_	30	cc	_	_
+27	με	με	ADP	ADP	_	30	case	_	_
+28	τους	ο	DET	DET	Case=Acc|Definite=Def|Gender=Masc|Number=Plur|PronType=Art	30	det	_	_
+29	κόκκινους	κόκκινος	ADJ	ADJ	Case=Acc|Gender=Masc|Number=Plur	30	amod	_	_
+30	διαβόλους	διάβολος	NOUN	NOUN	Case=Acc|Gender=Masc|Number=Plur	20	conj	_	_
+31	να	να	AUX	AUX	_	32	aux	_	_
+32	χρειάζονται	χρειάζομαι	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Plur|Person=3|Tense=Pres|VerbForm=Fin|Voice=Pass	30	acl	_	_
+33	νίκη	νίκη	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	32	obj	_	_
+34	με	με	ADP	ADP	_	35	case	_	_
+35	διαφορά	διαφορά	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	33	nmod	_	_
+36	δύο	δύο	NUM	NUM	Case=Gen|Gender=Neut|Number=Plur|NumType=Card	37	nummod	_	_
+37	τερμάτων	τέρμα	NOUN	NOUN	Case=Gen|Gender=Neut|Number=Plur	35	nmod	_	SpaceAfter=No
+38	.	.	PUNCT	PUNCT	_	5	punct	_	_
+
+# newdoc id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-1
+# text = Στον ποταμό Αχέροντα.
+1-2	Στον	_	_	_	_	_	_	_	_
+1	Σ	σε	ADP	ADP	_	3	case	_	_
+2	τον	ο	DET	DET	Case=Acc|Gender=Masc|Number=Sing	3	det	_	_
+3	ποταμό	ποταμός	NOUN	NOUN	Case=Acc|Gender=Masc|Number=Sing	0	root	_	_
+4	Αχέροντα	Αχέροντας	PROPN	PROPN	Case=Acc|Gender=Neut|Number=Sing	3	flat	_	SpaceAfter=No
+5	.	.	PUNCT	PUNCT	_	3	punct	_	_
+
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-2
+# text = Ο ποταμός Αχέροντας αποτελούσε κατά τη μυθολογία το δρόμο μέσω του οποίου ο Άδης μετέφερε τις ψυχές στο βασίλειό του, στη λίμνη Αχερουσία.
+1	Ο	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	2	det	_	_
+2	ποταμός	ποταμός	NOUN	NOUN	Case=Nom|Gender=Masc|Number=Sing	4	nsubj	_	_
+3	Αχέροντας	Αχέροντας	PROPN	PROPN	Case=Nom|Gender=Masc|Number=Sing	2	flat	_	_
+4	αποτελούσε	αποτελώ	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Act	0	root	_	_
+5	κατά	κατά	ADP	ADP	_	7	case	_	_
+6	τη	ο	DET	DET	Case=Acc|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	7	det	_	_
+7	μυθολογία	μυθολογία	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	4	obl	_	_
+8	το	ο	DET	DET	Case=Acc|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	9	det	_	_
+9	δρόμο	δρόμος	NOUN	NOUN	Case=Acc|Gender=Masc|Number=Sing	4	obj	_	_
+10	μέσω	μέσω	ADP	ADP	_	12	case	_	_
+11	του	ο	DET	DET	Case=Gen|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	12	det	_	_
+12	οποίου	οποίος	PRON	PRON	Case=Gen|Gender=Masc|Number=Sing|Person=3|PronType=Rel	15	obl	_	_
+13	ο	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	14	det	_	_
+14	Άδης	άδης	PROPN	PROPN	Case=Nom|Gender=Masc|Number=Sing	15	nsubj	_	_
+15	μετέφερε	μεταφέρω	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin|Voice=Act	9	acl:relcl	_	_
+16	τις	ο	DET	DET	Case=Acc|Definite=Def|Gender=Fem|Number=Plur|PronType=Art	17	det	_	_
+17	ψυχές	ψυχή	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Plur	15	obj	_	_
+18-19	στο	_	_	_	_	_	_	_	_
+18	σ	σε	ADP	ADP	_	20	case	_	_
+19	το	ο	DET	DET	Case=Acc|Gender=Neut|Number=Sing	20	det	_	_
+20	βασίλειό	βασίλειο	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	15	obl	_	_
+21	του	μου	PRON	PRON	Case=Gen|Gender=Masc|Number=Sing|Person=3|Poss=Yes|PronType=Prs	20	nmod	_	SpaceAfter=No
+22	,	,	PUNCT	PUNCT	_	25	punct	_	_
+23-24	στη	_	_	_	_	_	_	_	_
+23	σ	σε	ADP	ADP	_	25	case	_	_
+24	τη	ο	DET	DET	Case=Acc|Gender=Fem|Number=Sing	25	det	_	_
+25	λίμνη	λίμνη	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	20	nmod	_	_
+26	Αχερουσία	Αχερουσία	PROPN	PROPN	Case=Acc|Gender=Fem|Number=Sing	25	flat	_	SpaceAfter=No
+27	.	.	PUNCT	PUNCT	_	4	punct	_	_
+
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-3
+# text = Πηγάζει από τα ορεινά του Νομού Ιωαννίνων και έπειτα από διαδρομή 64 χιλιομέτρων εκβάλλει στο Ιόνιο Πέλαγος.
+1	Πηγάζει	πηγάζω	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Act	0	root	_	_
+2	από	από	ADP	ADP	_	4	case	_	_
+3	τα	ο	DET	DET	Case=Acc|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	4	det	_	_
+4	ορεινά	ορεινός	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Plur	1	obl	_	_
+5	του	ο	DET	DET	Case=Gen|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	6	det	_	_
+6	Νομού	νομός	NOUN	NOUN	Case=Gen|Gender=Masc|Number=Sing	4	nmod	_	_
+7	Ιωαννίνων	Ιωάννινα	PROPN	PROPN	Case=Gen|Gender=Neut|Number=Plur	6	nmod	_	_
+8	και	και	CCONJ	CCONJ	_	14	cc	_	_
+9	έπειτα	έπειτα	ADV	ADV	_	14	advmod	_	_
+10	από	από	ADP	ADP	_	11	case	_	_
+11	διαδρομή	διαδρομή	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	9	obl	_	_
+12	64	64	NUM	NUM	NumType=Card	13	nummod	_	_
+13	χιλιομέτρων	χιλιόμετρο	NOUN	NOUN	Case=Gen|Gender=Neut|Number=Plur	11	nmod	_	_
+14	εκβάλλει	εκβάλλω	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Act	1	conj	_	_
+15-16	στο	_	_	_	_	_	_	_	_
+15	σ	σε	ADP	ADP	_	18	case	_	_
+16	το	ο	DET	DET	Case=Acc|Gender=Neut|Number=Sing	18	det	_	_
+17	Ιόνιο	Ιόνιο	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Sing	18	amod	_	_
+18	Πέλαγος	πέλαγος	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	14	obl	_	SpaceAfter=No
+19	.	.	PUNCT	PUNCT	_	1	punct	_	_
+
+# sent_id = gdt-2005XXXX-ert-tourism_menoume_ellada_axerontas-4
+# text = Ο Αχέροντας διασχίζοντας την κοιλάδα που σχηματίζεται ανάμεσα στους ορεινούς όγκους της Δυτικής Ηπείρου διέρχεται από ένα στενό φαράγγι μεταξύ των βουνών Παραμυθίας και Σουλίου, το οποίο ονομάζεται "Στενά του Αχέροντα".
+1	Ο	ο	DET	DET	Case=Nom|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	2	det	_	_
+2	Αχέροντας	Αχέροντας	PROPN	PROPN	Case=Nom|Gender=Masc|Number=Sing	16	nsubj	_	_
+3	διασχίζοντας	διασχίζω	VERB	VERB	Aspect=Imp|VerbForm=Conv|Voice=Act	16	advcl	_	_
+4	την	ο	DET	DET	Case=Acc|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	5	det	_	_
+5	κοιλάδα	κοιλάδα	NOUN	NOUN	Case=Acc|Gender=Fem|Number=Sing	3	obj	_	_
+6	που	που	PRON	PRON	Case=Nom|Gender=Fem|Number=Sing|Person=3|PronType=Rel	7	nsubj:pass	_	_
+7	σχηματίζεται	σχηματίζω	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Pass	5	acl:relcl	_	_
+8	ανάμεσα	ανάμεσα	ADV	ADV	_	7	advmod	_	_
+9-10	στους	_	_	_	_	_	_	_	_
+9	σ	σε	ADP	ADP	_	12	case	_	_
+10	τους	ο	DET	DET	Case=Acc|Gender=Masc|Number=Plur	12	det	_	_
+11	ορεινούς	ορεινός	ADJ	ADJ	Case=Acc|Gender=Masc|Number=Plur	12	amod	_	_
+12	όγκους	όγκος	NOUN	NOUN	Case=Acc|Gender=Masc|Number=Plur	8	obl	_	_
+13	της	ο	DET	DET	Case=Gen|Definite=Def|Gender=Fem|Number=Sing|PronType=Art	15	det	_	_
+14	Δυτικής	δυτικός	ADJ	ADJ	Case=Gen|Gender=Fem|Number=Sing	15	amod	_	_
+15	Ηπείρου	Ήπειρος	PROPN	PROPN	Case=Gen|Gender=Fem|Number=Sing	12	nmod	_	_
+16	διέρχεται	διέρχομαι	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Pass	0	root	_	_
+17	από	από	ADP	ADP	_	20	case	_	_
+18	ένα	ένας	DET	DET	Case=Acc|Definite=Ind|Gender=Neut|Number=Sing|PronType=Art	20	det	_	_
+19	στενό	στενός	ADJ	ADJ	Case=Acc|Gender=Neut|Number=Sing	20	amod	_	_
+20	φαράγγι	φαράγγι	NOUN	NOUN	Case=Acc|Gender=Neut|Number=Sing	16	obl	_	_
+21	μεταξύ	μεταξύ	ADP	ADP	_	23	case	_	_
+22	των	ο	DET	DET	Case=Gen|Definite=Def|Gender=Neut|Number=Plur|PronType=Art	23	det	_	_
+23	βουνών	βουνό	NOUN	NOUN	Case=Gen|Gender=Neut|Number=Plur	20	nmod	_	_
+24	Παραμυθίας	παραμυθία	PROPN	PROPN	Case=Gen|Gender=Fem|Number=Sing	23	nmod	_	_
+25	και	και	CCONJ	CCONJ	_	26	cc	_	_
+26	Σουλίου	Σούλι	PROPN	PROPN	Case=Gen|Gender=Neut|Number=Sing	24	conj	_	SpaceAfter=No
+27	,	,	PUNCT	PUNCT	_	30	punct	_	_
+28	το	ο	DET	DET	Case=Nom|Definite=Def|Gender=Neut|Number=Sing|PronType=Art	29	det	_	_
+29	οποίο	οποίος	PRON	PRON	Case=Nom|Gender=Neut|Number=Sing|Person=3|PronType=Rel	30	nsubj:pass	_	_
+30	ονομάζεται	ονομάζω	VERB	VERB	Aspect=Imp|Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin|Voice=Pass	20	acl:relcl	_	_
+31	"	"	PUNCT	PUNCT	_	32	punct	_	SpaceAfter=No
+32	Στενά	στενά	NOUN	NOUN	Case=Nom|Gender=Neut|Number=Plur	30	xcomp	_	_
+33	του	ο	DET	DET	Case=Gen|Definite=Def|Gender=Masc|Number=Sing|PronType=Art	34	det	_	_
+34	Αχέροντα	Αχέροντας	PROPN	PROPN	Case=Gen|Gender=Masc|Number=Sing	32	nmod	_	SpaceAfter=No
+35	"	"	PUNCT	PUNCT	_	32	punct	_	SpaceAfter=No
+36	.	.	PUNCT	PUNCT	_	16	punct	_	_
+
diff --git a/tests/testdata/datasets/en_expected.conllu b/tests/testdata/datasets/en_expected.conllu
new file mode 100644
index 0000000..4b11596
--- /dev/null
+++ b/tests/testdata/datasets/en_expected.conllu
@@ -0,0 +1,246 @@
+# newdoc id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0001
+# newpar id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-p0001
+# text = Al-Zaman : American forces killed Shaikh Abdullah al-Ani, the preacher at the mosque in the town of Qaim, near the Syrian border.
+1	Al	al_	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	0	root	0:root	SpaceAfter=No
+2	-		PROPN	-LRB-	_	3	punct	3:punct	SpaceAfter=No
+3	Zaman	zbn	ADJ	-LRB-	_	1	flat	1:flat	_
+4	:	we	PUNCT	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	7	punct	7:punct	_
+5	American	americbe	DET	VB	_	6	amod	6:amod	_
+6	forces	forci	SYM	VB	Degree=Pos|NumForm=Word|NumType=Ord	7	nsubj	7:nsubj	_
+7	killed	kil	PROPN	-LRB-	_	1	parataxis	1:parataxis	_
+8	Shaikh	sha	PUNCT	-LRB-	VerbForm=Fin	7	obj	7:obj	_
+9	Abdullah	abdul	SYM	-LRB-	_	8	flat	8:flat	_
+10	al		PROPN	[PAD]	_	8	flat	8:flat	SpaceAfter=No
+11	-		PROPN	[PAD]	VerbForm=Fin	12	punct	12:punct	SpaceAfter=No
+12	Ani	ani]KNU[	SYM	_	Case=Acc|Number=Plur|Person=1|PronType=Prs	8	flat	8:flat	SpaceAfter=No
+13	,	i	PROPN	VB	VerbForm=Fin	15	punct	15:punct	_
+14	the	ti	PART	-LRB-	VerbForm=Fin	15	det	15:det	_
+15	preacher	preachi	PUNCT	-LRB-	_	8	appos	8:appos	_
+16	at	we	PUNCT	VB	Degree=Pos|NumForm=Word|NumType=Ord	18	case	18:case	_
+17	the	ti	PART	IN	Degree=Pos|NumForm=Word|NumType=Ord	18	det	18:det	_
+18	mosque	mosque]DAP[	INTJ	-LRB-	_	15	nmod	15:nmod:at	_
+19	in	i	SCONJ	-LRB-	VerbForm=Fin	21	case	21:case	_
+20	the	ti	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	21	det	21:det	_
+21	town	town	PROPN	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	18	nmod	18:nmod:in	_
+22	of	i	PART	-LRB-	VerbForm=Fin	23	case	23:case	_
+23	Qaim	qai	PROPN	-LRB-	VerbForm=Fin	21	nmod	21:nmod:of	SpaceAfter=No
+24	,	i	PROPN	-LRB-	VerbForm=Fin	28	punct	28:punct	_
+25	near	newe	ADJ	NNP	Tense=Pres|VerbForm=Part	28	case	28:case	_
+26	the	ti	PART	-LRB-	Tense=Pres|VerbForm=Part	28	det	28:det	_
+27	Syrian	syri	PROPN	PRP	VerbForm=Fin	28	amod	28:amod	_
+28	border	bordwe	PROPN	-LRB-	VerbForm=Fin	21	nmod	21:nmod:near	SpaceAfter=No
+29	.	i	INTJ	-LRB-	_	1	punct	1:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0002
+# text = [This killing of a respected cleric will be causing us trouble for years to come.]
+1	[	[_	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	10	punct	10:punct	SpaceAfter=No
+2	This	th	SCONJ	IN	Tense=Pres|VerbForm=Part	3	det	3:det	_
+3	killing	killi	[UNK]	CC	_	10	nsubj	10:nsubj	_
+4	of	we	SYM	IN	_	7	case	7:case	_
+5	a	i	PROPN	VB	Number=Plur	7	det	7:det	_
+6	respected	respecti	PART	RB	Degree=Pos|NumForm=Word|NumType=Ord	7	amod	7:amod	_
+7	cleric	cley	INTJ	VB	Case=Acc|Number=Plur|Person=1|PronType=Prs	3	nmod	3:nmod:of	_
+8	will	w	INTJ	VBD	Tense=Pres|VerbForm=Part	10	aux	10:aux	_
+9	be		DET	IN	_	10	aux	10:aux	_
+10	causing	causiwe	ADJ	VBD	VerbForm=Fin	0	root	0:root	_
+11	us	i	SYM	VBN	Case=Nom|Person=2|PronType=Prs	10	iobj	10:iobj	_
+12	trouble	troubwe	PART	VB	Number=Sing|PronType=Dem	10	obj	10:obj	_
+13	for	fi	PROPN	VBD	PronType=Rel	14	case	14:case	_
+14	years	yeabe	[UNK]	-LRB-	_	10	obl	10:obl:for	_
+15	to	be	DET	VBD	Tense=Pres|VerbForm=Part	16	mark	16:mark	_
+16	come	coi	PUNCT	VBD	Number=Plur	14	acl	14:acl:to	SpaceAfter=No
+17	.	we	PART	VBD	Definite=Def|PronType=Art	10	punct	10:punct	SpaceAfter=No
+18	]		SYM	-LRB-	Case=Acc|Number=Plur|Person=1|PronType=Prs	10	punct	10:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0003
+# text = DPA: Iraqi authorities announced that they had busted up 3 terrorist cells operating in Baghdad.
+1	DPA	dwe	ADP	-LRB-	_	0	root	0:root	SpaceAfter=No
+2	:	we	ADJ	-LRB-	_	5	punct	5:punct	_
+3	Iraqi	irawe	PROPN	VB	VerbForm=Fin	4	amod	4:amod	_
+4	authorities	authoritii	SYM	VB	Degree=Pos|NumForm=Word|NumType=Ord	5	nsubj	5:nsubj	_
+5	announced	announcbe	ADJ	VBD	Degree=Pos|NumForm=Word|NumType=Ord	1	parataxis	1:parataxis	_
+6	that	thwe	ADJ	-LRB-	Tense=Pres|VerbForm=Part	9	mark	9:mark	_
+7	they	thi	ADJ	VB	Case=Nom|Person=2|PronType=Prs	9	nsubj	9:nsubj	_
+8	had	bd	SYM	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	9	aux	9:aux	_
+9	busted	bustbe	SYM	VB	_	5	ccomp	5:ccomp	_
+10	up	be	SYM	CC	_	9	compound:prt	9:compound:prt	_
+11	3	be	SYM	VB	_	13	nummod	13:nummod	_
+12	terrorist	terroribe	PROPN	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	13	amod	13:amod	_
+13	cells	cells]DAP[	PUNCT	CC	Degree=Pos|NumForm=Word|NumType=Ord	9	obj	9:obj	_
+14	operating	operatiwe	ADJ	CC	Degree=Pos|NumForm=Word|NumType=Ord	13	acl	13:acl	_
+15	in	be	PROPN	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	16	case	16:case	_
+16	Baghdad	baghdwe	ADJ	-LRB-	VerbForm=Fin	14	obl	14:obl:in	SpaceAfter=No
+17	.	i	PART	-LRB-	Case=Acc|Number=Plur|Person=1|PronType=Prs	1	punct	1:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0004
+# text = Two of them were being run by 2 officials of the Ministry of the Interior!
+1	Two	two_	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	6	nsubj:pass	6:nsubj:pass	_
+2	of	we	SYM	VBD	_	3	case	3:case	_
+3	them	thwe	SYM	-LRB-	_	1	nmod	1:nmod:of	_
+4	were	w	SCONJ	VBG	Degree=Pos|NumForm=Word|NumType=Ord	6	aux	6:aux	_
+5	being	bei	ADJ	PRP	Degree=Pos|NumForm=Word|NumType=Ord	6	aux:pass	6:aux:pass	_
+6	run	run	PROPN	VBD	Degree=Pos|NumForm=Word|NumType=Ord	0	root	0:root	_
+7	by	be	PART	VBD	Degree=Pos|NumForm=Word|NumType=Ord	9	case	9:case	_
+8	2	be	SYM	VBD	Number=Plur	9	nummod	9:nummod	_
+9	officials	officials]DAP[	SYM	NNS	Degree=Pos|NumForm=Word|NumType=Ord	6	obl:agent	6:obl:agent	_
+10	of	we	PART	VBD	VerbForm=Fin	12	case	12:case	_
+11	the	ti	PART	VBD	VerbForm=Fin	12	det	12:det	_
+12	Ministry	ministwe	PROPN	VBP	VerbForm=Fin	9	nmod	9:nmod:of	_
+13	of	be	INTJ	NNS	VerbForm=Fin	15	case	15:case	_
+14	the	ti	PART	-LRB-	VerbForm=Fin	15	det	15:det	_
+15	Interior	inter	[UNK]	-LRB-	VerbForm=Fin	12	nmod	12:nmod:of	SpaceAfter=No
+16	!		PUNCT	VBD	VerbForm=Fin	6	punct	6:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0005
+# text = The MoI in Iraq is equivalent to the US FBI, so this would be like having J. Edgar Hoover unwittingly employ at a high level members of the Weathermen bombers back in the 1960s.
+1	The	the_	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	2	det	2:det	_
+2	MoI	m	DET	VB	_	6	nsubj	6:nsubj	_
+3	in	be	SCONJ	VBD	_	4	case	4:case	_
+4	Iraq	irbe	SYM	-LRB-	VerbForm=Fin	2	nmod	2:nmod:in	_
+5	is		INTJ	IN	Number=Plur	6	cop	6:cop	_
+6	equivalent	equivalebe	INTJ	-LRB-	_	0	root	0:root	_
+7	to	be	PUNCT	-LRB-	_	10	case	10:case	_
+8	the	twe	PART	-LRB-	_	10	det	10:det	_
+9	US	we	DET	-LRB-	Tense=Pres|VerbForm=Part	10	compound	10:compound	_
+10	FBI		DET	CC	_	6	obl	6:obl:to	SpaceAfter=No
+11	,	be	PART	-LRB-	_	17	punct	17:punct	_
+12	so	we	PUNCT	VBD	Tense=Past|VerbForm=Part	17	advmod	17:advmod	_
+13	this	bs	DET	-LRB-	Tense=Pres|VerbForm=Part	17	nsubj:outer	17:nsubj:outer	_
+14	would	would]DAP[	ADJ	VBD	VerbForm=Fin	17	aux	17:aux	_
+15	be	be	PROPN	-LRB-	_	17	cop	17:cop	_
+16	like	libe	ADJ	NNPS	Number=Sing|PronType=Dem	17	mark	17:mark	_
+17	having	having_	SYM	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	6	parataxis	6:parataxis	_
+18	J.	i	PART	-LRB-	_	17	obj	17:obj|22:nsubj:xsubj	_
+19	Edgar	edg	ADV	-LRB-	Number=Plur	18	flat	18:flat	_
+20	Hoover	hoovbe	PUNCT	-LRB-	_	18	flat	18:flat	_
+21	unwittingly	unwittingbe	ADJ	NNP	VerbForm=Fin	22	advmod	22:advmod	_
+22	employ	emplwe	INTJ	CC	_	17	xcomp	17:xcomp	_
+23	at	we	PUNCT	NNP	_	26	case	26:case	_
+24	a	i	PUNCT	CC	Degree=Pos|NumForm=Word|NumType=Ord	26	det	26:det	_
+25	high	hibe	_	NNP	Degree=Pos|NumForm=Word|NumType=Ord	26	amod	26:amod	_
+26	level	levey	PUNCT	CC	Degree=Pos|NumForm=Word|NumType=Ord	22	obl	22:obl:at	_
+27	members	members]DAP[	PUNCT	-LRB-	Case=Nom|Person=2|PronType=Prs	22	obj	22:obj	_
+28	of	be	PART	NNS	_	31	case	31:case	_
+29	the	ti	PART	-LRB-	_	31	det	31:det	_
+30	Weathermen	weathermbe	PUNCT	-LRB-	_	31	compound	31:compound	_
+31	bombers	bombers]DAP[	INTJ	-LRB-	Tense=Pres|VerbForm=Part	27	nmod	27:nmod:of	_
+32	back	babe	PUNCT	NNP	_	35	advmod	35:advmod	_
+33	in	be	AUX	-LRB-	VerbForm=Fin	35	case	35:case	_
+34	the	ti	PART	-LRB-	Tense=Pres|VerbForm=Part	35	det	35:det	_
+35	1960s	19	PART	-LRB-	VerbForm=Fin	22	obl	22:obl:in	SpaceAfter=No
+36	.	we	PART	-LRB-	_	6	punct	6:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0006
+# text = The third was being run by the head of an investment firm.
+1	The	the_	PART	-LRB-	Case=Acc|Number=Plur|Person=1|PronType=Prs	2	det	2:det	_
+2	third	tbd	PUNCT	VBD	Number=Sing|PronType=Dem	5	nsubj:pass	5:nsubj:pass	_
+3	was	wbe	SCONJ	VB	_	5	aux	5:aux	_
+4	being	beibe	ADJ	CC	_	5	aux:pass	5:aux:pass	_
+5	run	run	ADJ	VBD	_	0	root	0:root	_
+6	by	be	ADJ	VBD	Degree=Pos|NumForm=Word|NumType=Ord	8	case	8:case	_
+7	the	ti	ADJ	VBD	_	8	det	8:det	_
+8	head	hebe	PART	VBD	Degree=Pos|NumForm=Word|NumType=Ord	5	obl:agent	5:obl:agent	_
+9	of	i	PART	VBD	VerbForm=Fin	12	case	12:case	_
+10	an	i	_	VBD	Tense=Pres|VerbForm=Part	12	det	12:det	_
+11	investment	investment]DAP[	ADJ	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	12	compound	12:compound	_
+12	firm	fiwe	PROPN	CC	_	8	nmod	8:nmod:of	SpaceAfter=No
+13	.	i	PART	VBD	_	5	punct	5:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0007
+# text = You wonder if he was manipulating the market with his bombing targets.
+1	You	you_	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	2	nsubj	2:nsubj	_
+2	wonder	wonder	[UNK]	VB	_	0	root	0:root	_
+3	if	iy	PROPN	_	Case=Acc|Number=Plur|Person=1|PronType=Prs	6	mark	6:mark	_
+4	he	be	SYM	-LRB-	_	6	nsubj	6:nsubj	_
+5	was		SYM	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	6	aux	6:aux	_
+6	manipulating	manipulatiwe	ADJ	VB	Tense=Pres|VerbForm=Part	2	ccomp	2:ccomp	_
+7	the	ti	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	8	det	8:det	_
+8	market	mar	SYM	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	6	obj	6:obj	_
+9	with	wibe	SYM	VB	Tense=Pres|VerbForm=Part	12	case	12:case	_
+10	his	hbe	[UNK]	VB	_	12	nmod:poss	12:nmod:poss	_
+11	bombing	bomb	ADJ	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	12	compound	12:compound	_
+12	targets	targewe	SYM	-LRB-	_	6	obl	6:obl:with	SpaceAfter=No
+13	.		PART	VBD	_	2	punct	2:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0008
+# text = The cells were operating in the Ghazaliyah and al-Jihad districts of the capital.
+1	The	the_	PART	-LRB-	Case=Acc|Number=Plur|Person=1|PronType=Prs	2	det	2:det	_
+2	cells	celwe	PUNCT	CC	Degree=Pos|NumForm=Word|NumType=Ord	4	nsubj	4:nsubj	_
+3	were	webe	SCONJ	VB	Degree=Pos|NumForm=Word|NumType=Ord	4	aux	4:aux	_
+4	operating	operatiwe	PUNCT	CC	Degree=Pos|NumForm=Word|NumType=Ord	0	root	0:root	_
+5	in	be	SCONJ	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	12	case	12:case	_
+6	the	ti	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	12	det	12:det	_
+7	Ghazaliyah	ghazaliyi	SYM	IN	VerbForm=Fin	12	compound	12:compound	_
+8	and	ai	SYM	VBD	_	11	cc	11:cc	_
+9	al		PROPN	PRP	Case=Nom|Gender=Neut|Number=Sing|Person=3|PronType=Prs	11	compound	11:compound	SpaceAfter=No
+10	-		PROPN	PRP	Case=Nom|Person=2|PronType=Prs	9	punct	9:punct	SpaceAfter=No
+11	Jihad	jihi	INTJ	IN	_	7	conj	7:conj:and|12:compound	_
+12	districts	districwe	PUNCT	VB	Degree=Pos|NumForm=Word|NumType=Ord	4	obl	4:obl:in	_
+13	of	we	SCONJ	VB	Tense=Pres|VerbForm=Part	15	case	15:case	_
+14	the	ti	PART	-LRB-	_	15	det	15:det	_
+15	capital	capitwe	ADJ	-LRB-	_	12	nmod	12:nmod:of	SpaceAfter=No
+16	.	be	INTJ	VBD	_	4	punct	4:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0009
+# text = Although the announcement was probably made to show progress in identifying and breaking up terror cells, I don't find the news that the Baathists continue to penetrate the Iraqi government very hopeful.
+1	Although	although_	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	6	mark	6:mark	_
+2	the	ti	PUNCT	VBD	_	3	det	3:det	_
+3	announcement	announcemebe	PUNCT	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	6	nsubj:pass	6:nsubj:pass|8:nsubj:xsubj	_
+4	was	wbe	SCONJ	IN	Number=Plur	6	aux:pass	6:aux:pass	_
+5	probably	proba	PUNCT	_	VerbForm=Fin	6	advmod	6:advmod	_
+6	made	mabe	SYM	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	21	advcl	21:advcl:although	_
+7	to	be	INTJ	VBD	Case=Acc|Number=Plur|Person=1|PronType=Prs	8	mark	8:mark	_
+8	show	shi	ADJ	-LRB-	Tense=Past|VerbForm=Part	6	xcomp	6:xcomp	_
+9	progress	progrei	PUNCT	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	8	obj	8:obj	_
+10	in	we	INTJ	VBP	Tense=Pres|VerbForm=Part	11	mark	11:mark	_
+11	identifying	identifyiwe	ADJ	CC	Degree=Pos|NumForm=Word|NumType=Ord	9	acl	9:acl:in	_
+12	and	ai	_	IN	Tense=Pres|VerbForm=Part	13	cc	13:cc	_
+13	breaking	breaki	SYM	CC	Degree=Pos|NumForm=Word|NumType=Ord	11	conj	9:acl:in|11:conj:and	_
+14	up	be	SYM	CC	Degree=Pos|NumForm=Word|NumType=Ord	13	compound:prt	13:compound:prt	_
+15	terror	terror]DAP[	PART	PRP	Degree=Pos|NumForm=Word|NumType=Ord	16	compound	16:compound	_
+16	cells	celbe	SYM	NNP	Degree=Pos|NumForm=Word|NumType=Ord	11	obj	11:obj|13:obj	SpaceAfter=No
+17	,		PART	VBD	Case=Acc|Number=Plur|Person=1|PronType=Prs	6	punct	6:punct	_
+18	I	i]DAP[	ADJ	PRP	Case=Nom|Person=2|PronType=Prs	21	nsubj	21:nsubj	_
+19-20	don't	do	ADP	NNP	Case=Nom|Number=Sing|Person=1|PronType=Prs	_	_	_	_
+19	do	i	PUNCT	VB	NumForm=Digit|NumType=Card	21	aux	21:aux	_
+20	n't	ni	SYM	VBD	Case=Nom|Number=Sing|Person=1|PronType=Prs	21	advmod	21:advmod	_
+21	find	fiwe	ADJ	VBD	Degree=Pos|NumForm=Word|NumType=Ord	0	root	0:root	_
+22	the	ti	ADJ	IN	Tense=Pres|VerbForm=Part	23	det	23:det	_
+23	news	n	ADJ	CC	Degree=Pos|NumForm=Word|NumType=Ord	21	obj	21:obj|34:nsubj:xsubj	_
+24	that	t	INTJ	NNS	Tense=Pres|VerbForm=Part	27	mark	27:mark	_
+25	the	ti	PART	-LRB-	Mood=Ind|Number=Sing|Person=2|Tense=Pres|VerbForm=Fin	26	det	26:det	_
+26	Baathists	baathisbe	INTJ	-LRB-	VerbForm=Fin	27	nsubj	27:nsubj|29:nsubj:xsubj	_
+27	continue	continuy	PROPN	VB	Tense=Past|VerbForm=Part	23	acl	23:acl:that	_
+28	to	we	PROPN	NNS	Tense=Pres|VerbForm=Part	29	mark	29:mark	_
+29	penetrate	penetrawe	[UNK]	NNP	VerbForm=Fin	27	xcomp	27:xcomp	_
+30	the	ti	PART	-LRB-	Case=Nom|Person=2|PronType=Prs	32	det	32:det	_
+31	Iraqi	irawe	PROPN	-LRB-	VerbForm=Fin	32	amod	32:amod	_
+32	government	governmewe	SYM	RB	Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin	29	obj	29:obj	_
+33	very	very	PART	PRP	VerbForm=Fin	34	advmod	34:advmod	_
+34	hopeful	hopefwe	PUNCT	VB	Case=Gen|Gender=Masc|Number=Sing|Person=3|Poss=Yes|PronType=Prs	21	xcomp	21:xcomp	SpaceAfter=No
+35	.	we	PART	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	21	punct	21:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0010
+# text = It reminds me too much of the ARVN officers who were secretly working for the other side in Vietnam.
+1	It	it_	PART	-LRB-	Case=Acc|Number=Plur|Person=1|PronType=Prs	2	nsubj	2:nsubj	_
+2	reminds	reminwe	INTJ	VBD	Tense=Pres|VerbForm=Part	0	root	0:root	_
+3	me	be	SYM	VBD	Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin	2	iobj	2:iobj	_
+4	too	tbe	PART	IN	_	5	advmod	5:advmod	_
+5	much	mube	PUNCT	NNP	Number=Plur	2	advmod	2:advmod	_
+6	of	we	PART	VB	Case=Nom|Person=2|PronType=Prs	9	case	9:case	_
+7	the	ti	PART	-LRB-	Case=Nom|Person=2|PronType=Prs	9	det	9:det	_
+8	ARVN	ar	DET	VB	_	9	compound	9:compound	_
+9	officers	officers]DAP[	SYM	VB	_	2	obl	2:obl:of|13:nsubj	_
+10	who		SYM	VB	Tense=Pres|VerbForm=Part	13	nsubj	9:ref	_
+11	were	were]DAP[	SYM	VB	Degree=Pos|NumForm=Word|NumType=Ord	13	aux	13:aux	_
+12	secretly	secretwe	ADJ	NNP	Degree=Pos|NumForm=Word|NumType=Ord	13	advmod	13:advmod	_
+13	working	workiwe	PROPN	NNS	Degree=Pos|NumForm=Word|NumType=Ord	9	acl:relcl	9:acl:relcl	Cxn=rc-wh-nsubj
+14	for	fwe	PROPN	NNP	Degree=Pos|NumForm=Word|NumType=Ord	17	case	17:case	_
+15	the	ti	PART	VB	_	17	det	17:det	_
+16	other	oty	PROPN	CC	Mood=Ind|Number=Sing|Person=2|Tense=Pres|VerbForm=Fin	17	amod	17:amod	_
+17	side	s	PROPN	-LRB-	_	13	obl	13:obl:for	_
+18	in	be	SCONJ	NNP	Tense=Pres|VerbForm=Part	19	case	19:case	_
+19	Vietnam	vietnam	DET	-LRB-	Degree=Pos|NumForm=Word|NumType=Ord	13	obl	13:obl:in	SpaceAfter=No
+20	.	i	PART	-LRB-	Case=Acc|Number=Plur|Person=1|PronType=Prs	2	punct	2:punct	_
+
diff --git a/tests/testdata/datasets/en_train.conllu b/tests/testdata/datasets/en_train.conllu
new file mode 100644
index 0000000..7ce3453
--- /dev/null
+++ b/tests/testdata/datasets/en_train.conllu
@@ -0,0 +1,246 @@
+# newdoc id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0001
+# newpar id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-p0001
+# text = Al-Zaman : American forces killed Shaikh Abdullah al-Ani, the preacher at the mosque in the town of Qaim, near the Syrian border.
+1	Al	Al	PROPN	NNP	Number=Sing	0	root	0:root	SpaceAfter=No
+2	-	-	PUNCT	HYPH	_	3	punct	3:punct	SpaceAfter=No
+3	Zaman	Zaman	PROPN	NNP	Number=Sing	1	flat	1:flat	_
+4	:	:	PUNCT	:	_	7	punct	7:punct	_
+5	American	American	ADJ	JJ	Degree=Pos	6	amod	6:amod	_
+6	forces	force	NOUN	NNS	Number=Plur	7	nsubj	7:nsubj	_
+7	killed	kill	VERB	VBD	Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin	1	parataxis	1:parataxis	_
+8	Shaikh	Shaikh	PROPN	NNP	Number=Sing	7	obj	7:obj	_
+9	Abdullah	Abdullah	PROPN	NNP	Number=Sing	8	flat	8:flat	_
+10	al	al	PROPN	NNP	Number=Sing	8	flat	8:flat	SpaceAfter=No
+11	-	-	PUNCT	HYPH	_	12	punct	12:punct	SpaceAfter=No
+12	Ani	Ani	PROPN	NNP	Number=Sing	8	flat	8:flat	SpaceAfter=No
+13	,	,	PUNCT	,	_	15	punct	15:punct	_
+14	the	the	DET	DT	Definite=Def|PronType=Art	15	det	15:det	_
+15	preacher	preacher	NOUN	NN	Number=Sing	8	appos	8:appos	_
+16	at	at	ADP	IN	_	18	case	18:case	_
+17	the	the	DET	DT	Definite=Def|PronType=Art	18	det	18:det	_
+18	mosque	mosque	NOUN	NN	Number=Sing	15	nmod	15:nmod:at	_
+19	in	in	ADP	IN	_	21	case	21:case	_
+20	the	the	DET	DT	Definite=Def|PronType=Art	21	det	21:det	_
+21	town	town	NOUN	NN	Number=Sing	18	nmod	18:nmod:in	_
+22	of	of	ADP	IN	_	23	case	23:case	_
+23	Qaim	Qaim	PROPN	NNP	Number=Sing	21	nmod	21:nmod:of	SpaceAfter=No
+24	,	,	PUNCT	,	_	28	punct	28:punct	_
+25	near	near	ADP	IN	_	28	case	28:case	_
+26	the	the	DET	DT	Definite=Def|PronType=Art	28	det	28:det	_
+27	Syrian	Syrian	ADJ	JJ	Degree=Pos	28	amod	28:amod	_
+28	border	border	NOUN	NN	Number=Sing	21	nmod	21:nmod:near	SpaceAfter=No
+29	.	.	PUNCT	.	_	1	punct	1:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0002
+# text = [This killing of a respected cleric will be causing us trouble for years to come.]
+1	[	[	PUNCT	-LRB-	_	10	punct	10:punct	SpaceAfter=No
+2	This	this	DET	DT	Number=Sing|PronType=Dem	3	det	3:det	_
+3	killing	killing	NOUN	NN	Number=Sing	10	nsubj	10:nsubj	_
+4	of	of	ADP	IN	_	7	case	7:case	_
+5	a	a	DET	DT	Definite=Ind|PronType=Art	7	det	7:det	_
+6	respected	respected	ADJ	JJ	Degree=Pos	7	amod	7:amod	_
+7	cleric	cleric	NOUN	NN	Number=Sing	3	nmod	3:nmod:of	_
+8	will	will	AUX	MD	VerbForm=Fin	10	aux	10:aux	_
+9	be	be	AUX	VB	VerbForm=Inf	10	aux	10:aux	_
+10	causing	cause	VERB	VBG	Tense=Pres|VerbForm=Part	0	root	0:root	_
+11	us	we	PRON	PRP	Case=Acc|Number=Plur|Person=1|PronType=Prs	10	iobj	10:iobj	_
+12	trouble	trouble	NOUN	NN	Number=Sing	10	obj	10:obj	_
+13	for	for	ADP	IN	_	14	case	14:case	_
+14	years	year	NOUN	NNS	Number=Plur	10	obl	10:obl:for	_
+15	to	to	PART	TO	_	16	mark	16:mark	_
+16	come	come	VERB	VB	VerbForm=Inf	14	acl	14:acl:to	SpaceAfter=No
+17	.	.	PUNCT	.	_	10	punct	10:punct	SpaceAfter=No
+18	]	]	PUNCT	-RRB-	_	10	punct	10:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0003
+# text = DPA: Iraqi authorities announced that they had busted up 3 terrorist cells operating in Baghdad.
+1	DPA	DPA	PROPN	NNP	Number=Sing	0	root	0:root	SpaceAfter=No
+2	:	:	PUNCT	:	_	5	punct	5:punct	_
+3	Iraqi	Iraqi	ADJ	JJ	Degree=Pos	4	amod	4:amod	_
+4	authorities	authority	NOUN	NNS	Number=Plur	5	nsubj	5:nsubj	_
+5	announced	announce	VERB	VBD	Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin	1	parataxis	1:parataxis	_
+6	that	that	SCONJ	IN	_	9	mark	9:mark	_
+7	they	they	PRON	PRP	Case=Nom|Number=Plur|Person=3|PronType=Prs	9	nsubj	9:nsubj	_
+8	had	have	AUX	VBD	Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin	9	aux	9:aux	_
+9	busted	bust	VERB	VBN	Tense=Past|VerbForm=Part	5	ccomp	5:ccomp	_
+10	up	up	ADP	RP	_	9	compound:prt	9:compound:prt	_
+11	3	3	NUM	CD	NumForm=Digit|NumType=Card	13	nummod	13:nummod	_
+12	terrorist	terrorist	ADJ	JJ	Degree=Pos	13	amod	13:amod	_
+13	cells	cell	NOUN	NNS	Number=Plur	9	obj	9:obj	_
+14	operating	operate	VERB	VBG	VerbForm=Ger	13	acl	13:acl	_
+15	in	in	ADP	IN	_	16	case	16:case	_
+16	Baghdad	Baghdad	PROPN	NNP	Number=Sing	14	obl	14:obl:in	SpaceAfter=No
+17	.	.	PUNCT	.	_	1	punct	1:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0004
+# text = Two of them were being run by 2 officials of the Ministry of the Interior!
+1	Two	two	NUM	CD	NumForm=Word|NumType=Card	6	nsubj:pass	6:nsubj:pass	_
+2	of	of	ADP	IN	_	3	case	3:case	_
+3	them	they	PRON	PRP	Case=Acc|Number=Plur|Person=3|PronType=Prs	1	nmod	1:nmod:of	_
+4	were	be	AUX	VBD	Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin	6	aux	6:aux	_
+5	being	be	AUX	VBG	Tense=Pres|VerbForm=Part	6	aux:pass	6:aux:pass	_
+6	run	run	VERB	VBN	Tense=Past|VerbForm=Part|Voice=Pass	0	root	0:root	_
+7	by	by	ADP	IN	_	9	case	9:case	_
+8	2	2	NUM	CD	NumForm=Digit|NumType=Card	9	nummod	9:nummod	_
+9	officials	official	NOUN	NNS	Number=Plur	6	obl:agent	6:obl:agent	_
+10	of	of	ADP	IN	_	12	case	12:case	_
+11	the	the	DET	DT	Definite=Def|PronType=Art	12	det	12:det	_
+12	Ministry	Ministry	PROPN	NNP	Number=Sing	9	nmod	9:nmod:of	_
+13	of	of	ADP	IN	_	15	case	15:case	_
+14	the	the	DET	DT	Definite=Def|PronType=Art	15	det	15:det	_
+15	Interior	Interior	PROPN	NNP	Number=Sing	12	nmod	12:nmod:of	SpaceAfter=No
+16	!	!	PUNCT	.	_	6	punct	6:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0005
+# text = The MoI in Iraq is equivalent to the US FBI, so this would be like having J. Edgar Hoover unwittingly employ at a high level members of the Weathermen bombers back in the 1960s.
+1	The	the	DET	DT	Definite=Def|PronType=Art	2	det	2:det	_
+2	MoI	MoI	PROPN	NNP	Number=Sing	6	nsubj	6:nsubj	_
+3	in	in	ADP	IN	_	4	case	4:case	_
+4	Iraq	Iraq	PROPN	NNP	Number=Sing	2	nmod	2:nmod:in	_
+5	is	be	AUX	VBZ	Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin	6	cop	6:cop	_
+6	equivalent	equivalent	ADJ	JJ	Degree=Pos	0	root	0:root	_
+7	to	to	ADP	IN	_	10	case	10:case	_
+8	the	the	DET	DT	Definite=Def|PronType=Art	10	det	10:det	_
+9	US	US	PROPN	NNP	Number=Sing	10	compound	10:compound	_
+10	FBI	FBI	PROPN	NNP	Number=Sing	6	obl	6:obl:to	SpaceAfter=No
+11	,	,	PUNCT	,	_	17	punct	17:punct	_
+12	so	so	ADV	RB	_	17	advmod	17:advmod	_
+13	this	this	PRON	DT	Number=Sing|PronType=Dem	17	nsubj:outer	17:nsubj:outer	_
+14	would	would	AUX	MD	VerbForm=Fin	17	aux	17:aux	_
+15	be	be	AUX	VB	VerbForm=Inf	17	cop	17:cop	_
+16	like	like	SCONJ	IN	_	17	mark	17:mark	_
+17	having	have	VERB	VBG	Tense=Pres|VerbForm=Part	6	parataxis	6:parataxis	_
+18	J.	J.	PROPN	NNP	Number=Sing	17	obj	17:obj|22:nsubj:xsubj	_
+19	Edgar	Edgar	PROPN	NNP	Number=Sing	18	flat	18:flat	_
+20	Hoover	Hoover	PROPN	NNP	Number=Sing	18	flat	18:flat	_
+21	unwittingly	unwittingly	ADV	RB	_	22	advmod	22:advmod	_
+22	employ	employ	VERB	VB	VerbForm=Inf	17	xcomp	17:xcomp	_
+23	at	at	ADP	IN	_	26	case	26:case	_
+24	a	a	DET	DT	Definite=Ind|PronType=Art	26	det	26:det	_
+25	high	high	ADJ	JJ	Degree=Pos	26	amod	26:amod	_
+26	level	level	NOUN	NN	Number=Sing	22	obl	22:obl:at	_
+27	members	member	NOUN	NNS	Number=Plur	22	obj	22:obj	_
+28	of	of	ADP	IN	_	31	case	31:case	_
+29	the	the	DET	DT	Definite=Def|PronType=Art	31	det	31:det	_
+30	Weathermen	Weathermen	PROPN	NNPS	Number=Plur	31	compound	31:compound	_
+31	bombers	bomber	NOUN	NNS	Number=Plur	27	nmod	27:nmod:of	_
+32	back	back	ADV	RB	_	35	advmod	35:advmod	_
+33	in	in	ADP	IN	_	35	case	35:case	_
+34	the	the	DET	DT	Definite=Def|PronType=Art	35	det	35:det	_
+35	1960s	1960s	NOUN	NNS	Number=Ptan|NumForm=Digit|NumType=Card	22	obl	22:obl:in	SpaceAfter=No
+36	.	.	PUNCT	.	_	6	punct	6:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0006
+# text = The third was being run by the head of an investment firm.
+1	The	the	DET	DT	Definite=Def|PronType=Art	2	det	2:det	_
+2	third	third	ADJ	JJ	Degree=Pos|NumForm=Word|NumType=Ord	5	nsubj:pass	5:nsubj:pass	_
+3	was	be	AUX	VBD	Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin	5	aux	5:aux	_
+4	being	be	AUX	VBG	Tense=Pres|VerbForm=Part	5	aux:pass	5:aux:pass	_
+5	run	run	VERB	VBN	Tense=Past|VerbForm=Part|Voice=Pass	0	root	0:root	_
+6	by	by	ADP	IN	_	8	case	8:case	_
+7	the	the	DET	DT	Definite=Def|PronType=Art	8	det	8:det	_
+8	head	head	NOUN	NN	Number=Sing	5	obl:agent	5:obl:agent	_
+9	of	of	ADP	IN	_	12	case	12:case	_
+10	an	a	DET	DT	Definite=Ind|PronType=Art	12	det	12:det	_
+11	investment	investment	NOUN	NN	Number=Sing	12	compound	12:compound	_
+12	firm	firm	NOUN	NN	Number=Sing	8	nmod	8:nmod:of	SpaceAfter=No
+13	.	.	PUNCT	.	_	5	punct	5:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0007
+# text = You wonder if he was manipulating the market with his bombing targets.
+1	You	you	PRON	PRP	Case=Nom|Person=2|PronType=Prs	2	nsubj	2:nsubj	_
+2	wonder	wonder	VERB	VBP	Mood=Ind|Number=Sing|Person=2|Tense=Pres|VerbForm=Fin	0	root	0:root	_
+3	if	if	SCONJ	IN	_	6	mark	6:mark	_
+4	he	he	PRON	PRP	Case=Nom|Gender=Masc|Number=Sing|Person=3|PronType=Prs	6	nsubj	6:nsubj	_
+5	was	be	AUX	VBD	Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin	6	aux	6:aux	_
+6	manipulating	manipulate	VERB	VBG	Tense=Pres|VerbForm=Part	2	ccomp	2:ccomp	_
+7	the	the	DET	DT	Definite=Def|PronType=Art	8	det	8:det	_
+8	market	market	NOUN	NN	Number=Sing	6	obj	6:obj	_
+9	with	with	ADP	IN	_	12	case	12:case	_
+10	his	his	PRON	PRP$	Case=Gen|Gender=Masc|Number=Sing|Person=3|Poss=Yes|PronType=Prs	12	nmod:poss	12:nmod:poss	_
+11	bombing	bombing	NOUN	NN	Number=Sing	12	compound	12:compound	_
+12	targets	target	NOUN	NNS	Number=Plur	6	obl	6:obl:with	SpaceAfter=No
+13	.	.	PUNCT	.	_	2	punct	2:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0008
+# text = The cells were operating in the Ghazaliyah and al-Jihad districts of the capital.
+1	The	the	DET	DT	Definite=Def|PronType=Art	2	det	2:det	_
+2	cells	cell	NOUN	NNS	Number=Plur	4	nsubj	4:nsubj	_
+3	were	be	AUX	VBD	Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin	4	aux	4:aux	_
+4	operating	operate	VERB	VBG	Tense=Pres|VerbForm=Part	0	root	0:root	_
+5	in	in	ADP	IN	_	12	case	12:case	_
+6	the	the	DET	DT	Definite=Def|PronType=Art	12	det	12:det	_
+7	Ghazaliyah	Ghazaliyah	PROPN	NNP	Number=Sing	12	compound	12:compound	_
+8	and	and	CCONJ	CC	_	11	cc	11:cc	_
+9	al	al	PROPN	NNP	Number=Sing	11	compound	11:compound	SpaceAfter=No
+10	-	-	PUNCT	HYPH	_	9	punct	9:punct	SpaceAfter=No
+11	Jihad	Jihad	PROPN	NNP	Number=Sing	7	conj	7:conj:and|12:compound	_
+12	districts	district	NOUN	NNS	Number=Plur	4	obl	4:obl:in	_
+13	of	of	ADP	IN	_	15	case	15:case	_
+14	the	the	DET	DT	Definite=Def|PronType=Art	15	det	15:det	_
+15	capital	capital	NOUN	NN	Number=Sing	12	nmod	12:nmod:of	SpaceAfter=No
+16	.	.	PUNCT	.	_	4	punct	4:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0009
+# text = Although the announcement was probably made to show progress in identifying and breaking up terror cells, I don't find the news that the Baathists continue to penetrate the Iraqi government very hopeful.
+1	Although	although	SCONJ	IN	_	6	mark	6:mark	_
+2	the	the	DET	DT	Definite=Def|PronType=Art	3	det	3:det	_
+3	announcement	announcement	NOUN	NN	Number=Sing	6	nsubj:pass	6:nsubj:pass|8:nsubj:xsubj	_
+4	was	be	AUX	VBD	Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin	6	aux:pass	6:aux:pass	_
+5	probably	probably	ADV	RB	_	6	advmod	6:advmod	_
+6	made	make	VERB	VBN	Tense=Past|VerbForm=Part|Voice=Pass	21	advcl	21:advcl:although	_
+7	to	to	PART	TO	_	8	mark	8:mark	_
+8	show	show	VERB	VB	VerbForm=Inf	6	xcomp	6:xcomp	_
+9	progress	progress	NOUN	NN	Number=Sing	8	obj	8:obj	_
+10	in	in	SCONJ	IN	_	11	mark	11:mark	_
+11	identifying	identify	VERB	VBG	VerbForm=Ger	9	acl	9:acl:in	_
+12	and	and	CCONJ	CC	_	13	cc	13:cc	_
+13	breaking	break	VERB	VBG	VerbForm=Ger	11	conj	9:acl:in|11:conj:and	_
+14	up	up	ADP	RP	_	13	compound:prt	13:compound:prt	_
+15	terror	terror	NOUN	NN	Number=Sing	16	compound	16:compound	_
+16	cells	cell	NOUN	NNS	Number=Plur	11	obj	11:obj|13:obj	SpaceAfter=No
+17	,	,	PUNCT	,	_	6	punct	6:punct	_
+18	I	I	PRON	PRP	Case=Nom|Number=Sing|Person=1|PronType=Prs	21	nsubj	21:nsubj	_
+19-20	don't	_	_	_	_	_	_	_	_
+19	do	do	AUX	VBP	Mood=Ind|Number=Sing|Person=1|Tense=Pres|VerbForm=Fin	21	aux	21:aux	_
+20	n't	not	PART	RB	_	21	advmod	21:advmod	_
+21	find	find	VERB	VB	VerbForm=Inf	0	root	0:root	_
+22	the	the	DET	DT	Definite=Def|PronType=Art	23	det	23:det	_
+23	news	news	NOUN	NN	Number=Sing	21	obj	21:obj|34:nsubj:xsubj	_
+24	that	that	SCONJ	IN	_	27	mark	27:mark	_
+25	the	the	DET	DT	Definite=Def|PronType=Art	26	det	26:det	_
+26	Baathists	Baathist	PROPN	NNPS	Number=Plur	27	nsubj	27:nsubj|29:nsubj:xsubj	_
+27	continue	continue	VERB	VBP	Mood=Ind|Number=Plur|Person=3|Tense=Pres|VerbForm=Fin	23	acl	23:acl:that	_
+28	to	to	PART	TO	_	29	mark	29:mark	_
+29	penetrate	penetrate	VERB	VB	VerbForm=Inf	27	xcomp	27:xcomp	_
+30	the	the	DET	DT	Definite=Def|PronType=Art	32	det	32:det	_
+31	Iraqi	Iraqi	ADJ	JJ	Degree=Pos	32	amod	32:amod	_
+32	government	government	NOUN	NN	Number=Sing	29	obj	29:obj	_
+33	very	very	ADV	RB	_	34	advmod	34:advmod	_
+34	hopeful	hopeful	ADJ	JJ	Degree=Pos	21	xcomp	21:xcomp	SpaceAfter=No
+35	.	.	PUNCT	.	_	21	punct	21:punct	_
+
+# sent_id = weblog-juancole.com_juancole_20051126063000_ENG_20051126_063000-0010
+# text = It reminds me too much of the ARVN officers who were secretly working for the other side in Vietnam.
+1	It	it	PRON	PRP	Case=Nom|Gender=Neut|Number=Sing|Person=3|PronType=Prs	2	nsubj	2:nsubj	_
+2	reminds	remind	VERB	VBZ	Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin	0	root	0:root	_
+3	me	I	PRON	PRP	Case=Acc|Number=Sing|Person=1|PronType=Prs	2	iobj	2:iobj	_
+4	too	too	ADV	RB	_	5	advmod	5:advmod	_
+5	much	much	ADV	RB	_	2	advmod	2:advmod	_
+6	of	of	ADP	IN	_	9	case	9:case	_
+7	the	the	DET	DT	Definite=Def|PronType=Art	9	det	9:det	_
+8	ARVN	ARVN	PROPN	NNP	Number=Sing	9	compound	9:compound	_
+9	officers	officer	NOUN	NNS	Number=Plur	2	obl	2:obl:of|13:nsubj	_
+10	who	who	PRON	WP	PronType=Rel	13	nsubj	9:ref	_
+11	were	be	AUX	VBD	Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin	13	aux	13:aux	_
+12	secretly	secretly	ADV	RB	_	13	advmod	13:advmod	_
+13	working	work	VERB	VBG	Tense=Pres|VerbForm=Part	9	acl:relcl	9:acl:relcl	Cxn=rc-wh-nsubj
+14	for	for	ADP	IN	_	17	case	17:case	_
+15	the	the	DET	DT	Definite=Def|PronType=Art	17	det	17:det	_
+16	other	other	ADJ	JJ	Degree=Pos	17	amod	17:amod	_
+17	side	side	NOUN	NN	Number=Sing	13	obl	13:obl:for	_
+18	in	in	ADP	IN	_	19	case	19:case	_
+19	Vietnam	Vietnam	PROPN	NNP	Number=Sing	13	obl	13:obl:in	SpaceAfter=No
+20	.	.	PUNCT	.	_	2	punct	2:punct	_
+
diff --git a/tests/testdata/datasets/ru_expected.conllu b/tests/testdata/datasets/ru_expected.conllu
new file mode 100644
index 0000000..f23a2e3
--- /dev/null
+++ b/tests/testdata/datasets/ru_expected.conllu
@@ -0,0 +1,255 @@
+# sent_id = 2003Anketa.xml_2
+# text = Начальник областного управления связи Семен Еремеевич был человек простой, приходил на работу всегда вовремя, здоровался с секретаршей за руку и иногда даже писал в стенгазету заметки под псевдонимом "Муха".
+1	Начальник	начальтьик	INTJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	8	nsubj	8:nsubj	_
+2	областного	областнон	PROPN	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	3	amod	3:amod	_
+3	управления	управленая	PUNCT	_	Aspect=Imp|Gender=Fem|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	1	nmod	1:nmod:gen	_
+4	связи	свяоз	PUNCT	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	3	nmod	3:nmod:gen	_
+5	Семен	сетьен	PROPN	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	1	appos	1:appos	_
+6	Еремеевич	еремеевий	PROPN	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	5	flat:name	5:flat:name	_
+7	был	лн	PROPN	_	Case=Nom|PronType=Ind	8	cop	8:cop	_
+8	человек	челокн	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	0	root	0:root	_
+9	простой	простьой	PART	_	Case=Acc|Number=Plur|PronType=Tot	8	amod	8:amod	SpaceAfter=No
+10	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	11	punct	11:punct	_
+11	приходил	приходиь	PUNCT	_	Case=Nom|PronType=Ind	8	conj	0:root|8:conj	_
+12	на	ой	PROPN	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	13	case	13:case	_
+13	работу	рабун	PROPN	_	Mood=Cnd	11	obl	11:obl:на:acc	_
+14	всегда	вседсь	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	11	advmod	11:advmod	_
+15	вовремя	вовртьмя	SYM	_	Case=Acc|Number=Plur|PronType=Tot	11	advmod	11:advmod	SpaceAfter=No
+16	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	17	punct	17:punct	_
+17	здоровался	здороваян	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	8	conj	0:root|8:conj	_
+18	с	о	SYM	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	19	case	19:case	_
+19	секретаршей	секретарйн	SYM	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	17	obl	17:obl:с:ins	_
+20	за	оз	SYM	_	Animacy=Inan|Case=Loc|Gender=Neut|Number=Sing|PronType=Dem	21	case	21:case	_
+21	руку	рукй	SYM	_	Mood=Cnd	17	obl	17:obl:за:acc	_
+22	и	сь	PART	_	Case=Acc|Number=Plur|PronType=Tot	25	cc	25:cc	_
+23	иногда	инодсь	PUNCT	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	25	advmod	25:advmod	_
+24	даже	джсь	ADJ	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	25	advmod	25:advmod	_
+25	писал	пилн	SYM	_	Case=Acc|Number=Plur|PronType=Tot	8	conj	0:root|8:conj	_
+26	в	вн	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	27	case	27:case	_
+27	стенгазету	стенгазтьту	SYM	_	Case=Acc|Number=Plur|PronType=Tot	25	obl	25:obl:в:acc	_
+28	заметки	замеин	PROPN	_	Mood=Cnd	25	obj	25:obj	_
+29	под	дн	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	30	case	30:case	_
+30	псевдонимом	псевдонимн	INTJ	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	25	obl	25:obl:под:ins	_
+31	"	"н	PART	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	32	punct	32:punct	SpaceAfter=No
+32	Муха	мтьха	PART	_	Case=Acc|Gender=Masc|Number=Sing|Person=3|PronType=Prs	30	nmod	30:nmod:nom	SpaceAfter=No
+33	"	ь	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	32	punct	32:punct	SpaceAfter=No
+34	.	ть.	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	8	punct	8:punct	_
+
+# sent_id = 2003Anketa.xml_3
+# text = В приемной его с утра ожидали посетители, - кое-кто с важными делами, а кое-кто и с такими, которые легко можно было решить в нижестоящих инстанциях, не затрудняя Семена Еремеевича.
+1	В	тьв	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	case	2:case	_
+2	приемной	приемтьой	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	6	obl	6:obl:в:loc	_
+3	его	он	[UNK]	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	6	obj	6:obj	_
+4	с	сь	ADV	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	5	case	5:case	_
+5	утра	утьра	ADJ	_	Case=Gen|Degree=Pos|Number=Plur	6	obl	6:obl:с:gen	_
+6	ожидали	ожидин	ADJ	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	0	root	0:root	_
+7	посетители	посетитеол	SYM	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	6	nsubj	6:nsubj	SpaceAfter=No
+8	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	13	punct	13:punct	_
+9	-	-]DAP[	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	13	punct	13:punct	_
+10	кое-кто	кое-он	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	13	nsubj	13:nsubj	_
+11	с	ая	INTJ	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	13	case	13:case	_
+12	важными	важныма	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	13	amod	13:amod	_
+13	делами	делама	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	7	parataxis	7:parataxis	SpaceAfter=No
+14	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	19	punct	19:punct	_
+15	а	ой	[UNK]	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	19	cc	19:cc	_
+16	кое-кто	кое-он	PROPN	_	Case=Ins|Degree=Pos|Number=Plur	19	nsubj	19:nsubj	_
+17	и	сь	PART	_	Case=Nom|Degree=Pos|Number=Plur	19	advmod	19:advmod	_
+18	с	ая	SYM	_	Case=Nom|Degree=Pos|Number=Plur	19	case	19:case	_
+19	такими	такой	PRON	_	Case=Gen|Degree=Pos|Number=Plur	13	det	13:det|25:obj	SpaceAfter=No
+20	,	ть,	PART	_	Case=Acc|Number=Plur|PronType=Tot	23	punct	23:punct	_
+21	которые	которые_	PART	_	Animacy=Inan|Case=Loc|Gender=Neut|Number=Sing|PronType=Dem	25	obj	19:ref	_
+22	легко	легко_	PROPN	_	Case=Nom|PronType=Ind	23	advmod	23:advmod	_
+23	можно	можнй	PROPN	_	Mood=Cnd	19	acl:relcl	19:acl:relcl	_
+24	было	быой	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	23	cop	23:cop	_
+25	решить	решитй	[UNK]	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	23	csubj	23:csubj	_
+26	в	тьв	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	28	case	28:case	_
+27	нижестоящих	нижестоятьих	SCONJ	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	28	amod	28:amod	_
+28	инстанциях	инстанцияй	[UNK]	_	Mood=Cnd	25	obl	25:obl:в:loc	SpaceAfter=No
+29	,	ть,	PART	_	Case=Acc|Number=Plur|PronType=Tot	31	punct	31:punct	_
+30	не	тьне	[UNK]	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Sing	31	advmod	31:advmod	_
+31	затрудняя	затрудтьяя	ADJ	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	25	advcl	25:advcl	_
+32	Семена	семтьна	ADJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	31	obl	31:obl:gen	_
+33	Еремеевича	еремеевтьча	ADJ	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	32	flat:name	32:flat:name	SpaceAfter=No
+34	.	ть.	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	6	punct	6:punct	_
+
+# sent_id = 2003Anketa.xml_4
+# text = Однако стиль работы Семена Еремеевича заключался в том, чтобы принимать всех желающих и лично вникать в дело.
+1	Однако	однтько	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	6	advmod	6:advmod	_
+2	стиль	сттьль	PUNCT	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	6	nsubj	6:nsubj	_
+3	работы	рабтьты	PUNCT	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	nmod	2:nmod:gen	_
+4	Семена	семтьна	SCONJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	3	nmod	3:nmod:gen	_
+5	Еремеевича	еремеевтьча	PUNCT	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	4	flat:name	4:flat:name	_
+6	заключался	заключаян	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	0	root	0:root	_
+7	в	вн	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	8	case	8:case	_
+8	том	мн	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	6	obl	6:obl:в:loc	SpaceAfter=No
+9	,	,н	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	11	punct	11:punct	_
+10	чтобы	чтын	PUNCT	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	11	mark	11:mark	_
+11	принимать	принимтьть	PUNCT	_	Case=Nom|PronType=Ind	8	advcl	8:advcl:чтобы	_
+12	всех	втьех	SYM	_	Case=Nom|PronType=Ind	13	det	13:det	_
+13	желающих	желаютьих	SYM	_	Case=Acc|Number=Plur|PronType=Tot	11	xcomp	11:xcomp	_
+14	и	ь	SYM	_	Case=Acc|Number=Plur|PronType=Tot	16	cc	16:cc	_
+15	лично	личон	SYM	_	Case=Acc|Number=Plur|PronType=Tot	16	advmod	16:advmod	_
+16	вникать	вникаая	[UNK]	_	Case=Acc|Number=Plur|PronType=Tot	11	conj	8:advcl:чтобы|11:conj	_
+17	в	й	PART	_	Case=Acc|Number=Plur|PronType=Tot	18	case	18:case	_
+18	дело	деол	[UNK]	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	16	obl	16:obl:в:acc	SpaceAfter=No
+19	.	ть.	INTJ	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	6	punct	6:punct	_
+
+# sent_id = 2003Anketa.xml_5
+# text = Приемная была обставлена просто, но по-деловому.
+1	Приемная	приемтьая	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	3	nsubj:pass	3:nsubj:pass	_
+2	была	бан	SCONJ	_	Case=Nom|PronType=Ind	3	aux:pass	3:aux:pass	_
+3	обставлена	обставлан	ADJ	_	Case=Nom|PronType=Ind	0	root	0:root	_
+4	просто	протьто	[UNK]	_	Case=Acc|Number=Plur|PronType=Tot	3	advmod	3:advmod	SpaceAfter=No
+5	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	7	punct	7:punct	_
+6	но	он	PROPN	_	Case=Acc|PronType=Int,Rel	7	cc	7:cc	_
+7	по-деловому	по-деловтьму	PART	_	Case=Nom|PronType=Ind	4	conj	3:advmod|4:conj	SpaceAfter=No
+8	.	ть.	PART	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_6
+# text = У двери стоял стол секретарши, на столе - пишущая машинка с широкой кареткой.
+1	У	тьу	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	case	2:case	_
+2	двери	двин	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	3	obl	3:obl:у:gen	_
+3	стоял	стой	[UNK]	_	Case=Acc|Number=Plur|PronType=Tot	0	root	0:root	_
+4	стол	стоа	INTJ	_	Animacy=Inan|Case=Gen|Gender=Fem|Number=Plur	3	nsubj	3:nsubj	_
+5	секретарши	секретаин	PUNCT	_	Mood=Cnd	4	nmod	4:nmod:gen	SpaceAfter=No
+6	,	ь	PART	_	Case=Acc|Number=Plur|PronType=Tot	8	punct	8:punct	_
+7	на	нь	PUNCT	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	8	case	8:case	_
+8	столе	стен	[UNK]	_	Mood=Cnd	3	conj	0:root|3:conj	_
+9	-	ь	PART	_	Case=Acc|Number=Plur|PronType=Tot	11	punct	11:punct	_
+10	пишущая	пишуян	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	11	amod	11:amod	_
+11	машинка	машиан	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	8	nsubj	8:nsubj	_
+12	с	сн	SCONJ	_	Case=Acc|Number=Plur|PronType=Tot	14	case	14:case	_
+13	широкой	широой	PART	_	Case=Acc|Number=Plur|PronType=Tot	14	amod	14:amod	_
+14	кареткой	каретйн	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	11	nmod	11:nmod:с:ins	SpaceAfter=No
+15	.	ть.	PART	_	Case=Acc|Number=Plur|PronType=Tot	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_7
+# text = В углу висел репродуктор и играло радио для развлечения ожидающих и еще для того, чтобы заглушать голос начальника, доносившийся из кабинета, так как, бесспорно, среди посетителей могли находиться и случайные люди.
+1	В	тьв	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	case	2:case	_
+2	углу	уун	SYM	_	Case=Acc|Number=Plur|PronType=Tot	3	obl	3:obl:в:loc	_
+3	висел	висеь	[UNK]	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+4	репродуктор	репродукрн	[UNK]	_	Mood=Cnd	3	nsubj	3:nsubj	_
+5	и	а	PART	_	Case=Acc|Number=Plur|PronType=Tot	6	cc	6:cc	_
+6	играло	играой	PUNCT	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	3	conj	0:root|3:conj	_
+7	радио	раон	INTJ	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	6	nsubj	6:nsubj	_
+8	для	дла	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	9	case	9:case	_
+9	развлечения	развлечеян	INTJ	_	Case=Nom|PronType=Ind	6	obl	6:obl:для:gen	_
+10	ожидающих	ожидаютьих	SYM	_	Case=Acc|Number=Plur|PronType=Tot	9	acl	9:acl	_
+11	и	ь	PART	_	Case=Acc|Number=Plur|PronType=Tot	14	cc	14:cc	_
+12	еще	ещь	PUNCT	_	Case=Nom|Degree=Pos|Number=Plur	14	obl	14:obl	_
+13	для	лсь	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	14	case	14:case	_
+14	того	того_	PROPN	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	9	conj	6:obl:для:gen|9:conj	SpaceAfter=No
+15	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	17	punct	17:punct	_
+16	чтобы	чтын	ADV	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	17	mark	17:mark	_
+17	заглушать	заглушьн	ADJ	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Sing	14	advcl	14:advcl:чтобы	_
+18	голос	госн	ADJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	17	obj	17:obj	_
+19	начальника	начальнан	INTJ	_	Case=Nom|Degree=Pos|Number=Plur	18	nmod	18:nmod:gen	SpaceAfter=No
+20	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	21	punct	21:punct	_
+21	доносившийся	доносившиян	SYM	_	Case=Acc|Number=Plur|PronType=Tot	18	acl	18:acl	_
+22	из	ий	SYM	_	Aspect=Imp|VerbForm=Inf|Voice=Act	23	case	23:case	_
+23	кабинета	кабинтьта	SYM	_	Mood=Cnd	21	obl	21:obl:из:gen	SpaceAfter=No
+24	,	ть,	PART	_	Case=Acc|Number=Plur|PronType=Tot	21	punct	21:punct	_
+25	так	так_	PART	_	Case=Acc|Number=Plur|PronType=Tot	32	mark	32:mark	_
+26	как	тьак	INTJ	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	25	fixed	25:fixed	SpaceAfter=No
+27	,	ть,	PART	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	25	punct	25:punct	_
+28	бесспорно	бесспотьно	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	32	parataxis	32:parataxis	SpaceAfter=No
+29	,	ть,	PART	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	28	punct	28:punct	_
+30	среди	сртьди	[UNK]	_	[PAD]	31	case	31:case	_
+31	посетителей	посетитетьей	SYM	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	33	obl	33:obl:среди:gen	_
+32	могли	могль	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	17	advcl	17:advcl:так_как	_
+33	находиться	находитссь	[UNK]	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Mid	32	xcomp	32:xcomp	_
+34	и	тьи	PART	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	35	advmod	35:advmod	_
+35	случайные	случайтьые	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	36	amod	36:amod	_
+36	люди	лин	PROPN	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	32	nsubj	32:nsubj|33:nsubj	SpaceAfter=No
+37	.	ть.	PART	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_8
+# text = Кабинет отличался скромностью, присущей Семену Еремеевичу.
+1	Кабинет	кабитьет	SYM	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	nsubj	2:nsubj	_
+2	отличался	отличалая	PROPN	_	Aspect=Imp|Gender=Fem|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+3	скромностью	скромностьью	INTJ	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	2	obl	2:obl:ins	SpaceAfter=No
+4	,	ть,	PART	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	5	punct	5:punct	_
+5	присущей	присутьей	PUNCT	_	Aspect=Perf|Case=Nom|Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass	3	amod	3:amod	_
+6	Семену	семтьну	ADJ	_	Mood=Cnd	5	iobj	5:iobj	_
+7	Еремеевичу	еремеевичй	PROPN	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	6	flat:name	6:flat:name	SpaceAfter=No
+8	.	ть.	INTJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	punct	2:punct	_
+
+# sent_id = 2003Anketa.xml_9
+# text = В глубине стоял широкий письменный стол с бронзовыми чернильницами и перед ним два кожаных кресла.
+1	В	тьв	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	case	2:case	_
+2	глубине	глубен	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	3	obl	3:obl:в:loc	_
+3	стоял	стой	[UNK]	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+4	широкий	широйн	INTJ	_	Mood=Cnd	6	amod	6:amod	_
+5	письменный	письменйн	PART	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	6	amod	6:amod	_
+6	стол	стоа	INTJ	_	Mood=Cnd	3	nsubj	3:nsubj	_
+7	с	ая	SCONJ	_	Case=Acc|Number=Plur|PronType=Tot	9	case	9:case	_
+8	бронзовыми	бронзовин	PART	_	Case=Acc|Number=Plur|PronType=Tot	9	amod	9:amod	_
+9	чернильницами	чернильницин	PART	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Sing	6	nmod	6:nmod:с:ins	_
+10	и	ин	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	12	cc	12:cc	_
+11	перед	перая	[UNK]	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	12	case	12:case	_
+12	ним	мн	SYM	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Mid	3	conj	0:root|3:conj	_
+13	два	дой	ADP	_	Case=Acc|Number=Plur|PronType=Tot	15	nummod:gov	15:nummod:gov	_
+14	кожаных	кожахн	SCONJ	_	Case=Acc|Number=Plur|PronType=Tot	15	amod	15:amod	_
+15	кресла	креан	[UNK]	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	12	nsubj	12:nsubj	SpaceAfter=No
+16	.	й	PART	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_10
+# text = Справа был стол для заседаний - длинный, накрытый зеленым сукном и с обеих сторон аккуратно заставленный стульями.
+1	Справа	спртьва	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	2	advmod	2:advmod	_
+2	был	лн	DET	_	Case=Acc|Number=Plur|PronType=Tot	0	root	0:root	_
+3	стол	стоа	INTJ	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	2	nsubj	2:nsubj	_
+4	для	ян	PROPN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	5	case	5:case	_
+5	заседаний	заседатьий	ADJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	3	nmod	3:nmod:для:gen	_
+6	-	-]DAP[	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	7	punct	7:punct	_
+7	длинный	длинныь	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	3	parataxis	3:parataxis	SpaceAfter=No
+8	,	,н	PART	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	9	punct	9:punct	_
+9	накрытый	накрыйн	PART	_	Aspect=Perf|Case=Nom|Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass	7	conj	3:parataxis|7:conj	_
+10	зеленым	зелемн	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	11	amod	11:amod	_
+11	сукном	суктьом	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	9	obl	9:obl:ins	_
+12	и	ин	PART	_	Case=Acc|Number=Plur|PronType=Tot	17	cc	17:cc	_
+13	с	о	SYM	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	15	case	15:case	_
+14	обеих	обхн	SYM	_	Aspect=Perf|Gender=Fem|Number=Sing|Tense=Past|Variant=Short|VerbForm=Part|Voice=Pass	15	nummod	15:nummod	_
+15	сторон	стонн	SYM	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	17	obl	17:obl:с:gen	_
+16	аккуратно	аккуратнй	ADJ	_	Case=Acc|Number=Plur|PronType=Tot	17	advmod	17:advmod	_
+17	заставленный	заставленная	ADJ	_	Animacy=Inan|Case=Ins|Gender=Neut|Number=Sing	7	conj	3:parataxis|7:conj	_
+18	стульями	стульин	PART	_	Case=Acc|Number=Plur|PronType=Tot	17	obl	17:obl:ins	SpaceAfter=No
+19	.	ть.	PART	_	Case=Nom|PronType=Ind	2	punct	2:punct	_
+
+# sent_id = 2003Anketa.xml_11
+# text = Семен Еремеевич очень не любил, когда за этот стол кто-нибудь садился, и если видел отодвинутый стул, то всегда собственноручно подвигал его на место, так чтобы спинки образовывали ровную прямую линию.
+1	Семен	сетьен	PART	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	5	nsubj	5:nsubj	_
+2	Еремеевич	еремеевий	ADJ	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	1	appos	1:appos	_
+3	очень	очень_	SYM	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	5	advmod	5:advmod	_
+4	не	тьне	INTJ	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	5	advmod	5:advmod	_
+5	любил	любиь	INTJ	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	0	root	0:root	SpaceAfter=No
+6	,	,_	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	12	punct	12:punct	_
+7	когда	коан	SCONJ	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	12	mark	12:mark	_
+8	за	оз	SYM	_	Animacy=Inan|Case=Ins|Gender=Neut|Number=Sing	10	case	10:case	_
+9	этот	этоа	ADJ	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	10	det	10:det	_
+10	стол	стьол	SYM	_	Case=Nom|Degree=Pos|Number=Plur	12	obl	12:obl:за:acc	_
+11	кто-нибудь	кто-нибьн	PROPN	_	Case=Acc|Number=Plur|PronType=Tot	12	nsubj	12:nsubj	_
+12	садился	садиян	PUNCT	_	Case=Acc|Number=Plur|PronType=Tot	5	ccomp	5:ccomp	SpaceAfter=No
+13	,	ть,	PART	_	Case=Acc|Number=Plur|PronType=Tot	23	punct	23:punct	_
+14	и	ин	INTJ	_	Case=Acc|Number=Plur|PronType=Tot	23	cc	23:cc	_
+15	если	еин	PART	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	16	mark	16:mark	_
+16	видел	видеь	ADJ	_	Aspect=Perf|Gender=Fem|Number=Sing|Tense=Past|Variant=Short|VerbForm=Part|Voice=Pass	23	advcl	23:advcl:если	_
+17	отодвинутый	отодвинуйн	SYM	_	Case=Acc|Number=Plur|PronType=Tot	18	amod	18:amod	_
+18	стул	слн	SYM	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	16	obj	16:obj	SpaceAfter=No
+19	,	,н	PART	_	Case=Acc|Number=Plur|PronType=Tot	16	punct	23:punct	_
+20	то	он	PUNCT	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	23	mark	23:mark	_
+21	всегда	вседсь	ADJ	_	Case=Acc|Number=Plur|PronType=Tot	23	advmod	23:advmod	_
+22	собственноручно	собственнорунсь	SYM	_	Case=Acc|Number=Plur|PronType=Tot	23	advmod	23:advmod	_
+23	подвигал	подвилн	SYM	_	Case=Acc|Number=Plur|PronType=Tot	5	conj	0:root|5:conj	_
+24	его	он	SYM	_	Mood=Cnd	23	obj	23:obj	_
+25	на	нй	PROPN	_	Animacy=Inan|Case=Loc|Gender=Neut|Number=Sing|PronType=Dem	26	case	26:case	_
+26	место	меон	SYM	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	23	obl	23:obl:на:acc	SpaceAfter=No
+27	,	ть,	PART	_	Case=Acc|Number=Plur|PronType=Tot	31	punct	31:punct	_
+28	так	так_	PART	_	Case=Acc|Number=Plur|PronType=Tot	31	mark	31:mark	_
+29	чтобы	чттьбы	INTJ	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	28	fixed	28:fixed	_
+30	спинки	спинок	SYM	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	31	nsubj	31:nsubj	_
+31	образовывали	образовываая	[UNK]	_	Case=Nom|PronType=Ind	23	advcl	23:advcl:чтобы	_
+32	ровную	ровюн	PART	_	Case=Acc|Number=Plur|PronType=Tot	34	amod	34:amod	_
+33	прямую	пряюн	PART	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	34	amod	34:amod	_
+34	линию	линий	SYM	_	Mood=Cnd	31	obj	31:obj	SpaceAfter=No
+35	.	ть.	INTJ	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	5	punct	5:punct	_
+
diff --git a/tests/testdata/datasets/ru_train.conllu b/tests/testdata/datasets/ru_train.conllu
new file mode 100644
index 0000000..126982a
--- /dev/null
+++ b/tests/testdata/datasets/ru_train.conllu
@@ -0,0 +1,255 @@
+# sent_id = 2003Anketa.xml_2
+# text = Начальник областного управления связи Семен Еремеевич был человек простой, приходил на работу всегда вовремя, здоровался с секретаршей за руку и иногда даже писал в стенгазету заметки под псевдонимом "Муха".
+1	Начальник	начальник	NOUN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Sing	8	nsubj	8:nsubj	_
+2	областного	областной	ADJ	_	Case=Gen|Degree=Pos|Gender=Neut|Number=Sing	3	amod	3:amod	_
+3	управления	управление	NOUN	_	Animacy=Inan|Case=Gen|Gender=Neut|Number=Sing	1	nmod	1:nmod:gen	_
+4	связи	связь	NOUN	_	Animacy=Inan|Case=Gen|Gender=Fem|Number=Sing	3	nmod	3:nmod:gen	_
+5	Семен	Семен	PROPN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Sing	1	appos	1:appos	_
+6	Еремеевич	Еремеевич	PROPN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Sing	5	flat:name	5:flat:name	_
+7	был	быть	AUX	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	8	cop	8:cop	_
+8	человек	человек	NOUN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Sing	0	root	0:root	_
+9	простой	простой	ADJ	_	Case=Nom|Degree=Pos|Gender=Masc|Number=Sing	8	amod	8:amod	SpaceAfter=No
+10	,	,	PUNCT	_	_	11	punct	11:punct	_
+11	приходил	приходить	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	8	conj	0:root|8:conj	_
+12	на	на	ADP	_	_	13	case	13:case	_
+13	работу	работа	NOUN	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	11	obl	11:obl:на:acc	_
+14	всегда	всегда	ADV	_	Degree=Pos	11	advmod	11:advmod	_
+15	вовремя	вовремя	ADV	_	Degree=Pos	11	advmod	11:advmod	SpaceAfter=No
+16	,	,	PUNCT	_	_	17	punct	17:punct	_
+17	здоровался	здороваться	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Mid	8	conj	0:root|8:conj	_
+18	с	с	ADP	_	_	19	case	19:case	_
+19	секретаршей	секретарша	NOUN	_	Animacy=Anim|Case=Ins|Gender=Fem|Number=Sing	17	obl	17:obl:с:ins	_
+20	за	за	ADP	_	_	21	case	21:case	_
+21	руку	рука	NOUN	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	17	obl	17:obl:за:acc	_
+22	и	и	CCONJ	_	_	25	cc	25:cc	_
+23	иногда	иногда	ADV	_	Degree=Pos	25	advmod	25:advmod	_
+24	даже	даже	PART	_	_	25	advmod	25:advmod	_
+25	писал	писать	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	8	conj	0:root|8:conj	_
+26	в	в	ADP	_	_	27	case	27:case	_
+27	стенгазету	стенгазета	NOUN	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	25	obl	25:obl:в:acc	_
+28	заметки	заметка	NOUN	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Plur	25	obj	25:obj	_
+29	под	под	ADP	_	_	30	case	30:case	_
+30	псевдонимом	псевдоним	NOUN	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Sing	25	obl	25:obl:под:ins	_
+31	"	"	PUNCT	_	_	32	punct	32:punct	SpaceAfter=No
+32	Муха	муха	NOUN	_	Animacy=Anim|Case=Nom|Gender=Fem|Number=Sing	30	nmod	30:nmod:nom	SpaceAfter=No
+33	"	"	PUNCT	_	_	32	punct	32:punct	SpaceAfter=No
+34	.	.	PUNCT	_	_	8	punct	8:punct	_
+
+# sent_id = 2003Anketa.xml_3
+# text = В приемной его с утра ожидали посетители, - кое-кто с важными делами, а кое-кто и с такими, которые легко можно было решить в нижестоящих инстанциях, не затрудняя Семена Еремеевича.
+1	В	в	ADP	_	_	2	case	2:case	_
+2	приемной	приемная	NOUN	_	Animacy=Inan|Case=Loc|Gender=Fem|Number=Sing	6	obl	6:obl:в:loc	_
+3	его	он	PRON	_	Case=Acc|Gender=Masc|Number=Sing|Person=3|PronType=Prs	6	obj	6:obj	_
+4	с	с	ADP	_	_	5	case	5:case	_
+5	утра	утро	NOUN	_	Animacy=Inan|Case=Gen|Gender=Neut|Number=Sing	6	obl	6:obl:с:gen	_
+6	ожидали	ожидать	VERB	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+7	посетители	посетитель	NOUN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	6	nsubj	6:nsubj	SpaceAfter=No
+8	,	,	PUNCT	_	_	13	punct	13:punct	_
+9	-	-	PUNCT	_	_	13	punct	13:punct	_
+10	кое-кто	кое-кто	PRON	_	Case=Nom|PronType=Ind	13	nsubj	13:nsubj	_
+11	с	с	ADP	_	_	13	case	13:case	_
+12	важными	важный	ADJ	_	Case=Ins|Degree=Pos|Number=Plur	13	amod	13:amod	_
+13	делами	дело	NOUN	_	Animacy=Inan|Case=Ins|Gender=Neut|Number=Plur	7	parataxis	7:parataxis	SpaceAfter=No
+14	,	,	PUNCT	_	_	19	punct	19:punct	_
+15	а	а	CCONJ	_	_	19	cc	19:cc	_
+16	кое-кто	кое-кто	PRON	_	Case=Nom|PronType=Ind	19	nsubj	19:nsubj	_
+17	и	и	PART	_	_	19	advmod	19:advmod	_
+18	с	с	ADP	_	_	19	case	19:case	_
+19	такими	такой	DET	_	Case=Ins|Number=Plur|PronType=Dem	13	det	13:det|25:obj	SpaceAfter=No
+20	,	,	PUNCT	_	_	23	punct	23:punct	_
+21	которые	который	PRON	_	Case=Acc|PronType=Int,Rel	25	obj	19:ref	_
+22	легко	легко	ADV	_	Degree=Pos	23	advmod	23:advmod	_
+23	можно	можно	ADV	_	Degree=Pos	19	acl:relcl	19:acl:relcl	_
+24	было	быть	AUX	_	Aspect=Imp|Gender=Neut|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	23	cop	23:cop	_
+25	решить	решить	VERB	_	Aspect=Perf|VerbForm=Inf|Voice=Act	23	csubj	23:csubj	_
+26	в	в	ADP	_	_	28	case	28:case	_
+27	нижестоящих	нижестоящий	ADJ	_	Case=Loc|Degree=Pos|Number=Plur	28	amod	28:amod	_
+28	инстанциях	инстанция	NOUN	_	Animacy=Inan|Case=Loc|Gender=Fem|Number=Plur	25	obl	25:obl:в:loc	SpaceAfter=No
+29	,	,	PUNCT	_	_	31	punct	31:punct	_
+30	не	не	PART	_	_	31	advmod	31:advmod	_
+31	затрудняя	затруднять	VERB	_	Aspect=Imp|Tense=Pres|VerbForm=Conv|Voice=Act	25	advcl	25:advcl	_
+32	Семена	Семен	PROPN	_	Animacy=Anim|Case=Gen|Gender=Masc|Number=Sing	31	obl	31:obl:gen	_
+33	Еремеевича	Еремеевич	PROPN	_	Animacy=Anim|Case=Gen|Gender=Masc|Number=Sing	32	flat:name	32:flat:name	SpaceAfter=No
+34	.	.	PUNCT	_	_	6	punct	6:punct	_
+
+# sent_id = 2003Anketa.xml_4
+# text = Однако стиль работы Семена Еремеевича заключался в том, чтобы принимать всех желающих и лично вникать в дело.
+1	Однако	однако	ADV	_	Degree=Pos	6	advmod	6:advmod	_
+2	стиль	стиль	NOUN	_	Animacy=Inan|Case=Nom|Gender=Masc|Number=Sing	6	nsubj	6:nsubj	_
+3	работы	работа	NOUN	_	Animacy=Inan|Case=Gen|Gender=Fem|Number=Sing	2	nmod	2:nmod:gen	_
+4	Семена	Семен	PROPN	_	Animacy=Anim|Case=Gen|Gender=Masc|Number=Sing	3	nmod	3:nmod:gen	_
+5	Еремеевича	Еремеевич	PROPN	_	Animacy=Anim|Case=Gen|Gender=Masc|Number=Sing	4	flat:name	4:flat:name	_
+6	заключался	заключаться	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Mid	0	root	0:root	_
+7	в	в	ADP	_	_	8	case	8:case	_
+8	том	то	PRON	_	Animacy=Inan|Case=Loc|Gender=Neut|Number=Sing|PronType=Dem	6	obl	6:obl:в:loc	SpaceAfter=No
+9	,	,	PUNCT	_	_	11	punct	11:punct	_
+10	чтобы	чтобы	SCONJ	_	Mood=Cnd	11	mark	11:mark	_
+11	принимать	принимать	VERB	_	Aspect=Imp|VerbForm=Inf|Voice=Act	8	advcl	8:advcl:чтобы	_
+12	всех	весь	DET	_	Case=Acc|Number=Plur|PronType=Tot	13	det	13:det	_
+13	желающих	желать	VERB	_	Animacy=Anim|Aspect=Imp|Case=Acc|Number=Plur|Tense=Pres|VerbForm=Part|Voice=Act	11	xcomp	11:xcomp	_
+14	и	и	CCONJ	_	_	16	cc	16:cc	_
+15	лично	лично	ADV	_	Degree=Pos	16	advmod	16:advmod	_
+16	вникать	вникать	VERB	_	Aspect=Imp|VerbForm=Inf|Voice=Act	11	conj	8:advcl:чтобы|11:conj	_
+17	в	в	ADP	_	_	18	case	18:case	_
+18	дело	дело	NOUN	_	Animacy=Inan|Case=Acc|Gender=Neut|Number=Sing	16	obl	16:obl:в:acc	SpaceAfter=No
+19	.	.	PUNCT	_	_	6	punct	6:punct	_
+
+# sent_id = 2003Anketa.xml_5
+# text = Приемная была обставлена просто, но по-деловому.
+1	Приемная	приемная	NOUN	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Sing	3	nsubj:pass	3:nsubj:pass	_
+2	была	быть	AUX	_	Aspect=Imp|Gender=Fem|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	3	aux:pass	3:aux:pass	_
+3	обставлена	обставить	VERB	_	Aspect=Perf|Gender=Fem|Number=Sing|Tense=Past|Variant=Short|VerbForm=Part|Voice=Pass	0	root	0:root	_
+4	просто	просто	ADV	_	Degree=Pos	3	advmod	3:advmod	SpaceAfter=No
+5	,	,	PUNCT	_	_	7	punct	7:punct	_
+6	но	но	CCONJ	_	_	7	cc	7:cc	_
+7	по-деловому	по-деловому	ADV	_	Degree=Pos	4	conj	3:advmod|4:conj	SpaceAfter=No
+8	.	.	PUNCT	_	_	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_6
+# text = У двери стоял стол секретарши, на столе - пишущая машинка с широкой кареткой.
+1	У	у	ADP	_	_	2	case	2:case	_
+2	двери	дверь	NOUN	_	Animacy=Inan|Case=Gen|Gender=Fem|Number=Sing	3	obl	3:obl:у:gen	_
+3	стоял	стоять	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+4	стол	стол	NOUN	_	Animacy=Inan|Case=Nom|Gender=Masc|Number=Sing	3	nsubj	3:nsubj	_
+5	секретарши	секретарша	NOUN	_	Animacy=Anim|Case=Gen|Gender=Fem|Number=Sing	4	nmod	4:nmod:gen	SpaceAfter=No
+6	,	,	PUNCT	_	_	8	punct	8:punct	_
+7	на	на	ADP	_	_	8	case	8:case	_
+8	столе	стол	NOUN	_	Animacy=Inan|Case=Loc|Gender=Masc|Number=Sing	3	conj	0:root|3:conj	_
+9	-	-	PUNCT	_	_	11	punct	11:punct	_
+10	пишущая	писать	VERB	_	Aspect=Imp|Case=Nom|Gender=Fem|Number=Sing|Tense=Pres|VerbForm=Part|Voice=Act	11	amod	11:amod	_
+11	машинка	машинка	NOUN	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Sing	8	nsubj	8:nsubj	_
+12	с	с	ADP	_	_	14	case	14:case	_
+13	широкой	широкий	ADJ	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	14	amod	14:amod	_
+14	кареткой	каретка	NOUN	_	Animacy=Inan|Case=Ins|Gender=Fem|Number=Sing	11	nmod	11:nmod:с:ins	SpaceAfter=No
+15	.	.	PUNCT	_	_	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_7
+# text = В углу висел репродуктор и играло радио для развлечения ожидающих и еще для того, чтобы заглушать голос начальника, доносившийся из кабинета, так как, бесспорно, среди посетителей могли находиться и случайные люди.
+1	В	в	ADP	_	_	2	case	2:case	_
+2	углу	угол	NOUN	_	Animacy=Inan|Case=Loc|Gender=Masc|Number=Sing	3	obl	3:obl:в:loc	_
+3	висел	висеть	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+4	репродуктор	репродуктор	NOUN	_	Animacy=Inan|Case=Nom|Gender=Masc|Number=Sing	3	nsubj	3:nsubj	_
+5	и	и	CCONJ	_	_	6	cc	6:cc	_
+6	играло	играть	VERB	_	Aspect=Imp|Gender=Neut|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	3	conj	0:root|3:conj	_
+7	радио	радио	NOUN	_	Animacy=Inan|Case=Nom|Gender=Neut|Number=Sing	6	nsubj	6:nsubj	_
+8	для	для	ADP	_	_	9	case	9:case	_
+9	развлечения	развлечение	NOUN	_	Animacy=Inan|Case=Gen|Gender=Neut|Number=Sing	6	obl	6:obl:для:gen	_
+10	ожидающих	ожидать	VERB	_	Aspect=Imp|Case=Gen|Number=Plur|Tense=Pres|VerbForm=Part|Voice=Act	9	acl	9:acl	_
+11	и	и	CCONJ	_	_	14	cc	14:cc	_
+12	еще	еще	ADV	_	Degree=Pos	14	obl	14:obl	_
+13	для	для	ADP	_	_	14	case	14:case	_
+14	того	то	PRON	_	Animacy=Inan|Case=Gen|Gender=Neut|Number=Sing|PronType=Dem	9	conj	6:obl:для:gen|9:conj	SpaceAfter=No
+15	,	,	PUNCT	_	_	17	punct	17:punct	_
+16	чтобы	чтобы	SCONJ	_	Mood=Cnd	17	mark	17:mark	_
+17	заглушать	заглушать	VERB	_	Aspect=Imp|VerbForm=Inf|Voice=Act	14	advcl	14:advcl:чтобы	_
+18	голос	голос	NOUN	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	17	obj	17:obj	_
+19	начальника	начальник	NOUN	_	Animacy=Anim|Case=Gen|Gender=Masc|Number=Sing	18	nmod	18:nmod:gen	SpaceAfter=No
+20	,	,	PUNCT	_	_	21	punct	21:punct	_
+21	доносившийся	доноситься	VERB	_	Animacy=Inan|Aspect=Imp|Case=Acc|Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part|Voice=Mid	18	acl	18:acl	_
+22	из	из	ADP	_	_	23	case	23:case	_
+23	кабинета	кабинет	NOUN	_	Animacy=Inan|Case=Gen|Gender=Masc|Number=Sing	21	obl	21:obl:из:gen	SpaceAfter=No
+24	,	,	PUNCT	_	_	21	punct	21:punct	_
+25	так	так	ADV	_	Degree=Pos	32	mark	32:mark	_
+26	как	как	SCONJ	_	_	25	fixed	25:fixed	SpaceAfter=No
+27	,	,	PUNCT	_	_	25	punct	25:punct	_
+28	бесспорно	бесспорно	ADV	_	Degree=Pos	32	parataxis	32:parataxis	SpaceAfter=No
+29	,	,	PUNCT	_	_	28	punct	28:punct	_
+30	среди	среди	ADP	_	_	31	case	31:case	_
+31	посетителей	посетитель	NOUN	_	Animacy=Anim|Case=Gen|Gender=Masc|Number=Plur	33	obl	33:obl:среди:gen	_
+32	могли	мочь	VERB	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	17	advcl	17:advcl:так_как	_
+33	находиться	находиться	VERB	_	Aspect=Imp|VerbForm=Inf|Voice=Mid	32	xcomp	32:xcomp	_
+34	и	и	PART	_	_	35	advmod	35:advmod	_
+35	случайные	случайный	ADJ	_	Case=Nom|Degree=Pos|Number=Plur	36	amod	36:amod	_
+36	люди	человек	NOUN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Plur	32	nsubj	32:nsubj|33:nsubj	SpaceAfter=No
+37	.	.	PUNCT	_	_	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_8
+# text = Кабинет отличался скромностью, присущей Семену Еремеевичу.
+1	Кабинет	кабинет	NOUN	_	Animacy=Inan|Case=Nom|Gender=Masc|Number=Sing	2	nsubj	2:nsubj	_
+2	отличался	отличаться	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Mid	0	root	0:root	_
+3	скромностью	скромность	NOUN	_	Animacy=Inan|Case=Ins|Gender=Fem|Number=Sing	2	obl	2:obl:ins	SpaceAfter=No
+4	,	,	PUNCT	_	_	5	punct	5:punct	_
+5	присущей	присущий	ADJ	_	Case=Ins|Degree=Pos|Gender=Fem|Number=Sing	3	amod	3:amod	_
+6	Семену	Семен	PROPN	_	Animacy=Anim|Case=Dat|Gender=Masc|Number=Sing	5	iobj	5:iobj	_
+7	Еремеевичу	Еремеевич	PROPN	_	Animacy=Anim|Case=Dat|Gender=Masc|Number=Sing	6	flat:name	6:flat:name	SpaceAfter=No
+8	.	.	PUNCT	_	_	2	punct	2:punct	_
+
+# sent_id = 2003Anketa.xml_9
+# text = В глубине стоял широкий письменный стол с бронзовыми чернильницами и перед ним два кожаных кресла.
+1	В	в	ADP	_	_	2	case	2:case	_
+2	глубине	глубина	NOUN	_	Animacy=Inan|Case=Loc|Gender=Fem|Number=Sing	3	obl	3:obl:в:loc	_
+3	стоял	стоять	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+4	широкий	широкий	ADJ	_	Case=Nom|Degree=Pos|Gender=Masc|Number=Sing	6	amod	6:amod	_
+5	письменный	письменный	ADJ	_	Case=Nom|Degree=Pos|Gender=Masc|Number=Sing	6	amod	6:amod	_
+6	стол	стол	NOUN	_	Animacy=Inan|Case=Nom|Gender=Masc|Number=Sing	3	nsubj	3:nsubj	_
+7	с	с	ADP	_	_	9	case	9:case	_
+8	бронзовыми	бронзовый	ADJ	_	Case=Ins|Degree=Pos|Number=Plur	9	amod	9:amod	_
+9	чернильницами	чернильница	NOUN	_	Animacy=Inan|Case=Ins|Gender=Fem|Number=Plur	6	nmod	6:nmod:с:ins	_
+10	и	и	CCONJ	_	_	12	cc	12:cc	_
+11	перед	перед	ADP	_	_	12	case	12:case	_
+12	ним	он	PRON	_	Case=Ins|Gender=Masc|Number=Sing|Person=3|PronType=Prs	3	conj	0:root|3:conj	_
+13	два	два	NUM	_	Case=Nom|Gender=Neut|NumType=Card	15	nummod:gov	15:nummod:gov	_
+14	кожаных	кожаный	ADJ	_	Case=Gen|Degree=Pos|Number=Plur	15	amod	15:amod	_
+15	кресла	кресло	NOUN	_	Animacy=Inan|Case=Gen|Gender=Neut|Number=Sing	12	nsubj	12:nsubj	SpaceAfter=No
+16	.	.	PUNCT	_	_	3	punct	3:punct	_
+
+# sent_id = 2003Anketa.xml_10
+# text = Справа был стол для заседаний - длинный, накрытый зеленым сукном и с обеих сторон аккуратно заставленный стульями.
+1	Справа	справа	ADV	_	Degree=Pos	2	advmod	2:advmod	_
+2	был	быть	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	_
+3	стол	стол	NOUN	_	Animacy=Inan|Case=Nom|Gender=Masc|Number=Sing	2	nsubj	2:nsubj	_
+4	для	для	ADP	_	_	5	case	5:case	_
+5	заседаний	заседание	NOUN	_	Animacy=Inan|Case=Gen|Gender=Neut|Number=Plur	3	nmod	3:nmod:для:gen	_
+6	-	-	PUNCT	_	_	7	punct	7:punct	_
+7	длинный	длинный	ADJ	_	Case=Nom|Degree=Pos|Gender=Masc|Number=Sing	3	parataxis	3:parataxis	SpaceAfter=No
+8	,	,	PUNCT	_	_	9	punct	9:punct	_
+9	накрытый	накрыть	VERB	_	Aspect=Perf|Case=Nom|Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass	7	conj	3:parataxis|7:conj	_
+10	зеленым	зеленый	ADJ	_	Case=Ins|Degree=Pos|Gender=Neut|Number=Sing	11	amod	11:amod	_
+11	сукном	сукно	NOUN	_	Animacy=Inan|Case=Ins|Gender=Neut|Number=Sing	9	obl	9:obl:ins	_
+12	и	и	CCONJ	_	_	17	cc	17:cc	_
+13	с	с	ADP	_	_	15	case	15:case	_
+14	обеих	оба	NUM	_	Case=Gen|Gender=Fem|NumType=Card	15	nummod	15:nummod	_
+15	сторон	сторона	NOUN	_	Animacy=Inan|Case=Gen|Gender=Fem|Number=Plur	17	obl	17:obl:с:gen	_
+16	аккуратно	аккуратно	ADV	_	Degree=Pos	17	advmod	17:advmod	_
+17	заставленный	заставить	VERB	_	Aspect=Perf|Case=Nom|Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass	7	conj	3:parataxis|7:conj	_
+18	стульями	стул	NOUN	_	Animacy=Inan|Case=Ins|Gender=Masc|Number=Plur	17	obl	17:obl:ins	SpaceAfter=No
+19	.	.	PUNCT	_	_	2	punct	2:punct	_
+
+# sent_id = 2003Anketa.xml_11
+# text = Семен Еремеевич очень не любил, когда за этот стол кто-нибудь садился, и если видел отодвинутый стул, то всегда собственноручно подвигал его на место, так чтобы спинки образовывали ровную прямую линию.
+1	Семен	Семен	PROPN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Sing	5	nsubj	5:nsubj	_
+2	Еремеевич	Еремеевич	PROPN	_	Animacy=Anim|Case=Nom|Gender=Masc|Number=Sing	1	appos	1:appos	_
+3	очень	очень	ADV	_	Degree=Pos	5	advmod	5:advmod	_
+4	не	не	PART	_	_	5	advmod	5:advmod	_
+5	любил	любить	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	0	root	0:root	SpaceAfter=No
+6	,	,	PUNCT	_	_	12	punct	12:punct	_
+7	когда	когда	SCONJ	_	_	12	mark	12:mark	_
+8	за	за	ADP	_	_	10	case	10:case	_
+9	этот	этот	DET	_	Case=Acc|Gender=Masc|Number=Sing|PronType=Dem	10	det	10:det	_
+10	стол	стол	NOUN	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	12	obl	12:obl:за:acc	_
+11	кто-нибудь	кто-нибудь	PRON	_	Case=Nom|PronType=Ind	12	nsubj	12:nsubj	_
+12	садился	садиться	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Mid	5	ccomp	5:ccomp	SpaceAfter=No
+13	,	,	PUNCT	_	_	23	punct	23:punct	_
+14	и	и	CCONJ	_	_	23	cc	23:cc	_
+15	если	если	SCONJ	_	_	16	mark	16:mark	_
+16	видел	видеть	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	23	advcl	23:advcl:если	_
+17	отодвинутый	отодвинуть	VERB	_	Animacy=Inan|Aspect=Perf|Case=Acc|Gender=Masc|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass	18	amod	18:amod	_
+18	стул	стул	NOUN	_	Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing	16	obj	16:obj	SpaceAfter=No
+19	,	,	PUNCT	_	_	16	punct	23:punct	_
+20	то	то	SCONJ	_	_	23	mark	23:mark	_
+21	всегда	всегда	ADV	_	Degree=Pos	23	advmod	23:advmod	_
+22	собственноручно	собственноручно	ADV	_	Degree=Pos	23	advmod	23:advmod	_
+23	подвигал	подвигать	VERB	_	Aspect=Imp|Gender=Masc|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act	5	conj	0:root|5:conj	_
+24	его	он	PRON	_	Case=Acc|Gender=Masc|Number=Sing|Person=3|PronType=Prs	23	obj	23:obj	_
+25	на	на	ADP	_	_	26	case	26:case	_
+26	место	место	NOUN	_	Animacy=Inan|Case=Acc|Gender=Neut|Number=Sing	23	obl	23:obl:на:acc	SpaceAfter=No
+27	,	,	PUNCT	_	_	31	punct	31:punct	_
+28	так	так	ADV	_	Degree=Pos	31	mark	31:mark	_
+29	чтобы	чтобы	SCONJ	_	Mood=Cnd	28	fixed	28:fixed	_
+30	спинки	спинка	NOUN	_	Animacy=Inan|Case=Nom|Gender=Fem|Number=Plur	31	nsubj	31:nsubj	_
+31	образовывали	образовывать	VERB	_	Aspect=Imp|Mood=Ind|Number=Plur|Tense=Past|VerbForm=Fin|Voice=Act	23	advcl	23:advcl:чтобы	_
+32	ровную	ровный	ADJ	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	34	amod	34:amod	_
+33	прямую	прямой	ADJ	_	Case=Acc|Degree=Pos|Gender=Fem|Number=Sing	34	amod	34:amod	_
+34	линию	линия	NOUN	_	Animacy=Inan|Case=Acc|Gender=Fem|Number=Sing	31	obj	31:obj	SpaceAfter=No
+35	.	.	PUNCT	_	_	5	punct	5:punct	_
+
diff --git a/tests/testdata/udtube_config.yaml b/tests/testdata/udtube_config.yaml
new file mode 100644
index 0000000..0759b38
--- /dev/null
+++ b/tests/testdata/udtube_config.yaml
@@ -0,0 +1,10 @@
+checkpoint:
+  save_last: 'link'  # Gives us a standard name.
+  monitor: val_loss
+data:
+  batch_size: 5  # 2 batches per epoch.
+trainer:
+  accelerator: cpu  # Because that's what CircleCI has.
+  enable_progress_bar: false
+  max_epochs: 5
+seed_everything: 42
diff --git a/tests/udtube_test.py b/tests/udtube_test.py
new file mode 100644
index 0000000..f300d8a
--- /dev/null
+++ b/tests/udtube_test.py
@@ -0,0 +1,105 @@
+"""Full tests of training and prediction.
+
+This runs five epochs of training over a small toy data set, attempting to
+overfit, then compares the resubstitution predictions on this set to
+previously computed results. As such this is essentially a change-detector
+test. Currently, English (en), Greek (el), and Russian (ru) are supported."""
+
+import difflib
+import os
+import tempfile
+import unittest
+
+from parameterized import parameterized
+
+from udtube import cli
+
+# Directory the unit test is located in, relative to the working directory.
+DIR = os.path.relpath(os.path.dirname(__file__), os.getcwd())
+CONFIG_PATH = os.path.join(DIR, "testdata/udtube_config.yaml")
+DATASETS_DIR = os.path.join(DIR, "testdata/datasets")
+
+
+class UDTubeTest(unittest.TestCase):
+    def assertFileExists(self, path: str):
+        self.assertTrue(os.path.exists(path), msg=f"file {path} not found")
+
+    def setUp(self):
+        self.tempdir = tempfile.TemporaryDirectory(prefix="udtube_test-")
+        self.assertFileExists(CONFIG_PATH)
+
+    def tearDown(self):
+        self.tempdir.cleanup()
+
+    @parameterized.expand(
+        [
+            # TODO: commented out to test CircleCI test resource limitations.
+            # ("el", True),
+            ("en", True),
+            # TODO: ditto.
+            # ("ru", False),  # Russian doesn't have XPOS.
+        ]
+    )
+    def test_model(self, langcode: str, use_xpos: bool):
+        # Fits model.
+        train_path = os.path.join(DATASETS_DIR, f"{langcode}_train.conllu")
+        self.assertFileExists(train_path)
+        expected_path = os.path.join(
+            DATASETS_DIR, f"{langcode}_expected.conllu"
+        )
+        self.assertFileExists(expected_path)
+        model_dir = os.path.join(self.tempdir.name, "model")
+        cli.udtube_python_interface(
+            [
+                "fit",
+                f"--config={CONFIG_PATH}",
+                f"--data.model_dir={model_dir}",
+                f"--data.train={train_path}",
+                # We are trying to overfit on the training data.
+                f"--data.val={train_path}",
+                f"--model.use_xpos={use_xpos}",
+            ]
+        )
+        checkpoint_path = (
+            f"{self.tempdir.name}/model/lightning_logs/"
+            "version_0/checkpoints/last.ckpt"
+        )
+        self.assertFileExists(checkpoint_path)
+        # Predicts on "expected" data.
+        predicted_path = os.path.join(
+            self.tempdir.name, f"{langcode}_predicted.conllu"
+        )
+        cli.udtube_python_interface(
+            [
+                "predict",
+                f"--config={CONFIG_PATH}",
+                f"--data.model_dir={model_dir}",
+                f"--data.predict={expected_path}",
+                f"--model.use_xpos={use_xpos}",
+                f"--prediction.path={predicted_path}",
+            ]
+        )
+        self.assertFileExists(predicted_path)
+        diff = self._diff(predicted_path, expected_path)
+        self.assertEqual(diff, [], f"Differences found:\n{diff}")
+
+    @staticmethod
+    def _diff(predicted_path: str, expected_path: str) -> list:
+        with (
+            open(predicted_path, "r") as predicted,
+            open(expected_path, "r") as expected,
+        ):
+            diff = list(
+                difflib.unified_diff(
+                    predicted.readlines(),
+                    expected.readlines(),
+                    fromfile=predicted_path,
+                    tofile=expected_path,
+                    n=1,
+                )
+            )
+        return diff
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/udtube/callbacks.py b/udtube/callbacks.py
index d06540a..969b7e5 100644
--- a/udtube/callbacks.py
+++ b/udtube/callbacks.py
@@ -34,8 +34,8 @@ def __init__(
         self.mapper = data.Mapper.read(model_dir)
 
     def __del__(self):
-        # This appears to be harmless on sys.stdout.
-        self.sink.close()
+        if self.sink is not sys.stdout:
+            self.sink.close()
 
     # Required API.
 
diff --git a/udtube/cli.py b/udtube/cli.py
index 950d87c..39235a7 100644
--- a/udtube/cli.py
+++ b/udtube/cli.py
@@ -7,6 +7,17 @@
 from . import callbacks, data, models, trainers
 
 
+def udtube_python_interface(args: cli.ArgsType = None):
+    """Interface to use models through Python."""
+    UDTubeCLI(
+        models.UDTube,
+        data.DataModule,
+        auto_configure_optimizers=False,
+        trainer_class=trainers.Trainer,
+        args=args,
+    )
+
+
 class UDTubeCLI(cli.LightningCLI):
     """The UDTube CLI interface.