Multiple sequence alignments in historical linguistics

Slide 1

Slide 1 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Multiple Sequence Alignment in Historical Linguistics A Sound Class Based Approach Johann-Mattis List∗ ∗Institute for Romance Languages and Literature Heinrich Heine University Düsseldorf 2011/01/06 1 / 32

Slide 2

Slide 2 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Structure of the Talk Introduction Sequences Alignments Automatic Alignment Analyses Pairwise Sequence Alignment Multiple Sequence Alignment Alignments in Historical Linguistics Similarity Sound Classes LingPy Main Ideas Working Principle Scoring Performance of the Method Usage Example TPPSR 2 / 32

Slide 3

Slide 3 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Sequences Alignments Introduction Introduction - Sequences - - Alignments - 3 / 32

Slide 4

Slide 4 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Sequences Alignments Sequences Sets Sets are unordered lists of unique objects. Sets are compared by comparing the objects of different sets. Sequences Sequences are ordered lists of non-unique objects. Sequences are compared by comparing both the objects (segments) and the structure of different sequences. 4 / 32

Slide 5

Slide 5 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Sequences Alignments Alignments Sequence Alignment In alignment analyses, the corresponding segments of two or more sequences are ordered in such a way that they are set against each other. Segments which do not correspond to any other segments are marked by gaps (-). In this way, both, the structure and the segments of two or more sequences can be compared. 5 / 32

Slide 6

Slide 6 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Sequences Alignments Alignments ʧ ɪ l ɐ vʲ ɛ k ʧ o v ɛ k 1 6 / 32

Slide 7

Slide 7 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Sequences Alignments Alignments ʧ ɪ l ɐ vʲ ɛ k ʧ o v ɛ k 1 6 / 32

Slide 8

Slide 8 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Sequences Alignments Alignments ʧ ɪ l ɐ vʲ ɛ k ʧ - - o v ɛ k 1 6 / 32

Slide 9

Slide 9 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Pairwise Sequence Alignment Multiple Sequence Alignment Automatic Alignment Analyses h j - ä r t a - h - e - r z - - h - e a r t - - c - - o r d i s hjärta herz heart cordis 1 7 / 32

Slide 10

Slide 10 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Pairwise Sequence Alignment Multiple Sequence Alignment Pairwise Sequence Alignment Create a matrix which confronts all segments of two sequences, either with each other, or with gaps. Seek the path through the matrix which is of the lowest cost (or the highest score). Calculate the cost (or the score) cumulatively by scoring the matching of segments with segments and with gaps by means of a specific scoring function. 8 / 32

Slide 11

Slide 11 text

Slide 12

Slide 12 text

Slide 13

Slide 13 text

Slide 14

Slide 14 text

Slide 15

Slide 15 text

Slide 16

Slide 16 text

Slide 17

Slide 17 text

Slide 18

Slide 18 text

Slide 19

Slide 19 text

Slide 20

Slide 20 text

Slide 21

Slide 21 text

Slide 22

Slide 22 text

Slide 23

Slide 23 text

Slide 24

Slide 24 text

Slide 25

Slide 25 text

Slide 26

Slide 26 text

Slide 27

Slide 27 text

Slide 28

Slide 28 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Pairwise Sequence Alignment Multiple Sequence Alignment Multiple Sequence Alignments Guide Tree Heuristics Due to computational restrictions, multiple sequence alignment (MSA) is based on heuristics. Heuristics based on guide-trees are the most common ones used in computational biology. Based on pairwise alignment scores, a guide-tree is reconstructed, and the sequences are stepwise added to the MSA along it (Feng & Dolittle 1987). 10 / 32

Slide 29

Slide 29 text

Slide 30

Slide 30 text

Slide 31

Slide 31 text

Slide 32

Slide 32 text

Slide 33

Slide 33 text

Slide 34

Slide 34 text

Slide 35

Slide 35 text

Slide 36

Slide 36 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Pairwise Sequence Alignment Multiple Sequence Alignment Multiple Sequence Alignment ʧɪlɐvʲɛk Russian ʧlovʲɛk Czech ʧwɔvʲɛk Polish ʧovɛk Bulgarian ʧ ɪ l ɐ vʲ ɛ k ʧ - l o vʲ ɛ k ʧ w ɔ vʲ ɛ k ʧ - o v ɛ k ʧ ɪ l ɐ vʲ ɛ k ʧ ˗ l o vʲ ɛ k ʧ ˗ w ɔ vʲ ɛ k ʧ - - o v ɛ k 11 / 32

Slide 37

Slide 37 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Pairwise Sequence Alignment Multiple Sequence Alignment Multiple Sequence Alignment Profiles The guide-tree heuristic can be enhanced by the application of profiles. A profile consists of the relative frequency of all segments of an MSA in all its positions, thus, a profile represents an MSA as a sequence of vectors. Aligning profiles to profiles instead of aligning two representative sequences of two given MSA yields better results, since more information can be taken into account. 12 / 32

Slide 38

Slide 38 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Pairwise Sequence Alignment Multiple Sequence Alignment Multiple Sequence Alignment ʧɪlɐvʲɛk Russian ʧlovʲɛk Czech ʧwɔvʲɛk Polish ʧovɛk Bulgarian ʧ ɪ l ɐ vʲ ɛ k ʧ - l o vʲ ɛ k ʧ w ɔ vʲ ɛ k ʧ - o v ɛ k ʧ ɪ l ɐ vʲ ɛ k ʧ ˗ l o vʲ ɛ k ʧ ˗ w ɔ vʲ ɛ k ʧ - - o v ɛ k 13 / 32

Slide 39

Slide 39 text

Slide 40

Slide 40 text

Slide 41

Slide 41 text

Slide 42

Slide 42 text

Slide 43

Slide 43 text

Slide 44

Slide 44 text

Slide 45

Slide 45 text

Slide 46

Slide 46 text

Slide 47

Slide 47 text

Slide 48

Slide 48 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Alignments in Historical Linguistics *ph2 tēr *faθēr father 1 14 / 32

Slide 49

Slide 49 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity Synchronic Similarity Sounds in different languages are judged to be similar, if they show resemblences regarding the way they are produced or perceived. Diachronic Similarity Sounds in different languages are judged to be similar, if they go back to a common ancestor. 15 / 32

Slide 50

Slide 50 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity Language Word Meaning Mandarin ma⁵⁵ma³ “mother” German mama “mother” Russian tak “in this way” German tʰaːk “day” 16 / 32

Slide 51

Slide 51 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity Language Word Meaning German ʦʰaːn “tooth” English tʊːθ “tooth” Italian dɛntɛ “tooth” French dɑ̃ “tooth” 16 / 32

Slide 52

Slide 52 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity . . German ʦʰ aː n - * Proto-Germanic t a n d English t ʊː θ - ** Proto-Indo-European d o n t Italian d ɛ n t ɛ * Proto-Romance d e n t French d ã - - funktionier endlich! 17 / 32

Slide 53

Slide 53 text

Slide 54

Slide 54 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity . . German ʦʰ aː n - * Proto-Germanic t a n d English t ʊː - θ ** Proto-Indo-European d o n t Italian d ɛ n t ə * Proto-Romance d e n t French d ã - - funktionier endlich! 17 / 32

Slide 55

Slide 55 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity . . German ʦʰ aː n - * Proto-Germanic t a n θ English t ʊː - θ ** Proto-Indo-European d o n t Italian d ɛ n t ə * Proto-Romance d e n t French d ã - - funktionier endlich! 17 / 32

Slide 56

Slide 56 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity . . German ʦʰ aː n - * Proto-Germanic t a n θ English t ʊː - θ ** Proto-Indo-European d o n t Italian d ɛ n t ə * Proto-Romance d e n t French d ã - - funktionier endlich! 17 / 32

Slide 57

Slide 57 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity . . German ʦʰ aː n - * Proto-Germanic t a n θ English t ʊː - θ ** Proto-Indo-European d o n t Italian d ɛ n t ə * Proto-Romance d e n t French d ã - - funktionier endlich! 17 / 32

Slide 58

Slide 58 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Similarity . . German ʦʰ aː n - * Proto-Germanic t a n d English t ʊː - θ ** Proto-Indo-European d o n t Italian d ɛ n t ə * Proto-Romance d e n t French d ã - - funktionier endlich! 17 / 32

Slide 59

Slide 59 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Sound Classes . . Correspondence Classes In sound class approaches, sounds are “divided into several types and thereby distinguished in such a way that phonetic correspondences inside a ‘type’ are more regular than those between different ‘types’” (Dolgopolsky 1986: 35). Diachronic Similarity Similarity is not based on synchronic resemblances of sounds but on class-membership: two sounds, how dissimilar they may be from a synchronic perspective, may still belong to the same class. Class membership indicates that the probability that sounds occur in a correspondence relationship in genetically related languages is considerably high. 18 / 32

Slide 60

Slide 60 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Sound Classes k g p b ʧ ʤ f v t d ʃ ʒ θ ð s z 19 / 32

Slide 61

Slide 61 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Sound Classes k g p b ʧ ʤ f v t d ʃ ʒ θ ð s z 19 / 32

Slide 62

Slide 62 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Sound Classes k g p b ʧ ʤ f v t d ʃ ʒ θ ð s z 19 / 32

Slide 63

Slide 63 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Similarity Sound Classes Sound Classes K T P S 19 / 32

Slide 64

Slide 64 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring LingPy 20 / 32

Slide 65

Slide 65 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring LingPy A Python Library for Sequence Alignment LingPy (www.lingulist.de/lingpy) is a suite of open source Python modules for sequence comparison, and distance analyses in quantitative historical linguistics. The library allows to carry out both pairwise and multiple alignments of strings encoded in IPA or X-Sampa, using different methods and algorithms, such as global (Needleman & Wunsch 1970) and local (Smith & Waterman 1981) pairwise alignments, multiple alignments based on guide trees (Feng & Doolittle 1987), profiles (Thompson et al. 1994), or iteration (Barton & Sternberg 1987). 21 / 32

Slide 66

Slide 66 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring Main Ideas . . Alignment of Sound Class Sequences In contrast to previous approaches, which base the alignment on the sequences as they are given from the input, within the sound class approach, the input strings are first converted to sound classes before they are aligned. Transitions Between Sound Classes In contrast to previous sound class approaches (cf. e.g. Turchin et al. 2010), which do not allow for transitions between sound classes, this approach is based on a specific scoring function, which defines (diachronic) similarity among different sound classes. 22 / 32

Slide 67

Slide 67 text

Slide 68

Slide 68 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring Working Principle INPUT ʧɪlɐvʲɛk ʧovɛk TOKENIZATION ʧ, ɪ, l, ɐ, vʲ, ɛ, k ʧ, o, v, ɛ, k 23 / 32

Slide 69

Slide 69 text

Slide 70

Slide 70 text

Slide 71

Slide 71 text

Slide 72

Slide 72 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring Scoring . . Directionality of Sound Changes One crucial characteristic of certain well-known sound changes is their directionality, i.e. if certain sounds change, this change will go into a certain direction and the reverse change can rarely be attested. Directionality and Sound Correspondences While the nature of certain sound changes may be directional, sound correspondences do not directly reflect this directionality, and neither do scoring functions for sequence alignments, since these are not directional per definitionem, since the distance or similarity between two segments is always the same, regardless from which segment we start to compare. 24 / 32

Slide 73

Slide 73 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring Scoring . . Reflecting Directionality in Undirected Networks In this approach, the directionality of certain sound changes is accounted for by creating a non-metric scoring function. While in a metric scoring function the distance between two segments A and B would depend on the distance of A and B to a third segment C in such a way that, according to the triangle inequality the distance from A to B could not exceed the sum of the distances from A to C and from B to C, this does not hold for the probability of those sound correspondences, which occur as a product of directional sound change. 25 / 32

Slide 74

Slide 74 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Main Ideas Working Principle Scoring Scoring dentals aﬀricates fricatives velars 8 6 8 6 0 10 10 26 / 32

Slide 75

Slide 75 text

Slide 76

Slide 76 text

Slide 77

Slide 77 text

Slide 78

Slide 78 text

Slide 79

Slide 79 text

Slide 80

Slide 80 text

Slide 81

Slide 81 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Performance of the Method * * * * * * * * * * * * * v o l - d e m o r t v - l a d i m i r - v a l - d e m a r - 1 27 / 32

Slide 82

Slide 82 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Usage Example . . 28 / 32

Slide 83

Slide 83 text

Slide 84

Slide 84 text

Slide 85

Slide 85 text

Slide 86

Slide 86 text

Slide 87

Slide 87 text

Slide 88

Slide 88 text

Slide 89

Slide 89 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Usage Example . . >>> from lingpy.compare.seqcom import Multiple >>> mult = Multiple(['ʧwovʲɛk', 'ʧovɛk',\ ... 'ʧlɔvʲɛk', 'ʧɪlɐvʲɛk']) >>> print ', '.join(mult.ipt_seqs) ʧwɔvʲɛk, ʧovɛk, ʧlovʲɛk, ʧɪlɐvʲɛk >>> mult.prog_align(method='sca',mode='profile') ʧ - w ɔ vʲ ɛ k ʧ - - o v ɛ k ʧ - l o vʲ ɛ k ʧ ɪ l ɐ vʲ ɛ k >>> mult.show_guide_tree() 28 / 32

Slide 90

Slide 90 text

Slide 91

Slide 91 text

Slide 92

Slide 92 text

Slide 93

Slide 93 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Usage Example . . 29 / 32

Slide 94

Slide 94 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Usage Example . . >>> mult.flat_cluster(0.3,method='sca') 29 / 32

Slide 95

Slide 95 text

Slide 96

Slide 96 text

Slide 97

Slide 97 text

Slide 98

Slide 98 text

Slide 99

Slide 99 text

Slide 100

Slide 100 text

Slide 101

Slide 101 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Usage Example . . >>> mult.flat_cluster(0.3,method='sca') [1, 1, 1, 1] >>> mult.prog_align(method='sca',mode='profile')\ ... # profile-based alignment ʧ - w ɔ vʲ ɛ k ʧ - - o v ɛ k ʧ - l o vʲ ɛ k ʧ ɪ l ɐ vʲ ɛ k >>> mult.sum_of_pairs() 39.666666666666664 >>> mult.prog_align(method='sca',mode='fd') \ ... # simple guide-tree alignment ʧ w - ɔ vʲ ɛ k ʧ - - o v ɛ k ʧ - l o vʲ ɛ k ʧ ɪ l ɐ vʲ ɛ k 29 / 32

Slide 102

Slide 102 text

Slide 103

Slide 103 text

Slide 104

Slide 104 text

Slide 105

Slide 105 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR . . IPA-Encoding of the TPPSR The Tableaux phonétiques des patois suisses romand (TPPSR, Gauchat et al. 1925) is a collection of phonetic dialect data, which was digitized in an earlier research project of the Institute for Romance Languages and Literature (Heinrich Heine University Düsseldorf). The original data was converted to IPA in order make it suitable for alignment analyses using the LingPy library. The dataset consists of 480 charts (480 words and phrases) which contain phonetic information for 62 dialect points. Analysis within LingPy The analysis within LingPy is done via a simple terminal-based interface which takes text-files as input and outputs the results of the alignment analyses as text-files. 30 / 32

Slide 106

Slide 106 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 69,sont tout près,pressu 2 sɔ̃.tɔ.priː 3 i.sɔ̃.tɔ.ʤeː 5 ei.səɔ̃.tɔ.prei 8 sɔ̃.pre 11 sɔ̃.tɔ.pruːʦɔ 18 sɔ̃.pre 19 sɔ̃.tɔ.pre 30 ʃʊn.pre 31 ʃɔ̃n.tɔ.prei 34 i.sɔ̃.tɔ.pre 54 ɛ.sɔ̃.tɔ.prɛ 55 prɛj 56 a.sãõ.tɔ.d.koːt 57 sɔ̃.tɔ.preː 58 a.sɔ̃.tɔ.preŋ Interesting Site! 31 / 32

Slide 107

Slide 107 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 69,sont tout près,pressu 2 1 - s ɔ̃ - t ɔ p r iː - - 3 2 i.sɔ̃.tɔ.ʤeː 5 1 ei s əɔ̃ - t ɔ p r ei - - 8 1 - s ɔ̃ - - - p r e - - 11 1 - s ɔ̃ - t ɔ p r uː ʦ ɔ 18 1 - s ɔ̃ - - - p r e - - 19 1 - s ɔ̃ - t ɔ p r e - - 30 1 - ʃ ʊ n - - p r e - - 31 1 - ʃ ɔ̃ n t ɔ p r ei - - 34 1 i s ɔ̃ - t ɔ p r e - - 54 1 ɛ s ɔ̃ - t ɔ p r ɛ - - 55 1 - - - - - - p r ɛ j - 56 3 a.sãõ.tɔ.d.koːt 57 1 - s ɔ̃ - t ɔ p r eː - - 58 1 a s ɔ̃ - t ɔ p r e ŋ - 31 / 32

Slide 108

Slide 108 text

Slide 109

Slide 109 text

Slide 110

Slide 110 text

Slide 111

Slide 111 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 69,sont tout près,pressu 2 1 - s ɔ̃ - t ɔ p r iː - - 5 1 ei s əɔ̃ - t ɔ p r ei - - 8 1 - s ɔ̃ - - - p r e - - 11 1 - s ɔ̃ - t ɔ p r uː ʦ ɔ 18 1 - s ɔ̃ - - - p r e - - 19 1 - s ɔ̃ - t ɔ p r e - - 30 1 - ʃ ʊ n - - p r e - - 31 1 - ʃ ɔ̃ n t ɔ p r ei - - 34 1 i s ɔ̃ - t ɔ p r e - - 54 1 ɛ s ɔ̃ - t ɔ p r ɛ - - 55 1 - - - - - - p r ɛ j - 57 1 - s ɔ̃ - t ɔ p r eː - - 58 1 a s ɔ̃ - t ɔ p r e ŋ - Taxon-ID Cluster-ID Singleton Singleton 31 / 32

Slide 112

Slide 112 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 69,sont tout près,pressu 2 1 - s ɔ̃ - t ɔ p r iː - - 5 1 ei s əɔ̃ - t ɔ p r ei - - 8 1 - s ɔ̃ - - - p r e - - 11 1 - s ɔ̃ - t ɔ p r uː ʦ ɔ 18 1 - s ɔ̃ - - - p r e - - 19 1 - s ɔ̃ - t ɔ p r e - - 30 1 - ʃ ʊ n - - p r e - - 31 1 - ʃ ɔ̃ n t ɔ p r ei - - 34 1 i s ɔ̃ - t ɔ p r e - - 54 1 ɛ s ɔ̃ - t ɔ p r ɛ - - 55 1 - - - - - - p r ɛ j - 57 1 - s ɔ̃ - t ɔ p r eː - - 58 1 a s ɔ̃ - t ɔ p r e ŋ - Boring Site! Taxon-ID Cluster-ID 31 / 32

Slide 113

Slide 113 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 69,sont tout près,pressu 2 1 - s ɔ̃ - t ɔ p r iː - - 5 1 ei s əɔ̃ - t ɔ p r ei - - 8 1 - s ɔ̃ - - - p r e - - 11 1 - s ɔ̃ - t ɔ p r uː ʦ ɔ 18 1 - s ɔ̃ - - - p r e - - 19 1 - s ɔ̃ - t ɔ p r e - - 30 1 - ʃ ʊ n - - p r e - - 31 1 - ʃ ɔ̃ n t ɔ p r ei - - 34 1 i s ɔ̃ - t ɔ p r e - - 54 1 ɛ s ɔ̃ - t ɔ p r ɛ - - 55 1 - - - - - - p r ɛ j - 57 1 - s ɔ̃ - t ɔ p r eː - - 58 1 a s ɔ̃ - t ɔ p r e ŋ - Interesting Site! 31 / 32

Slide 114

Slide 114 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 66,est étroite,stricta 1 ɛ.etrɑːt 2 ɛ.ɛtræːtɛ 3 ɛ.ɛtreːta 5 ɛ.ɛtraɛːta 8 ɛ.ɛtrɑːɛt 11 l.ɛ.ɛtræːtə 19 l.ɛ.etrɑːtə 30 l.ɛθ.ɛθreiti 31 lʲ.ɛ.ɛhriːti 34 ɛt.eːtraːto 55 ɛ.ɛtræit 56 ɛ.ɛtrɑːət 57 ɛ.ɛtrɛt 58 j.ɛ.ɛtreːt Interesting Site! 31 / 32

Slide 115

Slide 115 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 66,est étroite,stricta 1 1 - ɛ - e t r ɑː t - 2 1 - ɛ - ɛ t r æː t ɛ 3 1 - ɛ - ɛ t r eː t a 5 1 - ɛ - ɛ t r aɛː t a 8 1 - ɛ - ɛ t r ɑːɛ t - 11 1 l ɛ - ɛ t r æː t ə 19 1 l ɛ - e t r ɑː t ə 30 1 l ɛ θ ɛ θ r ei t i 31 1 lʲ ɛ - ɛ h r iː t i 34 1 - ɛ t eː t r aː t o 55 1 - ɛ - ɛ t r æi t - 56 1 - ɛ - ɛ t r ɑːə t - 57 1 - ɛ - ɛ t r ɛ t - 58 1 j ɛ - ɛ t r eː t - 31 / 32

Slide 116

Slide 116 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 195,une feuille,folia 1 ɔ̃na.fɔlʲ 2 na.folʲɛ 3 na.fɔlʲ 5 una.fɔlʲə 8 ɔ̃na.fɔjə 11 ɔ̃na.fɔlʲə 19 na.føðə 30 folʲe 31 fɔłe 34 na.fwolʲ 55 ɔn.fɔdʲ 56 ɛn.fuj 57 ɛn.fuj 58 ɛn.fœj 31 / 32

Slide 117

Slide 117 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR TPPSR tppsr 195,une feuille,folia 1 1 ɔ̃ n a f - ɔ lʲ - 2 1 - n a f - o lʲ ɛ 3 1 - n a f - ɔ lʲ - 5 1 u n a f - ɔ lʲ ə 8 1 ɔ̃ n a f - ɔ j ə 11 1 ɔ̃ n a f - ɔ lʲ ə 19 1 - n a f - ø ð ə 30 1 - - - f - o lʲ e 31 1 - - - f - ɔ ł e 34 1 - n a f w o lʲ - 55 1 ɔ n - f - ɔ dʲ - 56 1 ɛ n - f - u j - 57 1 ɛ n - f - u j - 58 1 ɛ n - f - œ j - 31 / 32

Slide 118

Slide 118 text

Introduction Automatic Alignment Analyses Alignments in Historical Linguistics LingPy Performance of the Method Usage Example TPPSR Thank You for Listening! Special thanks to the German Federal Mi- nistry of Education and Research (BMBF) for funding our research project on evolution and clas- siﬁcation in biology, linguistics, and the history of sci- ence (EvoClass). 1 32 / 32