diff --git a/milli/src/search/new/db_cache.rs b/milli/src/search/new/db_cache.rs
index 09845377c..aa1c11773 100644
--- a/milli/src/search/new/db_cache.rs
+++ b/milli/src/search/new/db_cache.rs
@@ -89,7 +89,7 @@ impl<'ctx> SearchContext<'ctx> {
     }
 
     /// Retrieve or insert the given value in the `word_docids` database.
-    pub fn get_db_word_docids(&mut self, word: Interned<String>) -> Result<Option<RoaringBitmap>> {
+    fn get_db_word_docids(&mut self, word: Interned<String>) -> Result<Option<RoaringBitmap>> {
         DatabaseCache::get_value(
             self.txn,
             word,
diff --git a/milli/src/search/new/logger/visual.rs b/milli/src/search/new/logger/visual.rs
index 17f7ef76c..068b5ad68 100644
--- a/milli/src/search/new/logger/visual.rs
+++ b/milli/src/search/new/logger/visual.rs
@@ -427,7 +427,7 @@ fill: \"#B6E2D3\"
                 )?;
 
                 for w in term_subset.all_single_words_except_prefix_db(ctx)? {
-                    let w = ctx.word_interner.get(w);
+                    let w = ctx.word_interner.get(w.interned());
                     writeln!(file, "{w}: word")?;
                 }
                 for p in term_subset.all_phrases(ctx)? {
diff --git a/milli/src/search/new/mod.rs b/milli/src/search/new/mod.rs
index f51d3771d..9f8d8699f 100644
--- a/milli/src/search/new/mod.rs
+++ b/milli/src/search/new/mod.rs
@@ -50,6 +50,8 @@ use ranking_rules::{BoxRankingRule, RankingRule};
 use resolve_query_graph::compute_query_graph_docids;
 use sort::Sort;
 
+use self::interner::Interned;
+
 /// A structure used throughout the execution of a search query.
 pub struct SearchContext<'ctx> {
     pub index: &'ctx Index,
diff --git a/milli/src/search/new/query_term/mod.rs b/milli/src/search/new/query_term/mod.rs
index 83320139b..0a0d1a7eb 100644
--- a/milli/src/search/new/query_term/mod.rs
+++ b/milli/src/search/new/query_term/mod.rs
@@ -3,18 +3,18 @@ mod ntypo_subset;
 mod parse_query;
 mod phrase;
 
-use super::interner::{DedupInterner, Interned};
-use super::{limits, SearchContext};
-use crate::Result;
 use std::collections::BTreeSet;
 use std::ops::RangeInclusive;
 
+use compute_derivations::partially_initialized_term_from_word;
 use either::Either;
 pub use ntypo_subset::NTypoTermSubset;
 pub use parse_query::{located_query_terms_from_string, make_ngram, number_of_typos_allowed};
 pub use phrase::Phrase;
 
-use compute_derivations::partially_initialized_term_from_word;
+use super::interner::{DedupInterner, Interned};
+use super::{limits, SearchContext, Word};
+use crate::Result;
 
 /// A set of word derivations attached to a location in the search query.
 #[derive(Clone, PartialEq, Eq, Hash)]
@@ -180,7 +180,7 @@ impl QueryTermSubset {
     pub fn all_single_words_except_prefix_db(
         &self,
         ctx: &mut SearchContext,
-    ) -> Result<BTreeSet<Interned<String>>> {
+    ) -> Result<BTreeSet<Word>> {
         let mut result = BTreeSet::default();
         // TODO: a compute_partially funtion
         if !self.one_typo_subset.is_empty() || !self.two_typo_subset.is_empty() {
@@ -197,8 +197,20 @@ impl QueryTermSubset {
                     synonyms: _,
                     use_prefix_db: _,
                 } = &original.zero_typo;
-                result.extend(zero_typo.iter().copied());
-                result.extend(prefix_of.iter().copied());
+                result.extend(zero_typo.iter().copied().map(|w| {
+                    if original.ngram_words.is_some() {
+                        Word::Derived(w)
+                    } else {
+                        Word::Original(w)
+                    }
+                }));
+                result.extend(prefix_of.iter().copied().map(|w| {
+                    if original.ngram_words.is_some() {
+                        Word::Derived(w)
+                    } else {
+                        Word::Original(w)
+                    }
+                }));
             }
             NTypoTermSubset::Subset { words, phrases: _ } => {
                 let ZeroTypoTerm {
@@ -210,10 +222,14 @@ impl QueryTermSubset {
                 } = &original.zero_typo;
                 if let Some(zero_typo) = zero_typo {
                     if words.contains(zero_typo) {
-                        result.insert(*zero_typo);
+                        if original.ngram_words.is_some() {
+                            result.insert(Word::Derived(*zero_typo));
+                        } else {
+                            result.insert(Word::Original(*zero_typo));
+                        }
                     }
                 }
-                result.extend(prefix_of.intersection(words).copied());
+                result.extend(prefix_of.intersection(words).copied().map(Word::Derived));
             }
             NTypoTermSubset::Nothing => {}
         }
@@ -223,13 +239,13 @@ impl QueryTermSubset {
                 let Lazy::Init(OneTypoTerm { split_words: _, one_typo }) = &original.one_typo else {
                     panic!()
                 };
-                result.extend(one_typo.iter().copied())
+                result.extend(one_typo.iter().copied().map(Word::Derived))
             }
             NTypoTermSubset::Subset { words, phrases: _ } => {
                 let Lazy::Init(OneTypoTerm { split_words: _, one_typo }) = &original.one_typo else {
                     panic!()
                 };
-                result.extend(one_typo.intersection(words));
+                result.extend(one_typo.intersection(words).copied().map(Word::Derived));
             }
             NTypoTermSubset::Nothing => {}
         };
@@ -239,13 +255,13 @@ impl QueryTermSubset {
                 let Lazy::Init(TwoTypoTerm { two_typos }) = &original.two_typo else {
                     panic!()
                 };
-                result.extend(two_typos.iter().copied());
+                result.extend(two_typos.iter().copied().map(Word::Derived));
             }
             NTypoTermSubset::Subset { words, phrases: _ } => {
                 let Lazy::Init(TwoTypoTerm { two_typos }) = &original.two_typo else {
                     panic!()
                 };
-                result.extend(two_typos.intersection(words));
+                result.extend(two_typos.intersection(words).copied().map(Word::Derived));
             }
             NTypoTermSubset::Nothing => {}
         };
diff --git a/milli/src/search/new/ranking_rule_graph/exactness/mod.rs b/milli/src/search/new/ranking_rule_graph/exactness/mod.rs
index 4a3dd6549..7455a7a17 100644
--- a/milli/src/search/new/ranking_rule_graph/exactness/mod.rs
+++ b/milli/src/search/new/ranking_rule_graph/exactness/mod.rs
@@ -3,7 +3,8 @@ use roaring::RoaringBitmap;
 use super::{ComputedCondition, RankingRuleGraphTrait};
 use crate::search::new::interner::{DedupInterner, Interned};
 use crate::search::new::query_term::{ExactTerm, LocatedQueryTermSubset};
-use crate::{Result, RoaringBitmapCodec, SearchContext};
+use crate::search::new::Word;
+use crate::{Result, SearchContext};
 
 #[derive(Clone, PartialEq, Eq, Hash)]
 pub enum ExactnessCondition {
@@ -26,7 +27,7 @@ fn compute_docids(
     let mut candidates = match exact_term {
         ExactTerm::Phrase(phrase) => ctx.get_phrase_docids(phrase)?.clone(),
         ExactTerm::Word(word) => {
-            if let Some(word_candidates) = ctx.get_db_word_docids(word)? {
+            if let Some(word_candidates) = ctx.word_docids(Word::Original(word))? {
                 word_candidates
             } else {
                 return Ok(Default::default());
diff --git a/milli/src/search/new/ranking_rule_graph/proximity/compute_docids.rs b/milli/src/search/new/ranking_rule_graph/proximity/compute_docids.rs
index b6f164f16..760c7272c 100644
--- a/milli/src/search/new/ranking_rule_graph/proximity/compute_docids.rs
+++ b/milli/src/search/new/ranking_rule_graph/proximity/compute_docids.rs
@@ -9,7 +9,7 @@ use crate::search::new::interner::Interned;
 use crate::search::new::query_term::{Phrase, QueryTermSubset};
 use crate::search::new::ranking_rule_graph::ComputedCondition;
 use crate::search::new::resolve_query_graph::compute_query_term_subset_docids;
-use crate::search::new::SearchContext;
+use crate::search::new::{SearchContext, Word};
 use crate::Result;
 
 pub fn compute_docids(
@@ -54,7 +54,7 @@ pub fn compute_docids(
         {
             compute_prefix_edges(
                 ctx,
-                left_word,
+                left_word.interned(),
                 right_prefix,
                 left_phrase,
                 forward_proximity,
@@ -91,7 +91,7 @@ pub fn compute_docids(
                 if universe.is_disjoint(ctx.get_phrase_docids(left_phrase)?) {
                     continue;
                 }
-            } else if let Some(left_word_docids) = ctx.get_db_word_docids(left_word)? {
+            } else if let Some(left_word_docids) = ctx.word_docids(left_word)? {
                 if universe.is_disjoint(&left_word_docids) {
                     continue;
                 }
@@ -101,7 +101,7 @@ pub fn compute_docids(
         for (right_word, right_phrase) in right_derivs {
             compute_non_prefix_edges(
                 ctx,
-                left_word,
+                left_word.interned(),
                 right_word,
                 left_phrase,
                 right_phrase,
@@ -243,7 +243,7 @@ fn compute_non_prefix_edges(
 fn last_words_of_term_derivations(
     ctx: &mut SearchContext,
     t: &QueryTermSubset,
-) -> Result<BTreeSet<(Option<Interned<Phrase>>, Interned<String>)>> {
+) -> Result<BTreeSet<(Option<Interned<Phrase>>, Word)>> {
     let mut result = BTreeSet::new();
 
     for w in t.all_single_words_except_prefix_db(ctx)? {
@@ -253,7 +253,7 @@ fn last_words_of_term_derivations(
         let phrase = ctx.phrase_interner.get(p);
         let last_term_of_phrase = phrase.words.last().unwrap();
         if let Some(last_word) = last_term_of_phrase {
-            result.insert((Some(p), *last_word));
+            result.insert((Some(p), Word::Original(*last_word)));
         }
     }
 
@@ -266,7 +266,7 @@ fn first_word_of_term_iter(
     let mut result = BTreeSet::new();
     let all_words = t.all_single_words_except_prefix_db(ctx)?;
     for w in all_words {
-        result.insert((w, None));
+        result.insert((w.interned(), None));
     }
     for p in t.all_phrases(ctx)? {
         let phrase = ctx.phrase_interner.get(p);
diff --git a/milli/src/search/new/resolve_query_graph.rs b/milli/src/search/new/resolve_query_graph.rs
index bca8b6268..c78f0c5ee 100644
--- a/milli/src/search/new/resolve_query_graph.rs
+++ b/milli/src/search/new/resolve_query_graph.rs
@@ -9,7 +9,7 @@ use super::interner::Interned;
 use super::query_graph::QueryNodeData;
 use super::query_term::{Phrase, QueryTermSubset};
 use super::small_bitmap::SmallBitmap;
-use super::{QueryGraph, SearchContext};
+use super::{QueryGraph, SearchContext, Word};
 use crate::search::new::query_term::LocatedQueryTermSubset;
 use crate::Result;
 
@@ -35,7 +35,7 @@ pub fn compute_query_term_subset_docids(
 ) -> Result<RoaringBitmap> {
     let mut docids = RoaringBitmap::new();
     for word in term.all_single_words_except_prefix_db(ctx)? {
-        if let Some(word_docids) = ctx.get_db_word_docids(word)? {
+        if let Some(word_docids) = ctx.word_docids(word)? {
             docids |= word_docids;
         }
     }
@@ -125,7 +125,7 @@ pub fn compute_phrase_docids(
     }
     if words.len() == 1 {
         if let Some(word) = &words[0] {
-            if let Some(word_docids) = ctx.get_db_word_docids(*word)? {
+            if let Some(word_docids) = ctx.word_docids(Word::Original(*word))? {
                 return Ok(word_docids);
             } else {
                 return Ok(RoaringBitmap::new());