diff --git a/spec/text-utils-spec.coffee b/spec/text-utils-spec.coffee
index aa36c5003..bae7f5997 100644
--- a/spec/text-utils-spec.coffee
+++ b/spec/text-utils-spec.coffee
@@ -75,22 +75,23 @@ describe 'text utilities', ->
 
       expect(textUtils.isKoreanCharacter("O")).toBe(false)
 
-  describe ".isCJKCharacter(character)", ->
-    it "returns true when the character is either a korean, half-width or double-width character", ->
-      expect(textUtils.isCJKCharacter("我")).toBe(true)
-      expect(textUtils.isCJKCharacter("私")).toBe(true)
-      expect(textUtils.isCJKCharacter("Ｂ")).toBe(true)
-      expect(textUtils.isCJKCharacter("，")).toBe(true)
-      expect(textUtils.isCJKCharacter("￠")).toBe(true)
-      expect(textUtils.isCJKCharacter("ﾊ")).toBe(true)
-      expect(textUtils.isCJKCharacter("ﾋ")).toBe(true)
-      expect(textUtils.isCJKCharacter("ﾬ")).toBe(true)
-      expect(textUtils.isCJKCharacter("￭")).toBe(true)
-      expect(textUtils.isCJKCharacter("우")).toBe(true)
-      expect(textUtils.isCJKCharacter("가")).toBe(true)
-      expect(textUtils.isCJKCharacter("ㅢ")).toBe(true)
-      expect(textUtils.isCJKCharacter("ㄼ")).toBe(true)
+  describe ".isWrapBoundary(previousCharacter, character)", ->
+    it "returns true when the character is CJK or when the previous character is a space/tab", ->
+      anyCharacter = 'x'
+      expect(textUtils.isWrapBoundary(anyCharacter, "我")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "私")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "Ｂ")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "，")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "￠")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "ﾊ")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "ﾋ")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "ﾬ")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "￭")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "우")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "가")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "ㅢ")).toBe(true)
+      expect(textUtils.isWrapBoundary(anyCharacter, "ㄼ")).toBe(true)
 
-      expect(textUtils.isDoubleWidthCharacter("a")).toBe(false)
-      expect(textUtils.isDoubleWidthCharacter("O")).toBe(false)
-      expect(textUtils.isDoubleWidthCharacter("z")).toBe(false)
+      expect(textUtils.isWrapBoundary(' ', 'h')).toBe(true)
+      expect(textUtils.isWrapBoundary('\t', 'h')).toBe(true)
+      expect(textUtils.isWrapBoundary('a', 'h')).toBe(false)
diff --git a/src/display-buffer.coffee b/src/display-buffer.coffee
index 4b869233a..eb4faa1a5 100644
--- a/src/display-buffer.coffee
+++ b/src/display-buffer.coffee
@@ -7,7 +7,7 @@ Model = require './model'
 Token = require './token'
 Decoration = require './decoration'
 LayerDecoration = require './layer-decoration'
-{isDoubleWidthCharacter, isHalfWidthCharacter, isKoreanCharacter} = require './text-utils'
+{isDoubleWidthCharacter, isHalfWidthCharacter, isKoreanCharacter, isWrapBoundary} = require './text-utils'
 
 class BufferToScreenConversionError extends Error
   constructor: (@message, @metadata) ->
@@ -122,6 +122,7 @@ class DisplayBuffer extends Model
       showIndentGuides: @config.get('editor.showIndentGuide', scope: scopeDescriptor)
       tabLength: @config.get('editor.tabLength', scope: scopeDescriptor),
       ratioForCharacter: @ratioForCharacter.bind(this)
+      isWrapBoundary: isWrapBoundary
     })
 
   updateAllScreenLines: ->
diff --git a/src/text-utils.coffee b/src/text-utils.coffee
index af17335aa..f4d62772e 100644
--- a/src/text-utils.coffee
+++ b/src/text-utils.coffee
@@ -94,6 +94,13 @@ isCJKCharacter = (character) ->
   isHalfWidthCharacter(character) or
   isKoreanCharacter(character)
 
+isWordStart = (previousCharacter, character) ->
+  (previousCharacter is ' ' or previousCharacter is '\t') and
+  (character isnt ' '  and character isnt '\t')
+
+isWrapBoundary = (previousCharacter, character) ->
+  isWordStart(previousCharacter, character) or isCJKCharacter(character)
+
 # Does the given string contain at least surrogate pair, variation sequence,
 # or combined character?
 #
@@ -107,4 +114,8 @@ hasPairedCharacter = (string) ->
     index++
   false
 
-module.exports = {isPairedCharacter, hasPairedCharacter, isDoubleWidthCharacter, isHalfWidthCharacter, isKoreanCharacter, isCJKCharacter}
+module.exports = {
+  isPairedCharacter, hasPairedCharacter,
+  isDoubleWidthCharacter, isHalfWidthCharacter, isKoreanCharacter,
+  isWrapBoundary
+}