diff --git a/src/api/col-get-item.js b/src/api/col-get-item.js
index 252f19e..7191cdf 100644
--- a/src/api/col-get-item.js
+++ b/src/api/col-get-item.js
@@ -44,7 +44,9 @@
 var colGetItemAt = function ( k, start, end, itemFn ) {
   // To handle relative indexing, compute actual `k` by adding `start`.
   var ak = k + start;
-  return ( ( ak < start || ak > end ) ? undefined : itemFn( ak ) );
+  if ( ak < start || ak > end ) {
+    throw Error( `wink-nlp: ${k} is an invalid or out of bounds index.`);
+  } else return itemFn( ak );
 }; // colGetItemAt()
 
 module.exports = colGetItemAt;
diff --git a/src/api/sel-get-item.js b/src/api/sel-get-item.js
index a66be5a..7236649 100644
--- a/src/api/sel-get-item.js
+++ b/src/api/sel-get-item.js
@@ -41,7 +41,9 @@
  * @private
  */
 var selGetItemAt = function ( k, selection, itemFn ) {
-  return ( ( k < 0 || k >= selection.length ) ? undefined : itemFn( selection[ k ] ) );
+  if ( k < 0 || k >= selection.length ) {
+    throw Error( `wink-nlp: wink-nlp: ${k} is an invalid or out of bounds index.`);
+  } else return itemFn( selection[ k ] );
 }; // selGetItemAt()
 
 module.exports = selGetItemAt;
diff --git a/test/apiA-specs.js b/test/apiA-specs.js
index 3db5e07..2c76074 100644
--- a/test/apiA-specs.js
+++ b/test/apiA-specs.js
@@ -124,20 +124,20 @@ describe( 'APIs — A', function () {
     // ItemAt boundary tests.
     // Will need a revamp once SBD is in place (TODO):
     describe( 'doc API out of range access test', function () {
-      it( '.sentences() should return undefined for out of range index', function () {
-        expect( doc1.sentences().itemAt( -1 ) ).to.equal( undefined );
-        expect( doc1.sentences().itemAt( doc1.sentences().length() ) ).to.equal( undefined );
+      it( '.sentences() should throw  out of range index error', function () {
+        expect( () => doc1.sentences().itemAt(-1) ).to.throw( /^wink-nlp: -1 is an invalid or out of bounds index./ );
+        expect( () => doc1.sentences().itemAt( doc1.sentences().length() ) ).to.throw( /^wink-nlp: 3 is an invalid or out of bounds index./ );
       } );
 
-      it( '.tokens().itemAt() should return undefined for out of range index', function () {
-        expect( doc1.tokens().itemAt( -1 ) ).to.equal( undefined );
-        expect( doc1.tokens().itemAt( doc1.tokens().length() ) ).to.equal( undefined );
+      it( '.tokens().itemAt() should throw out of range index error', function () {
+        expect( () => doc1.tokens().itemAt( -1 ) ).to.throw( /^wink-nlp: -1 is an invalid or out of bounds index./ );
+        expect( () => doc1.tokens().itemAt( doc1.tokens().length() ) ).to.throw( /^wink-nlp: 37 is an invalid or out of bounds index./ );
       } );
 
-      it( '.tokens().filter().itemAt() should return undefined for out of range index', function () {
+      it( '.tokens().filter().itemAt() should throw out of range index error', function () {
         const ftk1 = doc1.tokens().filter( ( t ) => ( t.out( its.type ) === 'word' ) );
-        expect( ftk1.itemAt( -1 ) ).to.equal( undefined );
-        expect( ftk1.itemAt( ftk1.length() ) ).to.equal( undefined );
+        expect( () => ftk1.itemAt( -1 ) ).to.throw( /^wink-nlp: wink-nlp: -1 is an invalid or out of bounds index./ );
+        expect( () => ftk1.itemAt( ftk1.length() ) ).to.throw( /^wink-nlp: wink-nlp: 14 is an invalid or out of bounds index./ );
       } );
     } );
 
@@ -232,7 +232,7 @@ describe( 'APIs — A', function () {
         const i22 = doc2.entities().itemAt( 5 );
         expect( i22.out( its.detail ) ).to.deep.equal( ae2[ 5 ] );
 
-        expect( doc2.entities().itemAt( 12 ) ).to.deep.equal( undefined );
+        expect( () => doc2.entities().itemAt( 12 ) ).to.throw( /^wink-nlp: 12 is an invalid or out of bounds index./ );
       } );
 
       it( '.filter() should return correctly filter entities', function () {
@@ -247,7 +247,7 @@ describe( 'APIs — A', function () {
         // Also check the parent document!
         expect( fe1.itemAt( 1 ).parentDocument() ).to.deep.equal( doc1 );
         // Out of range item test
-        expect( fe1.itemAt( 2 ) ).to.deep.equal( undefined );
+        expect( () => fe1.itemAt( 2 ) ).to.throw( /^wink-nlp: wink-nlp: 2 is an invalid or out of bounds index./ );
         // itemAt() api.
         fe1.each( ( e, k ) => {
           expect( e.out() ).to.deep.equal( fe1.itemAt( k ).out() );
@@ -260,7 +260,7 @@ describe( 'APIs — A', function () {
         expect( fe2.out( its.detail ) ).to.deep.equal( fae2 );
         expect( fe2.itemAt( 1 ).out( its.detail ) ).to.deep.equal( fae2[ 1 ] );
         expect( fe2.itemAt( 1 ).parentDocument() ).to.deep.equal( doc2 );
-        expect( fe2.itemAt( 3 ) ).to.deep.equal( undefined );
+        expect( () => fe2.itemAt( 3 ) ).to.throw( /^wink-nlp: wink-nlp: 3 is an invalid or out of bounds index./ );
         fe2.each( ( e, k ) => {
           expect( e.out() ).to.deep.equal( fe2.itemAt( k ).out() );
         } );
diff --git a/types/index.d.ts b/types/index.d.ts
index 2c1cc94..31fa2cf 100644
--- a/types/index.d.ts
+++ b/types/index.d.ts
@@ -4,6 +4,8 @@ declare module 'wink-nlp' {
   // turn off exporting by default since we don't want to expose internal details
   export { };
 
+  // *** BEGIN Language Model Specific Declarations ***
+  // These should be always in sync with the langauge model's type declarations.
   // these types are internal details of the implementing model
   type StemAddon = unknown;
   type LemmatizeAddon = unknown;
@@ -40,6 +42,7 @@ declare module 'wink-nlp' {
     featureFn: FeatureFn;
     addons: ModelAddons;
   }
+  // *** END Language Model Specific Declarations ***
 
   // its helpers
 
@@ -106,25 +109,25 @@ declare module 'wink-nlp' {
 
   // Its
   export interface ItsHelpers {
-    case(index: number, token: Token, cache: Cache): Case;
-    uniqueId(index: number, token: Token): number;
-    negationFlag(index: number, token: Token): boolean;
-    normal(index: number, token: Token, cache: Cache): string;
-    contractionFlag(index: number, token: Token): boolean;
-    pos(index: number, token: Token, cache: Cache): PartOfSpeech;
-    precedingSpaces(index: number, token: Token): string;
-    prefix(index: number, token: Token, cache: Cache): string;
-    shape(index: number, token: Token, cache: Cache): string;
-    stopWordFlag(index: number, token: Token, cache: Cache): boolean;
-    abbrevFlag(index: number, token: Token, cache: Cache): boolean;
-    suffix(index: number, token: Token, cache: Cache): string;
-    type(index: number, token: Token, cache: Cache): string;
-    value(index: number, token: Token, cache: Cache): string;
-    stem(index: number, token: Token, cache: Cache, addons: ModelAddons): string;
-    lemma(index: number, token: Token, cache: Cache, addons: ModelAddons): string;
+    case(index: number, rdd: RawDocumentData): Case;
+    uniqueId(index: number, rdd: RawDocumentData): number;
+    negationFlag(index: number, rdd: RawDocumentData): boolean;
+    normal(index: number, rdd: RawDocumentData): string;
+    contractionFlag(index: number, rdd: RawDocumentData): boolean;
+    pos(index: number, rdd: RawDocumentData): PartOfSpeech;
+    precedingSpaces(index: number, rdd: RawDocumentData): string;
+    prefix(index: number, rdd: RawDocumentData): string;
+    shape(index: number, rdd: RawDocumentData): string;
+    stopWordFlag(index: number, rdd: RawDocumentData): boolean;
+    abbrevFlag(index: number, rdd: RawDocumentData): boolean;
+    suffix(index: number, rdd: RawDocumentData): string;
+    type(index: number, rdd: RawDocumentData): string;
+    value(index: number, rdd: RawDocumentData): string;
+    stem(index: number, rdd: RawDocumentData, addons: ModelAddons): string;
+    lemma(index: number, rdd: RawDocumentData, addons: ModelAddons): string;
     vector(): number[];
     detail(): Detail;
-    markedUpText(index: number, token: Token, cache: Cache): string;
+    markedUpText(index: number, rdd: RawDocumentData): string;
     span(spanItem: number[]): number[];
     sentenceWiseImportance(rdd: RawDocumentData): SentenceImportance[];
     sentiment(spanItem: number[]): number;
@@ -134,7 +137,7 @@ declare module 'wink-nlp' {
     docBOWArray(tf: ModelTermFrequencies): Bow;
     bow(tf: ModelTermFrequencies): Bow;
     idf(tf: ModelTermFrequencies, idf: ModelInverseDocumentFrequencies): Array<[term: string, frequency: number]>;
-    tf(tf: ModelTermFrequencies, idf: ModelInverseDocumentFrequencies): Array<[term: string, frequency: number]>;
+    tf(tf: ModelTermFrequencies): Array<[term: string, frequency: number]>;
     modelJSON(tf: ModelTermFrequencies, idf: ModelInverseDocumentFrequencies): string;
   }
 
@@ -146,6 +149,7 @@ declare module 'wink-nlp' {
     freqTable<T>(tokens: T[]): Array<[token: T, freq: number]>;
     bigrams<T>(tokens: T[]): Array<[T, T]>;
     unique<T>(tokens: T[]): T[];
+    vector(token: string[]): number[];
   }
 
   // functions for use with document
@@ -168,9 +172,9 @@ declare module 'wink-nlp' {
   }
 
   export interface SelectedTokens {
-    each(f: (token: ItemToken) => void): void;
-    filter(f: (token: ItemToken) => boolean): SelectedTokens;
-    itemAt(k: number): ItemToken | undefined;
+    each(cb: ((item: ItemToken) => void) | ((item: ItemToken, index: number) => void)): void;
+    filter(cb: (item: ItemToken) => boolean): SelectedTokens;
+    itemAt(k: number): ItemToken;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -178,9 +182,9 @@ declare module 'wink-nlp' {
   }
 
   export interface Tokens {
-    each(f: (token: ItemToken) => void): void;
-    filter(f: (token: ItemToken) => boolean): SelectedTokens;
-    itemAt(k: number): ItemToken | undefined;
+    each(cb: ((item: ItemToken) => void) | ((item: ItemToken, index: number) => void)): void;
+    filter(cb: (item: ItemToken) => boolean): SelectedTokens;
+    itemAt(k: number): ItemToken;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -189,7 +193,7 @@ declare module 'wink-nlp' {
 
   export interface ItemEntity {
     parentDocument(): Document;
-    markup(beginMarker: string, endMarker: string): void;
+    markup(beginMarker?: string, endMarker?: string): void;
     out(): string;
     out<T>(itsf: ItsFunction<T>): T | string;
     parentSentence(): ItemSentence;
@@ -198,9 +202,9 @@ declare module 'wink-nlp' {
   }
 
   export interface SelectedEntities {
-    each(f: (entity: ItemEntity) => void): void;
-    filter(f: (entity: ItemEntity) => boolean): SelectedEntities;
-    itemAt(k: number): ItemEntity | undefined;
+    each(cb: ((item: ItemEntity) => void) | ((item: ItemEntity, index: number) => void)): void;
+    filter(cb: (item: ItemEntity) => boolean): SelectedEntities;
+    itemAt(k: number): ItemEntity;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -208,9 +212,9 @@ declare module 'wink-nlp' {
   }
 
   export interface Entities {
-    each(f: (entity: ItemEntity) => void): void;
-    filter(f: (entity: ItemEntity) => boolean): SelectedEntities;
-    itemAt(k: number): ItemEntity | undefined;
+    each(cb: ((item: ItemEntity) => void) | ((item: ItemEntity, index: number) => void)): void;
+    filter(cb: (item: ItemEntity) => boolean): SelectedEntities;
+    itemAt(k: number): ItemEntity;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -219,7 +223,7 @@ declare module 'wink-nlp' {
 
   export interface ItemCustomEntity {
     parentDocument(): Document;
-    markup(beginMarker: string, endMarker: string): void;
+    markup(beginMarker?: string, endMarker?: string): void;
     out(): string;
     out<T>(itsf: ItsFunction<T>): T | string;
     parentSentence(): ItemSentence;
@@ -228,9 +232,9 @@ declare module 'wink-nlp' {
   }
 
   export interface SelectedCustomEntities {
-    each(f: (entity: ItemCustomEntity) => void): void;
-    filter(f: (entity: ItemCustomEntity) => boolean): SelectedCustomEntities;
-    itemAt(k: number): ItemCustomEntity | undefined;
+    each(cb: ((item: ItemCustomEntity) => void) | ((item: ItemCustomEntity, index: number) => void)): void;
+    filter(cb: (item: ItemCustomEntity) => boolean): SelectedCustomEntities;
+    itemAt(k: number): ItemCustomEntity;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -238,9 +242,9 @@ declare module 'wink-nlp' {
   }
 
   export interface CustomEntities {
-    each(f: (entity: ItemCustomEntity) => void): void;
-    filter(f: (entity: ItemCustomEntity) => boolean): SelectedCustomEntities;
-    itemAt(k: number): ItemCustomEntity | undefined;
+    each(cb: ((item: ItemCustomEntity) => void) | ((item: ItemCustomEntity, index: number) => void)): void;
+    filter(cb: (item: ItemCustomEntity) => boolean): SelectedCustomEntities;
+    itemAt(k: number): ItemCustomEntity;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -249,7 +253,7 @@ declare module 'wink-nlp' {
 
   export interface ItemSentence {
     parentDocument(): Document;
-    markup(beginMarker: string, endMarker: string): void;
+    markup(beginMarker?: string, endMarker?: string): void;
     out(): string;
     out<T>(itsf: ItsFunction<T>): T | string;
     entities(): Entities;
@@ -259,8 +263,8 @@ declare module 'wink-nlp' {
   }
 
   export interface Sentences {
-    each(f: (entity: ItemSentence) => void): void;
-    itemAt(k: number): ItemSentence | undefined;
+    each(cb: ((item: ItemSentence) => void) | ((item: ItemSentence, index: number) => void)): void;
+    itemAt(k: number): ItemSentence;
     length(): number;
     out(): string[];
     out<T>(itsf: ItsFunction<T>): T[] | string[];
@@ -277,6 +281,8 @@ declare module 'wink-nlp' {
     sentences(): Sentences;
     tokens(): Tokens;
     printTokens(): void;
+    pipeConfig(): string[];
+    contextualVectors(lemma: boolean, specifcWordVectors: string[], similarWordVectors: boolean, wordVectorsLimit: number): string;
   }
 
   export interface CerExample {
@@ -295,6 +301,18 @@ declare module 'wink-nlp' {
     patterns: string[];
   }
 
+  // Wink word embeddings structure, should stay in sync with emdedding repo.
+  interface WordEmbedding {
+    precision: number;
+    l2NormIndex: number;
+    wordIndex: number;
+    dimensions: number;
+    unkVector: number[];
+    size: number;
+    words: string[];
+    vectors: Record<string, number[]>;
+  }
+
   export interface WinkMethods {
     readDoc(text: string): Document;
     // returns number of learned entities
@@ -303,7 +321,7 @@ declare module 'wink-nlp' {
     as: AsHelpers;
   }
 
-  export default function WinkFn(theModel: Model, pipe?: string[]): WinkMethods;
+  export default function WinkFn(theModel: Model, pipe?: string[], wordEmbeddings?: WordEmbedding): WinkMethods;
 }
 
 declare module 'wink-nlp/utilities/bm25-vectorizer' {
@@ -348,6 +366,9 @@ declare module 'wink-nlp/utilities/similarity' {
       tversky<T>(setA: Set<T>, setB: Set<T>, alpha?: number, beta?: number): number;
       oo<T>(setA: Set<T>, setB: Set<T>): number;
     };
+    vector: {
+      cosine(vectorA: number[], vectorB: number[]): number;
+    };
   }
 
   const similarity: SimilarityHelper;