Skip to content

Offline, privacy-first grammar checker. Fast, open-source, Rust-powered

dictionary curation 2026-06-05 (#3598)

Andrew Dunbar1e9212e6c59b
diff --git a/harper-core/dictionary.dict b/harper-core/dictionary.dict
index 3e1d2d119..5eedd5043 100644
--- a/harper-core/dictionary.dict
+++ b/harper-core/dictionary.dict
@@ -7098,7 +7098,7 @@ Moran/Og
Moravia/Og
Moravian/JNg
Mordred/Og
-More/Og
+More/Og # surname: Sir Thomas ~, Henry ~; language of Burkina Faso
Moreno/Og
Morgan/ONSg
Morgantown/Og
@@ -16444,6 +16444,8 @@ camelhair/N
camellia/NgS
cameo/~NgSV
camera/~NgS
+camera obscura/NgS
+camerae obscurae/N9
cameraman/~N0g
cameramen/N9
camerapeople/N9
@@ -24940,7 +24942,7 @@ fibulae/N9
fibular/J
fiche/NSg
fichu/NSg
-fickle/J^>pV
+fickle/JpV
fickleness/Ng
fiction/~NmgS
fictional/~JY
@@ -27086,6 +27088,7 @@ godlike/J
godliness/NmgU
godly/J>^pU
godmother/~NSgV
+godown/NSg₹
godparent/NSg
godsend/NSg
godson/NSg
@@ -27874,6 +27877,7 @@ halftone/NgSV
halfway/~J
halfwit/NSg
halibut/NSg
+halide/NSg
halite/Ng
halitosis/Ng
hall/~NgS
@@ -34231,7 +34235,7 @@ methamphetamine/NgS
methane/~Ng
methanol/~Ng
methinks/
-method/~NgSV
+method/~NgS
methodical/~JYp
methodicalness/Ng
methodological/~JY
@@ -36967,7 +36971,7 @@ oestrus/NgS!_₹
oeuvre/~NgS
of/~P
of course/R
-off/~J>PVGdSNZz
+off/~JPVGdSNZz
off-grid/J
off-limits/J
off-ramp/NgS
diff --git a/harper-core/tests/text/tagged/Alice's Adventures in Wonderland.md b/harper-core/tests/text/tagged/Alice's Adventures in Wonderland.md
index 19c8dd844..928452680 100644
--- a/harper-core/tests/text/tagged/Alice's Adventures in Wonderland.md
+++ b/harper-core/tests/text/tagged/Alice's Adventures in Wonderland.md
@@ -820,8 +820,8 @@
#
> The Mouse did not notice this question , but hurriedly went on , “ ‘ — found it
# D+ NSg/VB+ VXPt NSg/R/C NSg/VB I/Ddem+ NSg/VB+ . NSg/C/P R NSg/VPt J/P . . Unlintable . NSg/VP NPr/ISg+
-> advisable to go with Edgar Atheling to meet William and offer him the crown .
-# J P NSg/VB/J P NPr+ ? P NSg/VB/J NPr+ VB/C NSg/VB/JC ISg+ D NSg/VB/J+ .
+> advisable to go with Edgar Atheling to meet William and offer him the crown .
+# J P NSg/VB/J P NPr+ ? P NSg/VB/J NPr+ VB/C NSg/VB ISg+ D NSg/VB/J+ .
> William’s conduct at first was moderate . But the insolence of his Normans — ’ How
# NPr$ NSg/VB NSg/P NSg/J VLPt NSg/VB/J . NSg/C/P D Nᴹ/VB P ISg/D$+ NPrPl . . NSg/C
> are you getting on now , my dear ? ” it continued , turning to Alice as it spoke .
@@ -2746,8 +2746,8 @@
# . R+ NSg/VX3 I/R/Dq . . VP/J D NPr/VB+ NSg/VB/J+ .
>
#
-> “ Then it wasn’t very civil of you to offer it , ” said Alice angrily .
-# . NSg/J/R/C NPr/ISg+ VPt J/R J P ISgPl+ P NSg/VB/JC NPr/ISg+ . . VP/J NPr+ R .
+> “ Then it wasn’t very civil of you to offer it , ” said Alice angrily .
+# . NSg/J/R/C NPr/ISg+ VPt J/R J P ISgPl+ P NSg/VB NPr/ISg+ . . VP/J NPr+ R .
>
#
> “ It wasn’t very civil of you to sit down without being invited , ” said the March
diff --git a/harper-core/tests/text/tagged/Computer science.md b/harper-core/tests/text/tagged/Computer science.md
index 1ba6edf6f..2a84faa33 100644
--- a/harper-core/tests/text/tagged/Computer science.md
+++ b/harper-core/tests/text/tagged/Computer science.md
@@ -368,8 +368,8 @@
# NSg/J+ N🅪Sg/VB+ . NSg/C/P NSg/VB/J/C/P Nᴹ+ . Nᴹ+ . VB/C NSg . I/J/R/Dq P ISg/D$+
> unique forms of observation and experience do not fit a narrow stereotype of
# NSg/J NPl/V3 P N🅪Sg VB/C N🅪Sg/VB+ VXB NSg/R/C NSg/VBP/J D/P NSg/VB/J NSg/VB P
-> the experimental method . Nonetheless , they are experiments . Each new machine
-# D NSg/J NSg/VB+ . R . IPl+ VLB NPl/V3+ . Dq+ NSg/J+ NSg/VB+
+> the experimental method . Nonetheless , they are experiments . Each new machine
+# D NSg/J NSg+ . R . IPl+ VLB NPl/V3+ . Dq+ NSg/J+ NSg/VB+
> that is built is an experiment . Actually constructing the machine poses a
# I/C/Ddem+ VL3 VP/J VL3 D/P+ NSg/VB+ . R Nᴹ/Vg/J D+ NSg/VB+ NPl/V3+ D/P+
> question to nature ; and we listen for the answer by observing the machine in
@@ -407,7 +407,7 @@
> that computer programs are physical realizations of mathematical entities and
# I/C/Ddem NSg+ NPrPl/V3+ VLB NSg/J NPl/Comm/NoAm P J NPl VB/C
> programs that can be deductively reasoned through mathematical formal methods .
-# NPrPl/V3+ I/C/Ddem+ NPr/VXB NSg/VLXB R VP/J J/P J NSg/J NPl/V3+ .
+# NPrPl/V3+ I/C/Ddem+ NPr/VXB NSg/VLXB R VP/J J/P J NSg/J NPl+ .
> Computer scientists Edsger W. Dijkstra and Tony Hoare regard instructions for
# NSg+ NPl+ ? ? NSg VB/C NPr/J ? NSg/VB+ NPl R/C/P
> computer programs as mathematical sentences and interpret formal semantics for
@@ -441,7 +441,7 @@
> empirical perspective of natural sciences , identifiable in some branches of
# NSg/J NSg/J P NSg/J NPl/V3+ . J NPr/J/R/P I/J/R/Dq NPl/V3 P
> artificial intelligence ) . Computer science focuses on methods involved in
-# J N🅪Sg+ . . NSg+ N🅪Sg/VB+ NPl/V3 J/P NPl/V3+ VP/J NPr/J/R/P
+# J N🅪Sg+ . . NSg+ N🅪Sg/VB+ NPl/V3 J/P NPl+ VP/J NPr/J/R/P
> design , specification , programming , verification , implementation and testing of
# N🅪Sg/VB+ . NSg+ . Nᴹ/Vg/J . N🅪Sg+ . N🅪Sg VB/C Nᴹ/Vg/J P
> human - made computing systems .
@@ -545,7 +545,7 @@
> also for network coding . Codes are studied for the purpose of designing
# R/C R/C/P NSg/VB+ Nᴹ/Vg/J+ . NPl/V3+ VLB VP/J R/C/P D N🅪Sg/VB P Nᴹ/Vg/J
> efficient and reliable data transmission methods .
-# NSg/J VB/C NSg/J+ N🅪Pl+ N🅪Sg+ NPl/V3+ .
+# NSg/J VB/C NSg/J+ N🅪Pl+ N🅪Sg+ NPl+ .
>
#
> Data structures and algorithms
@@ -555,11 +555,11 @@
> Data structures and algorithms are the studies of commonly used computational
# N🅪Pl+ NPl/V3 VB/C NPl+ VLB D NPl/V3 P R VP/J J
> methods and their computational efficiency .
-# NPl/V3 VB/C D$+ J+ N🅪Sg+ .
+# NPl VB/C D$+ J+ N🅪Sg+ .
>
#
> Programming language theory and formal methods
-# HeadingStart Nᴹ/Vg/J+ N🅪Sg+ N🅪Sg VB/C NSg/J NPl/V3+
+# HeadingStart Nᴹ/Vg/J+ N🅪Sg+ N🅪Sg VB/C NSg/J NPl+
>
#
> Programming language theory is a branch of computer science that deals with the
@@ -577,11 +577,11 @@
>
#
> Formal methods are a particular kind of mathematically based technique for the
-# NSg/J+ NPl/V3+ VLB D/P NSg/J NSg/J P R VP/J N🅪Sg+ R/C/P D+
+# NSg/J+ NPl+ VLB D/P NSg/J NSg/J P R VP/J N🅪Sg+ R/C/P D+
> specification , development and verification of software and hardware systems .
# NSg+ . N🅪Sg VB/C N🅪Sg P Nᴹ VB/C Nᴹ+ NPl+ .
> The use of formal methods for software and hardware design is motivated by the
-# D N🅪Sg/VB P NSg/J NPl/V3 R/C/P Nᴹ VB/C Nᴹ+ N🅪Sg/VB+ VL3 VP/J NSg/P D+
+# D N🅪Sg/VB P NSg/J NPl R/C/P Nᴹ VB/C Nᴹ+ N🅪Sg/VB+ VL3 VP/J NSg/P D+
> expectation that , as in other engineering disciplines , performing appropriate
# N🅪Sg+ I/C/Ddem+ . R/C/P NPr/J/R/P NSg/VB/J+ Nᴹ/Vg/J+ NPl/V3+ . Nᴹ/Vg/J VB/J+
> mathematical analysis can contribute to the reliability and robustness of a
@@ -589,17 +589,17 @@
> design . They form an important theoretical underpinning for software
# N🅪Sg/VB+ . IPl+ N🅪Sg/VB D/P J J NSg/Vg R/C/P Nᴹ+
> engineering , especially where safety or security is involved . Formal methods are
-# Nᴹ/Vg/J+ . R NSg/R/C N🅪Sg/VB NPr/C Nᴹ+ VL3 VP/J . NSg/J+ NPl/V3+ VLB
+# Nᴹ/Vg/J+ . R NSg/R/C N🅪Sg/VB NPr/C Nᴹ+ VL3 VP/J . NSg/J+ NPl+ VLB
> a useful adjunct to software testing since they help avoid errors and can also
# D/P J NSg/VB/J P Nᴹ Nᴹ/Vg/J+ C/P IPl+ NSg/VB VB NPl/V3+ VB/C NPr/VXB R/C
> give a framework for testing . For industrial use , tool support is required .
# NSg/VB D/P NSg R/C/P Nᴹ/Vg/J+ . R/C/P NSg/J N🅪Sg/VB+ . NSg/VB+ N🅪Sg/VB+ VL3 VP/J .
> However , the high cost of using formal methods means that they are usually only
-# C . D NSg/VB/J/R N🅪Sg/VBP/J P Nᴹ/Vg/J NSg/J+ NPl/V3+ NPl/V3 I/C/Ddem IPl+ VLB R J/R/C
+# C . D NSg/VB/J/R N🅪Sg/VBP/J P Nᴹ/Vg/J NSg/J+ NPl+ NPl/V3 I/C/Ddem IPl+ VLB R J/R/C
> used in the development of high - integrity and life - critical systems , where
# VP/J NPr/J/R/P D N🅪Sg P NSg/VB/J/R . Nᴹ VB/C N🅪Sg/VB+ . NSg/J NPl . NSg/R/C
> safety or security is of utmost importance . Formal methods are best described as
-# N🅪Sg/VB NPr/C Nᴹ+ VL3 P NSg/J+ Nᴹ+ . NSg/J+ NPl/V3+ VLB NPr/VXB/JS VP/J R/C/P
+# N🅪Sg/VB NPr/C Nᴹ+ VL3 P NSg/J+ Nᴹ+ . NSg/J+ NPl+ VLB NPr/VXB/JS VP/J R/C/P
> the application of a fairly broad variety of theoretical computer science
# D NSg P D/P R NSg/J N🅪Sg P J+ NSg+ N🅪Sg/VB+
> fundamentals , in particular logic calculi , formal languages , automata theory ,
@@ -1019,7 +1019,7 @@
> " objects " , which may contain data , in the form of fields , often known as
# . NPl/V3+ . . I/C+ NPr/VXB VB N🅪Pl+ . NPr/J/R/P D N🅪Sg/VB P NPrPl/V3+ . R VPp/J R/C/P
> attributes ; and code , in the form of procedures , often known as methods . A
-# NPl/V3+ . VB/C N🅪Sg/VB+ . NPr/J/R/P D N🅪Sg/VB P NPl+ . R VPp/J R/C/P NPl/V3+ . D/P
+# NPl/V3+ . VB/C N🅪Sg/VB+ . NPr/J/R/P D N🅪Sg/VB P NPl+ . R VPp/J R/C/P NPl+ . D/P
> feature of objects is that an object's procedures can access and often modify
# NSg/VB P NPl/V3+ VL3 I/C/Ddem D/P NSg$ NPl+ NPr/VXB N🅪Sg/VB+ VB/C R VB
> the data fields of the object with which they are associated . Thus
@@ -1038,8 +1038,8 @@
# NPl VB/C NSg/VB+ NSg/J+ Nᴹ+ NPrPl/V3+ .
>
#
-> Many languages offer support for multiple paradigms , making the distinction more
-# NSg/I/J/Dq+ NPl+ NSg/VB/JC N🅪Sg/VB R/C/P NSg/J/Dq+ NPl+ . Nᴹ/Vg/J D+ N🅪Sg+ NPr/I/J/R/Dq
+> Many languages offer support for multiple paradigms , making the distinction more
+# NSg/I/J/Dq+ NPl+ NSg/VB N🅪Sg/VB R/C/P NSg/J/Dq+ NPl+ . Nᴹ/Vg/J D+ N🅪Sg+ NPr/I/J/R/Dq
> a matter of style than of technical capabilities .
# D/P N🅪Sg/VB P NSg/VB+ C/P P NSg/J+ NPl+ .
>
diff --git a/harper-core/tests/text/tagged/Part-of-speech tagging.md b/harper-core/tests/text/tagged/Part-of-speech tagging.md
index e9e40c6d2..7dce70566 100644
--- a/harper-core/tests/text/tagged/Part-of-speech tagging.md
+++ b/harper-core/tests/text/tagged/Part-of-speech tagging.md
@@ -105,7 +105,7 @@
> In part - of - speech tagging by computer , it is typical to distinguish from 50 to
# NPr/J/R/P NSg/VB/J+ . P . N🅪Sg/VB+ NSg/Vg NSg/P NSg+ . NPr/ISg+ VL3 NSg/J P VB P # P
> 150 separate parts of speech for English . Work on stochastic methods for tagging
-# # NSg/VB/J NPl/V3 P N🅪Sg/VB R/C/P NPr🅪Sg/VB/J+ . N🅪Sg/VB J/P J NPl/V3+ R/C/P NSg/Vg
+# # NSg/VB/J NPl/V3 P N🅪Sg/VB R/C/P NPr🅪Sg/VB/J+ . N🅪Sg/VB J/P J NPl+ R/C/P NSg/Vg
> Koine Greek ( DeRose 1990 ) has used over 1 , 000 parts of speech and found that
# ? NPr/VB/J . ? # . V3 VP/J NSg/J/P # . # NPl/V3 P N🅪Sg/VB+ VB/C NSg/VP I/C/Ddem
> about as many words were ambiguous in that language as in English . A
@@ -246,8 +246,8 @@
# NSg/VB+ # . P D N🅪Sg/VB/J+ . D/P NSg/VB/J+ # . . VB/C D/P N🅪Sg/VB/JC+ # . . NSg/Vg/J/P I/Ddem+ . D/P+
> program can decide that " can " in " the can " is far more likely to be a noun than
# NPr/VB+ NPr/VXB VB I/C/Ddem+ . NPr/VXB . NPr/J/R/P . D+ NPr/VXB . VL3 NSg/VB/J NPr/I/J/R/Dq NSg/J P NSg/VLXB D/P NSg/VB C/P
-> a verb or a modal . The same method can , of course , be used to benefit from
-# D/P+ NSg/VB+ NPr/C D/P NSg/J . D+ I/J+ NSg/VB+ NPr/VXB . P NSg/VB+ . NSg/VLXB VP/J P NSg/VB P
+> a verb or a modal . The same method can , of course , be used to benefit from
+# D/P+ NSg/VB+ NPr/C D/P NSg/J . D+ I/J+ NSg+ NPr/VXB . P NSg/VB+ . NSg/VLXB VP/J P NSg/VB P
> knowledge about the following words .
# Nᴹ+ J/P D+ N🅪Sg/Vg/J/P NPl/V3+ .
>
@@ -293,7 +293,7 @@
> expensive since it enumerated all possibilities . It sometimes had to resort to
# J C/P NPr/ISg+ VP/J NSg/I/J/C/Dq NPl+ . NPr/ISg+ R VP P NSg/VB P
> backup methods when there were simply too many options ( the Brown Corpus
-# NSg/J NPl/V3+ NSg/I/C R+ VLPt R R NSg/I/J/Dq NPl/V3 . D+ NPr🅪Sg/VB/J+ NSg+
+# NSg/J NPl+ NSg/I/C R+ VLPt R R NSg/I/J/Dq NPl/V3 . D+ NPr🅪Sg/VB/J+ NSg+
> contains a case with 17 ambiguous words in a row , and there are words such as
# V3 D/P NPr🅪Sg/VB+ P # J NPl/V3 NPr/J/R/P D/P+ NSg/VB+ . VB/C R+ VLB NPl/V3+ NSg/I R/C/P
> " still " that can represent as many as 7 distinct parts of speech .
@@ -309,7 +309,7 @@
>
#
> Dynamic programming methods
-# HeadingStart NSg/J+ Nᴹ/Vg/J+ NPl/V3+
+# HeadingStart NSg/J+ Nᴹ/Vg/J+ NPl+
>
#
> In 1987 , Steven DeRose and Kenneth W. Church independently developed dynamic
@@ -317,15 +317,15 @@
> programming algorithms to solve the same problem in vastly less time . Their
# Nᴹ/Vg/J+ NPl+ P NSg/VB D I/J NSg/J+ NPr/J/R/P R VB/J/R/C/P N🅪Sg/VB/J+ . D$+
> methods were similar to the Viterbi algorithm known for some time in other
-# NPl/V3+ VLPt NSg/J P D ? NSg VPp/J R/C/P I/J/R/Dq N🅪Sg/VB/J+ NPr/J/R/P NSg/VB/J
+# NPl+ VLPt NSg/J P D ? NSg VPp/J R/C/P I/J/R/Dq N🅪Sg/VB/J+ NPr/J/R/P NSg/VB/J
> fields . DeRose used a table of pairs , while Church used a table of triples and a
# NPrPl/V3+ . ? VP/J D/P NSg/VB P NPl/V3+ . NSg/VB/C/P NPr🅪Sg/VB+ VP/J D/P NSg/VB P NPl/V3 VB/C D/P
> method of estimating the values for triples that were rare or nonexistent in the
-# NSg/VB P Nᴹ/Vg/J D NPl/V3+ R/C/P NPl/V3 I/C/Ddem+ VLPt NSg/VB/J NPr/C NSg/J NPr/J/R/P D
+# NSg P Nᴹ/Vg/J D NPl/V3+ R/C/P NPl/V3 I/C/Ddem+ VLPt NSg/VB/J NPr/C NSg/J NPr/J/R/P D
> Brown Corpus ( an actual measurement of triple probabilities would require a much
# NPr🅪Sg/VB/J NSg+ . D/P NSg/J N🅪Sg P NSg/VB/J NPl+ VXB NSg/VB D/P NSg/I/J/R/Dq
> larger corpus ) . Both methods achieved an accuracy of over 95 % . DeRose's 1990
-# JC NSg+ . . I/C/Dq NPl/V3+ VP/J D/P N🅪Sg+ P NSg/J/P # . . ? #
+# JC NSg+ . . I/C/Dq NPl+ VP/J D/P N🅪Sg+ P NSg/J/P # . . ? #
> dissertation at Brown University included analyses of the specific error types ,
# NSg+ NSg/P NPr🅪Sg/VB/J NSg+ VP/J NPl/V3/Au/Br P D NSg/J NSg/VB+ NPl/V3+ .
> probabilities , and other related data , and replicated his work for Greek , where
@@ -343,7 +343,7 @@
> levels of linguistic analysis : syntax , morphology , semantics , and so on . CLAWS ,
# NPl/V3 P J N🅪Sg . Nᴹ+ . Nᴹ+ . NPl+ . VB/C NSg/I/J/R/C J/P . NPl/V3+ .
> DeRose's and Church's methods did fail for some of the known cases where
-# ? VB/C NPr$ NPl/V3+ VXPt NSg/VB/J R/C/P I/J/R/Dq P D VPp/J NPl/V3+ NSg/R/C
+# ? VB/C NPr$ NPl+ VXPt NSg/VB/J R/C/P I/J/R/Dq P D VPp/J NPl/V3+ NSg/R/C
> semantics is required , but those proved negligibly rare . This convinced many in
# NPl+ VL3 VP/J . NSg/C/P I/Ddem VP/J R NSg/VB/J . I/Ddem VP/J NSg/I/J/Dq NPr/J/R/P
> the field that part - of - speech tagging could usefully be separated from the other
@@ -352,8 +352,8 @@
# NPl/V3 P Nᴹ/Vg/J+ . I/Ddem+ . NPr/J/R/P NSg/VB . VP/J D N🅪Sg VB/C NSg/VB P
> computerized language analysis and encouraged researchers to find ways to
# VP/J N🅪Sg+ N🅪Sg+ VB/C VP/J NPl+ P NSg/VB NPl+ P
-> separate other pieces as well . Markov Models became the standard method for the
-# NSg/VB/J NSg/VB/J NPl/V3+ R/C/P NSg/VB/J/R . NPr NPl/V3+ VPt D NSg/J NSg/VB+ R/C/P D
+> separate other pieces as well . Markov Models became the standard method for the
+# NSg/VB/J NSg/VB/J NPl/V3+ R/C/P NSg/VB/J/R . NPr NPl/V3+ VPt D NSg/J NSg+ R/C/P D
> part - of - speech assignment .
# NSg/VB/J+ . P . N🅪Sg/VB+ NSg+ .
>
@@ -363,7 +363,7 @@
>
#
> The methods already discussed involve working from a pre - existing corpus to
-# D+ NPl/V3+ R VP/J VB Nᴹ/Vg/J P D/P+ NSg/VB/P+ . Nᴹ/Vg/J NSg+ P
+# D+ NPl+ R VP/J VB Nᴹ/Vg/J P D/P+ NSg/VB/P+ . Nᴹ/Vg/J NSg+ P
> learn tag probabilities . It is , however , also possible to bootstrap using
# NSg/VB NSg/VB+ NPl+ . NPr/ISg+ VL3 . C . R/C NSg/J P NSg/VB Nᴹ/Vg/J
> " unsupervised " tagging . Unsupervised tagging techniques use an untagged corpus
@@ -391,7 +391,7 @@
>
#
> Other taggers and methods
-# HeadingStart NSg/VB/J NPl VB/C NPl/V3+
+# HeadingStart NSg/VB/J NPl VB/C NPl+
>
#
> Some current major algorithms for part - of - speech tagging include the Viterbi
@@ -409,9 +409,9 @@
>
#
> Many machine learning methods have also been applied to the problem of POS
-# NSg/I/J/Dq+ NSg/VB+ Nᴹ/Vg/J+ NPl/V3+ NSg/VXB R/C NSg/VLPp VP/J P D NSg/J P NSg+
+# NSg/I/J/Dq+ NSg/VB+ Nᴹ/Vg/J+ NPl+ NSg/VXB R/C NSg/VLPp VP/J P D NSg/J P NSg+
> tagging . Methods such as SVM , maximum entropy classifier , perceptron , and
-# NSg/Vg . NPl/V3+ NSg/I R/C/P ? . NSg/J NSg NSg . NSg . VB/C
+# NSg/Vg . NPl+ NSg/I R/C/P ? . NSg/J NSg NSg . NSg . VB/C
> nearest - neighbor have all been tried , and most can achieve accuracy above
# JS . NSg/VB/J/Am+ NSg/VXB NSg/I/J/C/Dq NSg/VLPp VP/J . VB/C NSg/I/J/R/Dq NPr/VXB VB N🅪Sg+ NSg/J/P
> 95 % . [ citation needed ]
@@ -419,7 +419,7 @@
>
#
> A direct comparison of several methods is reported ( with references ) at the ACL
-# D/P VB/J NSg P J/Dq+ NPl/V3+ VL3 VP/J . P NPl/V3+ . NSg/P D NSg
+# D/P VB/J NSg P J/Dq+ NPl+ VL3 VP/J . P NPl/V3+ . NSg/P D NSg
> Wiki . This comparison uses the Penn tag set on some of the Penn Treebank data ,
# NSg/VB+ . I/Ddem+ NSg+ NPl/V3 D+ NPr+ NSg/VB+ NPr/VBP/J J/P I/J/R/Dq P D NPr+ ? N🅪Pl+ .
> so the results are directly comparable . However , many significant taggers are
@@ -435,6 +435,6 @@
>
#
> In 2014 , a paper reporting using the structure regularization method for
-# NPr/J/R/P # . D/P+ N🅪Sg/VB/J+ Nᴹ/Vg/J Nᴹ/Vg/J D N🅪Sg/VB+ N🅪Sg NSg/VB R/C/P
+# NPr/J/R/P # . D/P+ N🅪Sg/VB/J+ Nᴹ/Vg/J Nᴹ/Vg/J D N🅪Sg/VB+ N🅪Sg NSg R/C/P
> part - of - speech tagging , achieving 97.36 % on a standard benchmark dataset .
# NSg/VB/J+ . P . N🅪Sg/VB+ NSg/Vg . Nᴹ/Vg/J # . J/P D/P NSg/J NSg/VB NSg .
diff --git a/harper-core/tests/text/tagged/The Great Gatsby.md b/harper-core/tests/text/tagged/The Great Gatsby.md
index ffb55b4c5..54e68ccb8 100644
--- a/harper-core/tests/text/tagged/The Great Gatsby.md
+++ b/harper-core/tests/text/tagged/The Great Gatsby.md
@@ -5092,8 +5092,8 @@
#
> I realize now that under different circumstances that conversation might have
# ISg/#r+ VB/Comm/NoAm NSg/J/R/C I/C/Ddem NSg/J/P NSg/J+ NPl/V3+ I/C/Ddem+ N🅪Sg/VB+ Nᴹ/VXB/J NSg/VXB
-> been one of the crises of my life . But , because the offer was obviously and
-# NSg/VLPp NSg/I/J P D NPl P D$+ N🅪Sg/VB+ . NSg/C/P . C/P D+ NSg/VB/JC+ VLPt R VB/C
+> been one of the crises of my life . But , because the offer was obviously and
+# NSg/VLPp NSg/I/J P D NPl P D$+ N🅪Sg/VB+ . NSg/C/P . C/P D+ NSg/VB+ VLPt R VB/C
> tactlessly for a service to be rendered , I had no choice except to cut him off
# R R/C/P D/P NSg/VB+ P NSg/VLXB VP/J . ISg/#r+ VP NSg/Dq/P N🅪Sg/J+ VB/C/P P NSg/VBP/J ISg+ NSg/VB/J/P
> there .